From e0963f2304196ac1fb9e7bbc279d35b1ad903120 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Thu, 21 May 2026 14:41:36 -0700 Subject: [PATCH 01/25] Add --verifiable flag to stellar contract build. --- FULL_HELP_DOCS.md | 6 + cmd/crates/soroban-test/tests/it/build.rs | 82 +++ .../src/commands/contract/build.rs | 33 +- .../src/commands/contract/build/verifiable.rs | 629 ++++++++++++++++++ .../src/commands/contract/deploy/wasm.rs | 6 +- cmd/soroban-cli/src/commands/contract/mod.rs | 2 +- .../src/commands/contract/upload.rs | 6 +- 7 files changed, 755 insertions(+), 9 deletions(-) create mode 100644 cmd/soroban-cli/src/commands/contract/build/verifiable.rs diff --git a/FULL_HELP_DOCS.md b/FULL_HELP_DOCS.md index fbdd59db03..f0ca271564 100644 --- a/FULL_HELP_DOCS.md +++ b/FULL_HELP_DOCS.md @@ -384,6 +384,7 @@ To view the commands that will be executed, without executing them, use the --pr If ommitted, wasm files are written only to the cargo target directory. - `--locked` — Assert that `Cargo.lock` will remain unchanged +- `-d`, `--docker-host ` — Optional argument to override the default docker host. This is useful when you are using a non-standard docker host path for your Docker-compatible container runtime, e.g. Docker Desktop defaults to $HOME/.docker/run/docker.sock instead of /var/run/docker.sock - `--optimize ` — Optimize the generated wasm. Enabled by default; pass `--optimize=false` to disable. Requires the `additional-libs` feature Default value: `true` @@ -394,6 +395,11 @@ To view the commands that will be executed, without executing them, use the --pr - `--print-commands-only` — Print commands to build without executing them +###### **Verifiable:** + +- `--verifiable` — Build inside a trusted Docker container and record SEP-58 metadata (`bldimg`, `source_rev`, `bldopt`) so the resulting WASM can be reproduced and verified by third parties. Implies `--locked`. Requires a clean git working tree +- `--image ` — Override the auto-selected container image used by `--verifiable`. Must be digest-pinned, e.g. `docker.io/stellar/stellar-cli@sha256:...`. Tag-only refs are rejected because SEP-58 requires content addressing + ## `stellar contract extend` Extend the time to live ledger of a contract-data ledger entry. diff --git a/cmd/crates/soroban-test/tests/it/build.rs b/cmd/crates/soroban-test/tests/it/build.rs index b5f7631ca6..5936bfeacf 100644 --- a/cmd/crates/soroban-test/tests/it/build.rs +++ b/cmd/crates/soroban-test/tests/it/build.rs @@ -993,3 +993,85 @@ fn build_always_injects_cli_version() { "CLI version should not be empty" ); } + +// `--verifiable` cannot accept reserved `--meta` keys that the cli writes itself. +#[test] +fn verifiable_meta_conflict_errors() { + let sandbox = TestEnv::default(); + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); + + sandbox + .new_assert_cmd("contract") + .current_dir(fixture_path) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg("docker.io/stellar/stellar-cli@sha256:0000000000000000000000000000000000000000000000000000000000000000") + .arg("--meta") + .arg("bldimg=not-allowed") + .assert() + .failure() + .stderr(predicate::str::contains("reserved key: bldimg")); +} + +// `--image` must be content-addressed; tag-only refs are rejected. +#[test] +fn verifiable_image_must_be_digest_pinned() { + let sandbox = TestEnv::default(); + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); + + sandbox + .new_assert_cmd("contract") + .current_dir(fixture_path) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg("docker.io/stellar/stellar-cli:latest") + .assert() + .failure() + .stderr(predicate::str::contains("must be digest-pinned")); +} + +// A dirty git tree breaks the verifiability property because `source_rev` would +// record a commit whose bytes don't match the produced WASM. Hard fail. +#[test] +fn verifiable_dirty_tree_errors() { + let sandbox = TestEnv::default(); + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace"); + let temp = TempDir::new().unwrap(); + let dir_path = temp.path(); + fs_extra::dir::copy(fixture_path, dir_path, &CopyOptions::new()).unwrap(); + let workspace = dir_path.join("workspace"); + + // Bootstrap a clean git tree at the workspace root, then dirty it so the + // verifiable path's dirty-check trips before docker is touched. + let git = |args: &[&str]| { + std::process::Command::new("git") + .args(args) + .current_dir(&workspace) + .env("GIT_AUTHOR_NAME", "Test") + .env("GIT_AUTHOR_EMAIL", "test@example.com") + .env("GIT_COMMITTER_NAME", "Test") + .env("GIT_COMMITTER_EMAIL", "test@example.com") + .status() + .unwrap(); + }; + git(&["init", "-q", "-b", "main"]); + git(&["add", "-A"]); + git(&["commit", "-q", "-m", "init"]); + std::fs::write(workspace.join("dirty.txt"), b"uncommitted").unwrap(); + + sandbox + .new_assert_cmd("contract") + .current_dir(workspace.join("contracts").join("add")) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg("docker.io/stellar/stellar-cli@sha256:0000000000000000000000000000000000000000000000000000000000000000") + .assert() + .failure() + .stderr(predicate::str::contains("dirty").or(predicate::str::contains("clean tree"))); +} diff --git a/cmd/soroban-cli/src/commands/contract/build.rs b/cmd/soroban-cli/src/commands/contract/build.rs index ed3ce7e1bd..c6f1173f74 100644 --- a/cmd/soroban-cli/src/commands/contract/build.rs +++ b/cmd/soroban-cli/src/commands/contract/build.rs @@ -20,11 +20,13 @@ use stellar_xdr::{Limited, Limits, ScMetaEntry, ScMetaV0, StringM, WriteXdr}; #[cfg(feature = "additional-libs")] use crate::commands::contract::optimize; use crate::{ - commands::{global, version}, + commands::{container, global, version}, print::Print, wasm, }; +pub mod verifiable; + /// A built WASM artifact with its package name and file path. #[derive(Debug, Clone)] pub struct BuiltContract { @@ -96,6 +98,22 @@ pub struct Cmd { #[arg(long, conflicts_with = "out_dir", help_heading = "Other")] pub print_commands_only: bool, + /// Build inside a trusted Docker container and record SEP-58 metadata + /// (`bldimg`, `source_rev`, `bldopt`) so the resulting WASM can be + /// reproduced and verified by third parties. Implies `--locked`. + /// Requires a clean git working tree. + #[arg(long, help_heading = "Verifiable")] + pub verifiable: bool, + + /// Override the auto-selected container image used by `--verifiable`. + /// Must be digest-pinned, e.g. `docker.io/stellar/stellar-cli@sha256:...`. + /// Tag-only refs are rejected because SEP-58 requires content addressing. + #[arg(long, requires = "verifiable", help_heading = "Verifiable")] + pub image: Option, + + #[command(flatten)] + pub container_args: container::shared::Args, + #[command(flatten)] pub build_args: BuildArgs, } @@ -204,6 +222,9 @@ pub enum Error { #[error("wasm parsing error: {0}")] WasmParsing(String), + + #[error(transparent)] + Verifiable(#[from] verifiable::Error), } const WASM_TARGET: &str = "wasm32v1-none"; @@ -222,6 +243,9 @@ impl Default for Cmd { out_dir: None, locked: false, print_commands_only: false, + verifiable: false, + image: None, + container_args: container::shared::Args { docker_host: None }, build_args: BuildArgs::default(), } } @@ -230,8 +254,13 @@ impl Default for Cmd { impl Cmd { /// Builds the project and returns the built WASM artifacts. #[allow(clippy::too_many_lines)] - pub fn run(&self, global_args: &global::Args) -> Result, Error> { + pub async fn run(&self, global_args: &global::Args) -> Result, Error> { let print = Print::new(global_args.quiet); + + if self.verifiable { + return verifiable::run(self, global_args, &print).await; + } + let working_dir = env::current_dir().map_err(Error::GettingCurrentDir)?; let metadata = self.metadata()?; let packages = self.packages(&metadata)?; diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs new file mode 100644 index 0000000000..80fa1c11b7 --- /dev/null +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -0,0 +1,629 @@ +use std::{ + path::{Path, PathBuf}, + process::Command, +}; + +use bollard::{ + models::ContainerCreateBody, + query_parameters::{ + AttachContainerOptions, CreateContainerOptions, CreateImageOptions, StartContainerOptions, + WaitContainerOptions, + }, + service::HostConfig, + Docker, +}; +use cargo_metadata::MetadataCommand; +use futures_util::{StreamExt, TryStreamExt}; +use regex::Regex; +use semver::Version; +use serde::Deserialize; + +use crate::{ + commands::{container::shared::Error as ConnectionError, global}, + print::Print, +}; + +use super::{BuiltContract, Cmd, WASM_TARGET}; + +const REGISTRY: &str = "docker.io/stellar/stellar-cli"; +const HUB_TAGS_URL: &str = + "https://hub.docker.com/v2/repositories/stellar/stellar-cli/tags/?page_size=100"; +const RESERVED_META_KEYS: &[&str] = &["bldimg", "source_rev", "bldopt"]; + +#[derive(thiserror::Error, Debug)] +pub enum Error { + #[error("⛔ failed to connect to docker: {0}")] + DockerConnection(#[from] ConnectionError), + + #[error(transparent)] + Bollard(#[from] bollard::errors::Error), + + #[error("--image must be digest-pinned (got {value}); SEP-58 requires content-addressed images. Pass docker.io/stellar/stellar-cli@sha256:")] + ImageNotDigestPinned { value: String }, + + #[error("could not determine the running rustc version: {0}")] + RustcVersion(String), + + #[error("could not pull image {tag}: {source}\n\nAvailable tags for this CLI version: {available_for_cli}\nAll published cli/rust pairs: {all_grouped}\n\nFix: install a matching rustc, or pass --image docker.io/stellar/stellar-cli@sha256: with one of the listed tags resolved to a digest.")] + ImageNotFound { + tag: String, + available_for_cli: String, + all_grouped: String, + source: bollard::errors::Error, + }, + + #[error("could not list published images on docker hub: {0}")] + TagListUnavailable(String), + + #[error("image {tag} has no repo digest after pull; cannot record a content-addressed bldimg")] + NoRepoDigest { tag: String }, + + #[error("cargo metadata failed: {0}")] + Metadata(#[from] cargo_metadata::Error), + + #[error("could not read git state at {path}: {source}")] + GitInvoke { + path: PathBuf, + source: std::io::Error, + }, + + #[error( + "git working tree at {path} is dirty. Verifiable builds require a clean tree so the recorded source_rev matches the WASM bytes. Commit or stash your changes and try again." + )] + GitDirty { path: PathBuf }, + + #[error( + "the cli sets bldimg, source_rev, and bldopt automatically when --verifiable is used; remove them from --meta. Got reserved key: {key}" + )] + ReservedMetaKey { key: String }, + + #[error("container build exited with status {status}. To reproduce manually:\n docker run --rm -v {mount}:/source {image} contract build {args}")] + ContainerExit { + status: i64, + image: String, + mount: String, + args: String, + }, +} + +pub async fn run( + cmd: &Cmd, + global_args: &global::Args, + print: &Print, +) -> Result, super::Error> { + // Stage 1: pure validation, no I/O. + for (k, _) in &cmd.build_args.meta { + if RESERVED_META_KEYS.iter().any(|r| r == k) { + return Err(Error::ReservedMetaKey { key: k.clone() }.into()); + } + } + if let Some(img) = &cmd.image { + if !img.contains("@sha256:") { + return Err(Error::ImageNotDigestPinned { value: img.clone() }.into()); + } + } + + if !cmd.locked { + print.infoln("--verifiable implies --locked"); + } + + // Stage 2: local filesystem + git, no network. + let workspace_root = resolve_workspace_root(cmd)?; + let source_rev = git_source_rev(&workspace_root, print)?; + + // Stage 3: docker. + let docker = cmd + .container_args + .connect_to_docker(print) + .await + .map_err(Error::DockerConnection)?; + let image_ref = resolve_image(cmd, &docker, print).await?; + + let (forwarded_args, bldopts) = build_forwarded_args(cmd); + let metadata_args = build_metadata_args(&image_ref, &source_rev, &bldopts); + let container_cmd_args = compose_container_args(&forwarded_args, &metadata_args); + + run_in_container( + &image_ref, + &workspace_root, + &container_cmd_args, + &docker, + print, + ) + .await?; + + let _ = global_args; + collect_built_contracts(cmd, &workspace_root, print) +} + +fn resolve_workspace_root(cmd: &Cmd) -> Result { + let mut mc = MetadataCommand::new(); + mc.no_deps(); + if let Some(p) = &cmd.manifest_path { + mc.manifest_path(p); + } + let md = mc.exec()?; + Ok(md.workspace_root.into_std_path_buf()) +} + +fn git_source_rev(workspace_root: &Path, print: &Print) -> Result { + // Probe with rev-parse first to detect "not a git repo". + let rev = Command::new("git") + .arg("-C") + .arg(workspace_root) + .arg("rev-parse") + .arg("HEAD") + .output(); + let rev = match rev { + Ok(o) if o.status.success() => String::from_utf8_lossy(&o.stdout).trim().to_string(), + Ok(_) => { + print.warnln(format!( + "{} is not a git repository; recording empty source_rev (verifiability is degraded).", + workspace_root.display() + )); + return Ok(String::new()); + } + Err(e) => { + return Err(Error::GitInvoke { + path: workspace_root.to_path_buf(), + source: e, + }) + } + }; + + // Dirty check. + let status = Command::new("git") + .arg("-C") + .arg(workspace_root) + .arg("status") + .arg("--porcelain") + .output() + .map_err(|e| Error::GitInvoke { + path: workspace_root.to_path_buf(), + source: e, + })?; + if !status.stdout.is_empty() { + return Err(Error::GitDirty { + path: workspace_root.to_path_buf(), + }); + } + + Ok(rev) +} + +/// The flags forwarded to the container's `stellar contract build`, plus the +/// bldopt strings recorded into SEP-58 metadata. `--locked` is always present. +fn build_forwarded_args(cmd: &Cmd) -> (Vec, Vec) { + let mut forwarded: Vec = Vec::new(); + let mut bldopts: Vec = Vec::new(); + + forwarded.push("--locked".to_string()); + bldopts.push("--locked".to_string()); + + if cmd.profile != "release" { + let s = format!("--profile={}", cmd.profile); + forwarded.push(s.clone()); + bldopts.push(s); + } + if let Some(features) = &cmd.features { + let s = format!("--features={features}"); + forwarded.push(s.clone()); + bldopts.push(s); + } + if cmd.all_features { + forwarded.push("--all-features".to_string()); + bldopts.push("--all-features".to_string()); + } + if cmd.no_default_features { + forwarded.push("--no-default-features".to_string()); + bldopts.push("--no-default-features".to_string()); + } + if let Some(pkg) = &cmd.package { + let s = format!("--package={pkg}"); + forwarded.push(s.clone()); + bldopts.push(s); + } + + // User-supplied --meta entries (none of which can collide with reserved keys + // because we already errored on that). + for (k, v) in &cmd.build_args.meta { + forwarded.push("--meta".to_string()); + forwarded.push(format!("{k}={v}")); + } + + if !cmd.build_args.optimize { + forwarded.push("--optimize=false".to_string()); + } + + (forwarded, bldopts) +} + +fn build_metadata_args(image_ref: &str, source_rev: &str, bldopts: &[String]) -> Vec { + let mut out = Vec::new(); + for (k, v) in [("bldimg", image_ref), ("source_rev", source_rev)] { + out.push("--meta".to_string()); + out.push(format!("{k}={v}")); + } + for o in bldopts { + out.push("--meta".to_string()); + out.push(format!("bldopt={o}")); + } + out +} + +fn compose_container_args(forwarded: &[String], metadata: &[String]) -> Vec { + let mut args = vec!["contract".to_string(), "build".to_string()]; + args.extend_from_slice(forwarded); + args.extend_from_slice(metadata); + args +} + +pub async fn resolve_image(cmd: &Cmd, docker: &Docker, print: &Print) -> Result { + if let Some(s) = &cmd.image { + if !s.contains("@sha256:") { + return Err(Error::ImageNotDigestPinned { value: s.clone() }); + } + return Ok(s.clone()); + } + + let cli_v = env!("CARGO_PKG_VERSION"); + let rust_v = rustc_version::version() + .map_err(|e| Error::RustcVersion(e.to_string()))? + .to_string(); + let tag = format!("{REGISTRY}:{cli_v}-rust{rust_v}"); + + print.infoln(format!("Pulling verifiable build image {tag}")); + let pull = pull_image(docker, &tag, print).await; + + match pull { + Ok(()) => {} + Err(e) => { + let (available_for_cli, all_grouped) = match list_published_tags().await { + Ok(tags) => format_available(&tags, cli_v), + Err(list_err) => ( + "".to_string(), + format!(""), + ), + }; + return Err(Error::ImageNotFound { + tag, + available_for_cli, + all_grouped, + source: e, + }); + } + } + + let inspect = docker.inspect_image(&tag).await?; + let digest = inspect + .repo_digests + .and_then(|v| v.into_iter().next()) + .ok_or_else(|| Error::NoRepoDigest { tag: tag.clone() })?; + Ok(digest) +} + +async fn pull_image( + docker: &Docker, + tag: &str, + print: &Print, +) -> Result<(), bollard::errors::Error> { + let mut stream = docker.create_image( + Some(CreateImageOptions { + from_image: Some(tag.to_string()), + ..Default::default() + }), + None, + None, + ); + while let Some(item) = stream.try_next().await? { + if let Some(status) = item.status { + if status.contains("Pulling from") + || status.contains("Digest") + || status.contains("Status") + { + print.infoln(status); + } + } + } + Ok(()) +} + +#[derive(Debug, Clone)] +pub struct PublishedTag { + pub cli: Version, + pub rust: Version, + pub raw: String, +} + +#[derive(Deserialize)] +struct HubPage { + results: Vec, + next: Option, +} + +#[derive(Deserialize)] +struct HubTag { + name: String, +} + +pub async fn list_published_tags() -> Result, Error> { + let re = Regex::new(r"^(\d+\.\d+\.\d+)-rust(\d+\.\d+\.\d+)$").unwrap(); + let mut out = Vec::new(); + let mut next = Some(HUB_TAGS_URL.to_string()); + let client = reqwest::Client::builder() + .user_agent("stellar-cli") + .build() + .map_err(|e| Error::TagListUnavailable(e.to_string()))?; + while let Some(url) = next { + let page: HubPage = client + .get(&url) + .send() + .await + .map_err(|e| Error::TagListUnavailable(e.to_string()))? + .error_for_status() + .map_err(|e| Error::TagListUnavailable(e.to_string()))? + .json() + .await + .map_err(|e| Error::TagListUnavailable(e.to_string()))?; + for t in page.results { + if let Some(c) = re.captures(&t.name) { + let cli = Version::parse(&c[1]); + let rust = Version::parse(&c[2]); + if let (Ok(cli), Ok(rust)) = (cli, rust) { + out.push(PublishedTag { + cli, + rust, + raw: t.name, + }); + } + } + } + next = page.next; + } + Ok(out) +} + +fn format_available(tags: &[PublishedTag], current_cli: &str) -> (String, String) { + let current = Version::parse(current_cli).ok(); + let mut for_this_cli: Vec<&PublishedTag> = tags + .iter() + .filter(|t| Some(&t.cli) == current.as_ref()) + .collect(); + for_this_cli.sort_by(|a, b| b.rust.cmp(&a.rust)); + let available_for_cli = if for_this_cli.is_empty() { + "".to_string() + } else { + for_this_cli + .iter() + .map(|t| t.raw.as_str()) + .collect::>() + .join(", ") + }; + + let mut by_cli: std::collections::BTreeMap> = + std::collections::BTreeMap::new(); + for t in tags { + by_cli + .entry(t.cli.to_string()) + .or_default() + .push(t.rust.to_string()); + } + let all_grouped = by_cli + .into_iter() + .map(|(cli, rusts)| format!("{cli}: [{}]", rusts.join(", "))) + .collect::>() + .join("; "); + + (available_for_cli, all_grouped) +} + +async fn run_in_container( + image_ref: &str, + workspace_root: &Path, + container_cmd: &[String], + docker: &Docker, + print: &Print, +) -> Result<(), Error> { + let bind = format!("{}:/source", workspace_root.display()); + let config = ContainerCreateBody { + image: Some(image_ref.to_string()), + cmd: Some(container_cmd.to_vec()), + working_dir: Some("/source".to_string()), + attach_stdout: Some(true), + attach_stderr: Some(true), + host_config: Some(HostConfig { + auto_remove: Some(true), + binds: Some(vec![bind.clone()]), + ..Default::default() + }), + ..Default::default() + }; + + print.infoln(format!( + "Running verifiable build in {image_ref} (mount {bind})" + )); + + let created = docker + .create_container(None::, config) + .await?; + + let attached = docker + .attach_container( + &created.id, + Some(AttachContainerOptions { + stdout: true, + stderr: true, + stream: true, + ..Default::default() + }), + ) + .await?; + + docker + .start_container(&created.id, None::) + .await?; + + let mut output = attached.output; + while let Some(chunk) = output.next().await { + match chunk { + Ok( + bollard::container::LogOutput::StdOut { message } + | bollard::container::LogOutput::StdErr { message }, + ) => { + let s = String::from_utf8_lossy(&message); + print.blankln(s.trim_end()); + } + Ok(_) => {} + Err(e) => return Err(e.into()), + } + } + + let mut wait = docker.wait_container(&created.id, None::); + while let Some(item) = wait.next().await { + match item { + Ok(r) if r.status_code == 0 => {} + Ok(r) => { + return Err(Error::ContainerExit { + status: r.status_code, + image: image_ref.to_string(), + mount: workspace_root.display().to_string(), + args: container_cmd.join(" "), + }); + } + Err(bollard::errors::Error::DockerContainerWaitError { code: 0, .. }) => {} + Err(bollard::errors::Error::DockerContainerWaitError { code, .. }) => { + return Err(Error::ContainerExit { + status: code, + image: image_ref.to_string(), + mount: workspace_root.display().to_string(), + args: container_cmd.join(" "), + }); + } + Err(e) => return Err(e.into()), + } + } + + Ok(()) +} + +fn collect_built_contracts( + cmd: &Cmd, + workspace_root: &Path, + _print: &Print, +) -> Result, super::Error> { + let mut mc = MetadataCommand::new(); + mc.no_deps(); + if let Some(p) = &cmd.manifest_path { + mc.manifest_path(p); + } + let md = mc.exec().map_err(Error::Metadata)?; + let target_dir = md.target_directory.as_std_path(); + + let mut out = Vec::new(); + for p in &md.packages { + let is_cdylib = p + .targets + .iter() + .any(|t| t.crate_types.iter().any(|c| c == "cdylib")); + if !is_cdylib { + continue; + } + if let Some(name) = &cmd.package { + if &p.name != name { + continue; + } + } else if !md.workspace_default_members.contains(&p.id) { + continue; + } + let wasm_name = p.name.replace('-', "_"); + let path = Path::new(target_dir) + .join(WASM_TARGET) + .join(&cmd.profile) + .join(format!("{wasm_name}.wasm")); + if let Some(out_dir) = &cmd.out_dir { + let dest = out_dir.join(format!("{wasm_name}.wasm")); + if path.exists() { + std::fs::create_dir_all(out_dir).map_err(super::Error::CreatingOutDir)?; + std::fs::copy(&path, &dest).map_err(super::Error::CopyingWasmFile)?; + out.push(BuiltContract { + name: p.name.clone(), + path: dest, + }); + continue; + } + } + out.push(BuiltContract { + name: p.name.clone(), + path, + }); + } + let _ = workspace_root; + Ok(out) +} + +#[cfg(test)] +mod tests { + use super::*; + + #[test] + fn build_forwarded_args_defaults() { + let cmd = Cmd::default(); + let (forwarded, bldopts) = build_forwarded_args(&cmd); + assert_eq!(forwarded, vec!["--locked".to_string()]); + assert_eq!(bldopts, vec!["--locked".to_string()]); + } + + #[test] + fn build_forwarded_args_features_and_package() { + let cmd = Cmd { + features: Some("a,b".to_string()), + package: Some("contract-a".to_string()), + ..Cmd::default() + }; + let (forwarded, bldopts) = build_forwarded_args(&cmd); + assert!(forwarded.contains(&"--features=a,b".to_string())); + assert!(forwarded.contains(&"--package=contract-a".to_string())); + assert!(bldopts.contains(&"--features=a,b".to_string())); + assert!(bldopts.contains(&"--package=contract-a".to_string())); + assert!(bldopts.contains(&"--locked".to_string())); + } + + #[test] + fn build_metadata_args_orders_keys() { + let m = build_metadata_args( + "docker.io/stellar/stellar-cli@sha256:abc", + "deadbeef", + &["--locked".to_string(), "--features=a".to_string()], + ); + // bldimg, source_rev, then bldopts in order. + let pairs: Vec<(&str, &str)> = m + .chunks(2) + .map(|c| (c[0].as_str(), c[1].as_str())) + .collect(); + assert_eq!( + pairs[0], + ("--meta", "bldimg=docker.io/stellar/stellar-cli@sha256:abc") + ); + assert_eq!(pairs[1], ("--meta", "source_rev=deadbeef")); + assert_eq!(pairs[2], ("--meta", "bldopt=--locked")); + assert_eq!(pairs[3], ("--meta", "bldopt=--features=a")); + } + + #[test] + fn compose_container_args_prefixes_subcommand() { + let composed = compose_container_args( + &["--locked".to_string()], + &["--meta".to_string(), "bldimg=x".to_string()], + ); + assert_eq!(composed[..2], ["contract".to_string(), "build".to_string()]); + assert!(composed.contains(&"--locked".to_string())); + assert!(composed.contains(&"bldimg=x".to_string())); + } + + #[test] + fn reserved_meta_keys_list() { + for key in ["bldimg", "source_rev", "bldopt"] { + assert!(RESERVED_META_KEYS.contains(&key)); + } + } +} diff --git a/cmd/soroban-cli/src/commands/contract/deploy/wasm.rs b/cmd/soroban-cli/src/commands/contract/deploy/wasm.rs index 9afdada059..5ba2d37aa7 100644 --- a/cmd/soroban-cli/src/commands/contract/deploy/wasm.rs +++ b/cmd/soroban-cli/src/commands/contract/deploy/wasm.rs @@ -193,7 +193,7 @@ impl Cmd { return Err(Error::BuildOnlyNotSupported); } - let built_contracts = self.resolve_contracts(global_args)?; + let built_contracts = self.resolve_contracts(global_args).await?; // Aliases derived from workspace package names are assigned per-iteration // inside the deploy loop, so validate them all up front: a package named @@ -286,7 +286,7 @@ impl Cmd { Ok(()) } - fn resolve_contracts( + async fn resolve_contracts( &self, global_args: &global::Args, ) -> Result, Error> { @@ -309,7 +309,7 @@ impl Cmd { build_args: self.build_args.clone(), ..build::Cmd::default() }; - let contracts = build_cmd.run(global_args).map_err(|e| match e { + let contracts = build_cmd.run(global_args).await.map_err(|e| match e { build::Error::Metadata(_) => Error::NotInCargoProject, other => other.into(), })?; diff --git a/cmd/soroban-cli/src/commands/contract/mod.rs b/cmd/soroban-cli/src/commands/contract/mod.rs index a5e6ce181c..fc4499c029 100644 --- a/cmd/soroban-cli/src/commands/contract/mod.rs +++ b/cmd/soroban-cli/src/commands/contract/mod.rs @@ -164,7 +164,7 @@ impl Cmd { Cmd::Asset(asset) => asset.run(global_args).await?, Cmd::Bindings(bindings) => bindings.run().await?, Cmd::Build(build) => { - build.run(global_args)?; + build.run(global_args).await?; } Cmd::Extend(extend) => extend.run(global_args).await?, Cmd::Alias(alias) => alias.run(global_args)?, diff --git a/cmd/soroban-cli/src/commands/contract/upload.rs b/cmd/soroban-cli/src/commands/contract/upload.rs index 55b8e308b9..9a843880d4 100644 --- a/cmd/soroban-cli/src/commands/contract/upload.rs +++ b/cmd/soroban-cli/src/commands/contract/upload.rs @@ -144,7 +144,7 @@ impl Cmd { return Err(Error::BuildOnlyNotSupported); } - let wasm_paths = self.resolve_wasm_paths(global_args)?; + let wasm_paths = self.resolve_wasm_paths(global_args).await?; for wasm_path in &wasm_paths { let res = self @@ -181,7 +181,7 @@ impl Cmd { self.upload_wasm(&wasm_path, config, quiet, no_cache).await } - fn resolve_wasm_paths(&self, global_args: &global::Args) -> Result, Error> { + async fn resolve_wasm_paths(&self, global_args: &global::Args) -> Result, Error> { if let Some(wasm) = &self.wasm { Ok(vec![wasm.clone()]) } else { @@ -190,7 +190,7 @@ impl Cmd { build_args: self.build_args.clone(), ..build::Cmd::default() }; - let contracts = build_cmd.run(global_args).map_err(|e| match e { + let contracts = build_cmd.run(global_args).await.map_err(|e| match e { build::Error::Metadata(_) => Error::NotInCargoProject, other => other.into(), })?; From b04676dc66382e37461afba8b2d50f372558f854 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Thu, 21 May 2026 15:07:25 -0700 Subject: [PATCH 02/25] Record every build-affecting flag as bldopt. --- .../src/commands/contract/build/verifiable.rs | 88 +++++++++++++------ 1 file changed, 61 insertions(+), 27 deletions(-) diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 80fa1c11b7..7d61322c40 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -119,7 +119,7 @@ pub async fn run( .map_err(Error::DockerConnection)?; let image_ref = resolve_image(cmd, &docker, print).await?; - let (forwarded_args, bldopts) = build_forwarded_args(cmd); + let (forwarded_args, bldopts) = build_forwarded_args(cmd, &workspace_root); let metadata_args = build_metadata_args(&image_ref, &source_rev, &bldopts); let container_cmd_args = compose_container_args(&forwarded_args, &metadata_args); @@ -192,47 +192,51 @@ fn git_source_rev(workspace_root: &Path, print: &Print) -> Result } /// The flags forwarded to the container's `stellar contract build`, plus the -/// bldopt strings recorded into SEP-58 metadata. `--locked` is always present. -fn build_forwarded_args(cmd: &Cmd) -> (Vec, Vec) { +/// bldopt strings recorded into SEP-58 metadata. Every build-affecting flag +/// becomes one bldopt entry so a verifier can replay the same invocation. +/// `--locked` is always present. `manifest_path` (when set) is recorded +/// relative to the workspace root so it's valid inside `/source`. +fn build_forwarded_args(cmd: &Cmd, workspace_root: &Path) -> (Vec, Vec) { let mut forwarded: Vec = Vec::new(); let mut bldopts: Vec = Vec::new(); - forwarded.push("--locked".to_string()); - bldopts.push("--locked".to_string()); + let mut record = |arg: String| { + forwarded.push(arg.clone()); + bldopts.push(arg); + }; + + record("--locked".to_string()); + if let Some(path) = &cmd.manifest_path { + let abs = std::path::absolute(path).unwrap_or_else(|_| path.clone()); + let rel = abs + .strip_prefix(workspace_root) + .map(Path::to_path_buf) + .unwrap_or(abs); + record(format!("--manifest-path={}", rel.display())); + } if cmd.profile != "release" { - let s = format!("--profile={}", cmd.profile); - forwarded.push(s.clone()); - bldopts.push(s); + record(format!("--profile={}", cmd.profile)); } if let Some(features) = &cmd.features { - let s = format!("--features={features}"); - forwarded.push(s.clone()); - bldopts.push(s); + record(format!("--features={features}")); } if cmd.all_features { - forwarded.push("--all-features".to_string()); - bldopts.push("--all-features".to_string()); + record("--all-features".to_string()); } if cmd.no_default_features { - forwarded.push("--no-default-features".to_string()); - bldopts.push("--no-default-features".to_string()); + record("--no-default-features".to_string()); } if let Some(pkg) = &cmd.package { - let s = format!("--package={pkg}"); - forwarded.push(s.clone()); - bldopts.push(s); + record(format!("--package={pkg}")); } - - // User-supplied --meta entries (none of which can collide with reserved keys - // because we already errored on that). for (k, v) in &cmd.build_args.meta { - forwarded.push("--meta".to_string()); - forwarded.push(format!("{k}={v}")); + // Use the `--meta=key=value` form so each option is a single token, + // matching how clap re-parses on the container side. + record(format!("--meta={k}={v}")); } - if !cmd.build_args.optimize { - forwarded.push("--optimize=false".to_string()); + record("--optimize=false".to_string()); } (forwarded, bldopts) @@ -565,10 +569,14 @@ fn collect_built_contracts( mod tests { use super::*; + fn ws() -> &'static Path { + Path::new("/tmp/ws") + } + #[test] fn build_forwarded_args_defaults() { let cmd = Cmd::default(); - let (forwarded, bldopts) = build_forwarded_args(&cmd); + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws()); assert_eq!(forwarded, vec!["--locked".to_string()]); assert_eq!(bldopts, vec!["--locked".to_string()]); } @@ -580,7 +588,7 @@ mod tests { package: Some("contract-a".to_string()), ..Cmd::default() }; - let (forwarded, bldopts) = build_forwarded_args(&cmd); + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws()); assert!(forwarded.contains(&"--features=a,b".to_string())); assert!(forwarded.contains(&"--package=contract-a".to_string())); assert!(bldopts.contains(&"--features=a,b".to_string())); @@ -588,6 +596,32 @@ mod tests { assert!(bldopts.contains(&"--locked".to_string())); } + #[test] + fn build_forwarded_args_records_meta_optimize_and_manifest() { + let cmd = Cmd { + manifest_path: Some(PathBuf::from("/tmp/ws/contracts/add/Cargo.toml")), + build_args: super::super::BuildArgs { + meta: vec![ + ("home_domain".to_string(), "fnando.com".to_string()), + ("author".to_string(), "alice".to_string()), + ], + optimize: false, + }, + ..Cmd::default() + }; + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws()); + assert!(forwarded.contains(&"--meta=home_domain=fnando.com".to_string())); + assert!(forwarded.contains(&"--meta=author=alice".to_string())); + assert!(forwarded.contains(&"--optimize=false".to_string())); + assert!(forwarded.contains(&"--manifest-path=contracts/add/Cargo.toml".to_string())); + // Same set is captured into bldopts so a verifier can replay every + // build-affecting flag. + assert!(bldopts.contains(&"--meta=home_domain=fnando.com".to_string())); + assert!(bldopts.contains(&"--meta=author=alice".to_string())); + assert!(bldopts.contains(&"--optimize=false".to_string())); + assert!(bldopts.contains(&"--manifest-path=contracts/add/Cargo.toml".to_string())); + } + #[test] fn build_metadata_args_orders_keys() { let m = build_metadata_args( From 1a118d14b4d9cfe4622d01b2b2adef3ebc234ab3 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Thu, 21 May 2026 15:18:03 -0700 Subject: [PATCH 03/25] Remove duplicate 'contract build' in container error hint. --- cmd/soroban-cli/src/commands/contract/build/verifiable.rs | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 7d61322c40..2c77226701 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -77,7 +77,7 @@ pub enum Error { )] ReservedMetaKey { key: String }, - #[error("container build exited with status {status}. To reproduce manually:\n docker run --rm -v {mount}:/source {image} contract build {args}")] + #[error("container build exited with status {status}. To reproduce manually:\n docker run --rm -v {mount}:/source {image} {args}")] ContainerExit { status: i64, image: String, From 196405d5fb1b756275c6f4c6e7ba0bd8e115958c Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Thu, 21 May 2026 15:36:26 -0700 Subject: [PATCH 04/25] Probe container cli version for --optimize syntax. --- .../src/commands/contract/build/verifiable.rs | 162 ++++++++++++++++-- 1 file changed, 148 insertions(+), 14 deletions(-) diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 2c77226701..6d4b2ae6c3 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -30,6 +30,12 @@ const HUB_TAGS_URL: &str = "https://hub.docker.com/v2/repositories/stellar/stellar-cli/tags/?page_size=100"; const RESERVED_META_KEYS: &[&str] = &["bldimg", "source_rev", "bldopt"]; +/// First cli release that accepts `--optimize=false` as an explicit value +/// (added by commit `b17d3f0b`). Containers older than this only accept bare +/// `--optimize`; we probe the container's `stellar version --only-version` to +/// pick the right syntax for `--optimize=false`. +const OPTIMIZE_NEW_SYNTAX_MIN: &str = "26.1.0"; + #[derive(thiserror::Error, Debug)] pub enum Error { #[error("⛔ failed to connect to docker: {0}")] @@ -119,7 +125,17 @@ pub async fn run( .map_err(Error::DockerConnection)?; let image_ref = resolve_image(cmd, &docker, print).await?; - let (forwarded_args, bldopts) = build_forwarded_args(cmd, &workspace_root); + // Only probe the container's cli version when we need to pick between + // `--optimize=false` (new syntax) and not-forwarded-at-all (old default). + // Bare `--optimize` is universally accepted, so the true path skips this. + let supports_explicit_optimize_false = if cmd.build_args.optimize { + true + } else { + probe_supports_optimize_false_syntax(&image_ref, &docker, print).await + }; + + let (forwarded_args, bldopts) = + build_forwarded_args(cmd, &workspace_root, supports_explicit_optimize_false); let metadata_args = build_metadata_args(&image_ref, &source_rev, &bldopts); let container_cmd_args = compose_container_args(&forwarded_args, &metadata_args); @@ -196,7 +212,16 @@ fn git_source_rev(workspace_root: &Path, print: &Print) -> Result /// becomes one bldopt entry so a verifier can replay the same invocation. /// `--locked` is always present. `manifest_path` (when set) is recorded /// relative to the workspace root so it's valid inside `/source`. -fn build_forwarded_args(cmd: &Cmd, workspace_root: &Path) -> (Vec, Vec) { +/// +/// `supports_explicit_optimize_false`: whether the container's cli accepts +/// `--optimize=false`. When false, the optimize=false case records the flag +/// in bldopt but does not forward it (the older container's cli default of +/// `false` already produces the desired state). +fn build_forwarded_args( + cmd: &Cmd, + workspace_root: &Path, + supports_explicit_optimize_false: bool, +) -> (Vec, Vec) { let mut forwarded: Vec = Vec::new(); let mut bldopts: Vec = Vec::new(); @@ -235,7 +260,14 @@ fn build_forwarded_args(cmd: &Cmd, workspace_root: &Path) -> (Vec, Vec (String, String (available_for_cli, all_grouped) } +/// Probe the container's `stellar` binary for its self-reported version with +/// `stellar version --only-version`. Returns true if the parsed version is +/// at or above the cutoff where `--optimize=false` was accepted. On any +/// probe failure (network, unparseable output, missing subcommand), returns +/// false — the conservative assumption that the container is old. +async fn probe_supports_optimize_false_syntax( + image_ref: &str, + docker: &Docker, + print: &Print, +) -> bool { + match probe_cli_version(image_ref, docker).await { + Ok(v) => { + let cutoff = Version::parse(OPTIMIZE_NEW_SYNTAX_MIN).unwrap(); + v >= cutoff + } + Err(e) => { + print.warnln(format!( + "could not probe container cli version ({e}); assuming pre-{OPTIMIZE_NEW_SYNTAX_MIN} syntax" + )); + false + } + } +} + +async fn probe_cli_version(image_ref: &str, docker: &Docker) -> Result { + let config = ContainerCreateBody { + image: Some(image_ref.to_string()), + cmd: Some(vec!["version".to_string(), "--only-version".to_string()]), + attach_stdout: Some(true), + attach_stderr: Some(true), + host_config: Some(HostConfig { + auto_remove: Some(true), + ..Default::default() + }), + ..Default::default() + }; + let created = docker + .create_container(None::, config) + .await?; + let attached = docker + .attach_container( + &created.id, + Some(AttachContainerOptions { + stdout: true, + stderr: true, + stream: true, + ..Default::default() + }), + ) + .await?; + docker + .start_container(&created.id, None::) + .await?; + + let mut stdout = String::new(); + let mut output = attached.output; + while let Some(chunk) = output.next().await { + if let Ok(bollard::container::LogOutput::StdOut { message }) = chunk { + stdout.push_str(&String::from_utf8_lossy(&message)); + } + } + + let mut wait = docker.wait_container(&created.id, None::); + while wait.next().await.is_some() {} + + Version::parse(stdout.trim()) + .map_err(|e| Error::TagListUnavailable(format!("unparseable version {stdout:?}: {e}"))) +} + async fn run_in_container( image_ref: &str, workspace_root: &Path, @@ -576,9 +677,16 @@ mod tests { #[test] fn build_forwarded_args_defaults() { let cmd = Cmd::default(); - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws()); - assert_eq!(forwarded, vec!["--locked".to_string()]); - assert_eq!(bldopts, vec!["--locked".to_string()]); + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), true); + // Default optimize=true → bare `--optimize` recorded + forwarded. + assert_eq!( + forwarded, + vec!["--locked".to_string(), "--optimize".to_string()] + ); + assert_eq!( + bldopts, + vec!["--locked".to_string(), "--optimize".to_string()] + ); } #[test] @@ -588,7 +696,7 @@ mod tests { package: Some("contract-a".to_string()), ..Cmd::default() }; - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws()); + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), true); assert!