[zuudevs]

Summary

Adds the first incremental HTTP/1.1 response parser on top of the existing Headers, Response, and transport-oriented architecture.

Added

• include/cpp_request/response.hpp
  • public owning Response value type
  • status code and reason access
  • owned response headers
  • owned in-memory body with zero-copy string_view access
• src/http/response_parser.hpp / .cpp
  • incremental feed model that accepts arbitrary TCP chunk boundaries
  • strict HTTP/1.1 status-line parsing
  • owned response-header parsing with token/value validation
  • interim 1xx handling before the final response
  • Content-Length framing
  • HEAD / 101 / 204 / 205 / 304 no-body handling
  • close-delimited response bodies completed by EOF
  • Connection: close reuse semantics
  • preserves bytes beyond a completed Content-Length body for a following response
  • detects conflicting Transfer-Encoding + Content-Length
  • explicit NeedsChunkedDecoder handoff for valid Transfer-Encoding: chunked
  • upgraded 101 connections are never marked reusable as HTTP connections
• response parser tests
  • one-byte-at-a-time feeds
  • exact body completion
  • duplicate headers
  • extra-byte preservation
  • HEAD semantics
  • close-delimited EOF completion
  • connection reuse decisions
  • chunked handoff without consuming chunk bytes
  • malformed status/header/framing failures
  • interim 100 Continue
  • premature EOF
  • protocol upgrade non-reuse

Parser model

The parser owns only bytes that must survive across read_some() calls. It does not assume TCP packet boundaries match HTTP lines or body boundaries. For length-delimited responses it consumes exactly the declared body length and leaves any following bytes pending, which is required for safe keep-alive reuse.

Chunked boundary

This PR intentionally does not decode chunked bodies yet. When a valid chunked response head is recognized, the parser returns ResponseParseProgress::NeedsChunkedDecoder and leaves all body bytes untouched in pending_bytes(). The next PR can attach the chunk decoder to this explicit state without redesigning status/header parsing.

Scope

This PR does not execute socket reads itself, decode chunked bodies, implement redirects, or add the public Client request orchestration path.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: fb508a8b12

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]

Reject status codes above 599

When a peer sends a status line such as HTTP/1.1 600 Weird, this upper bound accepts it and exposes an invalid status as a successful response. HTTP status codes outside 100–599 are invalid, so this should return InvalidStatusLine rather than proceeding with body framing.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Require the separator after the status code

When the status line ends immediately after the code, for example HTTP/1.1 204\r\n, this conditional skips the separator check and accepts the response. The HTTP/1.1 status-line grammar requires a space after the three-digit code even when the reason phrase is empty, so this malformed input should produce InvalidStatusLine.

Useful? React with 👍 / 👎.