prdoc

Documentation

Overview

Package prdoc parses PR descriptions and commit messages into chunks for the PR corpus index. Each PR produces 1-3 chunks:

• ChunkPRBackground — the "Background" / "Context" section
• ChunkPRSolution — the "Solution" / "Changes" section
• ChunkCommitMessage — one per commit (concatenated into one chunk)

The parser reuses prregress.ExtractBackground for section splitting and expects the caller to have already fetched the PR metadata via the gh CLI.

Index

• func Parse(meta PRMeta) []types.Chunk
• type PRMeta

Functions

func Parse

func Parse(meta PRMeta) []types.Chunk

Parse splits a PR's description and commit messages into chunks. The File field on each chunk is set to a synthetic path "pr/<repo>#<number>" so retrieval citations point back to the PR.

Types

type PRMeta

type PRMeta struct {
	Repo           string
	PRNumber       int
	Title          string
	Body           string
	CommitMessages []string
	ChangedFiles   []string
	MergedAt       time.Time
}

PRMeta is the input from the gh CLI. Mirrors the subset of prregress.Meta that the parser needs, decoupled so we don't import the prregress package (which brings exec/os deps).