(forwarded.contains(&"--features=a,b".to_string())); assert!(forwarded.contains(&"--package=contract-a".to_string())); assert!(bldopts.contains(&"--features=a,b".to_string())); @@ -597,7 +705,7 @@ mod tests { } #[test] - fn build_forwarded_args_records_meta_optimize_and_manifest() { + fn build_forwarded_args_records_meta_and_manifest() { let cmd = Cmd { manifest_path: Some(PathBuf::from("/tmp/ws/contracts/add/Cargo.toml")), build_args: super::super::BuildArgs { @@ -605,23 +713,49 @@ mod tests { ("home_domain".to_string(), "fnando.com".to_string()), ("author".to_string(), "alice".to_string()), ], - optimize: false, + optimize: true, }, ..Cmd::default() }; - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws()); + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), true); assert!(forwarded.contains(&"--meta=home_domain=fnando.com".to_string())); assert!(forwarded.contains(&"--meta=author=alice".to_string())); - assert!(forwarded.contains(&"--optimize=false".to_string())); assert!(forwarded.contains(&"--manifest-path=contracts/add/Cargo.toml".to_string())); - // Same set is captured into bldopts so a verifier can replay every - // build-affecting flag. assert!(bldopts.contains(&"--meta=home_domain=fnando.com".to_string())); assert!(bldopts.contains(&"--meta=author=alice".to_string())); - assert!(bldopts.contains(&"--optimize=false".to_string())); assert!(bldopts.contains(&"--manifest-path=contracts/add/Cargo.toml".to_string())); } + #[test] + fn build_forwarded_args_optimize_false_new_container() { + let cmd = Cmd { + build_args: super::super::BuildArgs { + meta: vec![], + optimize: false, + }, + ..Cmd::default() + }; + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), true); + assert!(forwarded.contains(&"--optimize=false".to_string())); + assert!(bldopts.contains(&"--optimize=false".to_string())); + } + + #[test] + fn build_forwarded_args_optimize_false_old_container() { + let cmd = Cmd { + build_args: super::super::BuildArgs { + meta: vec![], + optimize: false, + }, + ..Cmd::default() + }; + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), false); + // Old container's default is already false; record nothing. + // Passing `--optimize=false` to a pre-26.1.0 cli would fail. + assert!(!forwarded.iter().any(|a| a.starts_with("--optimize"))); + assert!(!bldopts.iter().any(|a| a.starts_with("--optimize"))); + } + #[test] fn build_metadata_args_orders_keys() { let m = build_metadata_args( From c81c6d578ab5548a2f0510e9719aefdf3581fa04 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Thu, 21 May 2026 16:19:24 -0700 Subject: [PATCH 05/25] Add SEP-58 source-id flags to --verifiable. --- FULL_HELP_DOCS.md | 4 + cmd/crates/soroban-test/tests/it/build.rs | 178 ++++++-- .../src/commands/contract/build.rs | 46 ++ .../src/commands/contract/build/verifiable.rs | 397 +++++++++++++++--- 4 files changed, 548 insertions(+), 77 deletions(-) diff --git a/FULL_HELP_DOCS.md b/FULL_HELP_DOCS.md index f0ca271564..ba4ec18ac6 100644 --- a/FULL_HELP_DOCS.md +++ b/FULL_HELP_DOCS.md @@ -399,6 +399,10 @@ To view the commands that will be executed, without executing them, use the --pr - `--verifiable` — Build inside a trusted Docker container and record SEP-58 metadata (`bldimg`, `source_rev`, `bldopt`) so the resulting WASM can be reproduced and verified by third parties. Implies `--locked`. Requires a clean git working tree - `--image ` — Override the auto-selected container image used by `--verifiable`. Must be digest-pinned, e.g. `docker.io/stellar/stellar-cli@sha256:...`. Tag-only refs are rejected because SEP-58 requires content addressing +- `--source-repo ` — SEP-58 source identification: HTTPS URL (or `github:user/repo`) of the source repository. Must be passed together with `--source-rev` +- `--source-rev ` — SEP-58 source identification: 40-char SHA-1 of the source commit. The local workspace must be a git repo at this exact SHA with a clean working tree. Must be passed together with `--source-repo` +- `--tarball-url ` — SEP-58 source identification: URL where the source tarball can be downloaded +- `--tarball-sha256 ` — SEP-58 source identification: SHA-256 of the source tarball bytes ## `stellar contract extend` diff --git a/cmd/crates/soroban-test/tests/it/build.rs b/cmd/crates/soroban-test/tests/it/build.rs index 5936bfeacf..bb41346a3c 100644 --- a/cmd/crates/soroban-test/tests/it/build.rs +++ b/cmd/crates/soroban-test/tests/it/build.rs @@ -994,6 +994,41 @@ fn build_always_injects_cli_version() { ); } +const ZERO_DIGEST: &str = + "docker.io/stellar/stellar-cli@sha256:0000000000000000000000000000000000000000000000000000000000000000"; + +// Convenience: drive a git command in a fixture directory. +fn git_in(dir: &Path, args: &[&str]) { + std::process::Command::new("git") + .args(args) + .current_dir(dir) + .env("GIT_AUTHOR_NAME", "Test") + .env("GIT_AUTHOR_EMAIL", "test@example.com") + .env("GIT_COMMITTER_NAME", "Test") + .env("GIT_COMMITTER_EMAIL", "test@example.com") + .status() + .unwrap(); +} + +// Init a tempdir copy of the workspace fixture and return the workspace path. +fn fresh_workspace() -> (TempDir, PathBuf) { + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace"); + let temp = TempDir::new().unwrap(); + fs_extra::dir::copy(&fixture_path, temp.path(), &CopyOptions::new()).unwrap(); + let workspace = temp.path().join("workspace"); + (temp, workspace) +} + +fn git_head(dir: &Path) -> String { + let out = std::process::Command::new("git") + .args(["rev-parse", "HEAD"]) + .current_dir(dir) + .output() + .unwrap(); + String::from_utf8(out.stdout).unwrap().trim().to_string() +} + // `--verifiable` cannot accept reserved `--meta` keys that the cli writes itself. #[test] fn verifiable_meta_conflict_errors() { @@ -1007,7 +1042,9 @@ fn verifiable_meta_conflict_errors() { .arg("build") .arg("--verifiable") .arg("--image") - .arg("docker.io/stellar/stellar-cli@sha256:0000000000000000000000000000000000000000000000000000000000000000") + .arg(ZERO_DIGEST) + .arg("--tarball-url") + .arg("https://example.com/foo.tar.gz") .arg("--meta") .arg("bldimg=not-allowed") .assert() @@ -1015,7 +1052,7 @@ fn verifiable_meta_conflict_errors() { .stderr(predicate::str::contains("reserved key: bldimg")); } -// `--image` must be content-addressed; tag-only refs are rejected. +// `--image` is validated against the SEP-58 bldimg regex; tag-only refs fail. #[test] fn verifiable_image_must_be_digest_pinned() { let sandbox = TestEnv::default(); @@ -1029,39 +1066,118 @@ fn verifiable_image_must_be_digest_pinned() { .arg("--verifiable") .arg("--image") .arg("docker.io/stellar/stellar-cli:latest") + .arg("--tarball-url") + .arg("https://example.com/foo.tar.gz") .assert() .failure() - .stderr(predicate::str::contains("must be digest-pinned")); + .stderr(predicate::str::contains("bldimg format")); } -// A dirty git tree breaks the verifiability property because `source_rev` would -// record a commit whose bytes don't match the produced WASM. Hard fail. +// SEP-58 bldimg requires an explicit registry host (e.g. `docker.io/...`). +// Implicit Docker-Hub-style short refs are rejected. #[test] -fn verifiable_dirty_tree_errors() { +fn verifiable_image_requires_explicit_registry_host() { let sandbox = TestEnv::default(); let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); - let fixture_path = cargo_dir.join("tests/fixtures/workspace"); - let temp = TempDir::new().unwrap(); - let dir_path = temp.path(); - fs_extra::dir::copy(fixture_path, dir_path, &CopyOptions::new()).unwrap(); - let workspace = dir_path.join("workspace"); - - // Bootstrap a clean git tree at the workspace root, then dirty it so the - // verifiable path's dirty-check trips before docker is touched. - let git = |args: &[&str]| { - std::process::Command::new("git") - .args(args) - .current_dir(&workspace) - .env("GIT_AUTHOR_NAME", "Test") - .env("GIT_AUTHOR_EMAIL", "test@example.com") - .env("GIT_COMMITTER_NAME", "Test") - .env("GIT_COMMITTER_EMAIL", "test@example.com") - .status() - .unwrap(); - }; - git(&["init", "-q", "-b", "main"]); - git(&["add", "-A"]); - git(&["commit", "-q", "-m", "init"]); + let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); + + let short_ref = format!("stellar/stellar-cli@sha256:{}", "0".repeat(64)); + + sandbox + .new_assert_cmd("contract") + .current_dir(fixture_path) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg(short_ref) + .arg("--tarball-url") + .arg("https://example.com/foo.tar.gz") + .assert() + .failure() + .stderr(predicate::str::contains("bldimg format")); +} + +// `--verifiable` without any source-identification flag must error. +#[test] +fn verifiable_requires_source_id() { + let sandbox = TestEnv::default(); + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); + + sandbox + .new_assert_cmd("contract") + .current_dir(fixture_path) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg(ZERO_DIGEST) + .assert() + .failure() + .stderr(predicate::str::contains("source-identification")); +} + +// `--source-rev` value must match the 40-hex regex. +#[test] +fn verifiable_source_rev_format_errors() { + let sandbox = TestEnv::default(); + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); + + sandbox + .new_assert_cmd("contract") + .current_dir(fixture_path) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg(ZERO_DIGEST) + .arg("--source-repo") + .arg("https://github.com/foo/bar") + .arg("--source-rev") + .arg("not-a-sha") + .assert() + .failure() + .stderr(predicate::str::contains("source_rev format")); +} + +// `--source-rev` is cross-checked against local git HEAD; a mismatch is a hard +// fail before docker is touched. +#[test] +fn verifiable_source_rev_must_match_head() { + let sandbox = TestEnv::default(); + let (_temp, workspace) = fresh_workspace(); + git_in(&workspace, &["init", "-q", "-b", "main"]); + git_in(&workspace, &["add", "-A"]); + git_in(&workspace, &["commit", "-q", "-m", "init"]); + + let bogus = "a".repeat(40); + + sandbox + .new_assert_cmd("contract") + .current_dir(workspace.join("contracts").join("add")) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg(ZERO_DIGEST) + .arg("--source-repo") + .arg("https://github.com/foo/bar") + .arg("--source-rev") + .arg(bogus) + .assert() + .failure() + .stderr(predicate::str::contains("does not match local HEAD")); +} + +// A dirty git tree under `--source-rev` is a hard fail (the recorded rev would +// not describe the bytes built). +#[test] +fn verifiable_dirty_tree_errors_with_source_rev() { + let sandbox = TestEnv::default(); + let (_temp, workspace) = fresh_workspace(); + git_in(&workspace, &["init", "-q", "-b", "main"]); + git_in(&workspace, &["add", "-A"]); + git_in(&workspace, &["commit", "-q", "-m", "init"]); + let head = git_head(&workspace); + // Dirty the tree after committing so HEAD matches but status is non-empty. std::fs::write(workspace.join("dirty.txt"), b"uncommitted").unwrap(); sandbox @@ -1070,7 +1186,11 @@ fn verifiable_dirty_tree_errors() { .arg("build") .arg("--verifiable") .arg("--image") - .arg("docker.io/stellar/stellar-cli@sha256:0000000000000000000000000000000000000000000000000000000000000000") + .arg(ZERO_DIGEST) + .arg("--source-repo") + .arg("https://github.com/foo/bar") + .arg("--source-rev") + .arg(head) .assert() .failure() .stderr(predicate::str::contains("dirty").or(predicate::str::contains("clean tree"))); diff --git a/cmd/soroban-cli/src/commands/contract/build.rs b/cmd/soroban-cli/src/commands/contract/build.rs index c6f1173f74..7d912c7084 100644 --- a/cmd/soroban-cli/src/commands/contract/build.rs +++ b/cmd/soroban-cli/src/commands/contract/build.rs @@ -111,6 +111,48 @@ pub struct Cmd { #[arg(long, requires = "verifiable", help_heading = "Verifiable")] pub image: Option, + /// SEP-58 source identification: HTTPS URL (or `github:user/repo`) of the + /// source repository. Must be passed together with `--source-rev`. + #[arg( + long, + requires = "verifiable", + requires = "source_rev", + conflicts_with_all = ["tarball_url", "tarball_sha256"], + help_heading = "Verifiable" + )] + pub source_repo: Option, + + /// SEP-58 source identification: 40-char SHA-1 of the source commit. The + /// local workspace must be a git repo at this exact SHA with a clean + /// working tree. Must be passed together with `--source-repo`. + #[arg( + long, + requires = "verifiable", + requires = "source_repo", + conflicts_with_all = ["tarball_url", "tarball_sha256"], + help_heading = "Verifiable" + )] + pub source_rev: Option, + + /// SEP-58 source identification: URL where the source tarball can be + /// downloaded. + #[arg( + long, + requires = "verifiable", + conflicts_with_all = ["source_repo", "source_rev"], + help_heading = "Verifiable" + )] + pub tarball_url: Option, + + /// SEP-58 source identification: SHA-256 of the source tarball bytes. + #[arg( + long, + requires = "verifiable", + conflicts_with_all = ["source_repo", "source_rev"], + help_heading = "Verifiable" + )] + pub tarball_sha256: Option, + #[command(flatten)] pub container_args: container::shared::Args, @@ -245,6 +287,10 @@ impl Default for Cmd { print_commands_only: false, verifiable: false, image: None, + source_repo: None, + source_rev: None, + tarball_url: None, + tarball_sha256: None, container_args: container::shared::Args { docker_host: None }, build_args: BuildArgs::default(), } diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 6d4b2ae6c3..adf7233ec0 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -44,8 +44,8 @@ pub enum Error { #[error(transparent)] Bollard(#[from] bollard::errors::Error), - #[error("--image must be digest-pinned (got {value}); SEP-58 requires content-addressed images. Pass docker.io/stellar/stellar-cli@sha256:")] - ImageNotDigestPinned { value: String }, + #[error("--image value {value:?} does not match the SEP-58 bldimg format `/@sha256:<64-hex>`. Examples: docker.io/stellar/stellar-cli@sha256:<64-hex>, localhost:5000/foo@sha256:<64-hex>. Tag-only refs and implicit Docker-Hub short refs are not accepted.")] + BldimgFormat { value: String }, #[error("could not determine the running rustc version: {0}")] RustcVersion(String), @@ -74,7 +74,7 @@ pub enum Error { }, #[error( - "git working tree at {path} is dirty. Verifiable builds require a clean tree so the recorded source_rev matches the WASM bytes. Commit or stash your changes and try again." + "git working tree at {path} is dirty. --source-rev requires a clean tree so the recorded source_rev matches the WASM bytes. Commit or stash your changes and try again." )] GitDirty { path: PathBuf }, @@ -83,6 +83,27 @@ pub enum Error { )] ReservedMetaKey { key: String }, + #[error("--verifiable requires a SEP-58 source-identification combination. Pass one of: (--source-repo + --source-rev), (--tarball-url and/or --tarball-sha256).")] + MissingSourceId, + + #[error("--source-rev value {value:?} does not match the SEP-58 source_rev format `^[0-9a-f]{{40}}$` (full 40-char SHA-1 of the source commit).")] + SourceRevFormat { value: String }, + + #[error("--source-repo value {value:?} does not match the SEP-58 source_repo format `^(https?://\\S+|github:[^/\\s]+/[^/\\s]+)$`.")] + SourceRepoFormat { value: String }, + + #[error("--tarball-url value {value:?} does not match the SEP-58 tarball_url format `^https?://\\S+$`.")] + TarballUrlFormat { value: String }, + + #[error("--tarball-sha256 value {value:?} does not match the SEP-58 tarball_sha256 format `^[0-9a-f]{{64}}$`.")] + TarballSha256Format { value: String }, + + #[error("--source-rev requires a git workspace at {path}; `git rev-parse HEAD` failed there.")] + SourceRevNotGitRepo { path: PathBuf }, + + #[error("--source-rev {claimed} does not match local HEAD {head}. Commit, switch, or pass the correct rev.")] + SourceRevHeadMismatch { claimed: String, head: String }, + #[error("container build exited with status {status}. To reproduce manually:\n docker run --rm -v {mount}:/source {image} {args}")] ContainerExit { status: i64, @@ -104,8 +125,8 @@ pub async fn run( } } if let Some(img) = &cmd.image { - if !img.contains("@sha256:") { - return Err(Error::ImageNotDigestPinned { value: img.clone() }.into()); + if !bldimg_regex().is_match(img) { + return Err(Error::BldimgFormat { value: img.clone() }.into()); } } @@ -113,9 +134,11 @@ pub async fn run( print.infoln("--verifiable implies --locked"); } - // Stage 2: local filesystem + git, no network. + // Stage 2: local filesystem + git, no network. Resolve the workspace root + // first so the (optional) `--source-rev` git cross-check has a path to + // anchor on. let workspace_root = resolve_workspace_root(cmd)?; - let source_rev = git_source_rev(&workspace_root, print)?; + let source_ids = validate_source_ids(cmd, &workspace_root)?; // Stage 3: docker. let docker = cmd @@ -136,7 +159,7 @@ pub async fn run( let (forwarded_args, bldopts) = build_forwarded_args(cmd, &workspace_root, supports_explicit_optimize_false); - let metadata_args = build_metadata_args(&image_ref, &source_rev, &bldopts); + let metadata_args = build_metadata_args(&image_ref, &source_ids, &bldopts); let container_cmd_args = compose_container_args(&forwarded_args, &metadata_args); run_in_container( @@ -162,32 +185,91 @@ fn resolve_workspace_root(cmd: &Cmd) -> Result { Ok(md.workspace_root.into_std_path_buf()) } -fn git_source_rev(workspace_root: &Path, print: &Print) -> Result { - // Probe with rev-parse first to detect "not a git repo". - let rev = Command::new("git") +/// Source-identification fields, gathered from the corresponding CLI flags +/// after validation. Each is `Some` only when the user passed the flag and the +/// value matched the SEP-58 format regex. The four fields cannot all be +/// `None` — `validate_source_ids` rejects that case. +#[derive(Debug, Default, Clone)] +struct SourceIds { + source_repo: Option, + source_rev: Option, + tarball_url: Option, + tarball_sha256: Option, +} + +fn validate_source_ids(cmd: &Cmd, workspace_root: &Path) -> Result { + let ids = SourceIds { + source_repo: cmd.source_repo.clone(), + source_rev: cmd.source_rev.clone(), + tarball_url: cmd.tarball_url.clone(), + tarball_sha256: cmd.tarball_sha256.clone(), + }; + + if ids.source_repo.is_none() + && ids.source_rev.is_none() + && ids.tarball_url.is_none() + && ids.tarball_sha256.is_none() + { + return Err(Error::MissingSourceId); + } + + if let Some(v) = &ids.source_rev { + if !source_rev_regex().is_match(v) { + return Err(Error::SourceRevFormat { value: v.clone() }); + } + } + + if let Some(v) = &ids.source_repo { + if !source_repo_regex().is_match(v) { + return Err(Error::SourceRepoFormat { value: v.clone() }); + } + } + + if let Some(v) = &ids.tarball_url { + if !tarball_url_regex().is_match(v) { + return Err(Error::TarballUrlFormat { value: v.clone() }); + } + } + + if let Some(v) = &ids.tarball_sha256 { + if !tarball_sha256_regex().is_match(v) { + return Err(Error::TarballSha256Format { value: v.clone() }); + } + } + + if let Some(claimed) = &ids.source_rev { + cross_check_source_rev_against_git(workspace_root, claimed)?; + } + + Ok(ids) +} + +fn cross_check_source_rev_against_git(workspace_root: &Path, claimed: &str) -> Result<(), Error> { + let rev_out = Command::new("git") .arg("-C") .arg(workspace_root) .arg("rev-parse") .arg("HEAD") - .output(); - let rev = match rev { - Ok(o) if o.status.success() => String::from_utf8_lossy(&o.stdout).trim().to_string(), - Ok(_) => { - print.warnln(format!( - "{} is not a git repository; recording empty source_rev (verifiability is degraded).", - workspace_root.display() - )); - return Ok(String::new()); - } - Err(e) => { - return Err(Error::GitInvoke { - path: workspace_root.to_path_buf(), - source: e, - }) - } - }; + .output() + .map_err(|e| Error::GitInvoke { + path: workspace_root.to_path_buf(), + source: e, + })?; + + if !rev_out.status.success() { + return Err(Error::SourceRevNotGitRepo { + path: workspace_root.to_path_buf(), + }); + } + + let head = String::from_utf8_lossy(&rev_out.stdout).trim().to_string(); + if head != claimed { + return Err(Error::SourceRevHeadMismatch { + claimed: claimed.to_string(), + head, + }); + } - // Dirty check. let status = Command::new("git") .arg("-C") .arg(workspace_root) @@ -198,13 +280,35 @@ fn git_source_rev(workspace_root: &Path, print: &Print) -> Result path: workspace_root.to_path_buf(), source: e, })?; + if !status.stdout.is_empty() { return Err(Error::GitDirty { path: workspace_root.to_path_buf(), }); } - Ok(rev) + Ok(()) +} + +fn bldimg_regex() -> Regex { + Regex::new(r"^(?:localhost(?::\d+)?|[^\s@/]*[.:][^\s@/]*)/[^\s@]+@sha256:[0-9a-f]{64}$") + .unwrap() +} + +fn source_rev_regex() -> Regex { + Regex::new(r"^[0-9a-f]{40}$").unwrap() +} + +fn source_repo_regex() -> Regex { + Regex::new(r"^(https?://\S+|github:[^/\s]+/[^/\s]+)$").unwrap() +} + +fn tarball_url_regex() -> Regex { + Regex::new(r"^https?://\S+$").unwrap() +} + +fn tarball_sha256_regex() -> Regex { + Regex::new(r"^[0-9a-f]{64}$").unwrap() } /// The flags forwarded to the container's `stellar contract build`, plus the @@ -274,16 +378,33 @@ fn build_forwarded_args( (forwarded, bldopts) } -fn build_metadata_args(image_ref: &str, source_rev: &str, bldopts: &[String]) -> Vec { +fn build_metadata_args(image_ref: &str, ids: &SourceIds, bldopts: &[String]) -> Vec { let mut out = Vec::new(); - for (k, v) in [("bldimg", image_ref), ("source_rev", source_rev)] { + + let push = |out: &mut Vec, key: &str, val: &str| { out.push("--meta".to_string()); - out.push(format!("{k}={v}")); + out.push(format!("{key}={val}")); + }; + + push(&mut out, "bldimg", image_ref); + + if let Some(v) = &ids.source_repo { + push(&mut out, "source_repo", v); + } + if let Some(v) = &ids.source_rev { + push(&mut out, "source_rev", v); } + if let Some(v) = &ids.tarball_url { + push(&mut out, "tarball_url", v); + } + if let Some(v) = &ids.tarball_sha256 { + push(&mut out, "tarball_sha256", v); + } + for o in bldopts { - out.push("--meta".to_string()); - out.push(format!("bldopt={o}")); + push(&mut out, "bldopt", o); } + out } @@ -296,8 +417,8 @@ fn compose_container_args(forwarded: &[String], metadata: &[String]) -> Vec Result { if let Some(s) = &cmd.image { - if !s.contains("@sha256:") { - return Err(Error::ImageNotDigestPinned { value: s.clone() }); + if !bldimg_regex().is_match(s) { + return Err(Error::BldimgFormat { value: s.clone() }); } return Ok(s.clone()); } @@ -756,25 +877,205 @@ mod tests { assert!(!bldopts.iter().any(|a| a.starts_with("--optimize"))); } + fn pairs(args: &[String]) -> Vec<(&str, &str)> { + args.chunks(2) + .map(|c| (c[0].as_str(), c[1].as_str())) + .collect() + } + #[test] - fn build_metadata_args_orders_keys() { + fn build_metadata_args_source_repo_and_rev() { + let ids = SourceIds { + source_repo: Some("https://github.com/foo/bar".to_string()), + source_rev: Some("a".repeat(40)), + tarball_url: None, + tarball_sha256: None, + }; let m = build_metadata_args( "docker.io/stellar/stellar-cli@sha256:abc", - "deadbeef", + &ids, &["--locked".to_string(), "--features=a".to_string()], ); - // bldimg, source_rev, then bldopts in order. - let pairs: Vec<(&str, &str)> = m - .chunks(2) - .map(|c| (c[0].as_str(), c[1].as_str())) - .collect(); + let p = pairs(&m); + // bldimg first; source-ids only for what's set; bldopts last. assert_eq!( - pairs[0], + p[0], ("--meta", "bldimg=docker.io/stellar/stellar-cli@sha256:abc") ); - assert_eq!(pairs[1], ("--meta", "source_rev=deadbeef")); - assert_eq!(pairs[2], ("--meta", "bldopt=--locked")); - assert_eq!(pairs[3], ("--meta", "bldopt=--features=a")); + assert_eq!(p[1], ("--meta", "source_repo=https://github.com/foo/bar")); + assert_eq!(p[2].0, "--meta"); + assert!(p[2].1.starts_with("source_rev=")); + assert_eq!(p[3], ("--meta", "bldopt=--locked")); + assert_eq!(p[4], ("--meta", "bldopt=--features=a")); + // No tarball entries emitted when those fields are None. + assert!(!m.iter().any(|s| s.starts_with("tarball_"))); + } + + #[test] + fn build_metadata_args_tarball_url_only() { + let ids = SourceIds { + tarball_url: Some("https://example.com/foo.tar.gz".to_string()), + ..SourceIds::default() + }; + let m = build_metadata_args("docker.io/stellar/stellar-cli@sha256:abc", &ids, &[]); + assert!(m + .iter() + .any(|s| s == "tarball_url=https://example.com/foo.tar.gz")); + assert!(!m.iter().any(|s| s.starts_with("source_"))); + assert!(!m.iter().any(|s| s.starts_with("tarball_sha256="))); + } + + #[test] + fn build_metadata_args_tarball_pair() { + let ids = SourceIds { + tarball_url: Some("https://example.com/foo.tar.gz".to_string()), + tarball_sha256: Some("f".repeat(64)), + ..SourceIds::default() + }; + let m = build_metadata_args("docker.io/stellar/stellar-cli@sha256:abc", &ids, &[]); + assert!(m + .iter() + .any(|s| s == "tarball_url=https://example.com/foo.tar.gz")); + assert!(m + .iter() + .any(|s| s == &format!("tarball_sha256={}", "f".repeat(64)))); + } + + #[test] + fn validate_source_ids_missing_all_errors() { + let cmd = Cmd::default(); + let err = validate_source_ids(&cmd, ws()).unwrap_err(); + assert!(matches!(err, Error::MissingSourceId)); + } + + #[test] + fn validate_source_ids_rejects_bad_source_rev_format() { + let cmd = Cmd { + source_repo: Some("https://github.com/foo/bar".to_string()), + source_rev: Some("not-a-sha".to_string()), + ..Cmd::default() + }; + let err = validate_source_ids(&cmd, ws()).unwrap_err(); + assert!(matches!(err, Error::SourceRevFormat { .. })); + } + + #[test] + fn validate_source_ids_rejects_bad_source_repo_format() { + let cmd = Cmd { + source_repo: Some("foo/bar".to_string()), // missing scheme + source_rev: Some("a".repeat(40)), + ..Cmd::default() + }; + let err = validate_source_ids(&cmd, ws()).unwrap_err(); + assert!(matches!(err, Error::SourceRepoFormat { .. })); + } + + #[test] + fn validate_source_ids_rejects_bad_tarball_url() { + let cmd = Cmd { + tarball_url: Some("ftp://example.com/foo.tar.gz".to_string()), + ..Cmd::default() + }; + let err = validate_source_ids(&cmd, ws()).unwrap_err(); + assert!(matches!(err, Error::TarballUrlFormat { .. })); + } + + #[test] + fn validate_source_ids_rejects_short_tarball_sha256() { + let cmd = Cmd { + tarball_sha256: Some("abc".to_string()), + ..Cmd::default() + }; + let err = validate_source_ids(&cmd, ws()).unwrap_err(); + assert!(matches!(err, Error::TarballSha256Format { .. })); + } + + #[test] + fn validate_source_ids_accepts_tarball_url_alone() { + let cmd = Cmd { + tarball_url: Some("https://example.com/foo.tar.gz".to_string()), + ..Cmd::default() + }; + let ids = validate_source_ids(&cmd, ws()).unwrap(); + assert_eq!( + ids.tarball_url.as_deref(), + Some("https://example.com/foo.tar.gz") + ); + assert!(ids.source_repo.is_none()); + assert!(ids.source_rev.is_none()); + assert!(ids.tarball_sha256.is_none()); + } + + #[test] + fn validate_source_ids_accepts_tarball_sha256_alone() { + let cmd = Cmd { + tarball_sha256: Some("f".repeat(64)), + ..Cmd::default() + }; + let ids = validate_source_ids(&cmd, ws()).unwrap(); + assert_eq!( + ids.tarball_sha256.as_deref(), + Some("ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff") + ); + } + + #[test] + fn bldimg_regex_accepts_docker_hub_full_ref() { + assert!(bldimg_regex().is_match(&format!( + "docker.io/stellar/stellar-cli@sha256:{}", + "a".repeat(64) + ))); + } + + #[test] + fn bldimg_regex_accepts_localhost_registry() { + assert!(bldimg_regex().is_match(&format!("localhost:5000/foo@sha256:{}", "0".repeat(64)))); + } + + #[test] + fn bldimg_regex_rejects_implicit_hub_short_ref() { + // Implicit Docker Hub short ref: no registry host prefix. + assert!(!bldimg_regex().is_match(&format!("stellar/stellar-cli@sha256:{}", "a".repeat(64)))); + } + + #[test] + fn bldimg_regex_rejects_tag_only() { + assert!(!bldimg_regex().is_match("docker.io/stellar/stellar-cli:latest")); + } + + #[test] + fn bldimg_regex_rejects_short_sha() { + assert!(!bldimg_regex().is_match("docker.io/stellar/stellar-cli@sha256:abc")); + } + + #[test] + fn source_rev_regex_matches_40_hex() { + assert!(source_rev_regex().is_match(&"a".repeat(40))); + assert!(!source_rev_regex().is_match(&"a".repeat(39))); + assert!(!source_rev_regex().is_match(&"A".repeat(40))); // upper-case rejected + } + + #[test] + fn source_repo_regex_accepts_https_and_github_shorthand() { + assert!(source_repo_regex().is_match("https://github.com/foo/bar")); + assert!(source_repo_regex().is_match("http://example.com/foo.git")); + assert!(source_repo_regex().is_match("github:foo/bar")); + assert!(!source_repo_regex().is_match("foo/bar")); + assert!(!source_repo_regex().is_match("git@github.com:foo/bar.git")); + } + + #[test] + fn tarball_url_regex_accepts_http_only() { + assert!(tarball_url_regex().is_match("https://example.com/foo.tar.gz")); + assert!(tarball_url_regex().is_match("http://example.com/foo.tar.gz")); + assert!(!tarball_url_regex().is_match("ftp://example.com/foo.tar.gz")); + } + + #[test] + fn tarball_sha256_regex_matches_64_hex() { + assert!(tarball_sha256_regex().is_match(&"f".repeat(64))); + assert!(!tarball_sha256_regex().is_match(&"f".repeat(63))); + assert!(!tarball_sha256_regex().is_match(&"F".repeat(64))); } #[test] From e000d846069ef632f6425a6fe7b955268c42a9e2 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Thu, 21 May 2026 16:20:26 -0700 Subject: [PATCH 06/25] Move --locked info banner after validation. --- .../src/commands/contract/build/verifiable.rs | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index adf7233ec0..03a376fda0 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -130,16 +130,18 @@ pub async fn run( } } - if !cmd.locked { - print.infoln("--verifiable implies --locked"); - } - // Stage 2: local filesystem + git, no network. Resolve the workspace root // first so the (optional) `--source-rev` git cross-check has a path to // anchor on. let workspace_root = resolve_workspace_root(cmd)?; let source_ids = validate_source_ids(cmd, &workspace_root)?; + // Defer the info banner until every validation has passed, so it doesn't + // appear right before an error. + if !cmd.locked { + print.infoln("--verifiable implies --locked"); + } + // Stage 3: docker. let docker = cmd .container_args From 5f59fb856aaecbd56f5817ddb584492f85c4c9df Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Thu, 21 May 2026 17:27:49 -0700 Subject: [PATCH 07/25] Group --docker-host under Verifiable on contract build. --- FULL_HELP_DOCS.md | 2 +- cmd/soroban-cli/src/commands/contract/build.rs | 9 +++++---- .../src/commands/contract/build/verifiable.rs | 6 ++++-- 3 files changed, 10 insertions(+), 7 deletions(-) diff --git a/FULL_HELP_DOCS.md b/FULL_HELP_DOCS.md index ba4ec18ac6..58f5340421 100644 --- a/FULL_HELP_DOCS.md +++ b/FULL_HELP_DOCS.md @@ -384,7 +384,6 @@ To view the commands that will be executed, without executing them, use the --pr If ommitted, wasm files are written only to the cargo target directory. - `--locked` — Assert that `Cargo.lock` will remain unchanged -- `-d`, `--docker-host ` — Optional argument to override the default docker host. This is useful when you are using a non-standard docker host path for your Docker-compatible container runtime, e.g. Docker Desktop defaults to $HOME/.docker/run/docker.sock instead of /var/run/docker.sock - `--optimize ` — Optimize the generated wasm. Enabled by default; pass `--optimize=false` to disable. Requires the `additional-libs` feature Default value: `true` @@ -403,6 +402,7 @@ To view the commands that will be executed, without executing them, use the --pr - `--source-rev ` — SEP-58 source identification: 40-char SHA-1 of the source commit. The local workspace must be a git repo at this exact SHA with a clean working tree. Must be passed together with `--source-repo` - `--tarball-url ` — SEP-58 source identification: URL where the source tarball can be downloaded - `--tarball-sha256 ` — SEP-58 source identification: SHA-256 of the source tarball bytes +- `-d`, `--docker-host ` — Override the default docker host used by `--verifiable` ## `stellar contract extend` diff --git a/cmd/soroban-cli/src/commands/contract/build.rs b/cmd/soroban-cli/src/commands/contract/build.rs index 7d912c7084..51829fa56f 100644 --- a/cmd/soroban-cli/src/commands/contract/build.rs +++ b/cmd/soroban-cli/src/commands/contract/build.rs @@ -20,7 +20,7 @@ use stellar_xdr::{Limited, Limits, ScMetaEntry, ScMetaV0, StringM, WriteXdr}; #[cfg(feature = "additional-libs")] use crate::commands::contract::optimize; use crate::{ - commands::{container, global, version}, + commands::{global, version}, print::Print, wasm, }; @@ -153,8 +153,9 @@ pub struct Cmd { )] pub tarball_sha256: Option, - #[command(flatten)] - pub container_args: container::shared::Args, + /// Override the default docker host used by `--verifiable`. + #[arg(short = 'd', long, env = "DOCKER_HOST", help_heading = "Verifiable")] + pub docker_host: Option, #[command(flatten)] pub build_args: BuildArgs, @@ -291,7 +292,7 @@ impl Default for Cmd { source_rev: None, tarball_url: None, tarball_sha256: None, - container_args: container::shared::Args { docker_host: None }, + docker_host: None, build_args: BuildArgs::default(), } } diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 03a376fda0..0a84f19ca5 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -143,8 +143,10 @@ pub async fn run( } // Stage 3: docker. - let docker = cmd - .container_args + let docker_args = crate::commands::container::shared::Args { + docker_host: cmd.docker_host.clone(), + }; + let docker = docker_args .connect_to_docker(print) .await .map_err(Error::DockerConnection)?; From 17c2fc8ad4c04dffd9ed41852fa8d9e7c759e39d Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Thu, 21 May 2026 18:32:05 -0700 Subject: [PATCH 08/25] Plumb verbose flag through run_in_container. --- .../src/commands/contract/build/verifiable.rs | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 0a84f19ca5..90fa367ecb 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -166,12 +166,17 @@ pub async fn run( let metadata_args = build_metadata_args(&image_ref, &source_ids, &bldopts); let container_cmd_args = compose_container_args(&forwarded_args, &metadata_args); + // Always stream the container's cargo output during `contract build + // --verifiable`, matching how a non-verifiable `contract build` shows + // cargo output by default. The verify-side caller gates this on + // `--verbose` because verifications are run as part of pipelines. run_in_container( &image_ref, &workspace_root, &container_cmd_args, &docker, print, + true, ) .await?; @@ -653,6 +658,7 @@ async fn run_in_container( container_cmd: &[String], docker: &Docker, print: &Print, + verbose: bool, ) -> Result<(), Error> { let bind = format!("{}:/source", workspace_root.display()); let config = ContainerCreateBody { @@ -700,8 +706,10 @@ async fn run_in_container( bollard::container::LogOutput::StdOut { message } | bollard::container::LogOutput::StdErr { message }, ) => { - let s = String::from_utf8_lossy(&message); - print.blankln(s.trim_end()); + if verbose { + let s = String::from_utf8_lossy(&message); + print.blankln(s.trim_end()); + } } Ok(_) => {} Err(e) => return Err(e.into()), From 9c3b31f86f0f23b17e192f0b00ee9d0125ea4e54 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Thu, 21 May 2026 19:41:06 -0700 Subject: [PATCH 09/25] Capitalize info and warn messages on verifiable build. --- cmd/soroban-cli/src/commands/contract/build/verifiable.rs | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 90fa367ecb..9d45440d91 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -139,7 +139,7 @@ pub async fn run( // Defer the info banner until every validation has passed, so it doesn't // appear right before an error. if !cmd.locked { - print.infoln("--verifiable implies --locked"); + print.infoln("Implying --locked because --verifiable was passed"); } // Stage 3: docker. @@ -600,7 +600,7 @@ async fn probe_supports_optimize_false_syntax( } Err(e) => { print.warnln(format!( - "could not probe container cli version ({e}); assuming pre-{OPTIMIZE_NEW_SYNTAX_MIN} syntax" + "Could not probe container cli version ({e}); assuming pre-{OPTIMIZE_NEW_SYNTAX_MIN} syntax" )); false } From c20a1d5f373ad4b0f9dc37616ab1c11787d12c9c Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Thu, 21 May 2026 22:29:10 -0700 Subject: [PATCH 10/25] Rewrite docker pull status lines for clarity. --- .../src/commands/contract/build/verifiable.rs | 17 ++++++++++++----- 1 file changed, 12 insertions(+), 5 deletions(-) diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 9d45440d91..71ff2f95bf 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -483,11 +483,18 @@ async fn pull_image( ); while let Some(item) = stream.try_next().await? { if let Some(status) = item.status { - if status.contains("Pulling from") - || status.contains("Digest") - || status.contains("Status") - { - print.infoln(status); + // The docker daemon emits short status lines like: + // "Pulling from " + // "Digest: sha256:" + // "Status: Image is up to date for " + // Stand-alone "Digest" reads as an orphan. Rewrite each line so + // it makes sense outside the docker-pull context. + if let Some(repo) = status.strip_prefix("Pulling from ") { + print.infoln(format!("Pulling image {repo}")); + } else if let Some(digest) = status.strip_prefix("Digest: ") { + print.infoln(format!("Image digest: {digest}")); + } else if let Some(rest) = status.strip_prefix("Status: ") { + print.infoln(format!("Image: {rest}")); } } } From 4d7fcd633d895d2eb371f7544a4307a9d4a59d48 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Thu, 21 May 2026 23:49:54 -0700 Subject: [PATCH 11/25] Anchor verifiable build bind-mount to git root or cwd. --- .../src/commands/contract/build/verifiable.rs | 82 ++++++++++++++++++- 1 file changed, 80 insertions(+), 2 deletions(-) diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 71ff2f95bf..1e185380c6 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -136,6 +136,16 @@ pub async fn run( let workspace_root = resolve_workspace_root(cmd)?; let source_ids = validate_source_ids(cmd, &workspace_root)?; + // Pick the anchor the container bind-mounts and the `--manifest-path` + // bldopt is relativized against. A verifier will clone source_repo (or + // extract the tarball) into a fresh tempdir and bind-mount its root, so + // the build must do the symmetric thing on the host: bind-mount the local + // clone root (where `.git` lives) or, if there's no clone, the user's + // cwd. We do NOT validate that the local clone matches `--source-repo` — + // a wrong clone produces different bytes, and verify catches that at + // byte-comparison time. + let source_root = resolve_source_root(cmd); + // Defer the info banner until every validation has passed, so it doesn't // appear right before an error. if !cmd.locked { @@ -162,7 +172,7 @@ pub async fn run( }; let (forwarded_args, bldopts) = - build_forwarded_args(cmd, &workspace_root, supports_explicit_optimize_false); + build_forwarded_args(cmd, &source_root, supports_explicit_optimize_false); let metadata_args = build_metadata_args(&image_ref, &source_ids, &bldopts); let container_cmd_args = compose_container_args(&forwarded_args, &metadata_args); @@ -172,7 +182,7 @@ pub async fn run( // `--verbose` because verifications are run as part of pipelines. run_in_container( &image_ref, - &workspace_root, + &source_root, &container_cmd_args, &docker, print, @@ -194,6 +204,34 @@ fn resolve_workspace_root(cmd: &Cmd) -> Result { Ok(md.workspace_root.into_std_path_buf()) } +/// Pick the anchor for the container bind-mount and for relativizing +/// `--manifest-path` into the recorded `bldopt`. Walk up from the user's +/// `--manifest-path` (or cwd, if no manifest_path) looking for a `.git` +/// directory; return its parent. If none is found, fall back to cwd. +/// +/// This isn't a validation step — any `.git` will do. Wrong-clone mistakes +/// are caught later by the verify-side byte comparison. +fn resolve_source_root(cmd: &Cmd) -> PathBuf { + let start = if let Some(p) = &cmd.manifest_path { + let abs = std::path::absolute(p).unwrap_or_else(|_| p.clone()); + abs.parent().map(Path::to_path_buf).unwrap_or(abs) + } else { + std::env::current_dir().unwrap_or_else(|_| PathBuf::from(".")) + }; + + let mut p = start.clone(); + loop { + if p.join(".git").exists() { + return p; + } + if !p.pop() { + break; + } + } + + std::env::current_dir().unwrap_or(start) +} + /// Source-identification fields, gathered from the corresponding CLI flags /// after validation. Each is `Some` only when the user passed the flag and the /// value matched the SEP-58 format regex. The four fields cannot all be @@ -1097,6 +1135,46 @@ mod tests { assert!