[zuudevs]

Summary

Completes HTTP/1.1 response body framing for Transfer-Encoding: chunked and integrates it directly into the incremental response parser.

Added

• src/http/chunked_decoder.hpp / .cpp
  • incremental chunk-size parsing in hexadecimal
  • arbitrary TCP/feed boundaries
  • chunk payload accumulation without exposing framing bytes
  • mandatory CRLF validation after each chunk payload
  • chunk-extension consumption
  • terminal zero-size chunk handling
  • trailer-section validation and consumption
  • rejects malformed chunk sizes/framing and forbidden framing fields in trailers
  • preserves bytes following the terminal trailer block for connection reuse / the next response
• ResponseParser integration
  • valid Transfer-Encoding: chunked is decoded internally instead of returning a handoff state
  • parser now returns only NeedMore or Complete
  • decoded chunk bytes populate the normal owning Response body
  • premature EOF during chunked framing maps to UnexpectedEof
  • chunked responses remain reusable when normal HTTP connection semantics permit it
• tests
  • multiple chunks
  • one-byte-at-a-time decoding
  • chunk extensions
  • trailer consumption
  • following-response byte preservation
  • invalid hexadecimal size
  • invalid chunk-data CRLF
  • malformed/forbidden trailer fields
  • incomplete chunks
  • full ResponseParser chunked integration
  • premature EOF

Trailer policy

Trailer fields are syntax-validated and consumed so message completion is unambiguous, but they are not merged into Response::headers() in v1. Framing-critical trailer fields such as Content-Length and Transfer-Encoding are rejected.

Scope

This PR completes the v1 response framing primitives for Content-Length, close-delimited, no-body, and chunked responses. It still does not add socket orchestration, redirects, or the public Client execution path.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 364345a15f

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Strip BWS before parsing an extended chunk size

When a server sends a chunk line such as 4 ;foo=bar\r\n, the HTTP chunk-extension grammar permits the whitespace before ; as BWS and requires recipients to tolerate it, but size_text remains "4 ", so from_chars rejects an otherwise parseable response as InvalidChunkSize. Remove the permitted trailing BWS from the size portion before parsing it.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Reject syntactically invalid chunk extensions

When a malformed peer sends an extension such as 4;\r\n or 4;foo="unterminated\r\n, this validator accepts it because it only rejects control bytes, and the response completes successfully. Chunk extensions require a nonempty token name and, when present, a token or valid quoted-string value, so the decoder needs to validate that grammar to satisfy its malformed-framing error contract.

Useful? React with 👍 / 👎.