(!tarball_sha256_regex().is_match(&"F".repeat(64))); } + #[test] + fn resolve_source_root_finds_git_root_from_subdir() { + let temp = tempfile::TempDir::new().unwrap(); + let root = temp.path(); + std::fs::create_dir_all(root.join(".git")).unwrap(); + let nested = root.join("contracts").join("foo"); + std::fs::create_dir_all(&nested).unwrap(); + std::fs::write(nested.join("Cargo.toml"), b"# placeholder").unwrap(); + + let cmd = Cmd { + manifest_path: Some(nested.join("Cargo.toml")), + ..Cmd::default() + }; + // Use canonicalize on both sides — `tempfile` returns symlinked /var + // paths on macOS while resolve_source_root walks the same prefix. + let got = std::fs::canonicalize(resolve_source_root(&cmd)).unwrap(); + let want = std::fs::canonicalize(root).unwrap(); + assert_eq!(got, want); + } + + #[test] + fn resolve_source_root_falls_back_to_cwd_without_git() { + let temp = tempfile::TempDir::new().unwrap(); + let root = temp.path(); + let nested = root.join("noisy"); + std::fs::create_dir_all(&nested).unwrap(); + std::fs::write(nested.join("Cargo.toml"), b"# placeholder").unwrap(); + + let cmd = Cmd { + manifest_path: Some(nested.join("Cargo.toml")), + ..Cmd::default() + }; + // No `.git` anywhere up the tree, so we fall back to cwd. We can't + // assert what cwd is in a test runner (it varies), but we can assert + // that the returned path doesn't contain the manifest's parent and + // doesn't have `.git`. That's enough to confirm fallback kicked in. + let got = resolve_source_root(&cmd); + assert!(!got.join(".git").exists()); + } + #[test] fn compose_container_args_prefixes_subcommand() { let composed = compose_container_args( From d5a7209e0da9d7cc2c19945b7b3885753a28b750 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Fri, 22 May 2026 01:02:04 -0700 Subject: [PATCH 12/25] Pull bldimg when --image is set. --- cmd/soroban-cli/src/commands/contract/build/verifiable.rs | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 1e185380c6..66022ac826 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -467,6 +467,11 @@ pub async fn resolve_image(cmd: &Cmd, docker: &Docker, print: &Print) -> Result< if !bldimg_regex().is_match(s) { return Err(Error::BldimgFormat { value: s.clone() }); } + // Always pull, even when the digest is user-supplied. Docker requires + // the image to be locally present before `create_container` will + // accept it, and the user typically expects the cli to fetch + // whatever they asked for. + pull_image(docker, s, print).await?; return Ok(s.clone()); } From 38d09ee8eaea2701a4db7dc38ca6f11640d2efd7 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Fri, 22 May 2026 01:07:56 -0700 Subject: [PATCH 13/25] Avoid duplicate Image prefix in pull status. --- cmd/soroban-cli/src/commands/contract/build/verifiable.rs | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 66022ac826..f9240fdd87 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -537,7 +537,10 @@ async fn pull_image( } else if let Some(digest) = status.strip_prefix("Digest: ") { print.infoln(format!("Image digest: {digest}")); } else if let Some(rest) = status.strip_prefix("Status: ") { - print.infoln(format!("Image: {rest}")); + // Docker's status text already starts with "Image …" or + // "Downloaded …", so we forward it verbatim instead of + // prepending another "Image:". + print.infoln(rest); } } } From ae73bcc6508fcbc903a3de718dbb6053363dbc3a Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Fri, 22 May 2026 01:23:18 -0700 Subject: [PATCH 14/25] Factor enforce_hardened_tree out of fix_config_permissions. --- cmd/soroban-cli/src/config/locator.rs | 84 ++++++++++++++++----------- 1 file changed, 49 insertions(+), 35 deletions(-) diff --git a/cmd/soroban-cli/src/config/locator.rs b/cmd/soroban-cli/src/config/locator.rs index e557d30740..d3e6c41f4f 100644 --- a/cmd/soroban-cli/src/config/locator.rs +++ b/cmd/soroban-cli/src/config/locator.rs @@ -623,52 +623,66 @@ impl Pwd for Args { } } -#[cfg(unix)] -fn fix_config_permissions(root: std::path::PathBuf) { - use std::os::unix::fs::PermissionsExt; - - let mut bad_dirs = Vec::new(); - let mut bad_files = Vec::new(); - let mut stack = vec![root]; - - while let Some(dir) = stack.pop() { - if let Ok(meta) = std::fs::metadata(&dir) { - if meta.permissions().mode() & 0o777 != 0o700 { - bad_dirs.push(dir.clone()); +/// Walk `root` recursively. For every regular entry whose permissions don't +/// already match the hardened mode (0o700 for dirs, 0o600 for files), set +/// them. Returns the dirs and files that were changed so callers can decide +/// whether to surface a warning. Symlinks are skipped — mode bits aren't +/// meaningful for them and `set_permissions` would follow them. +/// +/// On non-unix platforms this is a no-op; tempdirs / config dirs there rely +/// on filesystem ACLs created by the higher-level APIs. +#[allow(clippy::unnecessary_wraps)] +pub(crate) fn enforce_hardened_tree(root: &Path) -> io::Result<(Vec, Vec)> { + #[cfg(unix)] + { + use std::os::unix::fs::PermissionsExt; + let mut changed_dirs = Vec::new(); + let mut changed_files = Vec::new(); + let mut stack = vec![root.to_path_buf()]; + while let Some(p) = stack.pop() { + let Ok(meta) = std::fs::symlink_metadata(&p) else { + continue; + }; + if meta.file_type().is_symlink() { + continue; } - } - - if let Ok(entries) = std::fs::read_dir(&dir) { - for entry in entries.filter_map(Result::ok) { - let path = entry.path(); - - if path.is_dir() { - stack.push(path); - } else if let Ok(meta) = std::fs::metadata(&path) { - if meta.permissions().mode() & 0o777 != 0o600 { - bad_files.push(path); + let current = meta.permissions().mode() & 0o777; + if meta.is_dir() { + if current != 0o700 { + set_hardened_permissions(&p)?; + changed_dirs.push(p.clone()); + } + if let Ok(entries) = std::fs::read_dir(&p) { + for entry in entries.filter_map(Result::ok) { + stack.push(entry.path()); } } + } else if current != 0o600 { + set_hardened_permissions(&p)?; + changed_files.push(p); } } + Ok((changed_dirs, changed_files)) } + #[cfg(not(unix))] + { + let _ = root; + Ok((Vec::new(), Vec::new())) + } +} - let print = Print::new(false); +#[cfg(unix)] +fn fix_config_permissions(root: std::path::PathBuf) { + let Ok((dirs, files)) = enforce_hardened_tree(&root) else { + return; + }; - if !bad_dirs.is_empty() { + let print = Print::new(false); + if !dirs.is_empty() { print.warnln("Updated config directories permissions to 0700."); - - for dir in bad_dirs { - let _ = set_hardened_permissions(&dir); - } } - - if !bad_files.is_empty() { + if !files.is_empty() { print.warnln("Updated config files permissions to 0600."); - - for file in bad_files { - let _ = set_hardened_permissions(&file); - } } } From 7014661551e4f20c51c605660fa1b5c97818f679 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Tue, 16 Jun 2026 19:53:41 -0700 Subject: [PATCH 15/25] Generate source archive for verifiable builds. --- Cargo.lock | 18 +- FULL_HELP_DOCS.md | 9 +- cmd/crates/soroban-test/tests/it/build.rs | 114 +-- cmd/soroban-cli/Cargo.toml | 3 +- .../src/commands/contract/build.rs | 69 +- .../src/commands/contract/build/verifiable.rs | 840 +++++++++++++----- cmd/soroban-cli/src/config/data.rs | 18 + 7 files changed, 725 insertions(+), 346 deletions(-) diff --git a/Cargo.lock b/Cargo.lock index 5c27f3f806..260212a41a 100644 --- a/Cargo.lock +++ b/Cargo.lock @@ -2791,7 +2791,7 @@ dependencies = [ "libc", "percent-encoding", "pin-project-lite", - "socket2 0.5.10", + "socket2 0.6.4", "system-configuration", "tokio", "tower-service", @@ -4305,7 +4305,7 @@ dependencies = [ "quinn-udp", "rustc-hash 2.1.1", "rustls", - "socket2 0.5.10", + "socket2 0.6.4", "thiserror 2.0.18", "tokio", "tracing", @@ -4342,7 +4342,7 @@ dependencies = [ "cfg_aliases", "libc", "once_cell", - "socket2 0.5.10", + "socket2 0.6.4", "tracing", "windows-sys 0.60.2", ] @@ -5450,6 +5450,7 @@ dependencies = [ "strsim", "strum 0.17.1", "strum_macros 0.17.1", + "tar", "tempfile", "termcolor", "termcolor_output", @@ -6088,6 +6089,17 @@ version = "1.0.1" source = "registry+https://github.com/rust-lang/crates.io-index" checksum = "55937e1799185b12863d447f42597ed69d9928686b8d88a1df17376a097d8369" +[[package]] +name = "tar" +version = "0.4.46" +source = "registry+https://github.com/rust-lang/crates.io-index" +checksum = "3f6221d9a6003c78398e3b239969f352578258df48c8eb051caadae0015bc840" +dependencies = [ + "filetime", + "libc", + "xattr", +] + [[package]] name = "temp-dir" version = "0.1.16" diff --git a/FULL_HELP_DOCS.md b/FULL_HELP_DOCS.md index 58f5340421..e6af930448 100644 --- a/FULL_HELP_DOCS.md +++ b/FULL_HELP_DOCS.md @@ -396,12 +396,11 @@ To view the commands that will be executed, without executing them, use the --pr ###### **Verifiable:** -- `--verifiable` — Build inside a trusted Docker container and record SEP-58 metadata (`bldimg`, `source_rev`, `bldopt`) so the resulting WASM can be reproduced and verified by third parties. Implies `--locked`. Requires a clean git working tree +- `--verifiable` — Build inside a trusted Docker container and record SEP-58 metadata (`bldimg`, `source_uri`, `source_sha256`, `bldopt`) so the resulting WASM can be reproduced and verified by third parties. Implies `--locked`. Requires a clean git working tree - `--image ` — Override the auto-selected container image used by `--verifiable`. Must be digest-pinned, e.g. `docker.io/stellar/stellar-cli@sha256:...`. Tag-only refs are rejected because SEP-58 requires content addressing -- `--source-repo ` — SEP-58 source identification: HTTPS URL (or `github:user/repo`) of the source repository. Must be passed together with `--source-rev` -- `--source-rev ` — SEP-58 source identification: 40-char SHA-1 of the source commit. The local workspace must be a git repo at this exact SHA with a clean working tree. Must be passed together with `--source-repo` -- `--tarball-url ` — SEP-58 source identification: URL where the source tarball can be downloaded -- `--tarball-sha256 ` — SEP-58 source identification: SHA-256 of the source tarball bytes +- `--source-sha256 ` — SEP-58 source identification: SHA-256 of the source archive/tree (recorded as the `source_sha256` meta entry). Required with `--verifiable` unless `--archive` is used, which generates the archive and computes this for you +- `--source-uri ` — SEP-58 source identification: URI where the source can be obtained, e.g. `https://example.com/src.tar.gz` (recorded as the `source_uri` meta entry). Optional; when set it must accompany `--source-sha256` +- `--archive ` — Generate a source archive for the verifiable build, then build from it and record its SHA-256 as the SEP-58 `source_sha256` meta entry. Pass a path to choose where the gzipped tarball is written; with no path it goes to the data dir's `archives/`. In a git repo the archive is `git archive HEAD`; otherwise the working directory is archived minus a built-in denylist (.git, .svn, .hg, target/, node_modules/, .DS_Store) - `-d`, `--docker-host ` — Override the default docker host used by `--verifiable` ## `stellar contract extend` diff --git a/cmd/crates/soroban-test/tests/it/build.rs b/cmd/crates/soroban-test/tests/it/build.rs index bb41346a3c..9733d37fb6 100644 --- a/cmd/crates/soroban-test/tests/it/build.rs +++ b/cmd/crates/soroban-test/tests/it/build.rs @@ -1020,15 +1020,6 @@ fn fresh_workspace() -> (TempDir, PathBuf) { (temp, workspace) } -fn git_head(dir: &Path) -> String { - let out = std::process::Command::new("git") - .args(["rev-parse", "HEAD"]) - .current_dir(dir) - .output() - .unwrap(); - String::from_utf8(out.stdout).unwrap().trim().to_string() -} - // `--verifiable` cannot accept reserved `--meta` keys that the cli writes itself. #[test] fn verifiable_meta_conflict_errors() { @@ -1043,8 +1034,8 @@ fn verifiable_meta_conflict_errors() { .arg("--verifiable") .arg("--image") .arg(ZERO_DIGEST) - .arg("--tarball-url") - .arg("https://example.com/foo.tar.gz") + .arg("--source-sha256") + .arg("a".repeat(64)) .arg("--meta") .arg("bldimg=not-allowed") .assert() @@ -1066,8 +1057,8 @@ fn verifiable_image_must_be_digest_pinned() { .arg("--verifiable") .arg("--image") .arg("docker.io/stellar/stellar-cli:latest") - .arg("--tarball-url") - .arg("https://example.com/foo.tar.gz") + .arg("--source-sha256") + .arg("a".repeat(64)) .assert() .failure() .stderr(predicate::str::contains("bldimg format")); @@ -1090,35 +1081,66 @@ fn verifiable_image_requires_explicit_registry_host() { .arg("--verifiable") .arg("--image") .arg(short_ref) - .arg("--tarball-url") - .arg("https://example.com/foo.tar.gz") + .arg("--source-sha256") + .arg("a".repeat(64)) .assert() .failure() .stderr(predicate::str::contains("bldimg format")); } -// `--verifiable` without any source-identification flag must error. +// `--verifiable` with neither `--source-sha256` nor `--archive` must error. +// Run in a fresh (non-git) workspace so the clean-tree check is skipped and the +// missing-source error is what surfaces. #[test] -fn verifiable_requires_source_id() { +fn verifiable_requires_source_sha256() { let sandbox = TestEnv::default(); - let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); - let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); + let (_temp, workspace) = fresh_workspace(); sandbox .new_assert_cmd("contract") - .current_dir(fixture_path) + .current_dir(workspace.join("contracts").join("add")) .arg("build") .arg("--verifiable") .arg("--image") .arg(ZERO_DIGEST) .assert() .failure() - .stderr(predicate::str::contains("source-identification")); + .stderr(predicate::str::contains("--source-sha256")); +} + +// `--archive` generates the source archive (and computes source_sha256) before +// the docker stage, so the file is written even though the build then fails to +// reach a real image. +#[test] +fn verifiable_archive_writes_source_archive() { + let sandbox = TestEnv::default(); + let (temp, workspace) = fresh_workspace(); + git_in(&workspace, &["init", "-q", "-b", "main"]); + git_in(&workspace, &["add", "-A"]); + git_in(&workspace, &["commit", "-q", "-m", "init"]); + + let archive_path = temp.path().join("src.tar.gz"); + + sandbox + .new_assert_cmd("contract") + .current_dir(workspace.join("contracts").join("add")) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg(ZERO_DIGEST) + .arg(format!("--archive={}", archive_path.display())) + .assert() + .failure(); + + assert!( + archive_path.exists(), + "source archive should be written before the docker stage" + ); } -// `--source-rev` value must match the 40-hex regex. +// `--source-sha256` value must match the 64-hex regex. #[test] -fn verifiable_source_rev_format_errors() { +fn verifiable_source_sha256_format_errors() { let sandbox = TestEnv::default(); let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); @@ -1130,54 +1152,46 @@ fn verifiable_source_rev_format_errors() { .arg("--verifiable") .arg("--image") .arg(ZERO_DIGEST) - .arg("--source-repo") - .arg("https://github.com/foo/bar") - .arg("--source-rev") + .arg("--source-sha256") .arg("not-a-sha") .assert() .failure() - .stderr(predicate::str::contains("source_rev format")); + .stderr(predicate::str::contains("source_sha256 format")); } -// `--source-rev` is cross-checked against local git HEAD; a mismatch is a hard -// fail before docker is touched. +// `--source-uri` value must be a URI with a scheme. #[test] -fn verifiable_source_rev_must_match_head() { +fn verifiable_source_uri_format_errors() { let sandbox = TestEnv::default(); - let (_temp, workspace) = fresh_workspace(); - git_in(&workspace, &["init", "-q", "-b", "main"]); - git_in(&workspace, &["add", "-A"]); - git_in(&workspace, &["commit", "-q", "-m", "init"]); - - let bogus = "a".repeat(40); + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); sandbox .new_assert_cmd("contract") - .current_dir(workspace.join("contracts").join("add")) + .current_dir(fixture_path) .arg("build") .arg("--verifiable") .arg("--image") .arg(ZERO_DIGEST) - .arg("--source-repo") - .arg("https://github.com/foo/bar") - .arg("--source-rev") - .arg(bogus) + .arg("--source-sha256") + .arg("a".repeat(64)) + .arg("--source-uri") + .arg("not a uri") .assert() .failure() - .stderr(predicate::str::contains("does not match local HEAD")); + .stderr(predicate::str::contains("source_uri format")); } -// A dirty git tree under `--source-rev` is a hard fail (the recorded rev would -// not describe the bytes built). +// A dirty git tree is a hard fail under `--verifiable` (the recorded +// source_sha256 would not describe the bytes built). #[test] -fn verifiable_dirty_tree_errors_with_source_rev() { +fn verifiable_dirty_tree_errors() { let sandbox = TestEnv::default(); let (_temp, workspace) = fresh_workspace(); git_in(&workspace, &["init", "-q", "-b", "main"]); git_in(&workspace, &["add", "-A"]); git_in(&workspace, &["commit", "-q", "-m", "init"]); - let head = git_head(&workspace); - // Dirty the tree after committing so HEAD matches but status is non-empty. + // Dirty the tree after committing so status is non-empty. std::fs::write(workspace.join("dirty.txt"), b"uncommitted").unwrap(); sandbox @@ -1187,10 +1201,8 @@ fn verifiable_dirty_tree_errors_with_source_rev() { .arg("--verifiable") .arg("--image") .arg(ZERO_DIGEST) - .arg("--source-repo") - .arg("https://github.com/foo/bar") - .arg("--source-rev") - .arg(head) + .arg("--source-sha256") + .arg("a".repeat(64)) .assert() .failure() .stderr(predicate::str::contains("dirty").or(predicate::str::contains("clean tree"))); diff --git a/cmd/soroban-cli/Cargo.toml b/cmd/soroban-cli/Cargo.toml index 76d6fc0323..a1d29e654f 100644 --- a/cmd/soroban-cli/Cargo.toml +++ b/cmd/soroban-cli/Cargo.toml @@ -128,6 +128,8 @@ keyring = { version = "3", features = ["apple-native", "windows-native", "sync-s whoami = "1.5.2" serde_with = "3.11.0" rustc_version = "0.4.1" +tar = "0.4.40" +walkdir = "2.5.0" [build-dependencies] crate-git-revision = "0.0.9" @@ -139,6 +141,5 @@ thiserror.workspace = true assert_cmd = "2.0.4" assert_fs = "1.0.7" predicates = { workspace = true } -walkdir = "2.5.0" mockito = "1.5.0" serial_test = "3.0.0" diff --git a/cmd/soroban-cli/src/commands/contract/build.rs b/cmd/soroban-cli/src/commands/contract/build.rs index 51829fa56f..9ad08ca7fd 100644 --- a/cmd/soroban-cli/src/commands/contract/build.rs +++ b/cmd/soroban-cli/src/commands/contract/build.rs @@ -99,9 +99,9 @@ pub struct Cmd { pub print_commands_only: bool, /// Build inside a trusted Docker container and record SEP-58 metadata - /// (`bldimg`, `source_rev`, `bldopt`) so the resulting WASM can be - /// reproduced and verified by third parties. Implies `--locked`. - /// Requires a clean git working tree. + /// (`bldimg`, `source_uri`, `source_sha256`, `bldopt`) so the resulting + /// WASM can be reproduced and verified by third parties. Implies + /// `--locked`. Requires a clean git working tree. #[arg(long, help_heading = "Verifiable")] pub verifiable: bool, @@ -111,47 +111,38 @@ pub struct Cmd { #[arg(long, requires = "verifiable", help_heading = "Verifiable")] pub image: Option, - /// SEP-58 source identification: HTTPS URL (or `github:user/repo`) of the - /// source repository. Must be passed together with `--source-rev`. - #[arg( - long, - requires = "verifiable", - requires = "source_rev", - conflicts_with_all = ["tarball_url", "tarball_sha256"], - help_heading = "Verifiable" - )] - pub source_repo: Option, - - /// SEP-58 source identification: 40-char SHA-1 of the source commit. The - /// local workspace must be a git repo at this exact SHA with a clean - /// working tree. Must be passed together with `--source-repo`. - #[arg( - long, - requires = "verifiable", - requires = "source_repo", - conflicts_with_all = ["tarball_url", "tarball_sha256"], - help_heading = "Verifiable" - )] - pub source_rev: Option, + /// SEP-58 source identification: SHA-256 of the source archive/tree + /// (recorded as the `source_sha256` meta entry). Required with + /// `--verifiable` unless `--archive` is used, which generates the archive + /// and computes this for you. + #[arg(long, requires = "verifiable", help_heading = "Verifiable")] + pub source_sha256: Option, - /// SEP-58 source identification: URL where the source tarball can be - /// downloaded. + /// SEP-58 source identification: URI where the source can be obtained, e.g. + /// `https://example.com/src.tar.gz` (recorded as the `source_uri` meta + /// entry). Optional; when set it must accompany `--source-sha256`. #[arg( long, requires = "verifiable", - conflicts_with_all = ["source_repo", "source_rev"], + requires = "source_sha256", help_heading = "Verifiable" )] - pub tarball_url: Option, - - /// SEP-58 source identification: SHA-256 of the source tarball bytes. + pub source_uri: Option, + + /// Generate a source archive for the verifiable build, then build from it + /// and record its SHA-256 as the SEP-58 `source_sha256` meta entry. Pass a + /// path to choose where the gzipped tarball is written; with no path it + /// goes to the data dir's `archives/`. In a git repo the archive is + /// `git archive HEAD`; otherwise the working directory is archived minus a + /// built-in denylist (.git, .svn, .hg, target/, node_modules/, .DS_Store). #[arg( long, + num_args = 0..=1, + require_equals = true, requires = "verifiable", - conflicts_with_all = ["source_repo", "source_rev"], help_heading = "Verifiable" )] - pub tarball_sha256: Option, + pub archive: Option>, /// Override the default docker host used by `--verifiable`. #[arg(short = 'd', long, env = "DOCKER_HOST", help_heading = "Verifiable")] @@ -288,10 +279,9 @@ impl Default for Cmd { print_commands_only: false, verifiable: false, image: None, - source_repo: None, - source_rev: None, - tarball_url: None, - tarball_sha256: None, + source_sha256: None, + source_uri: None, + archive: None, docker_host: None, build_args: BuildArgs::default(), } @@ -622,10 +612,7 @@ impl Cmd { &wasm_bytes }; - print.blankln(format!( - "Wasm File: {path} ({size_description})", - path = rel_path.display() - )); + print.blankln(format!("Wasm File: {path}", path = rel_path.display())); print.blankln(format!("Wasm Hash: {}", hex::encode(Sha256::digest(bytes)))); print.blankln(format!("Wasm Size: {size_description}")); diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index f9240fdd87..618ee15a97 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -1,4 +1,5 @@ use std::{ + io::Write, path::{Path, PathBuf}, process::Command, }; @@ -17,9 +18,12 @@ use futures_util::{StreamExt, TryStreamExt}; use regex::Regex; use semver::Version; use serde::Deserialize; +use sha2::{Digest, Sha256}; +use walkdir::WalkDir; use crate::{ commands::{container::shared::Error as ConnectionError, global}, + config::{data, locator::enforce_hardened_tree}, print::Print, }; @@ -28,7 +32,37 @@ use super::{BuiltContract, Cmd, WASM_TARGET}; const REGISTRY: &str = "docker.io/stellar/stellar-cli"; const HUB_TAGS_URL: &str = "https://hub.docker.com/v2/repositories/stellar/stellar-cli/tags/?page_size=100"; -const RESERVED_META_KEYS: &[&str] = &["bldimg", "source_rev", "bldopt"]; +const RESERVED_META_KEYS: &[&str] = &["bldimg", "source_uri", "source_sha256", "bldopt"]; + +/// Top-level names excluded when archiving a non-git working directory (we have +/// no tracked-files list to consult, so fall back to a fixed denylist of VCS +/// metadata, build/cache/transient dirs, and editor/OS/AI-assistant junk). +/// Matched against each path component, so a directory like `target/` prunes +/// its whole subtree. +const ARCHIVE_DENYLIST: &[&str] = &[ + // version control + ".git", + ".svn", + ".hg", + // build output / dependencies + "target", + "node_modules", + // transient + "log", + "logs", + "tmp", + "temp", + // OS / editor junk + ".DS_Store", + "Thumbs.db", + ".idea", + ".vscode", + // AI assistant dirs + ".claude", + ".cursor", + ".windsurf", + ".aider", +]; /// First cli release that accepts `--optimize=false` as an explicit value /// (added by commit `b17d3f0b`). Containers older than this only accept bare @@ -74,35 +108,41 @@ pub enum Error { }, #[error( - "git working tree at {path} is dirty. --source-rev requires a clean tree so the recorded source_rev matches the WASM bytes. Commit or stash your changes and try again." + "git working tree at {path} is dirty. --verifiable requires a clean tree so the recorded source_sha256 matches the WASM bytes. Commit or stash your changes and try again." )] GitDirty { path: PathBuf }, #[error( - "the cli sets bldimg, source_rev, and bldopt automatically when --verifiable is used; remove them from --meta. Got reserved key: {key}" + "the cli sets bldimg, source_uri, source_sha256, and bldopt automatically when --verifiable is used; remove them from --meta. Got reserved key: {key}" )] ReservedMetaKey { key: String }, - #[error("--verifiable requires a SEP-58 source-identification combination. Pass one of: (--source-repo + --source-rev), (--tarball-url and/or --tarball-sha256).")] - MissingSourceId, + #[error("--verifiable requires --source-sha256 (the SEP-58 source_sha256: 64-char hex SHA-256 of the source), or --archive to generate the source archive and compute it. --source-uri is optional.")] + MissingSourceSha256, - #[error("--source-rev value {value:?} does not match the SEP-58 source_rev format `^[0-9a-f]{{40}}$` (full 40-char SHA-1 of the source commit).")] - SourceRevFormat { value: String }, + #[error("--source-sha256 value {value:?} does not match the SEP-58 source_sha256 format `^[0-9a-f]{{64}}$` (64-char lower-case hex).")] + SourceSha256Format { value: String }, - #[error("--source-repo value {value:?} does not match the SEP-58 source_repo format `^(https?://\\S+|github:[^/\\s]+/[^/\\s]+)$`.")] - SourceRepoFormat { value: String }, + #[error("--source-uri value {value:?} does not match the SEP-58 source_uri format `^[a-zA-Z][a-zA-Z0-9+.-]*:\\S+$` (a URI with a scheme, e.g. https://example.com/src.tar.gz).")] + SourceUriFormat { value: String }, - #[error("--tarball-url value {value:?} does not match the SEP-58 tarball_url format `^https?://\\S+$`.")] - TarballUrlFormat { value: String }, + #[error("--source-sha256 {provided} does not match the SHA-256 of the generated archive {computed}. Omit --source-sha256 to record the computed value, or fix the value.")] + SourceSha256Mismatch { provided: String, computed: String }, - #[error("--tarball-sha256 value {value:?} does not match the SEP-58 tarball_sha256 format `^[0-9a-f]{{64}}$`.")] - TarballSha256Format { value: String }, + #[error("`git archive` failed in {path}: {stderr}")] + GitArchive { path: PathBuf, stderr: String }, - #[error("--source-rev requires a git workspace at {path}; `git rev-parse HEAD` failed there.")] - SourceRevNotGitRepo { path: PathBuf }, + #[error("could not write source archive to {path}: {source}")] + ArchiveWrite { + path: PathBuf, + source: std::io::Error, + }, - #[error("--source-rev {claimed} does not match local HEAD {head}. Commit, switch, or pass the correct rev.")] - SourceRevHeadMismatch { claimed: String, head: String }, + #[error("could not extract source archive: {0}")] + ArchiveExtract(std::io::Error), + + #[error(transparent)] + Data(#[from] data::Error), #[error("container build exited with status {status}. To reproduce manually:\n docker run --rm -v {mount}:/source {image} {args}")] ContainerExit { @@ -130,22 +170,58 @@ pub async fn run( } } - // Stage 2: local filesystem + git, no network. Resolve the workspace root - // first so the (optional) `--source-rev` git cross-check has a path to - // anchor on. + // Stage 2: local filesystem + git, no network. let workspace_root = resolve_workspace_root(cmd)?; - let source_ids = validate_source_ids(cmd, &workspace_root)?; - - // Pick the anchor the container bind-mounts and the `--manifest-path` - // bldopt is relativized against. A verifier will clone source_repo (or - // extract the tarball) into a fresh tempdir and bind-mount its root, so - // the build must do the symmetric thing on the host: bind-mount the local - // clone root (where `.git` lives) or, if there's no clone, the user's - // cwd. We do NOT validate that the local clone matches `--source-repo` — - // a wrong clone produces different bytes, and verify catches that at - // byte-comparison time. + validate_source_formats(cmd)?; + + // Pick the anchor for the local source: the `--manifest-path` bldopt is + // relativized against it, and (when `--archive` is not used) it's also what + // gets bind-mounted into the container. We do NOT validate that it matches + // source_uri — a wrong source produces different bytes, and verify catches + // that at byte-comparison time. let source_root = resolve_source_root(cmd); + // A dirty working tree would make the recorded source_sha256 fail to + // describe the bytes actually built, so refuse to proceed. Skipped when + // the source root isn't a git repo (we can't check, e.g. archive sources). + enforce_clean_tree(&source_root)?; + + // Resolve the recorded source_sha256 and the directory the container mounts + // at /source. With `--archive`, the CLI builds the source archive, records + // its hash, and builds from the *extracted* archive (in a hardened tempdir) + // so the WASM is produced from exactly the bytes that were hashed. Without + // it, the user supplies --source-sha256 and we mount the working tree. + let resolved = match &cmd.archive { + Some(_) => { + let a = resolve_archive(cmd, &source_root, print)?; + // The extracted `source/` dir mirrors `source_root` exactly and is + // both the container mount and the tree the build writes `target/` + // into, so it's what `collect_built_contracts` resolves artifacts + // against. + let mount_root = a.extracted_root.join("source"); + ResolvedSource { + source_sha256: a.source_sha256, + extracted_root: Some(mount_root.clone()), + mount_root, + _tmp: Some(a.tmp), + } + } + None => ResolvedSource { + source_sha256: cmd + .source_sha256 + .clone() + .ok_or(Error::MissingSourceSha256)?, + mount_root: source_root.clone(), + extracted_root: None, + _tmp: None, + }, + }; + + let source_ids = SourceIds { + source_uri: cmd.source_uri.clone(), + source_sha256: Some(resolved.source_sha256.clone()), + }; + // Defer the info banner until every validation has passed, so it doesn't // appear right before an error. if !cmd.locked { @@ -171,8 +247,20 @@ pub async fn run( probe_supports_optimize_false_syntax(&image_ref, &docker, print).await }; - let (forwarded_args, bldopts) = - build_forwarded_args(cmd, &source_root, supports_explicit_optimize_false); + let package = resolve_build_package(cmd)?; + if cmd.package.is_none() { + if let Some(pkg) = &package { + print.infoln(format!( + "Inferred --package={pkg} and using it as a build option." + )); + } + } + let (forwarded_args, bldopts) = build_forwarded_args( + cmd, + &source_root, + package.as_deref(), + supports_explicit_optimize_false, + ); let metadata_args = build_metadata_args(&image_ref, &source_ids, &bldopts); let container_cmd_args = compose_container_args(&forwarded_args, &metadata_args); @@ -182,7 +270,7 @@ pub async fn run( // `--verbose` because verifications are run as part of pipelines. run_in_container( &image_ref, - &source_root, + &resolved.mount_root, &container_cmd_args, &docker, print, @@ -191,7 +279,18 @@ pub async fn run( .await?; let _ = global_args; - collect_built_contracts(cmd, &workspace_root, print) + let _ = workspace_root; + collect_built_contracts(cmd, &source_root, resolved.extracted_root.as_deref(), print) +} + +/// The recorded `source_sha256`, the directory bind-mounted at `/source`, and +/// (when `--archive` is used) the extracted-archive root plus its tempdir guard +/// — held so the temp dir outlives the container build and artifact collection. +struct ResolvedSource { + source_sha256: String, + mount_root: PathBuf, + extracted_root: Option, + _tmp: Option, } fn resolve_workspace_root(cmd: &Cmd) -> Result { @@ -209,7 +308,7 @@ fn resolve_workspace_root(cmd: &Cmd) -> Result { /// `--manifest-path` (or cwd, if no manifest_path) looking for a `.git` /// directory; return its parent. If none is found, fall back to cwd. /// -/// This isn't a validation step — any `.git` will do. Wrong-clone mistakes +/// This isn't a validation step — any `.git` will do. Wrong-source mistakes /// are caught later by the verify-side byte comparison. fn resolve_source_root(cmd: &Cmd) -> PathBuf { let start = if let Some(p) = &cmd.manifest_path { @@ -232,105 +331,258 @@ fn resolve_source_root(cmd: &Cmd) -> PathBuf { std::env::current_dir().unwrap_or(start) } -/// Source-identification fields, gathered from the corresponding CLI flags -/// after validation. Each is `Some` only when the user passed the flag and the -/// value matched the SEP-58 format regex. The four fields cannot all be -/// `None` — `validate_source_ids` rejects that case. +/// Source-identification fields recorded as SEP-58 meta. `source_sha256` is +/// always `Some` by the time these are built in `run()` (resolved from +/// `--source-sha256` or computed from the generated archive). `source_uri` is +/// `Some` only when the user passed `--source-uri`. #[derive(Debug, Default, Clone)] struct SourceIds { - source_repo: Option, - source_rev: Option, - tarball_url: Option, - tarball_sha256: Option, + source_uri: Option, + source_sha256: Option, } -fn validate_source_ids(cmd: &Cmd, workspace_root: &Path) -> Result { - let ids = SourceIds { - source_repo: cmd.source_repo.clone(), - source_rev: cmd.source_rev.clone(), - tarball_url: cmd.tarball_url.clone(), - tarball_sha256: cmd.tarball_sha256.clone(), - }; - - if ids.source_repo.is_none() - && ids.source_rev.is_none() - && ids.tarball_url.is_none() - && ids.tarball_sha256.is_none() - { - return Err(Error::MissingSourceId); - } - - if let Some(v) = &ids.source_rev { - if !source_rev_regex().is_match(v) { - return Err(Error::SourceRevFormat { value: v.clone() }); +/// Format-validate the user-supplied source flags. Requiredness is enforced in +/// `run()` (it depends on whether `--archive` is used), not here. +fn validate_source_formats(cmd: &Cmd) -> Result<(), Error> { + if let Some(sha) = &cmd.source_sha256 { + if !source_sha256_regex().is_match(sha) { + return Err(Error::SourceSha256Format { value: sha.clone() }); } } - - if let Some(v) = &ids.source_repo { - if !source_repo_regex().is_match(v) { - return Err(Error::SourceRepoFormat { value: v.clone() }); + if let Some(uri) = &cmd.source_uri { + if !source_uri_regex().is_match(uri) { + return Err(Error::SourceUriFormat { value: uri.clone() }); } } + Ok(()) +} - if let Some(v) = &ids.tarball_url { - if !tarball_url_regex().is_match(v) { - return Err(Error::TarballUrlFormat { value: v.clone() }); - } - } +/// Outcome of `--archive`: the generated archive's SHA-256 and the directory it +/// was extracted into (held alive by `tmp`). +struct ArchiveResult { + source_sha256: String, + extracted_root: PathBuf, + tmp: tempfile::TempDir, +} - if let Some(v) = &ids.tarball_sha256 { - if !tarball_sha256_regex().is_match(v) { - return Err(Error::TarballSha256Format { value: v.clone() }); +/// Build the source archive, record its hash, write it out, and extract it into +/// a permission-hardened tempdir that the container then builds from. +fn resolve_archive(cmd: &Cmd, source_root: &Path, print: &Print) -> Result { + let bytes = build_source_archive(source_root, print)?; + let computed = hex::encode(Sha256::digest(&bytes)); + + // If the user pinned a hash, it must match what we produced. + if let Some(provided) = &cmd.source_sha256 { + if provided != &computed { + return Err(Error::SourceSha256Mismatch { + provided: provided.clone(), + computed, + }); } } - if let Some(claimed) = &ids.source_rev { - cross_check_source_rev_against_git(workspace_root, claimed)?; + // `Some(Some(path))` → write there; `Some(None)` → content-addressed name + // under the managed archives dir. + let out_path = match &cmd.archive { + Some(Some(p)) => p.clone(), + Some(None) => data::archives_dir()?.join(format!("{computed}.tar.gz")), + None => unreachable!("resolve_archive is only called when --archive is set"), + }; + if let Some(parent) = out_path.parent() { + std::fs::create_dir_all(parent).map_err(|source| Error::ArchiveWrite { + path: out_path.clone(), + source, + })?; } + std::fs::write(&out_path, &bytes).map_err(|source| Error::ArchiveWrite { + path: out_path.clone(), + source, + })?; + print.infoln(format!( + "Wrote source archive {} (source_sha256 {computed})", + out_path.display() + )); - Ok(ids) + // Extract and harden, then build from the extracted copy so the WASM is + // produced from exactly the archived bytes. + // + // Extract under the data dir, NOT the OS temp dir: on macOS `$TMPDIR` lives + // under /var/folders, which container VMs (Docker Desktop, Colima, …) don't + // share by default, so a bind mount of it would be empty inside the + // container. The data dir lives under the user's home, which is shared. + let base = data::data_local_dir()?; + std::fs::create_dir_all(&base).map_err(|source| Error::ArchiveWrite { + path: base.clone(), + source, + })?; + let tmp = tempfile::Builder::new() + .prefix("verifiable-src-") + .tempdir_in(&base) + .map_err(Error::ArchiveExtract)?; + unpack_targz(&bytes, tmp.path())?; + enforce_hardened_tree(tmp.path()).map_err(Error::ArchiveExtract)?; + + let extracted_root = tmp.path().to_path_buf(); + Ok(ArchiveResult { + source_sha256: computed, + extracted_root, + tmp, + }) } -fn cross_check_source_rev_against_git(workspace_root: &Path, claimed: &str) -> Result<(), Error> { - let rev_out = Command::new("git") +/// Whether `source_root` is inside a git work tree. +fn is_git_repo(source_root: &Path) -> bool { + Command::new("git") .arg("-C") - .arg(workspace_root) + .arg(source_root) .arg("rev-parse") + .arg("--is-inside-work-tree") + .output() + .is_ok_and(|o| o.status.success()) +} + +/// Produce the gzipped source tarball bytes. Entries are rooted under a +/// top-level `source/` prefix (so the archive extracts to a `source/` dir, +/// mirroring the container's `/source` mount). In a git repo this is `git +/// archive HEAD` (the committed tree); otherwise the working directory is +/// walked and tarred, skipping `ARCHIVE_DENYLIST` entries, after warning. +fn build_source_archive(source_root: &Path, print: &Print) -> Result, Error> { + let tar = if is_git_repo(source_root) { + git_archive_tar(source_root)? + } else { + print.warnln(format!( + "{} is not a git repository; archiving the working directory. Inspect the generated archive to confirm its contents.", + source_root.display(), + )); + walk_tar(source_root)? + }; + gzip(&tar) +} + +/// `git archive --format=tar --prefix=source/ HEAD`, returning the tar bytes. +fn git_archive_tar(source_root: &Path) -> Result, Error> { + let out = Command::new("git") + .arg("-C") + .arg(source_root) + .arg("archive") + .arg("--format=tar") + .arg("--prefix=source/") .arg("HEAD") .output() - .map_err(|e| Error::GitInvoke { - path: workspace_root.to_path_buf(), - source: e, + .map_err(|source| Error::GitInvoke { + path: source_root.to_path_buf(), + source, })?; - - if !rev_out.status.success() { - return Err(Error::SourceRevNotGitRepo { - path: workspace_root.to_path_buf(), + if !out.status.success() { + return Err(Error::GitArchive { + path: source_root.to_path_buf(), + stderr: String::from_utf8_lossy(&out.stderr).trim().to_string(), }); } + Ok(out.stdout) +} - let head = String::from_utf8_lossy(&rev_out.stdout).trim().to_string(); - if head != claimed { - return Err(Error::SourceRevHeadMismatch { - claimed: claimed.to_string(), - head, - }); +/// Tar the working tree under `source_root`, skipping denylisted path +/// components, with entries sorted and headers normalized (deterministic mode) +/// so the bytes are reproducible. Each entry is prefixed with `source/`. +fn walk_tar(source_root: &Path) -> Result, Error> { + let mut files: Vec = Vec::new(); + let walk = WalkDir::new(source_root) + .sort_by_file_name() + .into_iter() + .filter_entry(|e| !is_denylisted(e.file_name())); + for entry in walk { + let entry = entry.map_err(|e| Error::ArchiveWrite { + path: source_root.to_path_buf(), + source: e.into(), + })?; + if entry.file_type().is_file() { + files.push(entry.path().to_path_buf()); + } } + files.sort(); + let mut builder = tar::Builder::new(Vec::new()); + builder.mode(tar::HeaderMode::Deterministic); + for path in &files { + let rel = path.strip_prefix(source_root).unwrap_or(path); + let name = Path::new("source").join(rel); + let mut f = std::fs::File::open(path).map_err(|source| Error::ArchiveWrite { + path: path.clone(), + source, + })?; + builder + .append_file(&name, &mut f) + .map_err(|source| Error::ArchiveWrite { + path: path.clone(), + source, + })?; + } + builder.into_inner().map_err(|source| Error::ArchiveWrite { + path: source_root.to_path_buf(), + source, + }) +} + +/// A path component is denylisted if it equals a denylist entry, or — for +/// dotted entries, which double as extension filters (e.g. `.swp`, `.log`) — if +/// it ends with that entry. Plain names (`target`, `node_modules`) match +/// exactly only, so `mytarget` is not excluded. +fn is_denylisted(name: &std::ffi::OsStr) -> bool { + let name = name.to_string_lossy(); + ARCHIVE_DENYLIST + .iter() + .any(|d| name == *d || (d.starts_with('.') && name.ends_with(d))) +} + +/// Gzip with a default (mtime-zeroed) header so the same tar bytes always hash +/// the same. +fn gzip(bytes: &[u8]) -> Result, Error> { + let mut enc = flate2::write::GzEncoder::new(Vec::new(), flate2::Compression::default()); + enc.write_all(bytes).map_err(|source| Error::ArchiveWrite { + path: PathBuf::new(), + source, + })?; + enc.finish().map_err(|source| Error::ArchiveWrite { + path: PathBuf::new(), + source, + }) +} + +/// Decompress gzip and unpack the tar into `dest`. Entries are `source/…`, so +/// they land at `/source/…`. +fn unpack_targz(bytes: &[u8], dest: &Path) -> Result<(), Error> { + let dec = flate2::read::GzDecoder::new(bytes); + tar::Archive::new(dec) + .unpack(dest) + .map_err(Error::ArchiveExtract) +} + +/// Refuse to run a verifiable build against a dirty git working tree: the +/// bind-mounted source must match the recorded source_sha256 for the build to +/// be reproducible. When the source root isn't a git repo (e.g. an extracted +/// archive) we can't check, so we proceed — the user owns the source_sha256 +/// they pass, and verify catches a mismatch at byte-comparison time. +fn enforce_clean_tree(source_root: &Path) -> Result<(), Error> { let status = Command::new("git") .arg("-C") - .arg(workspace_root) + .arg(source_root) .arg("status") .arg("--porcelain") .output() .map_err(|e| Error::GitInvoke { - path: workspace_root.to_path_buf(), + path: source_root.to_path_buf(), source: e, })?; + // Not a git repo (or git refused): can't verify cleanliness, proceed. + if !status.status.success() { + return Ok(()); + } + if !status.stdout.is_empty() { return Err(Error::GitDirty { - path: workspace_root.to_path_buf(), + path: source_root.to_path_buf(), }); } @@ -342,20 +594,45 @@ fn bldimg_regex() -> Regex { .unwrap() } -fn source_rev_regex() -> Regex { - Regex::new(r"^[0-9a-f]{40}$").unwrap() -} - -fn source_repo_regex() -> Regex { - Regex::new(r"^(https?://\S+|github:[^/\s]+/[^/\s]+)$").unwrap() +fn source_sha256_regex() -> Regex { + Regex::new(r"^[0-9a-f]{64}$").unwrap() } -fn tarball_url_regex() -> Regex { - Regex::new(r"^https?://\S+$").unwrap() +fn source_uri_regex() -> Regex { + Regex::new(r"^[a-zA-Z][a-zA-Z0-9+.-]*:\S+$").unwrap() } -fn tarball_sha256_regex() -> Regex { - Regex::new(r"^[0-9a-f]{64}$").unwrap() +/// Resolve the package to pin as `--package`. An explicit `--package` wins. +/// Otherwise, when the workspace builds exactly one cdylib by default, return +/// its name so the recorded bldopt is reproducible even if the workspace's +/// default members change later. Returns `None` when the selection is +/// ambiguous (zero or multiple default cdylibs) — the build then keeps cargo's +/// default behavior of building them all, which `--package` can't express +/// (the container's flag is singular). +fn resolve_build_package(cmd: &Cmd) -> Result, Error> { + if cmd.package.is_some() { + return Ok(cmd.package.clone()); + } + let mut mc = MetadataCommand::new(); + mc.no_deps(); + if let Some(p) = &cmd.manifest_path { + mc.manifest_path(p); + } + let md = mc.exec().map_err(Error::Metadata)?; + let mut names: Vec = md + .packages + .iter() + .filter(|p| md.workspace_default_members.contains(&p.id)) + .filter(|p| { + p.targets + .iter() + .any(|t| t.crate_types.iter().any(|c| c == "cdylib")) + }) + .map(|p| p.name.clone()) + .collect(); + names.sort(); + names.dedup(); + Ok((names.len() == 1).then(|| names.remove(0))) } /// The flags forwarded to the container's `stellar contract build`, plus the @@ -371,6 +648,7 @@ fn tarball_sha256_regex() -> Regex { fn build_forwarded_args( cmd: &Cmd, workspace_root: &Path, + package: Option<&str>, supports_explicit_optimize_false: bool, ) -> (Vec, Vec) { let mut forwarded: Vec = Vec::new(); @@ -403,7 +681,10 @@ fn build_forwarded_args( if cmd.no_default_features { record("--no-default-features".to_string()); } - if let Some(pkg) = &cmd.package { + // Always pin the package when it can be resolved (explicit `--package`, or + // a workspace that builds exactly one cdylib by default) so the recorded + // bldopt stays reproducible even if workspace default members change later. + if let Some(pkg) = package { record(format!("--package={pkg}")); } for (k, v) in &cmd.build_args.meta { @@ -435,17 +716,11 @@ fn build_metadata_args(image_ref: &str, ids: &SourceIds, bldopts: &[String]) -> push(&mut out, "bldimg", image_ref); - if let Some(v) = &ids.source_repo { - push(&mut out, "source_repo", v); - } - if let Some(v) = &ids.source_rev { - push(&mut out, "source_rev", v); + if let Some(v) = &ids.source_uri { + push(&mut out, "source_uri", v); } - if let Some(v) = &ids.tarball_url { - push(&mut out, "tarball_url", v); - } - if let Some(v) = &ids.tarball_sha256 { - push(&mut out, "tarball_sha256", v); + if let Some(v) = &ids.source_sha256 { + push(&mut out, "source_sha256", v); } for o in bldopts { @@ -797,9 +1072,16 @@ async fn run_in_container( Ok(()) } +/// Collect the built WASM artifacts. Package names and the host target dir come +/// from host `cargo metadata`. `extracted_root` is set when the build ran +/// against an extracted archive (step: `--archive`): the artifacts then live +/// under that tree's target dir and must be copied out before its tempdir +/// drops. `source_root` is the host source root the extracted tree mirrors, so +/// the target dir's position relative to it carries over. fn collect_built_contracts( cmd: &Cmd, - workspace_root: &Path, + source_root: &Path, + extracted_root: Option<&Path>, _print: &Print, ) -> Result, super::Error> { let mut mc = MetadataCommand::new(); @@ -808,7 +1090,15 @@ fn collect_built_contracts( mc.manifest_path(p); } let md = mc.exec().map_err(Error::Metadata)?; - let target_dir = md.target_directory.as_std_path(); + let host_target = md.target_directory.as_std_path(); + + // Where the build actually wrote artifacts. For an extracted-archive build + // that's `/`; otherwise the + // host target dir (the working tree was bind-mounted directly). + let src_target = match extracted_root { + Some(er) => er.join(host_target.strip_prefix(source_root).unwrap_or(host_target)), + None => host_target.to_path_buf(), + }; let mut out = Vec::new(); for p in &md.packages { @@ -827,28 +1117,39 @@ fn collect_built_contracts( continue; } let wasm_name = p.name.replace('-', "_"); - let path = Path::new(target_dir) - .join(WASM_TARGET) + let rel = Path::new(WASM_TARGET) .join(&cmd.profile) .join(format!("{wasm_name}.wasm")); - if let Some(out_dir) = &cmd.out_dir { - let dest = out_dir.join(format!("{wasm_name}.wasm")); - if path.exists() { - std::fs::create_dir_all(out_dir).map_err(super::Error::CreatingOutDir)?; - std::fs::copy(&path, &dest).map_err(super::Error::CopyingWasmFile)?; - out.push(BuiltContract { - name: p.name.clone(), - path: dest, - }); - continue; + let src = src_target.join(&rel); + + // Destination: --out-dir wins; else if the build ran in a tempdir, copy + // into the host target dir so the artifact survives; else leave in + // place (the working tree was mounted, so it's already on the host). + let dest = if let Some(out_dir) = &cmd.out_dir { + Some(out_dir.join(format!("{wasm_name}.wasm"))) + } else if extracted_root.is_some() { + Some(host_target.join(&rel)) + } else { + None + }; + + let path = match dest { + Some(dest) if src.exists() => { + if let Some(parent) = dest.parent() { + std::fs::create_dir_all(parent).map_err(super::Error::CreatingOutDir)?; + } + std::fs::copy(&src, &dest).map_err(super::Error::CopyingWasmFile)?; + dest } - } + // Source missing: report the intended dest (matches prior leniency). + Some(dest) => dest, + None => src, + }; out.push(BuiltContract { name: p.name.clone(), path, }); } - let _ = workspace_root; Ok(out) } @@ -863,7 +1164,7 @@ mod tests { #[test] fn build_forwarded_args_defaults() { let cmd = Cmd::default(); - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), true); + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true); // Default optimize=true → bare `--optimize` recorded + forwarded. assert_eq!( forwarded, @@ -882,7 +1183,7 @@ mod tests { package: Some("contract-a".to_string()), ..Cmd::default() }; - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), true); + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true); assert!(forwarded.contains(&"--features=a,b".to_string())); assert!(forwarded.contains(&"--package=contract-a".to_string())); assert!(bldopts.contains(&"--features=a,b".to_string())); @@ -890,6 +1191,25 @@ mod tests { assert!(bldopts.contains(&"--locked".to_string())); } + #[test] + fn build_forwarded_args_records_resolved_package_when_unspecified() { + // No `--package` on the cmd, but the caller resolved one (single + // default cdylib); it must still be forwarded and recorded. + let cmd = Cmd::default(); + assert!(cmd.package.is_none()); + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), Some("hello-world"), true); + assert!(forwarded.contains(&"--package=hello-world".to_string())); + assert!(bldopts.contains(&"--package=hello-world".to_string())); + } + + #[test] + fn build_forwarded_args_omits_package_when_unresolved() { + let cmd = Cmd::default(); + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), None, true); + assert!(!forwarded.iter().any(|a| a.starts_with("--package"))); + assert!(!bldopts.iter().any(|a| a.starts_with("--package"))); + } + #[test] fn build_forwarded_args_records_meta_and_manifest() { let cmd = Cmd { @@ -903,7 +1223,7 @@ mod tests { }, ..Cmd::default() }; - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), true); + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true); assert!(forwarded.contains(&"--meta=home_domain=fnando.com".to_string())); assert!(forwarded.contains(&"--meta=author=alice".to_string())); assert!(forwarded.contains(&"--manifest-path=contracts/add/Cargo.toml".to_string())); @@ -921,7 +1241,7 @@ mod tests { }, ..Cmd::default() }; - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), true); + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true); assert!(forwarded.contains(&"--optimize=false".to_string())); assert!(bldopts.contains(&"--optimize=false".to_string())); } @@ -935,7 +1255,7 @@ mod tests { }, ..Cmd::default() }; - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), false); + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), false); // Old container's default is already false; record nothing. // Passing `--optimize=false` to a pre-26.1.0 cli would fail. assert!(!forwarded.iter().any(|a| a.starts_with("--optimize"))); @@ -949,12 +1269,10 @@ mod tests { } #[test] - fn build_metadata_args_source_repo_and_rev() { + fn build_metadata_args_uri_and_sha256() { let ids = SourceIds { - source_repo: Some("https://github.com/foo/bar".to_string()), - source_rev: Some("a".repeat(40)), - tarball_url: None, - tarball_sha256: None, + source_uri: Some("https://example.com/src.tar.gz".to_string()), + source_sha256: Some("a".repeat(64)), }; let m = build_metadata_args( "docker.io/stellar/stellar-cli@sha256:abc", @@ -962,126 +1280,171 @@ mod tests { &["--locked".to_string(), "--features=a".to_string()], ); let p = pairs(&m); - // bldimg first; source-ids only for what's set; bldopts last. + // bldimg first; source_uri then source_sha256; bldopts last. assert_eq!( p[0], ("--meta", "bldimg=docker.io/stellar/stellar-cli@sha256:abc") ); - assert_eq!(p[1], ("--meta", "source_repo=https://github.com/foo/bar")); + assert_eq!( + p[1], + ("--meta", "source_uri=https://example.com/src.tar.gz") + ); assert_eq!(p[2].0, "--meta"); - assert!(p[2].1.starts_with("source_rev=")); + assert!(p[2].1.starts_with("source_sha256=")); assert_eq!(p[3], ("--meta", "bldopt=--locked")); assert_eq!(p[4], ("--meta", "bldopt=--features=a")); - // No tarball entries emitted when those fields are None. - assert!(!m.iter().any(|s| s.starts_with("tarball_"))); - } - - #[test] - fn build_metadata_args_tarball_url_only() { - let ids = SourceIds { - tarball_url: Some("https://example.com/foo.tar.gz".to_string()), - ..SourceIds::default() - }; - let m = build_metadata_args("docker.io/stellar/stellar-cli@sha256:abc", &ids, &[]); - assert!(m - .iter() - .any(|s| s == "tarball_url=https://example.com/foo.tar.gz")); - assert!(!m.iter().any(|s| s.starts_with("source_"))); - assert!(!m.iter().any(|s| s.starts_with("tarball_sha256="))); } #[test] - fn build_metadata_args_tarball_pair() { + fn build_metadata_args_sha256_only_omits_uri() { let ids = SourceIds { - tarball_url: Some("https://example.com/foo.tar.gz".to_string()), - tarball_sha256: Some("f".repeat(64)), + source_sha256: Some("f".repeat(64)), ..SourceIds::default() }; let m = build_metadata_args("docker.io/stellar/stellar-cli@sha256:abc", &ids, &[]); assert!(m .iter() - .any(|s| s == "tarball_url=https://example.com/foo.tar.gz")); - assert!(m - .iter() - .any(|s| s == &format!("tarball_sha256={}", "f".repeat(64)))); + .any(|s| s == &format!("source_sha256={}", "f".repeat(64)))); + assert!(!m.iter().any(|s| s.starts_with("source_uri="))); } #[test] - fn validate_source_ids_missing_all_errors() { - let cmd = Cmd::default(); - let err = validate_source_ids(&cmd, ws()).unwrap_err(); - assert!(matches!(err, Error::MissingSourceId)); - } - - #[test] - fn validate_source_ids_rejects_bad_source_rev_format() { + fn validate_source_formats_rejects_bad_sha256() { let cmd = Cmd { - source_repo: Some("https://github.com/foo/bar".to_string()), - source_rev: Some("not-a-sha".to_string()), + source_sha256: Some("not-a-sha".to_string()), ..Cmd::default() }; - let err = validate_source_ids(&cmd, ws()).unwrap_err(); - assert!(matches!(err, Error::SourceRevFormat { .. })); + let err = validate_source_formats(&cmd).unwrap_err(); + assert!(matches!(err, Error::SourceSha256Format { .. })); } #[test] - fn validate_source_ids_rejects_bad_source_repo_format() { + fn validate_source_formats_rejects_bad_uri() { let cmd = Cmd { - source_repo: Some("foo/bar".to_string()), // missing scheme - source_rev: Some("a".repeat(40)), + source_uri: Some("not a uri".to_string()), // no scheme + source_sha256: Some("a".repeat(64)), ..Cmd::default() }; - let err = validate_source_ids(&cmd, ws()).unwrap_err(); - assert!(matches!(err, Error::SourceRepoFormat { .. })); + let err = validate_source_formats(&cmd).unwrap_err(); + assert!(matches!(err, Error::SourceUriFormat { .. })); } #[test] - fn validate_source_ids_rejects_bad_tarball_url() { + fn validate_source_formats_accepts_valid_and_absent() { + // Both absent is fine here — requiredness is enforced in run(). + validate_source_formats(&Cmd::default()).unwrap(); let cmd = Cmd { - tarball_url: Some("ftp://example.com/foo.tar.gz".to_string()), + source_uri: Some("https://example.com/src.tar.gz".to_string()), + source_sha256: Some("f".repeat(64)), ..Cmd::default() }; - let err = validate_source_ids(&cmd, ws()).unwrap_err(); - assert!(matches!(err, Error::TarballUrlFormat { .. })); + validate_source_formats(&cmd).unwrap(); } #[test] - fn validate_source_ids_rejects_short_tarball_sha256() { - let cmd = Cmd { - tarball_sha256: Some("abc".to_string()), - ..Cmd::default() - }; - let err = validate_source_ids(&cmd, ws()).unwrap_err(); - assert!(matches!(err, Error::TarballSha256Format { .. })); + fn is_denylisted_matches_names_and_dotted_suffixes() { + use std::ffi::OsStr; + // exact name matches + assert!(is_denylisted(OsStr::new("target"))); + assert!(is_denylisted(OsStr::new(".git"))); + assert!(is_denylisted(OsStr::new(".DS_Store"))); + // plain names match exactly only + assert!(!is_denylisted(OsStr::new("mytarget"))); + assert!(!is_denylisted(OsStr::new("targets"))); + // dotted entries also match as suffix (extension-style) + assert!(is_denylisted(OsStr::new("backup.git"))); + // unrelated files pass through + assert!(!is_denylisted(OsStr::new("Cargo.toml"))); + assert!(!is_denylisted(OsStr::new("lib.rs"))); + } + + // Initialize a git repo at `root` with one commit of everything present. + #[cfg(unix)] + fn git_init_commit(root: &Path) { + for args in [ + &["init", "-q", "-b", "main"][..], + &["add", "-A"][..], + &["commit", "-q", "-m", "init"][..], + ] { + let ok = Command::new("git") + .arg("-C") + .arg(root) + .args(args) + .env("GIT_AUTHOR_NAME", "T") + .env("GIT_AUTHOR_EMAIL", "t@e.x") + .env("GIT_COMMITTER_NAME", "T") + .env("GIT_COMMITTER_EMAIL", "t@e.x") + .status() + .unwrap() + .success(); + assert!(ok); + } } #[test] - fn validate_source_ids_accepts_tarball_url_alone() { - let cmd = Cmd { - tarball_url: Some("https://example.com/foo.tar.gz".to_string()), - ..Cmd::default() - }; - let ids = validate_source_ids(&cmd, ws()).unwrap(); - assert_eq!( - ids.tarball_url.as_deref(), - Some("https://example.com/foo.tar.gz") - ); - assert!(ids.source_repo.is_none()); - assert!(ids.source_rev.is_none()); - assert!(ids.tarball_sha256.is_none()); + #[cfg(unix)] + fn build_source_archive_git_is_prefixed_and_deterministic() { + use std::os::unix::fs::PermissionsExt; + let print = Print::new(true); + let temp = tempfile::TempDir::new().unwrap(); + let root = temp.path(); + std::fs::write(root.join("Cargo.toml"), b"# crate").unwrap(); + std::fs::create_dir_all(root.join("src")).unwrap(); + std::fs::write(root.join("src/lib.rs"), b"// code").unwrap(); + git_init_commit(root); + + let a = build_source_archive(root, &print).unwrap(); + let b = build_source_archive(root, &print).unwrap(); + assert!(!a.is_empty()); + assert_eq!(a, b, "same commit should produce identical bytes"); + + let sha = hex::encode(Sha256::digest(&a)); + assert_eq!(sha.len(), 64); + + // Unpack and confirm the `source/` prefix + hardened perms. + let dest = tempfile::TempDir::new().unwrap(); + unpack_targz(&a, dest.path()).unwrap(); + assert!(dest.path().join("source/Cargo.toml").exists()); + assert!(dest.path().join("source/src/lib.rs").exists()); + + enforce_hardened_tree(dest.path()).unwrap(); + let file_mode = std::fs::metadata(dest.path().join("source/Cargo.toml")) + .unwrap() + .permissions() + .mode() + & 0o777; + let dir_mode = std::fs::metadata(dest.path().join("source")) + .unwrap() + .permissions() + .mode() + & 0o777; + assert_eq!(file_mode, 0o600); + assert_eq!(dir_mode, 0o700); } #[test] - fn validate_source_ids_accepts_tarball_sha256_alone() { - let cmd = Cmd { - tarball_sha256: Some("f".repeat(64)), - ..Cmd::default() - }; - let ids = validate_source_ids(&cmd, ws()).unwrap(); - assert_eq!( - ids.tarball_sha256.as_deref(), - Some("ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff") - ); + fn build_source_archive_non_git_excludes_denylist() { + let print = Print::new(true); + let temp = tempfile::TempDir::new().unwrap(); + let root = temp.path(); + std::fs::write(root.join("Cargo.toml"), b"# crate").unwrap(); + std::fs::create_dir_all(root.join("src")).unwrap(); + std::fs::write(root.join("src/lib.rs"), b"// code").unwrap(); + // Planted dirs that must be excluded. + std::fs::create_dir_all(root.join("target/debug")).unwrap(); + std::fs::write(root.join("target/debug/x"), b"junk").unwrap(); + std::fs::create_dir_all(root.join(".git")).unwrap(); + std::fs::write(root.join(".git/config"), b"junk").unwrap(); + + let bytes = build_source_archive(root, &print).unwrap(); + let dest = tempfile::TempDir::new().unwrap(); + unpack_targz(&bytes, dest.path()).unwrap(); + + assert!(dest.path().join("source/Cargo.toml").exists()); + assert!(dest.path().join("source/src/lib.rs").exists()); + assert!(!dest.path().join("source/target").exists()); + assert!(!dest.path().join("source/.git").exists()); + assert_eq!(hex::encode(Sha256::digest(&bytes)).len(), 64); } #[test] @@ -1114,33 +1477,20 @@ mod tests { } #[test] - fn source_rev_regex_matches_40_hex() { - assert!(source_rev_regex().is_match(&"a".repeat(40))); - assert!(!source_rev_regex().is_match(&"a".repeat(39))); - assert!(!source_rev_regex().is_match(&"A".repeat(40))); // upper-case rejected - } - - #[test] - fn source_repo_regex_accepts_https_and_github_shorthand() { - assert!(source_repo_regex().is_match("https://github.com/foo/bar")); - assert!(source_repo_regex().is_match("http://example.com/foo.git")); - assert!(source_repo_regex().is_match("github:foo/bar")); - assert!(!source_repo_regex().is_match("foo/bar")); - assert!(!source_repo_regex().is_match("git@github.com:foo/bar.git")); - } - - #[test] - fn tarball_url_regex_accepts_http_only() { - assert!(tarball_url_regex().is_match("https://example.com/foo.tar.gz")); - assert!(tarball_url_regex().is_match("http://example.com/foo.tar.gz")); - assert!(!tarball_url_regex().is_match("ftp://example.com/foo.tar.gz")); + fn source_sha256_regex_matches_64_hex() { + assert!(source_sha256_regex().is_match(&"f".repeat(64))); + assert!(!source_sha256_regex().is_match(&"f".repeat(63))); + assert!(!source_sha256_regex().is_match(&"F".repeat(64))); // upper-case rejected } #[test] - fn tarball_sha256_regex_matches_64_hex() { - assert!(tarball_sha256_regex().is_match(&"f".repeat(64))); - assert!(!tarball_sha256_regex().is_match(&"f".repeat(63))); - assert!(!tarball_sha256_regex().is_match(&"F".repeat(64))); + fn source_uri_regex_accepts_any_scheme() { + assert!(source_uri_regex().is_match("https://example.com/src.tar.gz")); + assert!(source_uri_regex().is_match("http://example.com/foo.git")); + assert!(source_uri_regex().is_match("ipfs://Qm...abc")); + assert!(source_uri_regex().is_match("github:foo/bar")); + assert!(!source_uri_regex().is_match("foo/bar")); // no scheme + assert!(!source_uri_regex().is_match("https://has space")); // whitespace } #[test] @@ -1196,7 +1546,7 @@ mod tests { #[test] fn reserved_meta_keys_list() { - for key in ["bldimg", "source_rev", "bldopt"] { + for key in ["bldimg", "source_uri", "source_sha256", "bldopt"] { assert!(RESERVED_META_KEYS.contains(&key)); } } diff --git a/cmd/soroban-cli/src/config/data.rs b/cmd/soroban-cli/src/config/data.rs index e310ebf131..6e49f19262 100644 --- a/cmd/soroban-cli/src/config/data.rs +++ b/cmd/soroban-cli/src/config/data.rs @@ -58,6 +58,12 @@ pub fn bucket_dir() -> Result { Ok(dir) } +pub fn archives_dir() -> Result { + let dir = data_local_dir()?.join("archives"); + std::fs::create_dir_all(&dir)?; + Ok(dir) +} + pub fn write(action: Action, rpc_url: &Url) -> Result { let data = Data { action, @@ -211,6 +217,18 @@ mod test { use crate::test_utils::with_env_set; use serial_test::serial; + #[test] + #[serial] + fn archives_dir_under_data_home_and_created() { + let t = assert_fs::TempDir::new().unwrap(); + with_env_set("STELLAR_DATA_HOME", t.path(), || { + let dir = archives_dir().unwrap(); + assert!(dir.ends_with("archives")); + assert!(dir.starts_with(t.path())); + assert!(dir.is_dir(), "archives_dir() should create the directory"); + }); + } + #[test] #[serial] fn test_write_read() { From 5653489383f9ba8100c616cf10d3a7009f1923a4 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Tue, 16 Jun 2026 21:16:43 -0700 Subject: [PATCH 16/25] Record each contract's package in verifiable builds. --- .../src/commands/contract/build/verifiable.rs | 158 +++++++++++++----- 1 file changed, 115 insertions(+), 43 deletions(-) diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 618ee15a97..e278ec961e 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -144,13 +144,8 @@ pub enum Error { #[error(transparent)] Data(#[from] data::Error), - #[error("container build exited with status {status}. To reproduce manually:\n docker run --rm -v {mount}:/source {image} {args}")] - ContainerExit { - status: i64, - image: String, - mount: String, - args: String, - }, + #[error("container build exited with status {status}. To reproduce manually:\n {command}")] + ContainerExit { status: i64, command: String }, } pub async fn run( @@ -247,31 +242,38 @@ pub async fn run( probe_supports_optimize_false_syntax(&image_ref, &docker, print).await }; - let package = resolve_build_package(cmd)?; - if cmd.package.is_none() { - if let Some(pkg) = &package { - print.infoln(format!( - "Inferred --package={pkg} and using it as a build option." - )); - } + // Build once per package, each with its own `--package` forwarded and + // recorded as a `bldopt`, so every WASM is independently reproducible. With + // no explicit `--package` the targets are inferred like a regular build. + let packages = resolve_build_packages(cmd)?; + if cmd.package.is_none() && !packages.is_empty() { + print.infoln(format!("Inferred packages: {}", packages.join(", "))); } - let (forwarded_args, bldopts) = build_forwarded_args( - cmd, - &source_root, - package.as_deref(), - supports_explicit_optimize_false, - ); - let metadata_args = build_metadata_args(&image_ref, &source_ids, &bldopts); - let container_cmd_args = compose_container_args(&forwarded_args, &metadata_args); + let targets: Vec> = if packages.is_empty() { + vec![None] + } else { + packages.iter().map(|p| Some(p.as_str())).collect() + }; + let container_cmds: Vec> = targets + .iter() + .map(|target| { + let (forwarded_args, bldopts) = + build_forwarded_args(cmd, &source_root, *target, supports_explicit_optimize_false); + let metadata_args = build_metadata_args(&image_ref, &source_ids, &bldopts); + compose_container_args(&forwarded_args, &metadata_args) + }) + .collect(); // Always stream the container's cargo output during `contract build // --verifiable`, matching how a non-verifiable `contract build` shows // cargo output by default. The verify-side caller gates this on - // `--verbose` because verifications are run as part of pipelines. + // `--verbose` because verifications are run as part of pipelines. All + // per-package builds run in one container so the crates download, compiled + // deps, and target/ are shared. run_in_container( &image_ref, &resolved.mount_root, - &container_cmd_args, + &container_cmds, &docker, print, true, @@ -602,16 +604,16 @@ fn source_uri_regex() -> Regex { Regex::new(r"^[a-zA-Z][a-zA-Z0-9+.-]*:\S+$").unwrap() } -/// Resolve the package to pin as `--package`. An explicit `--package` wins. -/// Otherwise, when the workspace builds exactly one cdylib by default, return -/// its name so the recorded bldopt is reproducible even if the workspace's -/// default members change later. Returns `None` when the selection is -/// ambiguous (zero or multiple default cdylibs) — the build then keeps cargo's -/// default behavior of building them all, which `--package` can't express -/// (the container's flag is singular). -fn resolve_build_package(cmd: &Cmd) -> Result, Error> { - if cmd.package.is_some() { - return Ok(cmd.package.clone()); +/// Resolve every package the build will produce, so each can be pinned with its +/// own `--package` (and recorded as a `bldopt`) — making each WASM independently +/// reproducible even if the workspace's default members change later. An +/// explicit `--package` wins; otherwise infer the default-member cdylibs exactly +/// like a regular `stellar contract build` does. May be empty (no cdylib default +/// members), in which case the caller falls back to a single no-`--package` +/// build. +fn resolve_build_packages(cmd: &Cmd) -> Result, Error> { + if let Some(pkg) = &cmd.package { + return Ok(vec![pkg.clone()]); } let mut mc = MetadataCommand::new(); mc.no_deps(); @@ -632,7 +634,7 @@ fn resolve_build_package(cmd: &Cmd) -> Result, Error> { .collect(); names.sort(); names.dedup(); - Ok((names.len() == 1).then(|| names.remove(0))) + Ok(names) } /// The flags forwarded to the container's `stellar contract build`, plus the @@ -980,18 +982,57 @@ async fn probe_cli_version(image_ref: &str, docker: &Docker) -> Result]) -> String { + cmds.iter() + .map(|cmd| { + std::iter::once("stellar") + .chain(cmd.iter().map(String::as_str)) + .map(|tok| shell_escape::escape(tok.into()).into_owned()) + .collect::>() + .join(" ") + }) + .collect::>() + .join(" && ") +} + async fn run_in_container( image_ref: &str, workspace_root: &Path, - container_cmd: &[String], + container_cmds: &[Vec], docker: &Docker, print: &Print, verbose: bool, ) -> Result<(), Error> { let bind = format!("{}:/source", workspace_root.display()); + + // One package → run the image's default `stellar` entrypoint directly. + // Several → override the entrypoint to a shell and chain the builds so they + // all run in this one container. + let (entrypoint, cmd, reproduce) = if container_cmds.len() > 1 { + let chain = compose_shell_command(container_cmds); + let reproduce = format!( + "docker run --rm -v {bind} --entrypoint /bin/sh {image_ref} -c {}", + shell_escape::escape(chain.clone().into()) + ); + ( + Some(vec!["/bin/sh".to_string(), "-c".to_string()]), + vec![chain], + reproduce, + ) + } else { + let cmd = container_cmds.first().cloned().unwrap_or_default(); + let reproduce = format!("docker run --rm -v {bind} {image_ref} {}", cmd.join(" ")); + (None, cmd, reproduce) + }; + let config = ContainerCreateBody { image: Some(image_ref.to_string()), - cmd: Some(container_cmd.to_vec()), + entrypoint, + cmd: Some(cmd), working_dir: Some("/source".to_string()), attach_stdout: Some(true), attach_stderr: Some(true), @@ -1051,18 +1092,14 @@ async fn run_in_container( Ok(r) => { return Err(Error::ContainerExit { status: r.status_code, - image: image_ref.to_string(), - mount: workspace_root.display().to_string(), - args: container_cmd.join(" "), + command: reproduce.clone(), }); } Err(bollard::errors::Error::DockerContainerWaitError { code: 0, .. }) => {} Err(bollard::errors::Error::DockerContainerWaitError { code, .. }) => { return Err(Error::ContainerExit { status: code, - image: image_ref.to_string(), - mount: workspace_root.display().to_string(), - args: container_cmd.join(" "), + command: reproduce.clone(), }); } Err(e) => return Err(e.into()), @@ -1550,4 +1587,39 @@ mod tests { assert!(RESERVED_META_KEYS.contains(&key)); } } + + #[test] + fn compose_shell_command_chains_and_escapes() { + let a = vec![ + "contract".to_string(), + "build".to_string(), + "--package=another".to_string(), + "--meta".to_string(), + "home_domain=fnando.com".to_string(), + ]; + let b = vec![ + "contract".to_string(), + "build".to_string(), + "--package=hello-world".to_string(), + ]; + let s = compose_shell_command(&[a, b]); + assert_eq!( + s, + "stellar contract build --package=another --meta home_domain=fnando.com \ + && stellar contract build --package=hello-world" + ); + + // A meta value with a space must be quoted so it stays one token. + let c = vec![ + "contract".to_string(), + "build".to_string(), + "--meta".to_string(), + "note=added on build".to_string(), + ]; + let s = compose_shell_command(&[c]); + assert!( + s.contains("'note=added on build'") || s.contains("\"note=added on build\""), + "expected the spaced value to be quoted, got: {s}" + ); + } } From 4518c1acb8e675c1ff7639c60292bd73755ba6e8 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Tue, 16 Jun 2026 22:04:02 -0700 Subject: [PATCH 17/25] Document reproducible source archive guarantees. --- .../src/commands/contract/build/verifiable.rs | 19 +++++++++++++++++-- 1 file changed, 17 insertions(+), 2 deletions(-) diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index e278ec961e..faa45b9213 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -485,8 +485,18 @@ fn git_archive_tar(source_root: &Path) -> Result, Error> { } /// Tar the working tree under `source_root`, skipping denylisted path -/// components, with entries sorted and headers normalized (deterministic mode) -/// so the bytes are reproducible. Each entry is prefixed with `source/`. +/// components. Each entry is prefixed with `source/`. +/// +/// The output is reproducible, following GNU tar's reproducibility guidance +/// () +/// with the portable equivalents available via the `tar` crate (the system +/// `tar` can't be relied on — macOS ships bsdtar, which lacks `--sort`, +/// `--mtime`, `--pax-option`, …): entries are sorted by name (`--sort=name`) +/// using locale-independent path ordering (`LC_ALL=C`), and `HeaderMode::Deterministic` +/// zeroes mtime (`--mtime`/`--clamp-mtime`), sets uid/gid to 0 with empty owner +/// names (`--owner=0 --group=0 --numeric-owner`), and normalizes mode +/// (`--mode=go+u,go-w`). ustar headers carry no atime/ctime or tar PID. The gzip +/// wrapper (see `gzip`) is likewise deterministic. fn walk_tar(source_root: &Path) -> Result, Error> { let mut files: Vec = Vec::new(); let walk = WalkDir::new(source_root) @@ -1482,6 +1492,11 @@ mod tests { assert!(!dest.path().join("source/target").exists()); assert!(!dest.path().join("source/.git").exists()); assert_eq!(hex::encode(Sha256::digest(&bytes)).len(), 64); + + // Reproducible: a second run over the same tree yields identical bytes + // (sorted entries + zeroed header fields + deterministic gzip). + let again = build_source_archive(root, &print).unwrap(); + assert_eq!(bytes, again); } #[test] From 36552c9beff5731ab4a7c8f4db99298c8de102f8 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Wed, 17 Jun 2026 09:54:42 -0700 Subject: [PATCH 18/25] Add stellar contract archive command. --- FULL_HELP_DOCS.md | 16 +- cmd/crates/soroban-test/tests/it/build.rs | 113 +++- .../src/commands/contract/archive.rs | 126 +++++ .../src/commands/contract/build.rs | 26 +- .../commands/contract/build/source_archive.rs | 450 ++++++++++++++++ .../src/commands/contract/build/verifiable.rs | 484 ++---------------- .../src/commands/contract/fetch.rs | 4 + cmd/soroban-cli/src/commands/contract/mod.rs | 8 + 8 files changed, 743 insertions(+), 484 deletions(-) create mode 100644 cmd/soroban-cli/src/commands/contract/archive.rs create mode 100644 cmd/soroban-cli/src/commands/contract/build/source_archive.rs diff --git a/FULL_HELP_DOCS.md b/FULL_HELP_DOCS.md index e6af930448..d48ab89939 100644 --- a/FULL_HELP_DOCS.md +++ b/FULL_HELP_DOCS.md @@ -84,6 +84,7 @@ Tools for smart contract developers - `asset` — Utilities to deploy a Stellar Asset Contract or get its id - `alias` — Utilities to manage contract aliases - `bindings` — Generate code client bindings for a contract +- `archive` — Generate the reproducible source archive used by verifiable builds - `build` — Build a contract from source - `extend` — Extend the time to live ledger of a contract-data ledger entry - `deploy` — Deploy a wasm contract @@ -344,6 +345,18 @@ Generate PHP bindings **Usage:** `stellar contract bindings php` +## `stellar contract archive` + +Generate the reproducible source archive used by verifiable builds + +**Usage:** `stellar contract archive [OPTIONS]` + +###### **Options:** + +- `-o`, `--out-file ` — Where to write the gzipped tarball. Required unless `--dry-run` is used +- `--manifest-path ` — Path to Cargo.toml, used to locate the source root (its enclosing git repository, or the working directory) +- `--dry-run` — List the entries that would be archived and the computed source_sha256, without writing any file + ## `stellar contract build` Build a contract from source @@ -398,9 +411,8 @@ To view the commands that will be executed, without executing them, use the --pr - `--verifiable` — Build inside a trusted Docker container and record SEP-58 metadata (`bldimg`, `source_uri`, `source_sha256`, `bldopt`) so the resulting WASM can be reproduced and verified by third parties. Implies `--locked`. Requires a clean git working tree - `--image ` — Override the auto-selected container image used by `--verifiable`. Must be digest-pinned, e.g. `docker.io/stellar/stellar-cli@sha256:...`. Tag-only refs are rejected because SEP-58 requires content addressing -- `--source-sha256 ` — SEP-58 source identification: SHA-256 of the source archive/tree (recorded as the `source_sha256` meta entry). Required with `--verifiable` unless `--archive` is used, which generates the archive and computes this for you +- `--source-sha256 ` — SEP-58 source identification: SHA-256 of the source archive (recorded as the `source_sha256` meta entry). Optional with `--verifiable`: the archive is always generated and its SHA-256 computed for you. When supplied it's treated as a pin — the build fails if it doesn't match the generated archive - `--source-uri ` — SEP-58 source identification: URI where the source can be obtained, e.g. `https://example.com/src.tar.gz` (recorded as the `source_uri` meta entry). Optional; when set it must accompany `--source-sha256` -- `--archive ` — Generate a source archive for the verifiable build, then build from it and record its SHA-256 as the SEP-58 `source_sha256` meta entry. Pass a path to choose where the gzipped tarball is written; with no path it goes to the data dir's `archives/`. In a git repo the archive is `git archive HEAD`; otherwise the working directory is archived minus a built-in denylist (.git, .svn, .hg, target/, node_modules/, .DS_Store) - `-d`, `--docker-host ` — Override the default docker host used by `--verifiable` ## `stellar contract extend` diff --git a/cmd/crates/soroban-test/tests/it/build.rs b/cmd/crates/soroban-test/tests/it/build.rs index 9733d37fb6..dfe798e982 100644 --- a/cmd/crates/soroban-test/tests/it/build.rs +++ b/cmd/crates/soroban-test/tests/it/build.rs @@ -1088,13 +1088,16 @@ fn verifiable_image_requires_explicit_registry_host() { .stderr(predicate::str::contains("bldimg format")); } -// `--verifiable` with neither `--source-sha256` nor `--archive` must error. -// Run in a fresh (non-git) workspace so the clean-tree check is skipped and the -// missing-source error is what surfaces. +// `--verifiable` always generates the source archive (and computes +// source_sha256) before the docker stage, so the "Wrote source archive" line +// appears even though the build then fails to reach a real image. #[test] -fn verifiable_requires_source_sha256() { +fn verifiable_always_writes_source_archive() { let sandbox = TestEnv::default(); let (_temp, workspace) = fresh_workspace(); + git_in(&workspace, &["init", "-q", "-b", "main"]); + git_in(&workspace, &["add", "-A"]); + git_in(&workspace, &["commit", "-q", "-m", "init"]); sandbox .new_assert_cmd("contract") @@ -1105,39 +1108,111 @@ fn verifiable_requires_source_sha256() { .arg(ZERO_DIGEST) .assert() .failure() - .stderr(predicate::str::contains("--source-sha256")); + .stderr( + predicate::str::contains("Wrote source archive") + .and(predicate::str::contains("source_sha256")), + ); } -// `--archive` generates the source archive (and computes source_sha256) before -// the docker stage, so the file is written even though the build then fails to -// reach a real image. +// `contract archive --out` writes the gzipped tarball and prints its +// source_sha256. #[test] -fn verifiable_archive_writes_source_archive() { +fn contract_archive_writes_out() { let sandbox = TestEnv::default(); let (temp, workspace) = fresh_workspace(); git_in(&workspace, &["init", "-q", "-b", "main"]); git_in(&workspace, &["add", "-A"]); git_in(&workspace, &["commit", "-q", "-m", "init"]); - let archive_path = temp.path().join("src.tar.gz"); + let out = temp.path().join("src.tar.gz"); sandbox .new_assert_cmd("contract") - .current_dir(workspace.join("contracts").join("add")) - .arg("build") - .arg("--verifiable") - .arg("--image") - .arg(ZERO_DIGEST) - .arg(format!("--archive={}", archive_path.display())) + .current_dir(&workspace) + .arg("archive") + .arg("--out-file") + .arg(&out) .assert() - .failure(); + .success() + .stderr( + predicate::str::contains("Wrote source archive") + .and(predicate::str::contains("source_sha256")), + ); + assert!(out.exists(), "the archive should be written to --out"); assert!( - archive_path.exists(), - "source archive should be written before the docker stage" + std::fs::metadata(&out).unwrap().len() > 0, + "the archive should not be empty" ); } +// `contract archive --dry-run` lists the archived entries and the +// source_sha256 without writing any file. +#[test] +fn contract_archive_dry_run_lists_entries() { + let sandbox = TestEnv::default(); + let (temp, workspace) = fresh_workspace(); + git_in(&workspace, &["init", "-q", "-b", "main"]); + git_in(&workspace, &["add", "-A"]); + git_in(&workspace, &["commit", "-q", "-m", "init"]); + + let out = temp.path().join("should-not-exist.tar.gz"); + + sandbox + .new_assert_cmd("contract") + .current_dir(&workspace) + .arg("archive") + .arg("--dry-run") + .assert() + .success() + .stdout(predicate::str::contains("source/Cargo.toml")) + .stderr(predicate::str::contains("source_sha256")); + + assert!(!out.exists(), "--dry-run must not write an archive"); +} + +// `--out-file` must name a gzipped tarball (.tar.gz / .tgz). +#[test] +fn contract_archive_rejects_bad_out_file_extension() { + let sandbox = TestEnv::default(); + let (temp, workspace) = fresh_workspace(); + git_in(&workspace, &["init", "-q", "-b", "main"]); + git_in(&workspace, &["add", "-A"]); + git_in(&workspace, &["commit", "-q", "-m", "init"]); + + let out = temp.path().join("src.zip"); + + sandbox + .new_assert_cmd("contract") + .current_dir(&workspace) + .arg("archive") + .arg("--out-file") + .arg(&out) + .assert() + .failure() + .stderr(predicate::str::contains(".tar.gz or .tgz")); + + assert!( + !out.exists(), + "no archive should be written on a bad extension" + ); +} + +// `--out-file` is required unless `--dry-run` is passed. +#[test] +fn contract_archive_requires_out_file_without_dry_run() { + let sandbox = TestEnv::default(); + let (_temp, workspace) = fresh_workspace(); + + sandbox + .new_assert_cmd("contract") + .current_dir(&workspace) + .arg("archive") + .assert() + .failure() + .stderr(predicate::str::contains("--out-file")); +} + // `--source-sha256` value must match the 64-hex regex. #[test] fn verifiable_source_sha256_format_errors() { diff --git a/cmd/soroban-cli/src/commands/contract/archive.rs b/cmd/soroban-cli/src/commands/contract/archive.rs new file mode 100644 index 0000000000..d4fe4cef03 --- /dev/null +++ b/cmd/soroban-cli/src/commands/contract/archive.rs @@ -0,0 +1,126 @@ +use std::path::PathBuf; + +use clap::Parser; +use sha2::{Digest, Sha256}; + +use crate::{commands::global, print::Print}; + +use super::build::source_archive; + +/// Accepted `--out-file` suffixes (lower-case). The archive is always a gzipped +/// tarball, so the filename must say so. +const ARCHIVE_EXTENSIONS: &[&str] = &[".tar.gz", ".tgz"]; + +/// Generate (or inspect) the reproducible source archive for a contract. +/// +/// Produces the same gzipped tarball that `stellar contract build --verifiable` +/// builds from, and prints its SHA-256 (the SEP-58 `source_sha256`). Use +/// `--dry-run` to list exactly what would be archived without writing anything — +/// handy for confirming the contents before a verifiable build, or for +/// producing the archive to host at a `--source-uri`. +/// +/// In a git repo the archive is `git archive HEAD` (the committed tree); +/// otherwise the working directory is archived minus a built-in denylist (.git, +/// target/, node_modules/, .DS_Store, …). +#[derive(Parser, Debug, Clone)] +#[group(skip)] +pub struct Cmd { + /// Where to write the gzipped tarball. Required unless `--dry-run` is used. + #[arg(long, short = 'o', required_unless_present = "dry_run")] + pub out_file: Option, + + /// Path to Cargo.toml, used to locate the source root (its enclosing git + /// repository, or the working directory). + #[arg(long)] + pub manifest_path: Option, + + /// List the entries that would be archived and the computed source_sha256, + /// without writing any file. + #[arg(long)] + pub dry_run: bool, +} + +#[derive(thiserror::Error, Debug)] +pub enum Error { + #[error(transparent)] + SourceArchive(#[from] source_archive::Error), + + #[error( + "--out-file {0} must end in .tar.gz or .tgz (the archive is always a gzipped tarball)" + )] + OutFileExtension(String), +} + +impl Cmd { + pub fn run(&self, global_args: &global::Args) -> Result<(), Error> { + let print = Print::new(global_args.quiet); + + let source_root = source_archive::resolve_source_root(self.manifest_path.as_deref()); + + // The git path archives HEAD, so uncommitted changes are silently + // excluded. Warn (don't fail — this is an inspect/generate tool, not a + // build) so the printed source_sha256 isn't mistaken for the working + // tree's. + if source_archive::tree_is_dirty(&source_root)? { + print.warnln(format!( + "git working tree at {} is dirty; the archive reflects HEAD only and excludes uncommitted changes.", + source_root.display(), + )); + } + + // The dry-run listing itself reveals the contents, so skip the + // "not a git repository" warning there. + let bytes = source_archive::build_source_archive(&source_root, &print, !self.dry_run)?; + let sha = hex::encode(Sha256::digest(&bytes)); + + if self.dry_run { + let names = source_archive::entry_names(&bytes)?; + let prefix = print.compute_emoji("📄"); + + for name in &names { + println!("{prefix} {name}"); + } + print.infoln(format!("{} files", names.len())); + print.infoln(format!("source_sha256 {sha}")); + return Ok(()); + } + + // `--out-file` is required when not `--dry-run`, so this is always set here. + let out = self + .out_file + .as_ref() + .expect("--out-file is required without --dry-run"); + + // The output is always a gzipped tarball, so require a matching + // extension to keep the filename honest. + let name = out + .file_name() + .unwrap_or_default() + .to_string_lossy() + .to_ascii_lowercase(); + if !ARCHIVE_EXTENSIONS.iter().any(|ext| name.ends_with(ext)) { + return Err(Error::OutFileExtension(out.display().to_string())); + } + + if let Some(parent) = out.parent() { + if !parent.as_os_str().is_empty() { + std::fs::create_dir_all(parent).map_err(|source| { + source_archive::Error::ArchiveWrite { + path: out.clone(), + source, + } + })?; + } + } + std::fs::write(out, &bytes).map_err(|source| source_archive::Error::ArchiveWrite { + path: out.clone(), + source, + })?; + print.checkln(format!( + "Wrote source archive {} (source_sha256 {sha})", + out.display() + )); + + Ok(()) + } +} diff --git a/cmd/soroban-cli/src/commands/contract/build.rs b/cmd/soroban-cli/src/commands/contract/build.rs index 9ad08ca7fd..e8fb9c0c4b 100644 --- a/cmd/soroban-cli/src/commands/contract/build.rs +++ b/cmd/soroban-cli/src/commands/contract/build.rs @@ -25,6 +25,7 @@ use crate::{ wasm, }; +pub(crate) mod source_archive; pub mod verifiable; /// A built WASM artifact with its package name and file path. @@ -111,10 +112,11 @@ pub struct Cmd { #[arg(long, requires = "verifiable", help_heading = "Verifiable")] pub image: Option, - /// SEP-58 source identification: SHA-256 of the source archive/tree - /// (recorded as the `source_sha256` meta entry). Required with - /// `--verifiable` unless `--archive` is used, which generates the archive - /// and computes this for you. + /// SEP-58 source identification: SHA-256 of the source archive + /// (recorded as the `source_sha256` meta entry). Optional with + /// `--verifiable`: the archive is always generated and its SHA-256 computed + /// for you. When supplied it's treated as a pin — the build fails if it + /// doesn't match the generated archive. #[arg(long, requires = "verifiable", help_heading = "Verifiable")] pub source_sha256: Option, @@ -129,21 +131,6 @@ pub struct Cmd { )] pub source_uri: Option, - /// Generate a source archive for the verifiable build, then build from it - /// and record its SHA-256 as the SEP-58 `source_sha256` meta entry. Pass a - /// path to choose where the gzipped tarball is written; with no path it - /// goes to the data dir's `archives/`. In a git repo the archive is - /// `git archive HEAD`; otherwise the working directory is archived minus a - /// built-in denylist (.git, .svn, .hg, target/, node_modules/, .DS_Store). - #[arg( - long, - num_args = 0..=1, - require_equals = true, - requires = "verifiable", - help_heading = "Verifiable" - )] - pub archive: Option>, - /// Override the default docker host used by `--verifiable`. #[arg(short = 'd', long, env = "DOCKER_HOST", help_heading = "Verifiable")] pub docker_host: Option, @@ -281,7 +268,6 @@ impl Default for Cmd { image: None, source_sha256: None, source_uri: None, - archive: None, docker_host: None, build_args: BuildArgs::default(), } diff --git a/cmd/soroban-cli/src/commands/contract/build/source_archive.rs b/cmd/soroban-cli/src/commands/contract/build/source_archive.rs new file mode 100644 index 0000000000..2f20d516cd --- /dev/null +++ b/cmd/soroban-cli/src/commands/contract/build/source_archive.rs @@ -0,0 +1,450 @@ +//! Reproducible source-archive generation for verifiable builds. +//! +//! Produces a gzipped tarball of a contract's source tree, rooted under a +//! top-level `source/` prefix (so it extracts to a `source/` dir, mirroring the +//! container's `/source` mount). In a git repo this is `git archive HEAD` (the +//! committed tree); otherwise the working directory is walked and tarred, +//! skipping `ARCHIVE_DENYLIST` entries. The output is byte-reproducible, so the +//! same tree always hashes to the same `source_sha256`. +//! +//! Shared by `contract build --verifiable` (which builds from the extracted +//! archive) and the standalone `contract archive` command (which generates and +//! inspects it). + +use std::{ + io::Write, + path::{Path, PathBuf}, + process::Command, +}; + +use walkdir::WalkDir; + +use crate::print::Print; + +/// Top-level names excluded when archiving a non-git working directory (we have +/// no tracked-files list to consult, so fall back to a fixed denylist of VCS +/// metadata, build/cache/transient dirs, and editor/OS/AI-assistant junk). +/// Matched against each path component, so a directory like `target/` prunes +/// its whole subtree. +pub(crate) const ARCHIVE_DENYLIST: &[&str] = &[ + // version control + ".git", + ".gitignore", + ".svn", + ".hg", + // secrets / local environment + ".env", + // build output / dependencies + "target", + "node_modules", + // transient + "log", + "logs", + "tmp", + "temp", + // OS / editor junk + ".DS_Store", + "Thumbs.db", + ".idea", + ".vscode", + // AI assistant dirs + ".claude", + ".cursor", + ".windsurf", + ".aider", +]; + +#[derive(thiserror::Error, Debug)] +pub enum Error { + #[error("could not read git state at {path}: {source}")] + GitInvoke { + path: PathBuf, + source: std::io::Error, + }, + + #[error("`git archive` failed in {path}: {stderr}")] + GitArchive { path: PathBuf, stderr: String }, + + #[error("could not write source archive to {path}: {source}")] + ArchiveWrite { + path: PathBuf, + source: std::io::Error, + }, + + #[error("could not extract source archive: {0}")] + ArchiveExtract(std::io::Error), +} + +/// Pick the anchor for the source tree: the directory whose `.git` parent we +/// archive (and, for verifiable builds, relativize `--manifest-path` against). +/// Walk up from `manifest_path` (or cwd, if none) looking for a `.git` +/// directory; return its parent. If none is found, fall back to cwd. +/// +/// This isn't a validation step — any `.git` will do. Wrong-source mistakes are +/// caught later by the verify-side byte comparison. +pub(crate) fn resolve_source_root(manifest_path: Option<&Path>) -> PathBuf { + let start = if let Some(p) = manifest_path { + let abs = std::path::absolute(p).unwrap_or_else(|_| p.to_path_buf()); + abs.parent().map(Path::to_path_buf).unwrap_or(abs) + } else { + std::env::current_dir().unwrap_or_else(|_| PathBuf::from(".")) + }; + + let mut p = start.clone(); + loop { + if p.join(".git").exists() { + return p; + } + if !p.pop() { + break; + } + } + + std::env::current_dir().unwrap_or(start) +} + +/// Whether `source_root` is inside a git work tree. +pub(crate) fn is_git_repo(source_root: &Path) -> bool { + Command::new("git") + .arg("-C") + .arg(source_root) + .arg("rev-parse") + .arg("--is-inside-work-tree") + .output() + .is_ok_and(|o| o.status.success()) +} + +/// Whether `source_root` is a git work tree with uncommitted changes. Returns +/// `Ok(false)` when it isn't a git repo (git ran but refused) — callers can't +/// verify cleanliness there, so they proceed. Errors only when git can't be +/// invoked at all. +pub(crate) fn tree_is_dirty(source_root: &Path) -> Result { + let status = Command::new("git") + .arg("-C") + .arg(source_root) + .arg("status") + .arg("--porcelain") + .output() + .map_err(|source| Error::GitInvoke { + path: source_root.to_path_buf(), + source, + })?; + + // Not a git repo (or git refused): can't verify cleanliness, proceed. + if !status.status.success() { + return Ok(false); + } + + Ok(!status.stdout.is_empty()) +} + +/// Produce the gzipped source tarball bytes. Entries are rooted under a +/// top-level `source/` prefix. In a git repo this is `git archive HEAD` (the +/// committed tree); otherwise the working directory is walked and tarred, +/// skipping `ARCHIVE_DENYLIST` entries. +/// +/// When the source isn't a git repo, `warn_non_git` controls whether to warn +/// that the working directory is being archived. Callers that only inspect the +/// result (e.g. `contract archive --dry-run`) pass `false`, since the listing +/// itself reveals the contents. +pub(crate) fn build_source_archive( + source_root: &Path, + print: &Print, + warn_non_git: bool, +) -> Result, Error> { + let tar = if is_git_repo(source_root) { + git_archive_tar(source_root)? + } else { + if warn_non_git { + print.warnln(format!( + "{} is not a git repository; archiving the working directory. Inspect the generated archive to confirm its contents.", + source_root.display(), + )); + } + walk_tar(source_root)? + }; + gzip(&tar) +} + +/// Tar entry paths inside the gzipped archive bytes, in archive order. Used by +/// `contract archive --dry-run` to list exactly what the bytes that hash to +/// `source_sha256` contain. +pub(crate) fn entry_names(bytes: &[u8]) -> Result, Error> { + let dec = flate2::read::GzDecoder::new(bytes); + let mut archive = tar::Archive::new(dec); + let mut names = Vec::new(); + for entry in archive.entries().map_err(Error::ArchiveExtract)? { + let entry = entry.map_err(Error::ArchiveExtract)?; + let path = entry.path().map_err(Error::ArchiveExtract)?; + names.push(path.to_string_lossy().into_owned()); + } + Ok(names) +} + +/// `git archive --format=tar --prefix=source/ HEAD`, returning the tar bytes. +fn git_archive_tar(source_root: &Path) -> Result, Error> { + let out = Command::new("git") + .arg("-C") + .arg(source_root) + .arg("archive") + .arg("--format=tar") + .arg("--prefix=source/") + .arg("HEAD") + .output() + .map_err(|source| Error::GitInvoke { + path: source_root.to_path_buf(), + source, + })?; + if !out.status.success() { + return Err(Error::GitArchive { + path: source_root.to_path_buf(), + stderr: String::from_utf8_lossy(&out.stderr).trim().to_string(), + }); + } + Ok(out.stdout) +} + +/// Tar the working tree under `source_root`, skipping denylisted path +/// components. Each entry is prefixed with `source/`. +/// +/// The output is reproducible, following GNU tar's reproducibility guidance +/// () +/// with the portable equivalents available via the `tar` crate (the system +/// `tar` can't be relied on — macOS ships bsdtar, which lacks `--sort`, +/// `--mtime`, `--pax-option`, …): entries are sorted by name (`--sort=name`) +/// using locale-independent path ordering (`LC_ALL=C`), and `HeaderMode::Deterministic` +/// zeroes mtime (`--mtime`/`--clamp-mtime`), sets uid/gid to 0 with empty owner +/// names (`--owner=0 --group=0 --numeric-owner`), and normalizes mode +/// (`--mode=go+u,go-w`). ustar headers carry no atime/ctime or tar PID. The gzip +/// wrapper (see `gzip`) is likewise deterministic. +fn walk_tar(source_root: &Path) -> Result, Error> { + let mut files: Vec = Vec::new(); + let walk = WalkDir::new(source_root) + .sort_by_file_name() + .into_iter() + .filter_entry(|e| !is_denylisted(e.file_name())); + for entry in walk { + let entry = entry.map_err(|e| Error::ArchiveWrite { + path: source_root.to_path_buf(), + source: e.into(), + })?; + if entry.file_type().is_file() { + files.push(entry.path().to_path_buf()); + } + } + files.sort(); + + let mut builder = tar::Builder::new(Vec::new()); + builder.mode(tar::HeaderMode::Deterministic); + for path in &files { + let rel = path.strip_prefix(source_root).unwrap_or(path); + let name = Path::new("source").join(rel); + let mut f = std::fs::File::open(path).map_err(|source| Error::ArchiveWrite { + path: path.clone(), + source, + })?; + builder + .append_file(&name, &mut f) + .map_err(|source| Error::ArchiveWrite { + path: path.clone(), + source, + })?; + } + builder.into_inner().map_err(|source| Error::ArchiveWrite { + path: source_root.to_path_buf(), + source, + }) +} + +/// A path component is denylisted if it equals a denylist entry, or — for +/// dotted entries, which double as extension filters (e.g. `.swp`, `.log`) — if +/// it ends with that entry. Plain names (`target`, `node_modules`) match +/// exactly only, so `mytarget` is not excluded. +fn is_denylisted(name: &std::ffi::OsStr) -> bool { + let name = name.to_string_lossy(); + ARCHIVE_DENYLIST + .iter() + .any(|d| name == *d || (d.starts_with('.') && name.ends_with(d))) +} + +/// Gzip with a default (mtime-zeroed) header so the same tar bytes always hash +/// the same. +fn gzip(bytes: &[u8]) -> Result, Error> { + let mut enc = flate2::write::GzEncoder::new(Vec::new(), flate2::Compression::default()); + enc.write_all(bytes).map_err(|source| Error::ArchiveWrite { + path: PathBuf::new(), + source, + })?; + enc.finish().map_err(|source| Error::ArchiveWrite { + path: PathBuf::new(), + source, + }) +} + +/// Decompress gzip and unpack the tar into `dest`. Entries are `source/…`, so +/// they land at `/source/…`. +pub(crate) fn unpack_targz(bytes: &[u8], dest: &Path) -> Result<(), Error> { + let dec = flate2::read::GzDecoder::new(bytes); + tar::Archive::new(dec) + .unpack(dest) + .map_err(Error::ArchiveExtract) +} + +#[cfg(test)] +mod tests { + use super::*; + use crate::config::locator::enforce_hardened_tree; + use sha2::{Digest, Sha256}; + + #[test] + fn is_denylisted_matches_names_and_dotted_suffixes() { + use std::ffi::OsStr; + // exact name matches + assert!(is_denylisted(OsStr::new("target"))); + assert!(is_denylisted(OsStr::new(".git"))); + assert!(is_denylisted(OsStr::new(".gitignore"))); + assert!(is_denylisted(OsStr::new(".env"))); + assert!(is_denylisted(OsStr::new(".DS_Store"))); + // plain names match exactly only + assert!(!is_denylisted(OsStr::new("mytarget"))); + assert!(!is_denylisted(OsStr::new("targets"))); + // dotted entries also match as suffix (extension-style) + assert!(is_denylisted(OsStr::new("backup.git"))); + // unrelated files pass through + assert!(!is_denylisted(OsStr::new("Cargo.toml"))); + assert!(!is_denylisted(OsStr::new("lib.rs"))); + } + + // Initialize a git repo at `root` with one commit of everything present. + #[cfg(unix)] + fn git_init_commit(root: &Path) { + for args in [ + &["init", "-q", "-b", "main"][..], + &["add", "-A"][..], + &["commit", "-q", "-m", "init"][..], + ] { + let ok = Command::new("git") + .arg("-C") + .arg(root) + .args(args) + .env("GIT_AUTHOR_NAME", "T") + .env("GIT_AUTHOR_EMAIL", "t@e.x") + .env("GIT_COMMITTER_NAME", "T") + .env("GIT_COMMITTER_EMAIL", "t@e.x") + .status() + .unwrap() + .success(); + assert!(ok); + } + } + + #[test] + #[cfg(unix)] + fn build_source_archive_git_is_prefixed_and_deterministic() { + use std::os::unix::fs::PermissionsExt; + let print = Print::new(true); + let temp = tempfile::TempDir::new().unwrap(); + let root = temp.path(); + std::fs::write(root.join("Cargo.toml"), b"# crate").unwrap(); + std::fs::create_dir_all(root.join("src")).unwrap(); + std::fs::write(root.join("src/lib.rs"), b"// code").unwrap(); + git_init_commit(root); + + let a = build_source_archive(root, &print, true).unwrap(); + let b = build_source_archive(root, &print, true).unwrap(); + assert!(!a.is_empty()); + assert_eq!(a, b, "same commit should produce identical bytes"); + + let sha = hex::encode(Sha256::digest(&a)); + assert_eq!(sha.len(), 64); + + // The listing reflects exactly the archived entries. + let names = entry_names(&a).unwrap(); + assert!(names.iter().any(|n| n == "source/Cargo.toml")); + assert!(names.iter().any(|n| n == "source/src/lib.rs")); + + // Unpack and confirm the `source/` prefix + hardened perms. + let dest = tempfile::TempDir::new().unwrap(); + unpack_targz(&a, dest.path()).unwrap(); + assert!(dest.path().join("source/Cargo.toml").exists()); + assert!(dest.path().join("source/src/lib.rs").exists()); + + enforce_hardened_tree(dest.path()).unwrap(); + let file_mode = std::fs::metadata(dest.path().join("source/Cargo.toml")) + .unwrap() + .permissions() + .mode() + & 0o777; + let dir_mode = std::fs::metadata(dest.path().join("source")) + .unwrap() + .permissions() + .mode() + & 0o777; + assert_eq!(file_mode, 0o600); + assert_eq!(dir_mode, 0o700); + } + + #[test] + fn build_source_archive_non_git_excludes_denylist() { + let print = Print::new(true); + let temp = tempfile::TempDir::new().unwrap(); + let root = temp.path(); + std::fs::write(root.join("Cargo.toml"), b"# crate").unwrap(); + std::fs::create_dir_all(root.join("src")).unwrap(); + std::fs::write(root.join("src/lib.rs"), b"// code").unwrap(); + // Planted dirs that must be excluded. + std::fs::create_dir_all(root.join("target/debug")).unwrap(); + std::fs::write(root.join("target/debug/x"), b"junk").unwrap(); + std::fs::create_dir_all(root.join(".git")).unwrap(); + std::fs::write(root.join(".git/config"), b"junk").unwrap(); + + let bytes = build_source_archive(root, &print, true).unwrap(); + let dest = tempfile::TempDir::new().unwrap(); + unpack_targz(&bytes, dest.path()).unwrap(); + + assert!(dest.path().join("source/Cargo.toml").exists()); + assert!(dest.path().join("source/src/lib.rs").exists()); + assert!(!dest.path().join("source/target").exists()); + assert!(!dest.path().join("source/.git").exists()); + assert_eq!(hex::encode(Sha256::digest(&bytes)).len(), 64); + + // Reproducible: a second run over the same tree yields identical bytes + // (sorted entries + zeroed header fields + deterministic gzip). + let again = build_source_archive(root, &print, true).unwrap(); + assert_eq!(bytes, again); + } + + #[test] + fn resolve_source_root_finds_git_root_from_subdir() { + let temp = tempfile::TempDir::new().unwrap(); + let root = temp.path(); + std::fs::create_dir_all(root.join(".git")).unwrap(); + let nested = root.join("contracts").join("foo"); + std::fs::create_dir_all(&nested).unwrap(); + std::fs::write(nested.join("Cargo.toml"), b"# placeholder").unwrap(); + + let manifest = nested.join("Cargo.toml"); + // Use canonicalize on both sides — `tempfile` returns symlinked /var + // paths on macOS while resolve_source_root walks the same prefix. + let got = std::fs::canonicalize(resolve_source_root(Some(&manifest))).unwrap(); + let want = std::fs::canonicalize(root).unwrap(); + assert_eq!(got, want); + } + + #[test] + fn resolve_source_root_falls_back_to_cwd_without_git() { + let temp = tempfile::TempDir::new().unwrap(); + let root = temp.path(); + let nested = root.join("noisy"); + std::fs::create_dir_all(&nested).unwrap(); + std::fs::write(nested.join("Cargo.toml"), b"# placeholder").unwrap(); + + let manifest = nested.join("Cargo.toml"); + // No `.git` anywhere up the tree, so we fall back to cwd. We can't + // assert what cwd is in a test runner (it varies), but we can assert + // that the returned path doesn't have `.git`. That's enough to confirm + // fallback kicked in. + let got = resolve_source_root(Some(&manifest)); + assert!(!got.join(".git").exists()); + } +} diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index faa45b9213..75d89dad63 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -1,8 +1,4 @@ -use std::{ - io::Write, - path::{Path, PathBuf}, - process::Command, -}; +use std::path::{Path, PathBuf}; use bollard::{ models::ContainerCreateBody, @@ -19,7 +15,6 @@ use regex::Regex; use semver::Version; use serde::Deserialize; use sha2::{Digest, Sha256}; -use walkdir::WalkDir; use crate::{ commands::{container::shared::Error as ConnectionError, global}, @@ -27,43 +22,13 @@ use crate::{ print::Print, }; -use super::{BuiltContract, Cmd, WASM_TARGET}; +use super::{source_archive, BuiltContract, Cmd, WASM_TARGET}; const REGISTRY: &str = "docker.io/stellar/stellar-cli"; const HUB_TAGS_URL: &str = "https://hub.docker.com/v2/repositories/stellar/stellar-cli/tags/?page_size=100"; const RESERVED_META_KEYS: &[&str] = &["bldimg", "source_uri", "source_sha256", "bldopt"]; -/// Top-level names excluded when archiving a non-git working directory (we have -/// no tracked-files list to consult, so fall back to a fixed denylist of VCS -/// metadata, build/cache/transient dirs, and editor/OS/AI-assistant junk). -/// Matched against each path component, so a directory like `target/` prunes -/// its whole subtree. -const ARCHIVE_DENYLIST: &[&str] = &[ - // version control - ".git", - ".svn", - ".hg", - // build output / dependencies - "target", - "node_modules", - // transient - "log", - "logs", - "tmp", - "temp", - // OS / editor junk - ".DS_Store", - "Thumbs.db", - ".idea", - ".vscode", - // AI assistant dirs - ".claude", - ".cursor", - ".windsurf", - ".aider", -]; - /// First cli release that accepts `--optimize=false` as an explicit value /// (added by commit `b17d3f0b`). Containers older than this only accept bare /// `--optimize`; we probe the container's `stellar version --only-version` to @@ -101,11 +66,8 @@ pub enum Error { #[error("cargo metadata failed: {0}")] Metadata(#[from] cargo_metadata::Error), - #[error("could not read git state at {path}: {source}")] - GitInvoke { - path: PathBuf, - source: std::io::Error, - }, + #[error(transparent)] + SourceArchive(#[from] source_archive::Error), #[error( "git working tree at {path} is dirty. --verifiable requires a clean tree so the recorded source_sha256 matches the WASM bytes. Commit or stash your changes and try again." @@ -117,9 +79,6 @@ pub enum Error { )] ReservedMetaKey { key: String }, - #[error("--verifiable requires --source-sha256 (the SEP-58 source_sha256: 64-char hex SHA-256 of the source), or --archive to generate the source archive and compute it. --source-uri is optional.")] - MissingSourceSha256, - #[error("--source-sha256 value {value:?} does not match the SEP-58 source_sha256 format `^[0-9a-f]{{64}}$` (64-char lower-case hex).")] SourceSha256Format { value: String }, @@ -129,18 +88,6 @@ pub enum Error { #[error("--source-sha256 {provided} does not match the SHA-256 of the generated archive {computed}. Omit --source-sha256 to record the computed value, or fix the value.")] SourceSha256Mismatch { provided: String, computed: String }, - #[error("`git archive` failed in {path}: {stderr}")] - GitArchive { path: PathBuf, stderr: String }, - - #[error("could not write source archive to {path}: {source}")] - ArchiveWrite { - path: PathBuf, - source: std::io::Error, - }, - - #[error("could not extract source archive: {0}")] - ArchiveExtract(std::io::Error), - #[error(transparent)] Data(#[from] data::Error), @@ -174,42 +121,29 @@ pub async fn run( // gets bind-mounted into the container. We do NOT validate that it matches // source_uri — a wrong source produces different bytes, and verify catches // that at byte-comparison time. - let source_root = resolve_source_root(cmd); + let source_root = source_archive::resolve_source_root(cmd.manifest_path.as_deref()); // A dirty working tree would make the recorded source_sha256 fail to // describe the bytes actually built, so refuse to proceed. Skipped when // the source root isn't a git repo (we can't check, e.g. archive sources). enforce_clean_tree(&source_root)?; - // Resolve the recorded source_sha256 and the directory the container mounts - // at /source. With `--archive`, the CLI builds the source archive, records - // its hash, and builds from the *extracted* archive (in a hardened tempdir) - // so the WASM is produced from exactly the bytes that were hashed. Without - // it, the user supplies --source-sha256 and we mount the working tree. - let resolved = match &cmd.archive { - Some(_) => { - let a = resolve_archive(cmd, &source_root, print)?; - // The extracted `source/` dir mirrors `source_root` exactly and is - // both the container mount and the tree the build writes `target/` - // into, so it's what `collect_built_contracts` resolves artifacts - // against. - let mount_root = a.extracted_root.join("source"); - ResolvedSource { - source_sha256: a.source_sha256, - extracted_root: Some(mount_root.clone()), - mount_root, - _tmp: Some(a.tmp), - } + // Always build the source archive, record its hash, and build from the + // *extracted* archive (in a hardened tempdir) so the WASM is produced from + // exactly the bytes that were hashed. A `--source-sha256` passed by the user + // is treated as a pin and validated against the computed hash. + let resolved = { + let a = resolve_archive(cmd, &source_root, print)?; + // The extracted `source/` dir mirrors `source_root` exactly and is both + // the container mount and the tree the build writes `target/` into, so + // it's what `collect_built_contracts` resolves artifacts against. + let mount_root = a.extracted_root.join("source"); + ResolvedSource { + source_sha256: a.source_sha256, + extracted_root: Some(mount_root.clone()), + mount_root, + _tmp: Some(a.tmp), } - None => ResolvedSource { - source_sha256: cmd - .source_sha256 - .clone() - .ok_or(Error::MissingSourceSha256)?, - mount_root: source_root.clone(), - extracted_root: None, - _tmp: None, - }, }; let source_ids = SourceIds { @@ -285,9 +219,9 @@ pub async fn run( collect_built_contracts(cmd, &source_root, resolved.extracted_root.as_deref(), print) } -/// The recorded `source_sha256`, the directory bind-mounted at `/source`, and -/// (when `--archive` is used) the extracted-archive root plus its tempdir guard -/// — held so the temp dir outlives the container build and artifact collection. +/// The recorded `source_sha256`, the directory bind-mounted at `/source`, the +/// extracted-archive root, and its tempdir guard — held so the temp dir +/// outlives the container build and artifact collection. struct ResolvedSource { source_sha256: String, mount_root: PathBuf, @@ -305,34 +239,6 @@ fn resolve_workspace_root(cmd: &Cmd) -> Result { Ok(md.workspace_root.into_std_path_buf()) } -/// Pick the anchor for the container bind-mount and for relativizing -/// `--manifest-path` into the recorded `bldopt`. Walk up from the user's -/// `--manifest-path` (or cwd, if no manifest_path) looking for a `.git` -/// directory; return its parent. If none is found, fall back to cwd. -/// -/// This isn't a validation step — any `.git` will do. Wrong-source mistakes -/// are caught later by the verify-side byte comparison. -fn resolve_source_root(cmd: &Cmd) -> PathBuf { - let start = if let Some(p) = &cmd.manifest_path { - let abs = std::path::absolute(p).unwrap_or_else(|_| p.clone()); - abs.parent().map(Path::to_path_buf).unwrap_or(abs) - } else { - std::env::current_dir().unwrap_or_else(|_| PathBuf::from(".")) - }; - - let mut p = start.clone(); - loop { - if p.join(".git").exists() { - return p; - } - if !p.pop() { - break; - } - } - - std::env::current_dir().unwrap_or(start) -} - /// Source-identification fields recorded as SEP-58 meta. `source_sha256` is /// always `Some` by the time these are built in `run()` (resolved from /// `--source-sha256` or computed from the generated archive). `source_uri` is @@ -343,8 +249,9 @@ struct SourceIds { source_sha256: Option, } -/// Format-validate the user-supplied source flags. Requiredness is enforced in -/// `run()` (it depends on whether `--archive` is used), not here. +/// Format-validate the user-supplied source flags. Both are optional under +/// `--verifiable`; `--source-sha256`, when present, is validated as a pin in +/// `resolve_archive`. fn validate_source_formats(cmd: &Cmd) -> Result<(), Error> { if let Some(sha) = &cmd.source_sha256 { if !source_sha256_regex().is_match(sha) { @@ -359,7 +266,7 @@ fn validate_source_formats(cmd: &Cmd) -> Result<(), Error> { Ok(()) } -/// Outcome of `--archive`: the generated archive's SHA-256 and the directory it +/// Outcome of archiving: the generated archive's SHA-256 and the directory it /// was extracted into (held alive by `tmp`). struct ArchiveResult { source_sha256: String, @@ -367,10 +274,12 @@ struct ArchiveResult { tmp: tempfile::TempDir, } -/// Build the source archive, record its hash, write it out, and extract it into -/// a permission-hardened tempdir that the container then builds from. +/// Build the source archive, record its hash, write it to the managed archives +/// dir (content-addressed, so the bytes are available to upload for +/// `--source-uri`), and extract it into a permission-hardened tempdir that the +/// container then builds from. fn resolve_archive(cmd: &Cmd, source_root: &Path, print: &Print) -> Result { - let bytes = build_source_archive(source_root, print)?; + let bytes = source_archive::build_source_archive(source_root, print, true)?; let computed = hex::encode(Sha256::digest(&bytes)); // If the user pinned a hash, it must match what we produced. @@ -383,20 +292,15 @@ fn resolve_archive(cmd: &Cmd, source_root: &Path, print: &Print) -> Result p.clone(), - Some(None) => data::archives_dir()?.join(format!("{computed}.tar.gz")), - None => unreachable!("resolve_archive is only called when --archive is set"), - }; + // Content-addressed name under the managed archives dir. + let out_path = data::archives_dir()?.join(format!("{computed}.tar.gz")); if let Some(parent) = out_path.parent() { - std::fs::create_dir_all(parent).map_err(|source| Error::ArchiveWrite { + std::fs::create_dir_all(parent).map_err(|source| source_archive::Error::ArchiveWrite { path: out_path.clone(), source, })?; } - std::fs::write(&out_path, &bytes).map_err(|source| Error::ArchiveWrite { + std::fs::write(&out_path, &bytes).map_err(|source| source_archive::Error::ArchiveWrite { path: out_path.clone(), source, })?; @@ -413,16 +317,16 @@ fn resolve_archive(cmd: &Cmd, source_root: &Path, print: &Print) -> Result Result bool { - Command::new("git") - .arg("-C") - .arg(source_root) - .arg("rev-parse") - .arg("--is-inside-work-tree") - .output() - .is_ok_and(|o| o.status.success()) -} - -/// Produce the gzipped source tarball bytes. Entries are rooted under a -/// top-level `source/` prefix (so the archive extracts to a `source/` dir, -/// mirroring the container's `/source` mount). In a git repo this is `git -/// archive HEAD` (the committed tree); otherwise the working directory is -/// walked and tarred, skipping `ARCHIVE_DENYLIST` entries, after warning. -fn build_source_archive(source_root: &Path, print: &Print) -> Result, Error> { - let tar = if is_git_repo(source_root) { - git_archive_tar(source_root)? - } else { - print.warnln(format!( - "{} is not a git repository; archiving the working directory. Inspect the generated archive to confirm its contents.", - source_root.display(), - )); - walk_tar(source_root)? - }; - gzip(&tar) -} - -/// `git archive --format=tar --prefix=source/ HEAD`, returning the tar bytes. -fn git_archive_tar(source_root: &Path) -> Result, Error> { - let out = Command::new("git") - .arg("-C") - .arg(source_root) - .arg("archive") - .arg("--format=tar") - .arg("--prefix=source/") - .arg("HEAD") - .output() - .map_err(|source| Error::GitInvoke { - path: source_root.to_path_buf(), - source, - })?; - if !out.status.success() { - return Err(Error::GitArchive { - path: source_root.to_path_buf(), - stderr: String::from_utf8_lossy(&out.stderr).trim().to_string(), - }); - } - Ok(out.stdout) -} - -/// Tar the working tree under `source_root`, skipping denylisted path -/// components. Each entry is prefixed with `source/`. -/// -/// The output is reproducible, following GNU tar's reproducibility guidance -/// () -/// with the portable equivalents available via the `tar` crate (the system -/// `tar` can't be relied on — macOS ships bsdtar, which lacks `--sort`, -/// `--mtime`, `--pax-option`, …): entries are sorted by name (`--sort=name`) -/// using locale-independent path ordering (`LC_ALL=C`), and `HeaderMode::Deterministic` -/// zeroes mtime (`--mtime`/`--clamp-mtime`), sets uid/gid to 0 with empty owner -/// names (`--owner=0 --group=0 --numeric-owner`), and normalizes mode -/// (`--mode=go+u,go-w`). ustar headers carry no atime/ctime or tar PID. The gzip -/// wrapper (see `gzip`) is likewise deterministic. -fn walk_tar(source_root: &Path) -> Result, Error> { - let mut files: Vec = Vec::new(); - let walk = WalkDir::new(source_root) - .sort_by_file_name() - .into_iter() - .filter_entry(|e| !is_denylisted(e.file_name())); - for entry in walk { - let entry = entry.map_err(|e| Error::ArchiveWrite { - path: source_root.to_path_buf(), - source: e.into(), - })?; - if entry.file_type().is_file() { - files.push(entry.path().to_path_buf()); - } - } - files.sort(); - - let mut builder = tar::Builder::new(Vec::new()); - builder.mode(tar::HeaderMode::Deterministic); - for path in &files { - let rel = path.strip_prefix(source_root).unwrap_or(path); - let name = Path::new("source").join(rel); - let mut f = std::fs::File::open(path).map_err(|source| Error::ArchiveWrite { - path: path.clone(), - source, - })?; - builder - .append_file(&name, &mut f) - .map_err(|source| Error::ArchiveWrite { - path: path.clone(), - source, - })?; - } - builder.into_inner().map_err(|source| Error::ArchiveWrite { - path: source_root.to_path_buf(), - source, - }) -} - -/// A path component is denylisted if it equals a denylist entry, or — for -/// dotted entries, which double as extension filters (e.g. `.swp`, `.log`) — if -/// it ends with that entry. Plain names (`target`, `node_modules`) match -/// exactly only, so `mytarget` is not excluded. -fn is_denylisted(name: &std::ffi::OsStr) -> bool { - let name = name.to_string_lossy(); - ARCHIVE_DENYLIST - .iter() - .any(|d| name == *d || (d.starts_with('.') && name.ends_with(d))) -} - -/// Gzip with a default (mtime-zeroed) header so the same tar bytes always hash -/// the same. -fn gzip(bytes: &[u8]) -> Result, Error> { - let mut enc = flate2::write::GzEncoder::new(Vec::new(), flate2::Compression::default()); - enc.write_all(bytes).map_err(|source| Error::ArchiveWrite { - path: PathBuf::new(), - source, - })?; - enc.finish().map_err(|source| Error::ArchiveWrite { - path: PathBuf::new(), - source, - }) -} - -/// Decompress gzip and unpack the tar into `dest`. Entries are `source/…`, so -/// they land at `/source/…`. -fn unpack_targz(bytes: &[u8], dest: &Path) -> Result<(), Error> { - let dec = flate2::read::GzDecoder::new(bytes); - tar::Archive::new(dec) - .unpack(dest) - .map_err(Error::ArchiveExtract) -} - /// Refuse to run a verifiable build against a dirty git working tree: the /// bind-mounted source must match the recorded source_sha256 for the build to /// be reproducible. When the source root isn't a git repo (e.g. an extracted /// archive) we can't check, so we proceed — the user owns the source_sha256 /// they pass, and verify catches a mismatch at byte-comparison time. fn enforce_clean_tree(source_root: &Path) -> Result<(), Error> { - let status = Command::new("git") - .arg("-C") - .arg(source_root) - .arg("status") - .arg("--porcelain") - .output() - .map_err(|e| Error::GitInvoke { - path: source_root.to_path_buf(), - source: e, - })?; - - // Not a git repo (or git refused): can't verify cleanliness, proceed. - if !status.status.success() { - return Ok(()); - } - - if !status.stdout.is_empty() { + if source_archive::tree_is_dirty(source_root)? { return Err(Error::GitDirty { path: source_root.to_path_buf(), }); } - Ok(()) } @@ -1388,117 +1137,6 @@ mod tests { validate_source_formats(&cmd).unwrap(); } - #[test] - fn is_denylisted_matches_names_and_dotted_suffixes() { - use std::ffi::OsStr; - // exact name matches - assert!(is_denylisted(OsStr::new("target"))); - assert!(is_denylisted(OsStr::new(".git"))); - assert!(is_denylisted(OsStr::new(".DS_Store"))); - // plain names match exactly only - assert!(!is_denylisted(OsStr::new("mytarget"))); - assert!(!is_denylisted(OsStr::new("targets"))); - // dotted entries also match as suffix (extension-style) - assert!(is_denylisted(OsStr::new("backup.git"))); - // unrelated files pass through - assert!(!is_denylisted(OsStr::new("Cargo.toml"))); - assert!(!is_denylisted(OsStr::new("lib.rs"))); - } - - // Initialize a git repo at `root` with one commit of everything present. - #[cfg(unix)] - fn git_init_commit(root: &Path) { - for args in [ - &["init", "-q", "-b", "main"][..], - &["add", "-A"][..], - &["commit", "-q", "-m", "init"][..], - ] { - let ok = Command::new("git") - .arg("-C") - .arg(root) - .args(args) - .env("GIT_AUTHOR_NAME", "T") - .env("GIT_AUTHOR_EMAIL", "t@e.x") - .env("GIT_COMMITTER_NAME", "T") - .env("GIT_COMMITTER_EMAIL", "t@e.x") - .status() - .unwrap() - .success(); - assert!(ok); - } - } - - #[test] - #[cfg(unix)] - fn build_source_archive_git_is_prefixed_and_deterministic() { - use std::os::unix::fs::PermissionsExt; - let print = Print::new(true); - let temp = tempfile::TempDir::new().unwrap(); - let root = temp.path(); - std::fs::write(root.join("Cargo.toml"), b"# crate").unwrap(); - std::fs::create_dir_all(root.join("src")).unwrap(); - std::fs::write(root.join("src/lib.rs"), b"// code").unwrap(); - git_init_commit(root); - - let a = build_source_archive(root, &print).unwrap(); - let b = build_source_archive(root, &print).unwrap(); - assert!(!a.is_empty()); - assert_eq!(a, b, "same commit should produce identical bytes"); - - let sha = hex::encode(Sha256::digest(&a)); - assert_eq!(sha.len(), 64); - - // Unpack and confirm the `source/` prefix + hardened perms. - let dest = tempfile::TempDir::new().unwrap(); - unpack_targz(&a, dest.path()).unwrap(); - assert!(dest.path().join("source/Cargo.toml").exists()); - assert!(dest.path().join("source/src/lib.rs").exists()); - - enforce_hardened_tree(dest.path()).unwrap(); - let file_mode = std::fs::metadata(dest.path().join("source/Cargo.toml")) - .unwrap() - .permissions() - .mode() - & 0o777; - let dir_mode = std::fs::metadata(dest.path().join("source")) - .unwrap() - .permissions() - .mode() - & 0o777; - assert_eq!(file_mode, 0o600); - assert_eq!(dir_mode, 0o700); - } - - #[test] - fn build_source_archive_non_git_excludes_denylist() { - let print = Print::new(true); - let temp = tempfile::TempDir::new().unwrap(); - let root = temp.path(); - std::fs::write(root.join("Cargo.toml"), b"# crate").unwrap(); - std::fs::create_dir_all(root.join("src")).unwrap(); - std::fs::write(root.join("src/lib.rs"), b"// code").unwrap(); - // Planted dirs that must be excluded. - std::fs::create_dir_all(root.join("target/debug")).unwrap(); - std::fs::write(root.join("target/debug/x"), b"junk").unwrap(); - std::fs::create_dir_all(root.join(".git")).unwrap(); - std::fs::write(root.join(".git/config"), b"junk").unwrap(); - - let bytes = build_source_archive(root, &print).unwrap(); - let dest = tempfile::TempDir::new().unwrap(); - unpack_targz(&bytes, dest.path()).unwrap(); - - assert!(dest.path().join("source/Cargo.toml").exists()); - assert!(dest.path().join("source/src/lib.rs").exists()); - assert!(!dest.path().join("source/target").exists()); - assert!(!dest.path().join("source/.git").exists()); - assert_eq!(hex::encode(Sha256::digest(&bytes)).len(), 64); - - // Reproducible: a second run over the same tree yields identical bytes - // (sorted entries + zeroed header fields + deterministic gzip). - let again = build_source_archive(root, &print).unwrap(); - assert_eq!(bytes, again); - } - #[test] fn bldimg_regex_accepts_docker_hub_full_ref() { assert!(bldimg_regex().is_match(&format!( @@ -1545,46 +1183,6 @@ mod tests { assert!(!source_uri_regex().is_match("https://has space")); // whitespace } - #[test] - fn resolve_source_root_finds_git_root_from_subdir() { - let temp = tempfile::TempDir::new().unwrap(); - let root = temp.path(); - std::fs::create_dir_all(root.join(".git")).unwrap(); - let nested = root.join("contracts").join("foo"); - std::fs::create_dir_all(&nested).unwrap(); - std::fs::write(nested.join("Cargo.toml"), b"# placeholder").unwrap(); - - let cmd = Cmd { - manifest_path: Some(nested.join("Cargo.toml")), - ..Cmd::default() - }; - // Use canonicalize on both sides — `tempfile` returns symlinked /var - // paths on macOS while resolve_source_root walks the same prefix. - let got = std::fs::canonicalize(resolve_source_root(&cmd)).unwrap(); - let want = std::fs::canonicalize(root).unwrap(); - assert_eq!(got, want); - } - - #[test] - fn resolve_source_root_falls_back_to_cwd_without_git() { - let temp = tempfile::TempDir::new().unwrap(); - let root = temp.path(); - let nested = root.join("noisy"); - std::fs::create_dir_all(&nested).unwrap(); - std::fs::write(nested.join("Cargo.toml"), b"# placeholder").unwrap(); - - let cmd = Cmd { - manifest_path: Some(nested.join("Cargo.toml")), - ..Cmd::default() - }; - // No `.git` anywhere up the tree, so we fall back to cwd. We can't - // assert what cwd is in a test runner (it varies), but we can assert - // that the returned path doesn't contain the manifest's parent and - // doesn't have `.git`. That's enough to confirm fallback kicked in. - let got = resolve_source_root(&cmd); - assert!(!got.join(".git").exists()); - } - #[test] fn compose_container_args_prefixes_subcommand() { let composed = compose_container_args( diff --git a/cmd/soroban-cli/src/commands/contract/fetch.rs b/cmd/soroban-cli/src/commands/contract/fetch.rs index a02bf98869..2905b8e375 100644 --- a/cmd/soroban-cli/src/commands/contract/fetch.rs +++ b/cmd/soroban-cli/src/commands/contract/fetch.rs @@ -22,14 +22,18 @@ pub struct Cmd { /// Contract ID to fetch #[arg(long = "id", env = "STELLAR_CONTRACT_ID")] pub contract_id: Option, + /// Wasm to fetch #[arg(long = "wasm-hash", conflicts_with = "contract_id")] pub wasm_hash: Option, + /// Where to write output otherwise stdout is used #[arg(long, short = 'o')] pub out_file: Option, + #[command(flatten)] pub locator: locator::Args, + #[command(flatten)] pub network: network::Args, } diff --git a/cmd/soroban-cli/src/commands/contract/mod.rs b/cmd/soroban-cli/src/commands/contract/mod.rs index fc4499c029..ee140be938 100644 --- a/cmd/soroban-cli/src/commands/contract/mod.rs +++ b/cmd/soroban-cli/src/commands/contract/mod.rs @@ -1,4 +1,5 @@ pub mod alias; +pub mod archive; pub mod arg_parsing; pub mod asset; pub mod bindings; @@ -33,6 +34,9 @@ pub enum Cmd { #[command(subcommand)] Bindings(bindings::Cmd), + /// Generate the reproducible source archive used by verifiable builds + Archive(archive::Cmd), + Build(build::Cmd), /// Extend the time to live ledger of a contract-data ledger entry. @@ -113,6 +117,9 @@ pub enum Error { #[error(transparent)] Bindings(#[from] bindings::Error), + #[error(transparent)] + Archive(#[from] archive::Error), + #[error(transparent)] Build(#[from] build::Error), @@ -163,6 +170,7 @@ impl Cmd { match &self { Cmd::Asset(asset) => asset.run(global_args).await?, Cmd::Bindings(bindings) => bindings.run().await?, + Cmd::Archive(archive) => archive.run(global_args)?, Cmd::Build(build) => { build.run(global_args).await?; } From 9892d6eea566883b5686919b14d490336dd42af1 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Wed, 17 Jun 2026 11:34:22 -0700 Subject: [PATCH 19/25] Add --env to set build environment variables. --- FULL_HELP_DOCS.md | 4 + cmd/crates/soroban-test/tests/it/build.rs | 41 ++++ .../src/commands/contract/build.rs | 95 +++++++++ .../src/commands/contract/build/verifiable.rs | 181 ++++++++++++++++-- 4 files changed, 304 insertions(+), 17 deletions(-) diff --git a/FULL_HELP_DOCS.md b/FULL_HELP_DOCS.md index d48ab89939..b5a68062d5 100644 --- a/FULL_HELP_DOCS.md +++ b/FULL_HELP_DOCS.md @@ -397,6 +397,7 @@ To view the commands that will be executed, without executing them, use the --pr If ommitted, wasm files are written only to the cargo target directory. - `--locked` — Assert that `Cargo.lock` will remain unchanged +- `--env ` — Set an environment variable for the build (repeatable), e.g. `--env NAME=VALUE`. It's set on the build process; for a verifiable build it's passed to the container and recorded as a `bldopt`, so avoid secrets there - `--optimize ` — Optimize the generated wasm. Enabled by default; pass `--optimize=false` to disable. Requires the `additional-libs` feature Default value: `true` @@ -502,6 +503,7 @@ Deploy a wasm contract Default value: `false` - `--alias ` — The alias that will be used to save the contract's id. Whenever used, `--alias` will always overwrite the existing contract id configuration without asking for confirmation +- `--env ` — Set an environment variable for the build (repeatable), e.g. `--env NAME=VALUE`. It's set on the build process; for a verifiable build it's passed to the container and recorded as a `bldopt`, so avoid secrets there - `--optimize ` — Optimize the generated wasm. Enabled by default; pass `--optimize=false` to disable. Requires the `additional-libs` feature Default value: `true` @@ -876,6 +878,7 @@ Install a WASM file to the ledger without creating a contract instance Default value: `false` +- `--env ` — Set an environment variable for the build (repeatable), e.g. `--env NAME=VALUE`. It's set on the build process; for a verifiable build it's passed to the container and recorded as a `bldopt`, so avoid secrets there - `--optimize ` — Optimize the generated wasm. Enabled by default; pass `--optimize=false` to disable. Requires the `additional-libs` feature Default value: `true` @@ -939,6 +942,7 @@ Install a WASM file to the ledger without creating a contract instance Default value: `false` +- `--env ` — Set an environment variable for the build (repeatable), e.g. `--env NAME=VALUE`. It's set on the build process; for a verifiable build it's passed to the container and recorded as a `bldopt`, so avoid secrets there - `--optimize ` — Optimize the generated wasm. Enabled by default; pass `--optimize=false` to disable. Requires the `additional-libs` feature Default value: `true` diff --git a/cmd/crates/soroban-test/tests/it/build.rs b/cmd/crates/soroban-test/tests/it/build.rs index dfe798e982..df4f395868 100644 --- a/cmd/crates/soroban-test/tests/it/build.rs +++ b/cmd/crates/soroban-test/tests/it/build.rs @@ -69,6 +69,47 @@ fn build_package_by_current_dir() { )); } +// `--env` is repeatable and sets env vars on the local cargo process; they +// surface in the printed command in --print-commands-only. +#[test] +fn build_with_env_vars() { + let sandbox = TestEnv::default(); + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); + sandbox + .new_assert_cmd("contract") + .current_dir(fixture_path) + .arg("build") + .arg("--print-commands-only") + .arg("--env") + .arg("FOO=bar") + .arg("--env") + .arg("BAZ=qux") + .assert() + .success() + .stdout(predicate::str::contains("FOO=bar").and(predicate::str::contains("BAZ=qux"))); +} + +// An invalid `--env` name is rejected before building. +#[test] +fn build_rejects_invalid_env_name() { + let sandbox = TestEnv::default(); + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); + sandbox + .new_assert_cmd("contract") + .current_dir(fixture_path) + .arg("build") + .arg("--print-commands-only") + .arg("--env") + .arg("1FOO=bar") + .assert() + .failure() + .stderr(predicate::str::contains( + "not a valid environment variable name", + )); +} + #[test] fn build_with_locked() { let sandbox = TestEnv::default(); diff --git a/cmd/soroban-cli/src/commands/contract/build.rs b/cmd/soroban-cli/src/commands/contract/build.rs index e8fb9c0c4b..56178b4742 100644 --- a/cmd/soroban-cli/src/commands/contract/build.rs +++ b/cmd/soroban-cli/src/commands/contract/build.rs @@ -146,6 +146,18 @@ pub struct BuildArgs { #[arg(long, num_args=1, value_parser=parse_meta_arg, action=clap::ArgAction::Append, help_heading = "Metadata")] pub meta: Vec<(String, String)>, + /// Set an environment variable for the build (repeatable), e.g. + /// `--env NAME=VALUE`. It's set on the build process; for a verifiable build + /// it's passed to the container and recorded as a `bldopt`, so avoid secrets + /// there. + #[arg( + long = "env", + num_args = 1, + value_parser = parse_env_arg, + action = clap::ArgAction::Append + )] + pub env: Vec<(String, String)>, + /// Optimize the generated wasm. Enabled by default; pass `--optimize=false` to disable. Requires the `additional-libs` feature. #[arg( long, @@ -163,6 +175,7 @@ impl Default for BuildArgs { fn default() -> Self { Self { meta: Vec::new(), + env: Vec::new(), optimize: true, } } @@ -179,6 +192,35 @@ pub fn parse_meta_arg(s: &str) -> Result<(String, String), Error> { Ok((key.to_string(), value.to_string())) } +/// Parse a `--env NAME=VALUE` argument. The name must be a valid environment +/// variable name (`[A-Za-z_][A-Za-z0-9_]*`, no surrounding whitespace); the +/// value is kept verbatim, since the shell has already resolved any quoting and +/// env values can carry significant whitespace. +pub fn parse_env_arg(s: &str) -> Result<(String, String), Error> { + let (name, value) = s + .split_once('=') + .ok_or_else(|| Error::EnvArg(format!("{s:?} must be in the form 'NAME=VALUE'")))?; + + if !is_valid_env_name(name) { + return Err(Error::EnvArg(format!( + "{name:?} is not a valid environment variable name (expected [A-Za-z_][A-Za-z0-9_]*)" + ))); + } + + Ok((name.to_string(), value.to_string())) +} + +/// Whether `name` is a valid environment variable name: a leading letter or +/// underscore followed by letters, digits, or underscores. +fn is_valid_env_name(name: &str) -> bool { + let mut chars = name.chars(); + match chars.next() { + Some(c) if c.is_ascii_alphabetic() || c == '_' => {} + _ => return false, + } + chars.all(|c| c.is_ascii_alphanumeric() || c == '_') +} + #[derive(thiserror::Error, Debug)] pub enum Error { #[error(transparent)] @@ -220,6 +262,9 @@ pub enum Error { #[error("invalid meta entry: {0}")] MetaArg(String), + #[error("invalid env entry: {0}")] + EnvArg(String), + #[error( "use a rust version other than 1.81, 1.82, 1.83 or 1.91.0 to build contracts (got {0})" )] @@ -348,6 +393,11 @@ impl Cmd { // optimization using markers. cmd.env("SOROBAN_SDK_BUILD_SYSTEM_SUPPORTS_SPEC_SHAKING_V2", "1"); + // User-supplied build env vars (--env NAME=VALUE). + for (name, value) in &self.build_args.env { + cmd.env(name, value); + } + let cmd_str = serialize_command(&cmd); if self.print_commands_only { @@ -911,4 +961,49 @@ mod tests { "shlex round-trip failed: {raw_arg:?} not found as a single token in {tokens:?}" ); } + + #[test] + fn parse_env_arg_parses_name_value() { + assert_eq!( + parse_env_arg("FOO=bar").unwrap(), + ("FOO".to_string(), "bar".to_string()) + ); + assert_eq!( + parse_env_arg("_FOO_BAR2=bar").unwrap(), + ("_FOO_BAR2".to_string(), "bar".to_string()) + ); + // Only the first `=` splits; the value keeps the rest verbatim. + assert_eq!( + parse_env_arg("FOO=a=b=c").unwrap(), + ("FOO".to_string(), "a=b=c".to_string()) + ); + // An empty value is allowed. + assert_eq!( + parse_env_arg("FOO=").unwrap(), + ("FOO".to_string(), String::new()) + ); + // The value is kept verbatim (the shell already handled quoting), so + // significant whitespace survives. + assert_eq!( + parse_env_arg("FOO= 1 ").unwrap(), + ("FOO".to_string(), " 1 ".to_string()) + ); + } + + #[test] + fn parse_env_arg_rejects_invalid() { + for bad in [ + "FOO", // no `=` + "=bar", // empty name + " FOO = 1 ", // whitespace in name + "1FOO=x", // leading digit + "FO-O=x", // invalid char + "FOO BAR=x", // space in name + ] { + assert!( + matches!(parse_env_arg(bad).unwrap_err(), Error::EnvArg(_)), + "expected {bad:?} to be rejected" + ); + } + } } diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 75d89dad63..17326fa5f2 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -204,10 +204,18 @@ pub async fn run( // `--verbose` because verifications are run as part of pipelines. All // per-package builds run in one container so the crates download, compiled // deps, and target/ are shared. + let env: Vec = cmd + .build_args + .env + .iter() + .map(|(name, value)| format!("{name}={value}")) + .collect(); + run_in_container( &image_ref, &resolved.mount_root, &container_cmds, + &env, &docker, print, true, @@ -415,12 +423,26 @@ fn build_forwarded_args( let mut forwarded: Vec = Vec::new(); let mut bldopts: Vec = Vec::new(); - let mut record = |arg: String| { - forwarded.push(arg.clone()); - bldopts.push(arg); + // Record a build option. `None` means a bare flag (`--locked`); `Some(v)` + // means `--flag=v`. The forwarded copy keeps the value raw (the container + // gets it as argv, and `compose_shell_command` re-escapes it for the + // multi-package `sh -c`); the bldopt copy shell-escapes the value once, here + // at the source, so every recorded option is valid shell on its own and no + // consumer has to split a flag from its value later. For `key=value` + // payloads (`--meta`, `--env`) the key goes in `key` (`--meta=home_domain`) + // and only the value is escaped, keeping `--env=B='nice value'` rather than + // `'--env=B=nice value'`. + let mut record = |key: &str, value: Option<&str>| { + if let Some(v) = value { + forwarded.push(format!("{key}={v}")); + bldopts.push(format!("{key}={}", shell_escape::escape(v.into()))); + } else { + forwarded.push(key.to_string()); + bldopts.push(key.to_string()); + } }; - record("--locked".to_string()); + record("--locked", None); if let Some(path) = &cmd.manifest_path { let abs = std::path::absolute(path).unwrap_or_else(|_| path.clone()); @@ -428,30 +450,28 @@ fn build_forwarded_args( .strip_prefix(workspace_root) .map(Path::to_path_buf) .unwrap_or(abs); - record(format!("--manifest-path={}", rel.display())); + record("--manifest-path", Some(rel.display().to_string().as_str())); } if cmd.profile != "release" { - record(format!("--profile={}", cmd.profile)); + record("--profile", Some(cmd.profile.as_str())); } if let Some(features) = &cmd.features { - record(format!("--features={features}")); + record("--features", Some(features.as_str())); } if cmd.all_features { - record("--all-features".to_string()); + record("--all-features", None); } if cmd.no_default_features { - record("--no-default-features".to_string()); + record("--no-default-features", None); } // Always pin the package when it can be resolved (explicit `--package`, or // a workspace that builds exactly one cdylib by default) so the recorded // bldopt stays reproducible even if workspace default members change later. if let Some(pkg) = package { - record(format!("--package={pkg}")); + record("--package", Some(pkg)); } for (k, v) in &cmd.build_args.meta { - // Use the `--meta=key=value` form so each option is a single token, - // matching how clap re-parses on the container side. - record(format!("--meta={k}={v}")); + record(&format!("--meta={k}"), Some(v.as_str())); } // `--optimize` true is recorded as a bare flag (universally accepted). @@ -459,9 +479,21 @@ fn build_forwarded_args( // (added in `b17d3f0b`); on older containers, false is the default and // we record/forward nothing — passing `--optimize=false` there would fail. if cmd.build_args.optimize { - record("--optimize".to_string()); + record("--optimize", None); } else if supports_explicit_optimize_false { - record("--optimize=false".to_string()); + record("--optimize", Some("false")); + } + + // Build env vars are applied via docker `-e` (see run_in_container), not as + // arguments to the inner `stellar contract build`, so they're recorded as + // bldopts only — never forwarded. A verifier replays them with `--env`. The + // value is escaped (the name is a validated identifier) so the recorded + // option stays valid shell. + for (name, value) in &cmd.build_args.env { + bldopts.push(format!( + "--env={name}={}", + shell_escape::escape(value.as_str().into()) + )); } (forwarded, bldopts) @@ -484,6 +516,10 @@ fn build_metadata_args(image_ref: &str, ids: &SourceIds, bldopts: &[String]) -> push(&mut out, "source_sha256", v); } + // bldopts already arrive as valid shell (escaped at the source in + // `build_forwarded_args`), so they're recorded verbatim: a verifier + // reconstructs the build by joining the recorded values and running them + // through a shell. for o in bldopts { push(&mut out, "bldopt", o); } @@ -758,23 +794,44 @@ fn compose_shell_command(cmds: &[Vec]) -> String { .join(" && ") } +/// Shell-escape each token of a single-package container command so a value +/// with spaces (a `--meta` value, or an `--env=` recorded as a `bldopt`) +/// survives when the reproduce line is copy-pasted into a shell. The +/// single-package path runs the image's default `stellar` entrypoint directly, +/// so there's no `sh -c` wrapper as in `compose_shell_command`. +fn escape_container_args(cmd: &[String]) -> String { + cmd.iter() + .map(|tok| shell_escape::escape(tok.into()).into_owned()) + .collect::>() + .join(" ") +} + async fn run_in_container( image_ref: &str, workspace_root: &Path, container_cmds: &[Vec], + env: &[String], docker: &Docker, print: &Print, verbose: bool, ) -> Result<(), Error> { let bind = format!("{}:/source", workspace_root.display()); + // `-e KEY=VALUE` flags for the reproduce command, mirroring the env passed + // to the container below. + let mut env_flags = String::new(); + for e in env { + env_flags.push_str(" -e "); + env_flags.push_str(&shell_escape::escape(e.as_str().into())); + } + // One package → run the image's default `stellar` entrypoint directly. // Several → override the entrypoint to a shell and chain the builds so they // all run in this one container. let (entrypoint, cmd, reproduce) = if container_cmds.len() > 1 { let chain = compose_shell_command(container_cmds); let reproduce = format!( - "docker run --rm -v {bind} --entrypoint /bin/sh {image_ref} -c {}", + "docker run --rm -v {bind}{env_flags} --entrypoint /bin/sh {image_ref} -c {}", shell_escape::escape(chain.clone().into()) ); ( @@ -784,7 +841,10 @@ async fn run_in_container( ) } else { let cmd = container_cmds.first().cloned().unwrap_or_default(); - let reproduce = format!("docker run --rm -v {bind} {image_ref} {}", cmd.join(" ")); + let reproduce = format!( + "docker run --rm -v {bind}{env_flags} {image_ref} {}", + escape_container_args(&cmd) + ); (None, cmd, reproduce) }; @@ -792,6 +852,7 @@ async fn run_in_container( image: Some(image_ref.to_string()), entrypoint, cmd: Some(cmd), + env: (!env.is_empty()).then(|| env.to_vec()), working_dir: Some("/source".to_string()), attach_stdout: Some(true), attach_stderr: Some(true), @@ -806,6 +867,9 @@ async fn run_in_container( print.infoln(format!( "Running verifiable build in {image_ref} (mount {bind})" )); + if verbose { + print.infoln(format!("Running: {reproduce}")); + } let created = docker .create_container(None::, config) @@ -1015,6 +1079,7 @@ mod tests { ("home_domain".to_string(), "fnando.com".to_string()), ("author".to_string(), "alice".to_string()), ], + env: vec![], optimize: true, }, ..Cmd::default() @@ -1028,11 +1093,32 @@ mod tests { assert!(bldopts.contains(&"--manifest-path=contracts/add/Cargo.toml".to_string())); } + #[test] + fn build_forwarded_args_records_env_as_bldopt_only() { + let cmd = Cmd { + build_args: super::super::BuildArgs { + env: vec![ + ("FOO".to_string(), "bar".to_string()), + ("BAZ".to_string(), "qux".to_string()), + ], + ..super::super::BuildArgs::default() + }, + ..Cmd::default() + }; + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true); + // Env vars are applied via docker `-e`, so they're recorded as bldopts + // for the verifier but never forwarded as build arguments. + assert!(bldopts.contains(&"--env=FOO=bar".to_string())); + assert!(bldopts.contains(&"--env=BAZ=qux".to_string())); + assert!(!forwarded.iter().any(|a| a.starts_with("--env"))); + } + #[test] fn build_forwarded_args_optimize_false_new_container() { let cmd = Cmd { build_args: super::super::BuildArgs { meta: vec![], + env: vec![], optimize: false, }, ..Cmd::default() @@ -1047,6 +1133,7 @@ mod tests { let cmd = Cmd { build_args: super::super::BuildArgs { meta: vec![], + env: vec![], optimize: false, }, ..Cmd::default() @@ -1091,6 +1178,39 @@ mod tests { assert_eq!(p[4], ("--meta", "bldopt=--features=a")); } + #[test] + fn build_forwarded_args_escapes_bldopt_values_as_shell() { + // Values with shell metacharacters are escaped at the source so each + // recorded bldopt is valid shell on its own. Only the value side is + // quoted: `--env=B='this is very nice'`, never `'--env=B=this is very + // nice'` (which would quote the flag and key too). + let cmd = Cmd { + features: Some("a,b".to_string()), + build_args: super::super::BuildArgs { + meta: vec![("note".to_string(), "added on build".to_string())], + env: vec![ + ("B".to_string(), "this is very nice".to_string()), + ("C".to_string(), "it's a \"trap\"".to_string()), + ], + optimize: true, + }, + ..Cmd::default() + }; + let (_forwarded, bldopts) = build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true); + + // The flag and key stay outside the quotes; only the value is escaped. + assert!(bldopts.contains(&"--env=B='this is very nice'".to_string())); + assert!(bldopts.contains(&"--meta=note='added on build'".to_string())); + // No-metacharacter values stay verbatim. + assert!(bldopts.contains(&"--features=a,b".to_string())); + + // Every recorded bldopt is valid shell that parses back to one argv token. + for o in &bldopts { + let tokens = shlex::split(o).expect("each bldopt must be valid shell"); + assert_eq!(tokens.len(), 1, "{o} must be a single shell token"); + } + } + #[test] fn build_metadata_args_sha256_only_omits_uri() { let ids = SourceIds { @@ -1235,4 +1355,31 @@ mod tests { "expected the spaced value to be quoted, got: {s}" ); } + + #[test] + fn escape_container_args_quotes_spaced_tokens() { + // An `--env=` recorded as a bldopt carries the env value verbatim, so a + // spaced value lands in a single `--meta bldopt=…` token. The reproduce + // line must quote it so a copy-paste round-trips back to one argv token. + let cmd = vec![ + "contract".to_string(), + "build".to_string(), + "--package=hello-world".to_string(), + "--meta".to_string(), + "bldopt=--env=B=this is very nice".to_string(), + ]; + let s = escape_container_args(&cmd); + let tokens = shlex::split(&s).expect("reproduce args must be valid shell"); + assert_eq!( + tokens, + vec![ + "contract", + "build", + "--package=hello-world", + "--meta", + "bldopt=--env=B=this is very nice", + ], + "spaced token must survive a shlex round-trip as one argument" + ); + } } From 3af186c0ffaee1b59e332b0c541fdf5cd688b842 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Thu, 18 Jun 2026 16:44:45 -0700 Subject: [PATCH 20/25] Pin the rust toolchain in verifiable builds. --- .../src/commands/contract/build/verifiable.rs | 106 +++++++++++++++--- 1 file changed, 88 insertions(+), 18 deletions(-) diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 17326fa5f2..1c6c15ba0a 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -777,6 +777,61 @@ async fn probe_cli_version(image_ref: &str, docker: &Docker) -> Result Option { + let config = ContainerCreateBody { + image: Some(image_ref.to_string()), + entrypoint: Some(vec!["rustup".to_string()]), + cmd: Some(vec!["show".to_string(), "active-toolchain".to_string()]), + attach_stdout: Some(true), + attach_stderr: Some(true), + host_config: Some(HostConfig { + auto_remove: Some(true), + ..Default::default() + }), + ..Default::default() + }; + let created = docker + .create_container(None::, config) + .await + .ok()?; + let attached = docker + .attach_container( + &created.id, + Some(AttachContainerOptions { + stdout: true, + stderr: true, + stream: true, + ..Default::default() + }), + ) + .await + .ok()?; + docker + .start_container(&created.id, None::) + .await + .ok()?; + + let mut stdout = String::new(); + let mut output = attached.output; + while let Some(chunk) = output.next().await { + if let Ok(bollard::container::LogOutput::StdOut { message }) = chunk { + stdout.push_str(&String::from_utf8_lossy(&message)); + } + } + + let mut wait = docker.wait_container(&created.id, None::); + while wait.next().await.is_some() {} + + stdout.split_whitespace().next().map(str::to_string) +} + /// Render the per-package `stellar contract build …` commands into a single /// `sh -c` script (`stellar … && stellar …`), shell-escaping every token so meta /// values with spaces survive. Used when more than one package is built so they @@ -817,10 +872,22 @@ async fn run_in_container( ) -> Result<(), Error> { let bind = format!("{}:/source", workspace_root.display()); + // Pin rustup to the image's own toolchain (per SEP-58): without this, a + // `rust-toolchain.toml` in the source could make rustup switch toolchains + // mid-build, defeating the digest-pinned image. Probe the image for its + // active toolchain and pass it through with `-e`, unless the caller already + // set RUSTUP_TOOLCHAIN. Skipped silently when the image has no rustup. + let mut env = env.to_vec(); + if !env.iter().any(|e| e.starts_with("RUSTUP_TOOLCHAIN=")) { + if let Some(toolchain) = probe_active_toolchain(image_ref, docker).await { + env.push(format!("RUSTUP_TOOLCHAIN={toolchain}")); + } + } + // `-e KEY=VALUE` flags for the reproduce command, mirroring the env passed // to the container below. let mut env_flags = String::new(); - for e in env { + for e in &env { env_flags.push_str(" -e "); env_flags.push_str(&shell_escape::escape(e.as_str().into())); } @@ -852,7 +919,7 @@ async fn run_in_container( image: Some(image_ref.to_string()), entrypoint, cmd: Some(cmd), - env: (!env.is_empty()).then(|| env.to_vec()), + env: (!env.is_empty()).then(|| env.clone()), working_dir: Some("/source".to_string()), attach_stdout: Some(true), attach_stderr: Some(true), @@ -908,24 +975,27 @@ async fn run_in_container( } } - let mut wait = docker.wait_container(&created.id, None::); + wait_for_container_exit(docker, &created.id, &reproduce).await +} + +/// Block until the container exits, mapping a non-zero exit code (whether +/// reported as a successful wait or as a `DockerContainerWaitError`) to +/// `ContainerExit` carrying the reproduce command. +async fn wait_for_container_exit(docker: &Docker, id: &str, reproduce: &str) -> Result<(), Error> { + let mut wait = docker.wait_container(id, None::); while let Some(item) = wait.next().await { - match item { - Ok(r) if r.status_code == 0 => {} - Ok(r) => { - return Err(Error::ContainerExit { - status: r.status_code, - command: reproduce.clone(), - }); - } - Err(bollard::errors::Error::DockerContainerWaitError { code: 0, .. }) => {} - Err(bollard::errors::Error::DockerContainerWaitError { code, .. }) => { - return Err(Error::ContainerExit { - status: code, - command: reproduce.clone(), - }); - } + // Both a successful wait and a `DockerContainerWaitError` carry an exit + // code; normalize to it (other errors are genuine failures). + let status = match item { + Ok(r) => r.status_code, + Err(bollard::errors::Error::DockerContainerWaitError { code, .. }) => code, Err(e) => return Err(e.into()), + }; + if status != 0 { + return Err(Error::ContainerExit { + status, + command: reproduce.to_string(), + }); } } From bf7cf3abdf427914f6c9aa8dd57ba25673be14d9 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Fri, 19 Jun 2026 12:14:08 -0700 Subject: [PATCH 21/25] Build source archives from the working directory. --- FULL_HELP_DOCS.md | 1 - cmd/crates/soroban-test/tests/it/build.rs | 30 ++ cmd/soroban-cli/Cargo.toml | 2 +- .../src/commands/contract/archive.rs | 29 +- .../commands/contract/build/source_archive.rs | 331 ++++++++++-------- .../src/commands/contract/build/verifiable.rs | 40 +-- 6 files changed, 228 insertions(+), 205 deletions(-) diff --git a/FULL_HELP_DOCS.md b/FULL_HELP_DOCS.md index b5a68062d5..4d10b321b9 100644 --- a/FULL_HELP_DOCS.md +++ b/FULL_HELP_DOCS.md @@ -354,7 +354,6 @@ Generate the reproducible source archive used by verifiable builds ###### **Options:** - `-o`, `--out-file ` — Where to write the gzipped tarball. Required unless `--dry-run` is used -- `--manifest-path ` — Path to Cargo.toml, used to locate the source root (its enclosing git repository, or the working directory) - `--dry-run` — List the entries that would be archived and the computed source_sha256, without writing any file ## `stellar contract build` diff --git a/cmd/crates/soroban-test/tests/it/build.rs b/cmd/crates/soroban-test/tests/it/build.rs index df4f395868..6d1ba0290c 100644 --- a/cmd/crates/soroban-test/tests/it/build.rs +++ b/cmd/crates/soroban-test/tests/it/build.rs @@ -1254,6 +1254,36 @@ fn contract_archive_requires_out_file_without_dry_run() { .stderr(predicate::str::contains("--out-file")); } +// A dirty git tree is a hard fail for `contract archive` too, matching +// `--verifiable`: the source_sha256 must describe a committed state. +#[test] +fn contract_archive_dirty_tree_errors() { + let sandbox = TestEnv::default(); + let (temp, workspace) = fresh_workspace(); + git_in(&workspace, &["init", "-q", "-b", "main"]); + git_in(&workspace, &["add", "-A"]); + git_in(&workspace, &["commit", "-q", "-m", "init"]); + // Dirty the tree after committing so status is non-empty. + std::fs::write(workspace.join("dirty.txt"), b"uncommitted").unwrap(); + + let out = temp.path().join("src.tar.gz"); + + sandbox + .new_assert_cmd("contract") + .current_dir(&workspace) + .arg("archive") + .arg("--out-file") + .arg(&out) + .assert() + .failure() + .stderr(predicate::str::contains("dirty")); + + assert!( + !out.exists(), + "no archive should be written for a dirty tree" + ); +} + // `--source-sha256` value must match the 64-hex regex. #[test] fn verifiable_source_sha256_format_errors() { diff --git a/cmd/soroban-cli/Cargo.toml b/cmd/soroban-cli/Cargo.toml index a1d29e654f..9cf393047c 100644 --- a/cmd/soroban-cli/Cargo.toml +++ b/cmd/soroban-cli/Cargo.toml @@ -129,7 +129,7 @@ whoami = "1.5.2" serde_with = "3.11.0" rustc_version = "0.4.1" tar = "0.4.40" -walkdir = "2.5.0" +ignore = "0.4.26" [build-dependencies] crate-git-revision = "0.0.9" diff --git a/cmd/soroban-cli/src/commands/contract/archive.rs b/cmd/soroban-cli/src/commands/contract/archive.rs index d4fe4cef03..34248fb5f1 100644 --- a/cmd/soroban-cli/src/commands/contract/archive.rs +++ b/cmd/soroban-cli/src/commands/contract/archive.rs @@ -19,9 +19,9 @@ const ARCHIVE_EXTENSIONS: &[&str] = &[".tar.gz", ".tgz"]; /// handy for confirming the contents before a verifiable build, or for /// producing the archive to host at a `--source-uri`. /// -/// In a git repo the archive is `git archive HEAD` (the committed tree); -/// otherwise the working directory is archived minus a built-in denylist (.git, -/// target/, node_modules/, .DS_Store, …). +/// The archive is the current working directory, honoring the project's +/// `.gitignore` and `.ignore` files (the `.git` directory itself is always +/// skipped). Run this from the project (or workspace) root you want archived. #[derive(Parser, Debug, Clone)] #[group(skip)] pub struct Cmd { @@ -29,11 +29,6 @@ pub struct Cmd { #[arg(long, short = 'o', required_unless_present = "dry_run")] pub out_file: Option, - /// Path to Cargo.toml, used to locate the source root (its enclosing git - /// repository, or the working directory). - #[arg(long)] - pub manifest_path: Option, - /// List the entries that would be archived and the computed source_sha256, /// without writing any file. #[arg(long)] @@ -55,18 +50,12 @@ impl Cmd { pub fn run(&self, global_args: &global::Args) -> Result<(), Error> { let print = Print::new(global_args.quiet); - let source_root = source_archive::resolve_source_root(self.manifest_path.as_deref()); - - // The git path archives HEAD, so uncommitted changes are silently - // excluded. Warn (don't fail — this is an inspect/generate tool, not a - // build) so the printed source_sha256 isn't mistaken for the working - // tree's. - if source_archive::tree_is_dirty(&source_root)? { - print.warnln(format!( - "git working tree at {} is dirty; the archive reflects HEAD only and excludes uncommitted changes.", - source_root.display(), - )); - } + let source_root = source_archive::resolve_source_root(); + + // The archive is the working tree, so a dirty repo would bake uncommitted + // changes into the bytes and the printed source_sha256 — refuse it, so the + // hash always corresponds to a committed state (matching --verifiable). + source_archive::ensure_clean_tree(&source_root, &print)?; // The dry-run listing itself reveals the contents, so skip the // "not a git repository" warning there. diff --git a/cmd/soroban-cli/src/commands/contract/build/source_archive.rs b/cmd/soroban-cli/src/commands/contract/build/source_archive.rs index 2f20d516cd..b266cc81c4 100644 --- a/cmd/soroban-cli/src/commands/contract/build/source_archive.rs +++ b/cmd/soroban-cli/src/commands/contract/build/source_archive.rs @@ -2,10 +2,10 @@ //! //! Produces a gzipped tarball of a contract's source tree, rooted under a //! top-level `source/` prefix (so it extracts to a `source/` dir, mirroring the -//! container's `/source` mount). In a git repo this is `git archive HEAD` (the -//! committed tree); otherwise the working directory is walked and tarred, -//! skipping `ARCHIVE_DENYLIST` entries. The output is byte-reproducible, so the -//! same tree always hashes to the same `source_sha256`. +//! container's `/source` mount). The working directory is walked and tarred, +//! honoring the project's own `.gitignore`/`.ignore` files (the `.git` directory +//! itself is always skipped). The output is byte-reproducible, so the same tree +//! always hashes to the same `source_sha256`. //! //! Shared by `contract build --verifiable` (which builds from the extracted //! archive) and the standalone `contract archive` command (which generates and @@ -17,19 +17,18 @@ use std::{ process::Command, }; -use walkdir::WalkDir; +use ignore::WalkBuilder; use crate::print::Print; -/// Top-level names excluded when archiving a non-git working directory (we have -/// no tracked-files list to consult, so fall back to a fixed denylist of VCS -/// metadata, build/cache/transient dirs, and editor/OS/AI-assistant junk). -/// Matched against each path component, so a directory like `target/` prunes -/// its whole subtree. -pub(crate) const ARCHIVE_DENYLIST: &[&str] = &[ - // version control - ".git", - ".gitignore", +/// Names that usually shouldn't end up in a source archive — VCS metadata of +/// other systems, secrets/local env, build/cache/transient dirs, and editor/OS/ +/// AI-assistant junk. These don't *exclude* anything (selection is driven +/// entirely by `.gitignore`/`.ignore`); instead, if any of them slip into the +/// archive because the project didn't ignore them, we warn the user so they can +/// add an ignore rule. Matched against each path component. +pub(crate) const ARCHIVE_WARN_LIST: &[&str] = &[ + // version control (other systems) ".svn", ".hg", // secrets / local environment @@ -62,8 +61,10 @@ pub enum Error { source: std::io::Error, }, - #[error("`git archive` failed in {path}: {stderr}")] - GitArchive { path: PathBuf, stderr: String }, + #[error( + "refusing to archive a dirty git working tree at {path}; commit or stash your changes and try again." + )] + GitDirty { path: PathBuf }, #[error("could not write source archive to {path}: {source}")] ArchiveWrite { @@ -75,50 +76,41 @@ pub enum Error { ArchiveExtract(std::io::Error), } -/// Pick the anchor for the source tree: the directory whose `.git` parent we -/// archive (and, for verifiable builds, relativize `--manifest-path` against). -/// Walk up from `manifest_path` (or cwd, if none) looking for a `.git` -/// directory; return its parent. If none is found, fall back to cwd. -/// -/// This isn't a validation step — any `.git` will do. Wrong-source mistakes are -/// caught later by the verify-side byte comparison. -pub(crate) fn resolve_source_root(manifest_path: Option<&Path>) -> PathBuf { - let start = if let Some(p) = manifest_path { - let abs = std::path::absolute(p).unwrap_or_else(|_| p.to_path_buf()); - abs.parent().map(Path::to_path_buf).unwrap_or(abs) - } else { - std::env::current_dir().unwrap_or_else(|_| PathBuf::from(".")) - }; - - let mut p = start.clone(); - loop { - if p.join(".git").exists() { - return p; - } - if !p.pop() { - break; - } - } - - std::env::current_dir().unwrap_or(start) +/// The source tree's root: always the current working directory. The archive is +/// rooted there as-is — we do NOT search upward for a git repository or anchor on +/// `--manifest-path`'s directory, since for a workspace member the build needs +/// the whole workspace (its root `Cargo.toml`/`Cargo.lock`), which lives at the +/// cwd, not the member's directory. So run `contract archive`/`build +/// --verifiable` from the project (or workspace) root you want archived; +/// `--manifest-path`, when given, is interpreted relative to it. +pub(crate) fn resolve_source_root() -> PathBuf { + std::env::current_dir().unwrap_or_else(|_| PathBuf::from(".")) } -/// Whether `source_root` is inside a git work tree. -pub(crate) fn is_git_repo(source_root: &Path) -> bool { - Command::new("git") - .arg("-C") - .arg(source_root) - .arg("rev-parse") - .arg("--is-inside-work-tree") - .output() - .is_ok_and(|o| o.status.success()) +/// Warn about and reject a dirty git working tree. Both `contract archive` and +/// `build --verifiable` archive the working tree as-is, so uncommitted changes +/// would be baked into the recorded `source_sha256`; refuse them (after +/// explaining why) so an archive always corresponds to a committed state. A +/// no-op when `source_root` isn't a git repo (we can't check, e.g. archive +/// sources) — the user owns the bytes they produce there. +pub(crate) fn ensure_clean_tree(source_root: &Path, print: &Print) -> Result<(), Error> { + if tree_is_dirty(source_root)? { + print.warnln(format!( + "git working tree at {} is dirty; the archive would include uncommitted changes.", + source_root.display(), + )); + return Err(Error::GitDirty { + path: source_root.to_path_buf(), + }); + } + Ok(()) } /// Whether `source_root` is a git work tree with uncommitted changes. Returns /// `Ok(false)` when it isn't a git repo (git ran but refused) — callers can't /// verify cleanliness there, so they proceed. Errors only when git can't be /// invoked at all. -pub(crate) fn tree_is_dirty(source_root: &Path) -> Result { +fn tree_is_dirty(source_root: &Path) -> Result { let status = Command::new("git") .arg("-C") .arg(source_root) @@ -138,31 +130,20 @@ pub(crate) fn tree_is_dirty(source_root: &Path) -> Result { Ok(!status.stdout.is_empty()) } -/// Produce the gzipped source tarball bytes. Entries are rooted under a -/// top-level `source/` prefix. In a git repo this is `git archive HEAD` (the -/// committed tree); otherwise the working directory is walked and tarred, -/// skipping `ARCHIVE_DENYLIST` entries. +/// Produce the gzipped source tarball bytes. The working directory under +/// `source_root` is walked and tarred, honoring the project's `.gitignore`/ +/// `.ignore` files; entries are rooted under a top-level `source/` prefix. /// -/// When the source isn't a git repo, `warn_non_git` controls whether to warn -/// that the working directory is being archived. Callers that only inspect the -/// result (e.g. `contract archive --dry-run`) pass `false`, since the listing -/// itself reveals the contents. +/// `warn` controls whether to warn about archived paths that usually shouldn't +/// be shipped (see `ARCHIVE_WARN_LIST`). Callers that only inspect the result +/// (e.g. `contract archive --dry-run`) pass `false`, since the listing itself +/// reveals the contents. pub(crate) fn build_source_archive( source_root: &Path, print: &Print, - warn_non_git: bool, + warn: bool, ) -> Result, Error> { - let tar = if is_git_repo(source_root) { - git_archive_tar(source_root)? - } else { - if warn_non_git { - print.warnln(format!( - "{} is not a git repository; archiving the working directory. Inspect the generated archive to confirm its contents.", - source_root.display(), - )); - } - walk_tar(source_root)? - }; + let tar = walk_tar(source_root, print, warn)?; gzip(&tar) } @@ -181,31 +162,16 @@ pub(crate) fn entry_names(bytes: &[u8]) -> Result, Error> { Ok(names) } -/// `git archive --format=tar --prefix=source/ HEAD`, returning the tar bytes. -fn git_archive_tar(source_root: &Path) -> Result, Error> { - let out = Command::new("git") - .arg("-C") - .arg(source_root) - .arg("archive") - .arg("--format=tar") - .arg("--prefix=source/") - .arg("HEAD") - .output() - .map_err(|source| Error::GitInvoke { - path: source_root.to_path_buf(), - source, - })?; - if !out.status.success() { - return Err(Error::GitArchive { - path: source_root.to_path_buf(), - stderr: String::from_utf8_lossy(&out.stderr).trim().to_string(), - }); - } - Ok(out.stdout) -} - -/// Tar the working tree under `source_root`, skipping denylisted path -/// components. Each entry is prefixed with `source/`. +/// Tar the working tree under `source_root`, honoring the project's `.gitignore`/ +/// `.ignore` files and always skipping the `.git` directory. Each entry is +/// prefixed with `source/`. When `warn` is set, archived paths matching +/// `ARCHIVE_WARN_LIST` (e.g. `.env`, `target/`) trigger a warning so the user can +/// add an ignore rule. +/// +/// Selection depends only on the in-tree files plus the `.gitignore`/`.ignore` +/// files inside the archived tree — never on machine-specific state (the global +/// gitignore, `.git/info/exclude`, or ignore files in parent directories are not +/// consulted) — so the archive stays byte-reproducible across machines. /// /// The output is reproducible, following GNU tar's reproducibility guidance /// () @@ -217,23 +183,34 @@ fn git_archive_tar(source_root: &Path) -> Result, Error> { /// names (`--owner=0 --group=0 --numeric-owner`), and normalizes mode /// (`--mode=go+u,go-w`). ustar headers carry no atime/ctime or tar PID. The gzip /// wrapper (see `gzip`) is likewise deterministic. -fn walk_tar(source_root: &Path) -> Result, Error> { +fn walk_tar(source_root: &Path, print: &Print, warn: bool) -> Result, Error> { + let walk = WalkBuilder::new(source_root) + .hidden(false) // include dotfiles; let .gitignore decide + .git_ignore(true) // honor in-tree .gitignore + .ignore(true) // honor .ignore + .git_global(false) // not the machine's global gitignore (not reproducible) + .git_exclude(false) // not .git/info/exclude (not in the archive) + .require_git(false) // apply .gitignore/.ignore even without a .git dir + .parents(false) // only ignore files inside the archived tree + .filter_entry(|e| e.file_name() != ".git") // never archive VCS internals + .build(); + let mut files: Vec = Vec::new(); - let walk = WalkDir::new(source_root) - .sort_by_file_name() - .into_iter() - .filter_entry(|e| !is_denylisted(e.file_name())); for entry in walk { - let entry = entry.map_err(|e| Error::ArchiveWrite { + let entry = entry.map_err(|source| Error::ArchiveWrite { path: source_root.to_path_buf(), - source: e.into(), + source: std::io::Error::other(source), })?; - if entry.file_type().is_file() { + if entry.file_type().is_some_and(|t| t.is_file()) { files.push(entry.path().to_path_buf()); } } files.sort(); + if warn { + warn_unexpected_paths(&files, source_root, print); + } + let mut builder = tar::Builder::new(Vec::new()); builder.mode(tar::HeaderMode::Deterministic); for path in &files { @@ -256,17 +233,51 @@ fn walk_tar(source_root: &Path) -> Result, Error> { }) } -/// A path component is denylisted if it equals a denylist entry, or — for -/// dotted entries, which double as extension filters (e.g. `.swp`, `.log`) — if -/// it ends with that entry. Plain names (`target`, `node_modules`) match -/// exactly only, so `mytarget` is not excluded. -fn is_denylisted(name: &std::ffi::OsStr) -> bool { +/// Whether a path component matches the warn list: it equals an entry, or — for +/// dotted entries, which double as extension filters (e.g. `.swp`, `.log`) — it +/// ends with that entry. Plain names (`target`, `node_modules`) match exactly +/// only, so `mytarget` is not flagged. +fn is_warned(name: &std::ffi::OsStr) -> bool { let name = name.to_string_lossy(); - ARCHIVE_DENYLIST + ARCHIVE_WARN_LIST .iter() .any(|d| name == *d || (d.starts_with('.') && name.ends_with(d))) } +/// Warn about archived paths that usually shouldn't be shipped (secrets, build +/// output, editor/OS junk; see `ARCHIVE_WARN_LIST`). Selection is driven by +/// `.gitignore`/`.ignore`, so these slipped in only because the project didn't +/// ignore them — point that out so the user can add a rule. Reports the path up +/// to each matched component once (so a flagged directory is named once, not per +/// file under it), each on its own line since paths can be long. +fn warn_unexpected_paths(files: &[PathBuf], source_root: &Path, print: &Print) { + let mut hits: Vec = Vec::new(); + for path in files { + let rel = path.strip_prefix(source_root).unwrap_or(path); + let mut prefix = PathBuf::new(); + for comp in rel.components() { + prefix.push(comp); + if is_warned(comp.as_os_str()) { + let hit = prefix.to_string_lossy().into_owned(); + if !hits.contains(&hit) { + hits.push(hit); + } + break; + } + } + } + if hits.is_empty() { + return; + } + hits.sort(); + print.warnln( + "archive includes paths usually excluded; add them to .gitignore or .ignore if unintended:", + ); + for hit in &hits { + print.blankln(hit); + } +} + /// Gzip with a default (mtime-zeroed) header so the same tar bytes always hash /// the same. fn gzip(bytes: &[u8]) -> Result, Error> { @@ -297,22 +308,24 @@ mod tests { use sha2::{Digest, Sha256}; #[test] - fn is_denylisted_matches_names_and_dotted_suffixes() { + fn is_warned_matches_names_and_dotted_suffixes() { use std::ffi::OsStr; // exact name matches - assert!(is_denylisted(OsStr::new("target"))); - assert!(is_denylisted(OsStr::new(".git"))); - assert!(is_denylisted(OsStr::new(".gitignore"))); - assert!(is_denylisted(OsStr::new(".env"))); - assert!(is_denylisted(OsStr::new(".DS_Store"))); + assert!(is_warned(OsStr::new("target"))); + assert!(is_warned(OsStr::new(".env"))); + assert!(is_warned(OsStr::new(".DS_Store"))); // plain names match exactly only - assert!(!is_denylisted(OsStr::new("mytarget"))); - assert!(!is_denylisted(OsStr::new("targets"))); + assert!(!is_warned(OsStr::new("mytarget"))); + assert!(!is_warned(OsStr::new("targets"))); // dotted entries also match as suffix (extension-style) - assert!(is_denylisted(OsStr::new("backup.git"))); + assert!(is_warned(OsStr::new("backup.svn"))); + // `.git`/`.gitignore` are not warned: `.git` is skipped structurally and + // `.gitignore` is legitimately archived like any other tracked file. + assert!(!is_warned(OsStr::new(".git"))); + assert!(!is_warned(OsStr::new(".gitignore"))); // unrelated files pass through - assert!(!is_denylisted(OsStr::new("Cargo.toml"))); - assert!(!is_denylisted(OsStr::new("lib.rs"))); + assert!(!is_warned(OsStr::new("Cargo.toml"))); + assert!(!is_warned(OsStr::new("lib.rs"))); } // Initialize a git repo at `root` with one commit of everything present. @@ -353,7 +366,13 @@ mod tests { let a = build_source_archive(root, &print, true).unwrap(); let b = build_source_archive(root, &print, true).unwrap(); assert!(!a.is_empty()); - assert_eq!(a, b, "same commit should produce identical bytes"); + assert_eq!(a, b, "same tree should produce identical bytes"); + + // The `.git` dir git_init_commit created is never archived. + assert!(entry_names(&a) + .unwrap() + .iter() + .all(|n| !n.starts_with("source/.git/"))); let sha = hex::encode(Sha256::digest(&a)); assert_eq!(sha.len(), 64); @@ -385,18 +404,20 @@ mod tests { } #[test] - fn build_source_archive_non_git_excludes_denylist() { + fn build_source_archive_skips_git_dir_and_is_reproducible() { let print = Print::new(true); let temp = tempfile::TempDir::new().unwrap(); let root = temp.path(); std::fs::write(root.join("Cargo.toml"), b"# crate").unwrap(); std::fs::create_dir_all(root.join("src")).unwrap(); std::fs::write(root.join("src/lib.rs"), b"// code").unwrap(); - // Planted dirs that must be excluded. - std::fs::create_dir_all(root.join("target/debug")).unwrap(); - std::fs::write(root.join("target/debug/x"), b"junk").unwrap(); + // A `.git` dir is always skipped, even without a real repo. std::fs::create_dir_all(root.join(".git")).unwrap(); std::fs::write(root.join(".git/config"), b"junk").unwrap(); + // No `.gitignore`, so `target/` is NOT excluded — selection is driven by + // ignore files only. + std::fs::create_dir_all(root.join("target/debug")).unwrap(); + std::fs::write(root.join("target/debug/x"), b"junk").unwrap(); let bytes = build_source_archive(root, &print, true).unwrap(); let dest = tempfile::TempDir::new().unwrap(); @@ -404,8 +425,9 @@ mod tests { assert!(dest.path().join("source/Cargo.toml").exists()); assert!(dest.path().join("source/src/lib.rs").exists()); - assert!(!dest.path().join("source/target").exists()); assert!(!dest.path().join("source/.git").exists()); + // Un-ignored `target/` is included (and would have triggered a warning). + assert!(dest.path().join("source/target/debug/x").exists()); assert_eq!(hex::encode(Sha256::digest(&bytes)).len(), 64); // Reproducible: a second run over the same tree yields identical bytes @@ -415,36 +437,37 @@ mod tests { } #[test] - fn resolve_source_root_finds_git_root_from_subdir() { + fn build_source_archive_respects_gitignore_and_dot_ignore() { + let print = Print::new(true); let temp = tempfile::TempDir::new().unwrap(); let root = temp.path(); - std::fs::create_dir_all(root.join(".git")).unwrap(); - let nested = root.join("contracts").join("foo"); - std::fs::create_dir_all(&nested).unwrap(); - std::fs::write(nested.join("Cargo.toml"), b"# placeholder").unwrap(); - - let manifest = nested.join("Cargo.toml"); - // Use canonicalize on both sides — `tempfile` returns symlinked /var - // paths on macOS while resolve_source_root walks the same prefix. - let got = std::fs::canonicalize(resolve_source_root(Some(&manifest))).unwrap(); - let want = std::fs::canonicalize(root).unwrap(); - assert_eq!(got, want); + std::fs::write(root.join("Cargo.toml"), b"# crate").unwrap(); + std::fs::create_dir_all(root.join("src")).unwrap(); + std::fs::write(root.join("src/lib.rs"), b"// code").unwrap(); + // `.gitignore` and `.ignore` are honored even without a git repo. + std::fs::write(root.join(".gitignore"), b"target/\n").unwrap(); + std::fs::write(root.join(".ignore"), b"secret.txt\n").unwrap(); + std::fs::create_dir_all(root.join("target/debug")).unwrap(); + std::fs::write(root.join("target/debug/x"), b"junk").unwrap(); + std::fs::write(root.join("secret.txt"), b"shh").unwrap(); + + let bytes = build_source_archive(root, &print, true).unwrap(); + let dest = tempfile::TempDir::new().unwrap(); + unpack_targz(&bytes, dest.path()).unwrap(); + + assert!(dest.path().join("source/Cargo.toml").exists()); + assert!(dest.path().join("source/src/lib.rs").exists()); + // Excluded by the in-tree ignore files. + assert!(!dest.path().join("source/target").exists()); + assert!(!dest.path().join("source/secret.txt").exists()); + // The ignore files themselves are archived like any other tracked file. + assert!(dest.path().join("source/.gitignore").exists()); } #[test] - fn resolve_source_root_falls_back_to_cwd_without_git() { - let temp = tempfile::TempDir::new().unwrap(); - let root = temp.path(); - let nested = root.join("noisy"); - std::fs::create_dir_all(&nested).unwrap(); - std::fs::write(nested.join("Cargo.toml"), b"# placeholder").unwrap(); - - let manifest = nested.join("Cargo.toml"); - // No `.git` anywhere up the tree, so we fall back to cwd. We can't - // assert what cwd is in a test runner (it varies), but we can assert - // that the returned path doesn't have `.git`. That's enough to confirm - // fallback kicked in. - let got = resolve_source_root(Some(&manifest)); - assert!(!got.join(".git").exists()); + fn resolve_source_root_is_cwd() { + // The root is always the current working directory — no upward search, + // no manifest anchoring. + assert_eq!(resolve_source_root(), std::env::current_dir().unwrap()); } } diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 1c6c15ba0a..ecad61fc38 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -69,11 +69,6 @@ pub enum Error { #[error(transparent)] SourceArchive(#[from] source_archive::Error), - #[error( - "git working tree at {path} is dirty. --verifiable requires a clean tree so the recorded source_sha256 matches the WASM bytes. Commit or stash your changes and try again." - )] - GitDirty { path: PathBuf }, - #[error( "the cli sets bldimg, source_uri, source_sha256, and bldopt automatically when --verifiable is used; remove them from --meta. Got reserved key: {key}" )] @@ -116,17 +111,18 @@ pub async fn run( let workspace_root = resolve_workspace_root(cmd)?; validate_source_formats(cmd)?; - // Pick the anchor for the local source: the `--manifest-path` bldopt is - // relativized against it, and (when `--archive` is not used) it's also what - // gets bind-mounted into the container. We do NOT validate that it matches - // source_uri — a wrong source produces different bytes, and verify catches - // that at byte-comparison time. - let source_root = source_archive::resolve_source_root(cmd.manifest_path.as_deref()); + // The source root is the current working directory: it's bind-mounted into + // the container and the `--manifest-path` bldopt is relativized against it. + // Run from the project/workspace root you want built. We do NOT validate that + // it matches source_uri — a wrong source produces different bytes, and verify + // catches that at byte-comparison time. + let source_root = source_archive::resolve_source_root(); - // A dirty working tree would make the recorded source_sha256 fail to - // describe the bytes actually built, so refuse to proceed. Skipped when - // the source root isn't a git repo (we can't check, e.g. archive sources). - enforce_clean_tree(&source_root)?; + // The archive is the working tree, so refuse a dirty repo: a verifiable build + // should be deliberate, off a committed state, not whatever happens to be on + // disk. Skipped when the source root isn't a git repo (we can't check, e.g. + // archive sources). + source_archive::ensure_clean_tree(&source_root, print).map_err(Error::from)?; // Always build the source archive, record its hash, and build from the // *extracted* archive (in a hardened tempdir) so the WASM is produced from @@ -344,20 +340,6 @@ fn resolve_archive(cmd: &Cmd, source_root: &Path, print: &Print) -> Result Result<(), Error> { - if source_archive::tree_is_dirty(source_root)? { - return Err(Error::GitDirty { - path: source_root.to_path_buf(), - }); - } - Ok(()) -} - fn bldimg_regex() -> Regex { Regex::new(r"^(?:localhost(?::\d+)?|[^\s@/]*[.:][^\s@/]*)/[^\s@]+@sha256:[0-9a-f]{64}$") .unwrap() From 183360666226329601824b66d2d7647745ebc9ac Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Wed, 8 Jul 2026 11:27:06 -0300 Subject: [PATCH 22/25] Skip --locked on build images that lack it. --- .../src/commands/contract/build/verifiable.rs | 191 ++++++++++++------ 1 file changed, 125 insertions(+), 66 deletions(-) diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index ecad61fc38..ebda60f7d1 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -147,12 +147,6 @@ pub async fn run( source_sha256: Some(resolved.source_sha256.clone()), }; - // Defer the info banner until every validation has passed, so it doesn't - // appear right before an error. - if !cmd.locked { - print.infoln("Implying --locked because --verifiable was passed"); - } - // Stage 3: docker. let docker_args = crate::commands::container::shared::Args { docker_host: cmd.docker_host.clone(), @@ -172,6 +166,23 @@ pub async fn run( probe_supports_optimize_false_syntax(&image_ref, &docker, print).await }; + // `--locked` is implied by `--verifiable`, but it was only added to + // `contract build` in cli 25.2.0. Probe the image before adding it so a + // build against an older, still-valid bldimg doesn't fail on an unknown + // flag. When the flag is unavailable we drop it and warn that the rebuild + // can't be pinned against dependency drift. + let supports_locked = probe_supports_locked(&image_ref, &docker, print).await; + if supports_locked { + if !cmd.locked { + print.infoln("Implying --locked because --verifiable was passed"); + } + } else { + print.warnln( + "The build image's `contract build` does not support --locked; \ + building without it. Dependency drift may affect reproducibility.", + ); + } + // Build once per package, each with its own `--package` forwarded and // recorded as a `bldopt`, so every WASM is independently reproducible. With // no explicit `--package` the targets are inferred like a regular build. @@ -187,8 +198,13 @@ pub async fn run( let container_cmds: Vec> = targets .iter() .map(|target| { - let (forwarded_args, bldopts) = - build_forwarded_args(cmd, &source_root, *target, supports_explicit_optimize_false); + let (forwarded_args, bldopts) = build_forwarded_args( + cmd, + &source_root, + *target, + supports_explicit_optimize_false, + supports_locked, + ); let metadata_args = build_metadata_args(&image_ref, &source_ids, &bldopts); compose_container_args(&forwarded_args, &metadata_args) }) @@ -389,8 +405,13 @@ fn resolve_build_packages(cmd: &Cmd) -> Result, Error> { /// The flags forwarded to the container's `stellar contract build`, plus the /// bldopt strings recorded into SEP-58 metadata. Every build-affecting flag /// becomes one bldopt entry so a verifier can replay the same invocation. -/// `--locked` is always present. `manifest_path` (when set) is recorded -/// relative to the workspace root so it's valid inside `/source`. +/// `manifest_path` (when set) is recorded relative to the workspace root so it's +/// valid inside `/source`. +/// +/// `supports_locked`: whether the container's `contract build` accepts +/// `--locked` (added in cli 25.2.0). When false the flag is neither forwarded +/// nor recorded, so a build against an older image doesn't fail on an unknown +/// argument. /// /// `supports_explicit_optimize_false`: whether the container's cli accepts /// `--optimize=false`. When false, the optimize=false case records the flag @@ -401,6 +422,7 @@ fn build_forwarded_args( workspace_root: &Path, package: Option<&str>, supports_explicit_optimize_false: bool, + supports_locked: bool, ) -> (Vec, Vec) { let mut forwarded: Vec = Vec::new(); let mut bldopts: Vec = Vec::new(); @@ -424,7 +446,9 @@ fn build_forwarded_args( } }; - record("--locked", None); + if supports_locked { + record("--locked", None); + } if let Some(path) = &cmd.manifest_path { let abs = std::path::absolute(path).unwrap_or_else(|_| path.clone()); @@ -714,10 +738,20 @@ async fn probe_supports_optimize_false_syntax( } } -async fn probe_cli_version(image_ref: &str, docker: &Docker) -> Result { +/// Run `cmd` in a throwaway container (optionally overriding the entrypoint) and +/// return its captured stdout. The container auto-removes; stderr is attached so +/// the daemon streams it, but only stdout is collected. Shared by every image +/// probe (cli version, active toolchain, flag support). +async fn run_probe( + image_ref: &str, + docker: &Docker, + entrypoint: Option>, + cmd: Vec, +) -> Result { let config = ContainerCreateBody { image: Some(image_ref.to_string()), - cmd: Some(vec!["version".to_string(), "--only-version".to_string()]), + entrypoint, + cmd: Some(cmd), attach_stdout: Some(true), attach_stderr: Some(true), host_config: Some(HostConfig { @@ -755,10 +789,50 @@ async fn probe_cli_version(image_ref: &str, docker: &Docker) -> Result); while wait.next().await.is_some() {} + Ok(stdout) +} + +async fn probe_cli_version(image_ref: &str, docker: &Docker) -> Result { + let stdout = run_probe( + image_ref, + docker, + None, + vec!["version".to_string(), "--only-version".to_string()], + ) + .await?; Version::parse(stdout.trim()) .map_err(|e| Error::TagListUnavailable(format!("unparseable version {stdout:?}: {e}"))) } +/// Probe whether the container's `stellar contract build` accepts `--locked`. +/// The flag was added in cli 25.2.0 (commit `6115b818`); older images reject it +/// outright, which would fail the build. Rather than map versions, ask the +/// container's own `contract build --help` whether the flag exists. On any probe +/// failure returns false — the conservative assumption that the flag is absent, +/// so the build proceeds without it rather than erroring. +pub(crate) async fn probe_supports_locked(image_ref: &str, docker: &Docker, print: &Print) -> bool { + match run_probe( + image_ref, + docker, + None, + vec![ + "contract".to_string(), + "build".to_string(), + "--help".to_string(), + ], + ) + .await + { + Ok(help) => help.contains("--locked"), + Err(e) => { + print.warnln(format!( + "Could not probe whether the container's `contract build` supports --locked ({e}); building without it" + )); + false + } + } +} + /// Probe the image for the toolchain rustup uses by default, so it can be /// pinned via `RUSTUP_TOOLCHAIN` (see `run_in_container`). Overrides the /// entrypoint to run `rustup show active-toolchain` and returns the toolchain @@ -767,50 +841,14 @@ async fn probe_cli_version(image_ref: &str, docker: &Docker) -> Result Option { - let config = ContainerCreateBody { - image: Some(image_ref.to_string()), - entrypoint: Some(vec!["rustup".to_string()]), - cmd: Some(vec!["show".to_string(), "active-toolchain".to_string()]), - attach_stdout: Some(true), - attach_stderr: Some(true), - host_config: Some(HostConfig { - auto_remove: Some(true), - ..Default::default() - }), - ..Default::default() - }; - let created = docker - .create_container(None::, config) - .await - .ok()?; - let attached = docker - .attach_container( - &created.id, - Some(AttachContainerOptions { - stdout: true, - stderr: true, - stream: true, - ..Default::default() - }), - ) - .await - .ok()?; - docker - .start_container(&created.id, None::) - .await - .ok()?; - - let mut stdout = String::new(); - let mut output = attached.output; - while let Some(chunk) = output.next().await { - if let Ok(bollard::container::LogOutput::StdOut { message }) = chunk { - stdout.push_str(&String::from_utf8_lossy(&message)); - } - } - - let mut wait = docker.wait_container(&created.id, None::); - while wait.next().await.is_some() {} - + let stdout = run_probe( + image_ref, + docker, + Some(vec!["rustup".to_string()]), + vec!["show".to_string(), "active-toolchain".to_string()], + ) + .await + .ok()?; stdout.split_whitespace().next().map(str::to_string) } @@ -1076,7 +1114,8 @@ mod tests { #[test] fn build_forwarded_args_defaults() { let cmd = Cmd::default(); - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true); + let (forwarded, bldopts) = + build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true, true); // Default optimize=true → bare `--optimize` recorded + forwarded. assert_eq!( forwarded, @@ -1088,6 +1127,19 @@ mod tests { ); } + #[test] + fn build_forwarded_args_omits_locked_when_unsupported() { + // Older images (< cli 25.2.0) reject `--locked`; when the probe reports + // it's unsupported, the flag is neither forwarded nor recorded. + let cmd = Cmd::default(); + let (forwarded, bldopts) = + build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true, false); + assert!(!forwarded.iter().any(|a| a == "--locked")); + assert!(!bldopts.iter().any(|a| a == "--locked")); + // Everything else is still recorded (default optimize=true here). + assert!(forwarded.contains(&"--optimize".to_string())); + } + #[test] fn build_forwarded_args_features_and_package() { let cmd = Cmd { @@ -1095,7 +1147,8 @@ mod tests { package: Some("contract-a".to_string()), ..Cmd::default() }; - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true); + let (forwarded, bldopts) = + build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true, true); assert!(forwarded.contains(&"--features=a,b".to_string())); assert!(forwarded.contains(&"--package=contract-a".to_string())); assert!(bldopts.contains(&"--features=a,b".to_string())); @@ -1109,7 +1162,8 @@ mod tests { // default cdylib); it must still be forwarded and recorded. let cmd = Cmd::default(); assert!(cmd.package.is_none()); - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), Some("hello-world"), true); + let (forwarded, bldopts) = + build_forwarded_args(&cmd, ws(), Some("hello-world"), true, true); assert!(forwarded.contains(&"--package=hello-world".to_string())); assert!(bldopts.contains(&"--package=hello-world".to_string())); } @@ -1117,7 +1171,7 @@ mod tests { #[test] fn build_forwarded_args_omits_package_when_unresolved() { let cmd = Cmd::default(); - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), None, true); + let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), None, true, true); assert!(!forwarded.iter().any(|a| a.starts_with("--package"))); assert!(!bldopts.iter().any(|a| a.starts_with("--package"))); } @@ -1136,7 +1190,8 @@ mod tests { }, ..Cmd::default() }; - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true); + let (forwarded, bldopts) = + build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true, true); assert!(forwarded.contains(&"--meta=home_domain=fnando.com".to_string())); assert!(forwarded.contains(&"--meta=author=alice".to_string())); assert!(forwarded.contains(&"--manifest-path=contracts/add/Cargo.toml".to_string())); @@ -1157,7 +1212,8 @@ mod tests { }, ..Cmd::default() }; - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true); + let (forwarded, bldopts) = + build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true, true); // Env vars are applied via docker `-e`, so they're recorded as bldopts // for the verifier but never forwarded as build arguments. assert!(bldopts.contains(&"--env=FOO=bar".to_string())); @@ -1175,7 +1231,8 @@ mod tests { }, ..Cmd::default() }; - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true); + let (forwarded, bldopts) = + build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true, true); assert!(forwarded.contains(&"--optimize=false".to_string())); assert!(bldopts.contains(&"--optimize=false".to_string())); } @@ -1190,7 +1247,8 @@ mod tests { }, ..Cmd::default() }; - let (forwarded, bldopts) = build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), false); + let (forwarded, bldopts) = + build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), false, true); // Old container's default is already false; record nothing. // Passing `--optimize=false` to a pre-26.1.0 cli would fail. assert!(!forwarded.iter().any(|a| a.starts_with("--optimize"))); @@ -1248,7 +1306,8 @@ mod tests { }, ..Cmd::default() }; - let (_forwarded, bldopts) = build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true); + let (_forwarded, bldopts) = + build_forwarded_args(&cmd, ws(), cmd.package.as_deref(), true, true); // The flag and key stay outside the quotes; only the value is escaped. assert!(bldopts.contains(&"--env=B='this is very nice'".to_string())); From ef4993c2162b0efee32c7c1a92f0dc02a6b9bdf8 Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Thu, 16 Jul 2026 11:13:49 -0300 Subject: [PATCH 23/25] Run verifiable builds through the docker CLI. --- Cargo.lock | 10 +- .../src/commands/container/shared.rs | 62 +++- .../src/commands/contract/build/verifiable.rs | 310 ++++++------------ 3 files changed, 171 insertions(+), 211 deletions(-) diff --git a/Cargo.lock b/Cargo.lock index 260212a41a..aba5091f09 100644 --- a/Cargo.lock +++ b/Cargo.lock @@ -2382,9 +2382,9 @@ checksum = "0cc23270f6e1808e30a928bdc84dea0b9b4136a8bc82338574f23baf47bbd280" [[package]] name = "globset" -version = "0.4.16" +version = "0.4.19" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "54a1028dfc5f5df5da8a56a73e6c153c9a9708ec57232470703592a3f18e49f5" +checksum = "e47d37d2ae4464254884b60ab7071be2b876a9c35b696bd018ddcc76847309cd" dependencies = [ "aho-corasick", "bstr", @@ -2959,9 +2959,9 @@ dependencies = [ [[package]] name = "ignore" -version = "0.4.23" +version = "0.4.29" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "6d89fd380afde86567dfba715db065673989d6253f42b88179abd3eae47bda4b" +checksum = "d4ffa3a0547a138e59ddd6fa3b7c672ed47e6ad6a3cd177984ff1116aa5ba742" dependencies = [ "crossbeam-deque", "globset", @@ -5410,6 +5410,7 @@ dependencies = [ "hex", "home", "humantime", + "ignore", "indexmap 2.11.0", "itertools 0.10.5", "jsonrpsee-types", @@ -5464,7 +5465,6 @@ dependencies = [ "tracing-subscriber", "ulid", "url", - "walkdir", "wasm-gen", "wasm-opt", "wasmparser 0.116.1", diff --git a/cmd/soroban-cli/src/commands/container/shared.rs b/cmd/soroban-cli/src/commands/container/shared.rs index b621aea9ca..d4435ab502 100644 --- a/cmd/soroban-cli/src/commands/container/shared.rs +++ b/cmd/soroban-cli/src/commands/container/shared.rs @@ -1,6 +1,8 @@ use core::fmt; +use std::process::Stdio; use clap::ValueEnum; +use tokio::io::{AsyncBufReadExt, AsyncReadExt, BufReader}; use tokio::process::Command; use crate::print::Print; @@ -20,6 +22,9 @@ pub enum Error { program: String, source: std::io::Error, }, + + #[error("could not pull image {image}: {stderr}")] + PullImageFailed { image: String, stderr: String }, } /// Container runtime to shell out to. @@ -191,7 +196,7 @@ impl Args { /// `--docker-host` (or `DOCKER_HOST` env) value is passed as `-H `; the /// `-H` flag outranks `DOCKER_CONTEXT`, so the override is honored even when a /// docker context is active. Host resolution is otherwise left to the CLI. - fn base_command(&self) -> Command { + pub(crate) fn base_command(&self) -> Command { let engine = self.engine(); let mut cmd = Command::new(engine.program()); if engine.supports_docker_host() { @@ -236,6 +241,59 @@ impl Args { }; cmd } + + /// Pull `image`, streaming the engine's high-level status lines ("Pulling + /// from", "Digest", "Status") through `print`. Per-layer progress written to + /// stderr is captured rather than shown and surfaced only when the pull + /// fails, as `PullImageFailed` — callers that need to explain a failed pull + /// (e.g. the verifiable build's tag-listing hint) rely on that captured text. + /// A missing engine binary surfaces via `io_error` as `NotFound`. + pub(crate) async fn pull_image(&self, image: &str, print: &Print) -> Result<(), Error> { + let mut child = self + .pull_command(image) + .stdout(Stdio::piped()) + .stderr(Stdio::piped()) + .spawn() + .map_err(|e| self.io_error(e))?; + + let stdout = child.stdout.take(); + let stderr = child.stderr.take(); + + let stream_stdout = async { + if let Some(stdout) = stdout { + let mut lines = BufReader::new(stdout).lines(); + while let Ok(Some(line)) = lines.next_line().await { + if line.contains("Pulling from") + || line.contains("Digest") + || line.contains("Status") + { + print.infoln(line); + } + } + } + }; + + let capture_stderr = async { + let mut buf = String::new(); + if let Some(mut stderr) = stderr { + let _ = stderr.read_to_string(&mut buf).await; + } + buf + }; + + // Drain both pipes concurrently so a full stderr buffer can't deadlock + // the child while we're reading stdout. + let ((), stderr) = tokio::join!(stream_stdout, capture_stderr); + + if child.wait().await.map_err(|e| self.io_error(e))?.success() { + Ok(()) + } else { + Err(Error::PullImageFailed { + image: image.to_string(), + stderr: stderr.trim().to_string(), + }) + } + } } /// Resource limits for commands that *run* a container (e.g. `container start`). @@ -458,7 +516,7 @@ mod test { let not_found = std::io::Error::from(std::io::ErrorKind::NotFound); match args(None, Some(Engine::AppleContainer)).io_error(not_found) { Error::NotFound { program, .. } => assert_eq!(program, "container"), - Error::Command { .. } => panic!("expected NotFound, got Command"), + other => panic!("expected NotFound, got {other:?}"), } } diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index ebda60f7d1..7ca08ec649 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -1,23 +1,17 @@ use std::path::{Path, PathBuf}; +use std::process::Stdio; -use bollard::{ - models::ContainerCreateBody, - query_parameters::{ - AttachContainerOptions, CreateContainerOptions, CreateImageOptions, StartContainerOptions, - WaitContainerOptions, - }, - service::HostConfig, - Docker, -}; use cargo_metadata::MetadataCommand; -use futures_util::{StreamExt, TryStreamExt}; use regex::Regex; use semver::Version; use serde::Deserialize; use sha2::{Digest, Sha256}; use crate::{ - commands::{container::shared::Error as ConnectionError, global}, + commands::{ + container::shared::{self, Error as ConnectionError}, + global, + }, config::{data, locator::enforce_hardened_tree}, print::Print, }; @@ -37,11 +31,8 @@ const OPTIMIZE_NEW_SYNTAX_MIN: &str = "26.1.0"; #[derive(thiserror::Error, Debug)] pub enum Error { - #[error("⛔ failed to connect to docker: {0}")] - DockerConnection(#[from] ConnectionError), - #[error(transparent)] - Bollard(#[from] bollard::errors::Error), + DockerConnection(#[from] ConnectionError), #[error("--image value {value:?} does not match the SEP-58 bldimg format `/@sha256:<64-hex>`. Examples: docker.io/stellar/stellar-cli@sha256:<64-hex>, localhost:5000/foo@sha256:<64-hex>. Tag-only refs and implicit Docker-Hub short refs are not accepted.")] BldimgFormat { value: String }, @@ -49,12 +40,12 @@ pub enum Error { #[error("could not determine the running rustc version: {0}")] RustcVersion(String), - #[error("could not pull image {tag}: {source}\n\nAvailable tags for this CLI version: {available_for_cli}\nAll published cli/rust pairs: {all_grouped}\n\nFix: install a matching rustc, or pass --image docker.io/stellar/stellar-cli@sha256: with one of the listed tags resolved to a digest.")] + #[error("could not pull image {tag}: {detail}\n\nAvailable tags for this CLI version: {available_for_cli}\nAll published cli/rust pairs: {all_grouped}\n\nFix: install a matching rustc, or pass --image docker.io/stellar/stellar-cli@sha256: with one of the listed tags resolved to a digest.")] ImageNotFound { tag: String, available_for_cli: String, all_grouped: String, - source: bollard::errors::Error, + detail: String, }, #[error("could not list published images on docker hub: {0}")] @@ -147,14 +138,15 @@ pub async fn run( source_sha256: Some(resolved.source_sha256.clone()), }; - // Stage 3: docker. - let docker_args = crate::commands::container::shared::Args { + // Stage 3: docker. Every docker interaction shells out to the container + // engine through this `Args` (honoring `--docker-host`). Verifiable builds + // pin the engine to docker (`engine: None` → the default): the probes, + // `inspect`, and reproduce lines below are docker-specific, and SEP-58 + // reproducibility depends on that exact toolchain. + let docker = shared::Args { docker_host: cmd.docker_host.clone(), + engine: None, }; - let docker = docker_args - .connect_to_docker(print) - .await - .map_err(Error::DockerConnection)?; let image_ref = resolve_image(cmd, &docker, print).await?; // Only probe the container's cli version when we need to pick between @@ -540,16 +532,20 @@ fn compose_container_args(forwarded: &[String], metadata: &[String]) -> Vec Result { +pub async fn resolve_image( + cmd: &Cmd, + docker: &shared::Args, + print: &Print, +) -> Result { if let Some(s) = &cmd.image { if !bldimg_regex().is_match(s) { return Err(Error::BldimgFormat { value: s.clone() }); } // Always pull, even when the digest is user-supplied. Docker requires - // the image to be locally present before `create_container` will - // accept it, and the user typically expects the cli to fetch - // whatever they asked for. - pull_image(docker, s, print).await?; + // the image to be locally present before `docker run` will accept it, + // and the user typically expects the cli to fetch whatever they asked + // for. + docker.pull_image(s, print).await?; return Ok(s.clone()); } @@ -560,11 +556,13 @@ pub async fn resolve_image(cmd: &Cmd, docker: &Docker, print: &Print) -> Result< let tag = format!("{REGISTRY}:{cli_v}-rust{rust_v}"); print.infoln(format!("Pulling verifiable build image {tag}")); - let pull = pull_image(docker, &tag, print).await; - match pull { + match docker.pull_image(&tag, print).await { Ok(()) => {} - Err(e) => { + // A failed pull of the derived cli/rust tag usually means no image was + // published for this pair; turn it into the tag-listing hint. A missing + // `docker` binary (or other connection failure) propagates as-is. + Err(ConnectionError::PullImageFailed { stderr, .. }) => { let (available_for_cli, all_grouped) = match list_published_tags().await { Ok(tags) => format_available(&tags, cli_v), Err(list_err) => ( @@ -576,53 +574,33 @@ pub async fn resolve_image(cmd: &Cmd, docker: &Docker, print: &Print) -> Result< tag, available_for_cli, all_grouped, - source: e, + detail: stderr, }); } + Err(e) => return Err(Error::DockerConnection(e)), } - let inspect = docker.inspect_image(&tag).await?; - let digest = inspect - .repo_digests - .and_then(|v| v.into_iter().next()) - .ok_or_else(|| Error::NoRepoDigest { tag: tag.clone() })?; - Ok(digest) + image_repo_digest(docker, &tag).await } -async fn pull_image( - docker: &Docker, - tag: &str, - print: &Print, -) -> Result<(), bollard::errors::Error> { - let mut stream = docker.create_image( - Some(CreateImageOptions { - from_image: Some(tag.to_string()), - ..Default::default() - }), - None, - None, - ); - while let Some(item) = stream.try_next().await? { - if let Some(status) = item.status { - // The docker daemon emits short status lines like: - // "Pulling from " - // "Digest: sha256:" - // "Status: Image is up to date for " - // Stand-alone "Digest" reads as an orphan. Rewrite each line so - // it makes sense outside the docker-pull context. - if let Some(repo) = status.strip_prefix("Pulling from ") { - print.infoln(format!("Pulling image {repo}")); - } else if let Some(digest) = status.strip_prefix("Digest: ") { - print.infoln(format!("Image digest: {digest}")); - } else if let Some(rest) = status.strip_prefix("Status: ") { - // Docker's status text already starts with "Image …" or - // "Downloaded …", so we forward it verbatim instead of - // prepending another "Image:". - print.infoln(rest); - } - } +/// Resolve a locally-present image to its content-addressed repo digest +/// (`@sha256:`) via `docker inspect`, so the recorded `bldimg` pins +/// the exact bytes that were pulled rather than a mutable tag. +async fn image_repo_digest(docker: &shared::Args, tag: &str) -> Result { + let output = docker + .base_command() + .args(["inspect", "--format", "{{index .RepoDigests 0}}", tag]) + .output() + .await + .map_err(|e| docker.io_error(e))?; + + let digest = String::from_utf8_lossy(&output.stdout).trim().to_string(); + if !output.status.success() || digest.is_empty() || digest == "" { + return Err(Error::NoRepoDigest { + tag: tag.to_string(), + }); } - Ok(()) + Ok(digest) } #[derive(Debug, Clone)] @@ -721,7 +699,7 @@ fn format_available(tags: &[PublishedTag], current_cli: &str) -> (String, String /// false — the conservative assumption that the container is old. async fn probe_supports_optimize_false_syntax( image_ref: &str, - docker: &Docker, + docker: &shared::Args, print: &Print, ) -> bool { match probe_cli_version(image_ref, docker).await { @@ -738,61 +716,30 @@ async fn probe_supports_optimize_false_syntax( } } -/// Run `cmd` in a throwaway container (optionally overriding the entrypoint) and -/// return its captured stdout. The container auto-removes; stderr is attached so -/// the daemon streams it, but only stdout is collected. Shared by every image -/// probe (cli version, active toolchain, flag support). +/// Run `cmd` in a throwaway `docker run --rm` container (optionally overriding +/// the entrypoint) and return its captured stdout. Only stdout is collected; +/// stderr and the exit status are ignored, matching how every probe treats a +/// missing subcommand or unexpected output as "unsupported". Shared by every +/// image probe (cli version, active toolchain, flag support). async fn run_probe( image_ref: &str, - docker: &Docker, - entrypoint: Option>, + docker: &shared::Args, + entrypoint: Option<&str>, cmd: Vec, ) -> Result { - let config = ContainerCreateBody { - image: Some(image_ref.to_string()), - entrypoint, - cmd: Some(cmd), - attach_stdout: Some(true), - attach_stderr: Some(true), - host_config: Some(HostConfig { - auto_remove: Some(true), - ..Default::default() - }), - ..Default::default() - }; - let created = docker - .create_container(None::, config) - .await?; - let attached = docker - .attach_container( - &created.id, - Some(AttachContainerOptions { - stdout: true, - stderr: true, - stream: true, - ..Default::default() - }), - ) - .await?; - docker - .start_container(&created.id, None::) - .await?; - - let mut stdout = String::new(); - let mut output = attached.output; - while let Some(chunk) = output.next().await { - if let Ok(bollard::container::LogOutput::StdOut { message }) = chunk { - stdout.push_str(&String::from_utf8_lossy(&message)); - } + let mut command = docker.base_command(); + command.args(["run", "--rm"]); + if let Some(entrypoint) = entrypoint { + command.args(["--entrypoint", entrypoint]); } + command.arg(image_ref); + command.args(&cmd); - let mut wait = docker.wait_container(&created.id, None::); - while wait.next().await.is_some() {} - - Ok(stdout) + let output = command.output().await.map_err(|e| docker.io_error(e))?; + Ok(String::from_utf8_lossy(&output.stdout).into_owned()) } -async fn probe_cli_version(image_ref: &str, docker: &Docker) -> Result { +async fn probe_cli_version(image_ref: &str, docker: &shared::Args) -> Result { let stdout = run_probe( image_ref, docker, @@ -810,7 +757,11 @@ async fn probe_cli_version(image_ref: &str, docker: &Docker) -> Result bool { +pub(crate) async fn probe_supports_locked( + image_ref: &str, + docker: &shared::Args, + print: &Print, +) -> bool { match run_probe( image_ref, docker, @@ -840,11 +791,11 @@ pub(crate) async fn probe_supports_locked(image_ref: &str, docker: &Docker, prin /// `(default)` marker (e.g. `1.93.0-x86_64-unknown-linux-gnu`). Returns `None` /// on any failure (e.g. an image without rustup), so the build proceeds without /// the pin rather than failing. -async fn probe_active_toolchain(image_ref: &str, docker: &Docker) -> Option { +async fn probe_active_toolchain(image_ref: &str, docker: &shared::Args) -> Option { let stdout = run_probe( image_ref, docker, - Some(vec!["rustup".to_string()]), + Some("rustup"), vec!["show".to_string(), "active-toolchain".to_string()], ) .await @@ -886,7 +837,7 @@ async fn run_in_container( workspace_root: &Path, container_cmds: &[Vec], env: &[String], - docker: &Docker, + docker: &shared::Args, print: &Print, verbose: bool, ) -> Result<(), Error> { @@ -912,20 +863,18 @@ async fn run_in_container( env_flags.push_str(&shell_escape::escape(e.as_str().into())); } - // One package → run the image's default `stellar` entrypoint directly. - // Several → override the entrypoint to a shell and chain the builds so they - // all run in this one container. - let (entrypoint, cmd, reproduce) = if container_cmds.len() > 1 { + // One package → run the image's default `stellar` entrypoint directly, so + // `post_image` is just the `contract build …` argv. Several → override the + // entrypoint to a shell and chain the builds so they all run in this one + // container; `--entrypoint` takes only the executable, so the `-c ` + // arguments follow the image name in `post_image`. + let (entrypoint, post_image, reproduce) = if container_cmds.len() > 1 { let chain = compose_shell_command(container_cmds); let reproduce = format!( "docker run --rm -v {bind}{env_flags} --entrypoint /bin/sh {image_ref} -c {}", shell_escape::escape(chain.clone().into()) ); - ( - Some(vec!["/bin/sh".to_string(), "-c".to_string()]), - vec![chain], - reproduce, - ) + (Some("/bin/sh"), vec!["-c".to_string(), chain], reproduce) } else { let cmd = container_cmds.first().cloned().unwrap_or_default(); let reproduce = format!( @@ -935,22 +884,6 @@ async fn run_in_container( (None, cmd, reproduce) }; - let config = ContainerCreateBody { - image: Some(image_ref.to_string()), - entrypoint, - cmd: Some(cmd), - env: (!env.is_empty()).then(|| env.clone()), - working_dir: Some("/source".to_string()), - attach_stdout: Some(true), - attach_stderr: Some(true), - host_config: Some(HostConfig { - auto_remove: Some(true), - binds: Some(vec![bind.clone()]), - ..Default::default() - }), - ..Default::default() - }; - print.infoln(format!( "Running verifiable build in {image_ref} (mount {bind})" )); @@ -958,65 +891,34 @@ async fn run_in_container( print.infoln(format!("Running: {reproduce}")); } - let created = docker - .create_container(None::, config) - .await?; - - let attached = docker - .attach_container( - &created.id, - Some(AttachContainerOptions { - stdout: true, - stderr: true, - stream: true, - ..Default::default() - }), - ) - .await?; - - docker - .start_container(&created.id, None::) - .await?; - - let mut output = attached.output; - while let Some(chunk) = output.next().await { - match chunk { - Ok( - bollard::container::LogOutput::StdOut { message } - | bollard::container::LogOutput::StdErr { message }, - ) => { - if verbose { - let s = String::from_utf8_lossy(&message); - print.blankln(s.trim_end()); - } - } - Ok(_) => {} - Err(e) => return Err(e.into()), - } + let mut command = docker.base_command(); + command.args(["run", "--rm", "-v", &bind, "-w", "/source"]); + for e in &env { + command.args(["-e", e]); + } + if let Some(entrypoint) = entrypoint { + command.args(["--entrypoint", entrypoint]); } + command.arg(image_ref); + command.args(&post_image); - wait_for_container_exit(docker, &created.id, &reproduce).await -} + // Stream the build's cargo output straight to the terminal when verbose + // (matching a non-verifiable `contract build`); otherwise discard it (the + // verify pipeline suppresses per-build noise). `quiet` overrides verbose. + let show_output = verbose && !print.quiet; + let (stdout, stderr) = if show_output { + (Stdio::inherit(), Stdio::inherit()) + } else { + (Stdio::null(), Stdio::null()) + }; + command.stdout(stdout).stderr(stderr); -/// Block until the container exits, mapping a non-zero exit code (whether -/// reported as a successful wait or as a `DockerContainerWaitError`) to -/// `ContainerExit` carrying the reproduce command. -async fn wait_for_container_exit(docker: &Docker, id: &str, reproduce: &str) -> Result<(), Error> { - let mut wait = docker.wait_container(id, None::); - while let Some(item) = wait.next().await { - // Both a successful wait and a `DockerContainerWaitError` carry an exit - // code; normalize to it (other errors are genuine failures). - let status = match item { - Ok(r) => r.status_code, - Err(bollard::errors::Error::DockerContainerWaitError { code, .. }) => code, - Err(e) => return Err(e.into()), - }; - if status != 0 { - return Err(Error::ContainerExit { - status, - command: reproduce.to_string(), - }); - } + let status = command.status().await.map_err(|e| docker.io_error(e))?; + if !status.success() { + return Err(Error::ContainerExit { + status: status.code().unwrap_or(-1).into(), + command: reproduce, + }); } Ok(()) From b55638291f20fe2d877e1a20e1feb27ec7050c6c Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Thu, 16 Jul 2026 21:27:47 -0300 Subject: [PATCH 24/25] Support other container engines in verifiable builds. --- FULL_HELP_DOCS.md | 15 +- .../src/commands/container/shared.rs | 149 +++++++++++++++++- .../src/commands/contract/build.rs | 32 ++-- .../src/commands/contract/build/verifiable.rs | 66 ++++---- cmd/soroban-cli/src/commands/mod.rs | 2 + 5 files changed, 218 insertions(+), 46 deletions(-) diff --git a/FULL_HELP_DOCS.md b/FULL_HELP_DOCS.md index 4d10b321b9..f5b1b7dab5 100644 --- a/FULL_HELP_DOCS.md +++ b/FULL_HELP_DOCS.md @@ -368,6 +368,18 @@ To view the commands that will be executed, without executing them, use the --pr **Usage:** `stellar contract build [OPTIONS]` +###### **Container Options:** + +- `-d`, `--docker-host ` — Optional argument to override the default docker host. This is useful when you are using a non-standard docker host path for your Docker-compatible container runtime, e.g. Docker Desktop defaults to $HOME/.docker/run/docker.sock instead of /var/run/docker.sock +- `--engine ` — Container engine to use [default: docker] + + Possible values: + - `docker`: Docker, or any Docker-compatible CLI + - `apple-container`: Apple's `container` CLI (macOS 26+, Apple silicon) + +- `--cpus ` — Limit the number of CPUs available to the container, e.g. `2`. A whole number: Apple's `container` engine does not accept fractional CPUs +- `--memory ` — Limit the memory available to the container, e.g. `2g` or `512m` + ###### **Features:** - `--features ` — Build with the list of features activated, space or comma separated @@ -407,13 +419,12 @@ To view the commands that will be executed, without executing them, use the --pr - `--print-commands-only` — Print commands to build without executing them -###### **Verifiable:** +###### **Verifiable Options:** - `--verifiable` — Build inside a trusted Docker container and record SEP-58 metadata (`bldimg`, `source_uri`, `source_sha256`, `bldopt`) so the resulting WASM can be reproduced and verified by third parties. Implies `--locked`. Requires a clean git working tree - `--image ` — Override the auto-selected container image used by `--verifiable`. Must be digest-pinned, e.g. `docker.io/stellar/stellar-cli@sha256:...`. Tag-only refs are rejected because SEP-58 requires content addressing - `--source-sha256 ` — SEP-58 source identification: SHA-256 of the source archive (recorded as the `source_sha256` meta entry). Optional with `--verifiable`: the archive is always generated and its SHA-256 computed for you. When supplied it's treated as a pin — the build fails if it doesn't match the generated archive - `--source-uri ` — SEP-58 source identification: URI where the source can be obtained, e.g. `https://example.com/src.tar.gz` (recorded as the `source_uri` meta entry). Optional; when set it must accompany `--source-sha256` -- `-d`, `--docker-host ` — Override the default docker host used by `--verifiable` ## `stellar contract extend` diff --git a/cmd/soroban-cli/src/commands/container/shared.rs b/cmd/soroban-cli/src/commands/container/shared.rs index d4435ab502..000e9838ce 100644 --- a/cmd/soroban-cli/src/commands/container/shared.rs +++ b/cmd/soroban-cli/src/commands/container/shared.rs @@ -99,6 +99,44 @@ impl Engine { Engine::AppleContainer => stderr.contains("not found"), } } + + /// The `inspect`-family argv (after the engine binary and any host flag) + /// that prints an image's digest metadata: docker's `RepoDigests` Go + /// template vs Apple's `image inspect` JSON (Apple groups image operations + /// under the `image` subcommand and has no `--format` templates). + fn image_inspect_args(self, image: &str) -> Vec<&str> { + match self { + Engine::Docker => vec!["inspect", "--format", "{{index .RepoDigests 0}}", image], + Engine::AppleContainer => vec!["image", "inspect", image], + } + } + + /// Parse the stdout of [`image_inspect_args`] into a content-addressed + /// `@sha256:` reference, or `None` when the engine reports no + /// digest (e.g. a locally-built image never pushed or pulled). + fn parse_repo_digest(self, stdout: &[u8], image: &str) -> Option { + match self { + Engine::Docker => { + let digest = String::from_utf8_lossy(stdout).trim().to_string(); + (!digest.is_empty() && digest != "").then_some(digest) + } + // Apple emits a JSON array whose first entry carries the manifest-list + // descriptor at `configuration.descriptor.digest` — the equivalent of + // docker's `RepoDigests`. The per-platform `variants[].digest` is + // deliberately not used. `None` if the output doesn't have that shape. + Engine::AppleContainer => { + let value: serde_json::Value = serde_json::from_slice(stdout).ok()?; + let digest = value + .as_array()? + .first()? + .get("configuration")? + .get("descriptor")? + .get("digest")? + .as_str()?; + Some(format!("{}@{digest}", repo_of(image))) + } + } + } } impl fmt::Display for Engine { @@ -113,7 +151,7 @@ impl fmt::Display for Engine { } } -#[derive(Debug, clap::Parser, Clone)] +#[derive(Debug, clap::Parser, Clone, Default)] pub struct Args { /// Optional argument to override the default docker host. This is useful when you are using a non-standard docker host path for your Docker-compatible container runtime, e.g. Docker Desktop defaults to $HOME/.docker/run/docker.sock instead of /var/run/docker.sock #[arg(short = 'd', long, help = DOCKER_HOST_HELP, env = "DOCKER_HOST")] @@ -294,6 +332,46 @@ impl Args { }) } } + + /// The engine's executable name (`docker`, `container`), for rendering + /// copy-pasteable reproduce commands that name the same binary the CLI ran. + pub(crate) fn program(&self) -> &'static str { + self.engine().program() + } + + /// Resolve a locally-present image to its content-addressed repo digest + /// (`@sha256:`), so a caller can pin the exact bytes rather than a + /// mutable tag. Returns `Ok(None)` when the engine reports no digest (e.g. a + /// locally-built image that was never pushed or pulled). The per-engine + /// `inspect` argv and output parsing live on [`Engine`]; this owns only the + /// command execution (the engine binary and `--docker-host`). + pub(crate) async fn image_repo_digest(&self, image: &str) -> Result, Error> { + let engine = self.engine(); + let output = self + .base_command() + .args(engine.image_inspect_args(image)) + .output() + .await + .map_err(|e| self.io_error(e))?; + if !output.status.success() { + return Ok(None); + } + Ok(engine.parse_repo_digest(&output.stdout, image)) + } +} + +/// The repo portion of an image reference: everything before the `:tag` (or +/// `@digest`). A `:` only separates a tag when it appears after the last `/`, so +/// a registry host's `:port` (e.g. `localhost:5000/foo`) is preserved. +fn repo_of(image: &str) -> &str { + if let Some((repo, _)) = image.split_once('@') { + return repo; + } + let last_slash = image.rfind('/').map_or(0, |i| i + 1); + match image[last_slash..].find(':') { + Some(colon) => &image[..last_slash + colon], + None => image, + } } /// Resource limits for commands that *run* a container (e.g. `container start`). @@ -545,6 +623,63 @@ mod test { assert!(!apple.is_container_not_found("some unrelated failure")); } + #[test] + fn program_matches_engine_binary() { + assert_eq!(args(None, None).program(), "docker"); + assert_eq!( + args(None, Some(Engine::AppleContainer)).program(), + "container" + ); + } + + #[test] + fn repo_of_strips_tag_but_keeps_registry_port() { + assert_eq!( + repo_of("docker.io/stellar/stellar-cli:26.1.0-rust1.90.0"), + "docker.io/stellar/stellar-cli" + ); + assert_eq!(repo_of("localhost:5000/foo:bar"), "localhost:5000/foo"); + assert_eq!(repo_of("localhost:5000/foo"), "localhost:5000/foo"); + // An already digest-pinned ref keeps its repo. + assert_eq!( + repo_of(&format!( + "docker.io/stellar/stellar-cli@sha256:{}", + "a".repeat(64) + )), + "docker.io/stellar/stellar-cli" + ); + } + + #[test] + fn apple_repo_digest_reads_manifest_list_descriptor() { + // Shape mirrors real `container image inspect` output: the top-level + // manifest-list digest lives at [0].configuration.descriptor.digest, + // while the per-platform digest under variants[] must be ignored. + let list = format!("sha256:{}", "8d".repeat(32)); + let variant = format!("sha256:{}", "85".repeat(32)); + let json = format!( + r#"[{{"configuration":{{"descriptor":{{"digest":"{list}","mediaType":"application/vnd.docker.distribution.manifest.list.v2+json","size":743}},"name":"docker.io/stellar/quickstart:latest"}},"id":"8ddf","variants":[{{"digest":"{variant}","platform":{{"architecture":"arm64","os":"linux"}}}}]}}]"# + ); + assert_eq!( + Engine::AppleContainer + .parse_repo_digest(json.as_bytes(), "docker.io/stellar/quickstart:latest"), + Some(format!("docker.io/stellar/quickstart@{list}")) + ); + } + + #[test] + fn docker_parse_repo_digest_trims_and_rejects_no_value() { + assert_eq!( + Engine::Docker.parse_repo_digest(b" docker.io/stellar/cli@sha256:abc\n", "ignored"), + Some("docker.io/stellar/cli@sha256:abc".to_string()) + ); + assert_eq!( + Engine::Docker.parse_repo_digest(b"\n", "ignored"), + None + ); + assert_eq!(Engine::Docker.parse_repo_digest(b" \n", "ignored"), None); + } + #[test] fn run_args_flags_emit_only_set_limits() { assert!(RunArgs::default().flags().is_empty()); @@ -565,4 +700,16 @@ mod test { ["--cpus", "2", "--memory", "2g"] ); } + + #[test] + fn apple_repo_digest_none_when_shape_unexpected() { + let apple = Engine::AppleContainer; + assert_eq!(apple.parse_repo_digest(b"[]", "foo:bar"), None); + assert_eq!(apple.parse_repo_digest(b"not json", "foo:bar"), None); + // Missing the configuration.descriptor.digest path. + assert_eq!( + apple.parse_repo_digest(br#"[{"id":"8ddf"}]"#, "foo:bar"), + None + ); + } } diff --git a/cmd/soroban-cli/src/commands/contract/build.rs b/cmd/soroban-cli/src/commands/contract/build.rs index 56178b4742..b52bf057ed 100644 --- a/cmd/soroban-cli/src/commands/contract/build.rs +++ b/cmd/soroban-cli/src/commands/contract/build.rs @@ -20,7 +20,10 @@ use stellar_xdr::{Limited, Limits, ScMetaEntry, ScMetaV0, StringM, WriteXdr}; #[cfg(feature = "additional-libs")] use crate::commands::contract::optimize; use crate::{ - commands::{global, version}, + commands::{ + container::shared::{Args as ContainerArgs, RunArgs as ContainerRunArgs}, + global, version, HEADING_CONTAINER, HEADING_VERIFIABLE, + }, print::Print, wasm, }; @@ -103,13 +106,13 @@ pub struct Cmd { /// (`bldimg`, `source_uri`, `source_sha256`, `bldopt`) so the resulting /// WASM can be reproduced and verified by third parties. Implies /// `--locked`. Requires a clean git working tree. - #[arg(long, help_heading = "Verifiable")] + #[arg(long, help_heading = HEADING_VERIFIABLE)] pub verifiable: bool, /// Override the auto-selected container image used by `--verifiable`. /// Must be digest-pinned, e.g. `docker.io/stellar/stellar-cli@sha256:...`. /// Tag-only refs are rejected because SEP-58 requires content addressing. - #[arg(long, requires = "verifiable", help_heading = "Verifiable")] + #[arg(long, requires = "verifiable", help_heading = HEADING_VERIFIABLE)] pub image: Option, /// SEP-58 source identification: SHA-256 of the source archive @@ -117,7 +120,7 @@ pub struct Cmd { /// `--verifiable`: the archive is always generated and its SHA-256 computed /// for you. When supplied it's treated as a pin — the build fails if it /// doesn't match the generated archive. - #[arg(long, requires = "verifiable", help_heading = "Verifiable")] + #[arg(long, requires = "verifiable", help_heading = HEADING_VERIFIABLE)] pub source_sha256: Option, /// SEP-58 source identification: URI where the source can be obtained, e.g. @@ -127,16 +130,24 @@ pub struct Cmd { long, requires = "verifiable", requires = "source_sha256", - help_heading = "Verifiable" + help_heading = HEADING_VERIFIABLE )] pub source_uri: Option, - /// Override the default docker host used by `--verifiable`. - #[arg(short = 'd', long, env = "DOCKER_HOST", help_heading = "Verifiable")] - pub docker_host: Option, - #[command(flatten)] pub build_args: BuildArgs, + + // Declared last so their `next_help_heading` groups them under the Container + // heading without leaking it onto the ungrouped `build_args` flags above. + /// Container connection options (`--engine`, `--docker-host`) used by + /// `--verifiable`. `--docker-host` is honored only by the docker engine. + #[command(flatten, next_help_heading = HEADING_CONTAINER)] + pub container_args: ContainerArgs, + + /// Container resource limits (`--cpus`, `--memory`) applied to the + /// `--verifiable` build container. + #[command(flatten, next_help_heading = HEADING_CONTAINER)] + pub run_args: ContainerRunArgs, } /// Shared build options for meta and optimization, reused by deploy and upload. @@ -313,7 +324,8 @@ impl Default for Cmd { image: None, source_sha256: None, source_uri: None, - docker_host: None, + container_args: ContainerArgs::default(), + run_args: ContainerRunArgs::default(), build_args: BuildArgs::default(), } } diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 7ca08ec649..25077a9d68 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -138,15 +138,13 @@ pub async fn run( source_sha256: Some(resolved.source_sha256.clone()), }; - // Stage 3: docker. Every docker interaction shells out to the container - // engine through this `Args` (honoring `--docker-host`). Verifiable builds - // pin the engine to docker (`engine: None` → the default): the probes, - // `inspect`, and reproduce lines below are docker-specific, and SEP-58 - // reproducibility depends on that exact toolchain. - let docker = shared::Args { - docker_host: cmd.docker_host.clone(), - engine: None, - }; + // Stage 3: the container engine. Every interaction shells out through these + // `container_args`, which select the engine binary (`--engine`/ + // `STELLAR_CONTAINER_ENGINE`, default docker) and honor `--docker-host` where + // the engine supports it; `run_args` carries the build container's resource + // limits. + let docker = cmd.container_args.clone(); + docker.warn_if_host_ignored(print); let image_ref = resolve_image(cmd, &docker, print).await?; // Only probe the container's cli version when we need to pick between @@ -221,6 +219,7 @@ pub async fn run( &container_cmds, &env, &docker, + &cmd.run_args, print, true, ) @@ -580,27 +579,12 @@ pub async fn resolve_image( Err(e) => return Err(Error::DockerConnection(e)), } - image_repo_digest(docker, &tag).await -} - -/// Resolve a locally-present image to its content-addressed repo digest -/// (`@sha256:`) via `docker inspect`, so the recorded `bldimg` pins -/// the exact bytes that were pulled rather than a mutable tag. -async fn image_repo_digest(docker: &shared::Args, tag: &str) -> Result { - let output = docker - .base_command() - .args(["inspect", "--format", "{{index .RepoDigests 0}}", tag]) - .output() - .await - .map_err(|e| docker.io_error(e))?; - - let digest = String::from_utf8_lossy(&output.stdout).trim().to_string(); - if !output.status.success() || digest.is_empty() || digest == "" { - return Err(Error::NoRepoDigest { - tag: tag.to_string(), - }); - } - Ok(digest) + // Pin the mutable tag to the content-addressed digest the engine resolved, + // so the recorded `bldimg` names the exact bytes that were pulled. + docker + .image_repo_digest(&tag) + .await? + .ok_or(Error::NoRepoDigest { tag }) } #[derive(Debug, Clone)] @@ -832,12 +816,14 @@ fn escape_container_args(cmd: &[String]) -> String { .join(" ") } +#[allow(clippy::too_many_arguments)] async fn run_in_container( image_ref: &str, workspace_root: &Path, container_cmds: &[Vec], env: &[String], docker: &shared::Args, + run_args: &shared::RunArgs, print: &Print, verbose: bool, ) -> Result<(), Error> { @@ -863,6 +849,18 @@ async fn run_in_container( env_flags.push_str(&shell_escape::escape(e.as_str().into())); } + // Render the reproduce line against the engine binary the CLI actually ran + // (`docker`, `container`), so a third party can replay the exact build. + let program = docker.program(); + + // Resource limits go right after `--rm`, matching where they're applied to + // the spawned command below, so the reproduce line stays copy-paste faithful. + let mut run_flags = String::new(); + for f in run_args.flags() { + run_flags.push(' '); + run_flags.push_str(&shell_escape::escape(f.into())); + } + // One package → run the image's default `stellar` entrypoint directly, so // `post_image` is just the `contract build …` argv. Several → override the // entrypoint to a shell and chain the builds so they all run in this one @@ -871,14 +869,14 @@ async fn run_in_container( let (entrypoint, post_image, reproduce) = if container_cmds.len() > 1 { let chain = compose_shell_command(container_cmds); let reproduce = format!( - "docker run --rm -v {bind}{env_flags} --entrypoint /bin/sh {image_ref} -c {}", + "{program} run --rm{run_flags} -v {bind}{env_flags} --entrypoint /bin/sh {image_ref} -c {}", shell_escape::escape(chain.clone().into()) ); (Some("/bin/sh"), vec!["-c".to_string(), chain], reproduce) } else { let cmd = container_cmds.first().cloned().unwrap_or_default(); let reproduce = format!( - "docker run --rm -v {bind}{env_flags} {image_ref} {}", + "{program} run --rm{run_flags} -v {bind}{env_flags} {image_ref} {}", escape_container_args(&cmd) ); (None, cmd, reproduce) @@ -892,7 +890,9 @@ async fn run_in_container( } let mut command = docker.base_command(); - command.args(["run", "--rm", "-v", &bind, "-w", "/source"]); + command.args(["run", "--rm"]); + run_args.apply(&mut command); + command.args(["-v", &bind, "-w", "/source"]); for e in &env { command.args(["-e", e]); } diff --git a/cmd/soroban-cli/src/commands/mod.rs b/cmd/soroban-cli/src/commands/mod.rs index 9367463625..2e879e8f0d 100644 --- a/cmd/soroban-cli/src/commands/mod.rs +++ b/cmd/soroban-cli/src/commands/mod.rs @@ -32,6 +32,8 @@ pub const HEADING_ARCHIVE: &str = "Archive Options"; pub const HEADING_GLOBAL: &str = "Global Options"; pub const HEADING_SIGNING: &str = "Signing Options"; pub const HEADING_TRANSACTION: &str = "Transaction Options"; +pub const HEADING_CONTAINER: &str = "Container Options"; +pub const HEADING_VERIFIABLE: &str = "Verifiable Options"; const ABOUT: &str = "Work seamlessly with Stellar accounts, contracts, and assets from the command line. From 601814250968d1b083fc5324630a6bc13108e25b Mon Sep 17 00:00:00 2001 From: Nando Vieira Date: Fri, 17 Jul 2026 11:38:47 -0300 Subject: [PATCH 25/25] Stop the build container when the build is interrupted. --- .../src/commands/container/shared.rs | 10 +++ .../src/commands/contract/build/verifiable.rs | 79 ++++++++++++++++++- 2 files changed, 87 insertions(+), 2 deletions(-) diff --git a/cmd/soroban-cli/src/commands/container/shared.rs b/cmd/soroban-cli/src/commands/container/shared.rs index 000e9838ce..af44909d56 100644 --- a/cmd/soroban-cli/src/commands/container/shared.rs +++ b/cmd/soroban-cli/src/commands/container/shared.rs @@ -270,6 +270,16 @@ impl Args { cmd } + /// Immediately terminate a running container (SIGKILL), with no graceful + /// grace period — unlike `stop`, which waits up to 10s before force-killing. + /// Used to tear down a build container the instant the CLI is interrupted. + /// Both docker and Apple's `container` accept `kill `. + pub(crate) fn kill_command(&self, name: &str) -> Command { + let mut cmd = self.base_command(); + cmd.args(["kill", name]); + cmd + } + pub(crate) fn logs_command(&self, name: &str) -> Command { let mut cmd = self.base_command(); match self.engine() { diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs index 25077a9d68..9baea2cdf3 100644 --- a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -79,6 +79,9 @@ pub enum Error { #[error("container build exited with status {status}. To reproduce manually:\n {command}")] ContainerExit { status: i64, command: String }, + + #[error("verifiable build interrupted; stopped the build container")] + Interrupted, } pub async fn run( @@ -816,6 +819,48 @@ fn escape_container_args(cmd: &[String]) -> String { .join(" ") } +/// Resolve once the process receives any catchable signal that would otherwise +/// terminate it, so the caller can stop the build container before exiting. +/// `SIGKILL` can't be caught, so a `kill -9` still orphans the container — +/// nothing in-process can prevent that. On non-Unix platforms only Ctrl-C is +/// observable. +#[cfg(unix)] +async fn wait_for_termination_signal() { + use tokio::signal::unix::{signal, SignalKind}; + + // If any handler fails to install we simply never resolve on that signal; + // the build still runs, it just won't self-clean on that particular signal. + let mut sigint = signal(SignalKind::interrupt()); + let mut sigterm = signal(SignalKind::terminate()); + let mut sighup = signal(SignalKind::hangup()); + let mut sigquit = signal(SignalKind::quit()); + + tokio::select! { + () = recv_signal(&mut sigint) => {}, + () = recv_signal(&mut sigterm) => {}, + () = recv_signal(&mut sighup) => {}, + () = recv_signal(&mut sigquit) => {}, + } +} + +/// Await one delivery of an installed signal. When the handler failed to +/// install, never resolves, so it drops out of the `select!` above rather than +/// firing spuriously. +#[cfg(unix)] +async fn recv_signal(s: &mut std::io::Result) { + match s { + Ok(s) => { + s.recv().await; + } + Err(_) => std::future::pending().await, + } +} + +#[cfg(not(unix))] +async fn wait_for_termination_signal() { + let _ = tokio::signal::ctrl_c().await; +} + #[allow(clippy::too_many_arguments)] async fn run_in_container( image_ref: &str, @@ -889,8 +934,22 @@ async fn run_in_container( print.infoln(format!("Running: {reproduce}")); } + // Name the build container so it can be stopped if the CLI is interrupted. + // Without a name there's no handle to target: on a termination signal the + // CLI process dies, but the container the daemon owns keeps running (the + // engine client exiting doesn't stop it, and signal-forwarding through the + // client is unreliable for a long cargo build). `--rm` removes it once + // stopped. The name is unique per invocation (pid + random) so concurrent + // builds don't collide, and it's kept out of the reproduce line — a fixed + // name there would clash on re-run. + let container_name = format!( + "stellar-verifiable-build-{}-{:08x}", + std::process::id(), + rand::random::() + ); + let mut command = docker.base_command(); - command.args(["run", "--rm"]); + command.args(["run", "--rm", "--name", &container_name]); run_args.apply(&mut command); command.args(["-v", &bind, "-w", "/source"]); for e in &env { @@ -913,7 +972,23 @@ async fn run_in_container( }; command.stdout(stdout).stderr(stderr); - let status = command.status().await.map_err(|e| docker.io_error(e))?; + let mut child = command.spawn().map_err(|e| docker.io_error(e))?; + + // Race the build against any catchable termination signal. On a signal, + // stop the named container (best-effort) so it doesn't outlive the CLI, + // kill the engine client we spawned, then surface the interruption. + let status = tokio::select! { + result = child.wait() => result.map_err(|e| docker.io_error(e))?, + () = wait_for_termination_signal() => { + print.warnln("Interrupted; stopping build container"); + // `kill` (SIGKILL, immediate) rather than `stop` (SIGTERM + 10s + // grace): otherwise the container keeps building for the whole grace + // period while we block here. + let _ = docker.kill_command(&container_name).output().await; + let _ = child.start_kill(); + return Err(Error::Interrupted); + } + }; if !status.success() { return Err(Error::ContainerExit { status: status.code().unwrap_or(-1).into(),