src/extractors

Module summary

Module src/extractors

Exports

Name	Kind	Default
CURSOR_REGEX	variable	no
ISSUES_QUERY	variable	no
ISSUE_COMMENTS_QUERY	variable	no
_formatJsDoc	function	no
classifyCommit	function	no
extractAst	function	no
extractAstAndJsDoc	function	no
extractChangelog	function	no
extractConfigComments	function	no
extractIssues	function	no
extractJsDoc	function	no
extractReadmes	function	no
notifyParseFailure	function	no
parseSource	function	no
readCursor	function	no
validatePath	function	no
writeCursor	function	no

Detail

CURSOR_REGEX

CURSOR_REGEX

Params

None.

Returns

None.

Examples

None.

ISSUES_QUERY

ISSUES_QUERY

Params

None.

Returns

None.

Examples

None.

ISSUE_COMMENTS_QUERY

ISSUE_COMMENTS_QUERY

Params

None.

Returns

None.

Examples

None.

_formatJsDoc

_formatJsDoc(parsed): object

Params

• parsed

Returns

object

Examples

None.

classifyCommit

classifyCommit(subject)

Pure: classify a single conventional-commit subject line.

Params

• subject

Returns

None.

Examples

None.

extractAst

extractAst({ path, source })

Params

• { path, source }

Returns

None.

Examples

None.

extractAstAndJsDoc

extractAstAndJsDoc({ path, source }): object

Params

• { path, source }

Returns

object

Examples

None.

extractChangelog

async extractChangelog({ since = '30.days.ago', repoDir = process.cwd() }): Promise<unknown>

Run git log --since=<since> in repoDir and return classified commits.

Params

• { since = '30.days.ago', repoDir = process.cwd() } = {}

Returns

Promise<unknown>

Examples

None.

extractConfigComments

async extractConfigComments(rootDir): Promise<object>

Extract inline-comment annotations from project config files.

Looked-up files (best-effort — missing files yield empty maps):

• vitest.config.ts (or .js, .mts, .cts) — top-level keys inside the test: { ... } block.
• package.json + sibling package.scripts.md — comment for each entry in scripts.
• .env.example — comment for each KEY=... line.

Params

• rootDir = process.cwd()

Returns

Promise<object>

Examples

None.

extractIssues

async extractIssues({
  owner,
  repo,
  client,
  cursorPath = DEFAULT_CURSOR_PATH,
  pageSize = DEFAULT_PAGE_SIZE,
  maxPages = DEFAULT_MAX_PAGES,
  reset = false,
  now = () => new Date(),
}): Promise<object>

Paginated GraphQL extractor for issues. See file header for cursor invariants.

Cursor-invalidation fallback (must_fix_in_build): when the pagination loop is resuming mid-run (pageCursor !== null) and the GraphQL call throws an error matched by isCursorError (explicit “cursor” mention, or pagination-argument error referencing “after”/“before” with “invalid”), the loop resets pageCursor to null, restores sinceForRun to the original cursor.last_updated_at, and retries the call once. On retry success, it immediately persists a cleared end_cursor (writeCursor with end_cursor: null) so the stale cursor on disk cannot re-trigger the fallback on the next run. If the retry also throws, the error propagates. Non-cursor errors (auth, scope, rate-limit) bypass the retry and propagate immediately.

Per-issue comment pagination (#1391): each matched issue’s comments connection is walked via ISSUE_COMMENTS_QUERY until pageInfo.hasNextPage is false or MAX_COMMENTS_PAGES_PER_ISSUE is hit. Issues with <= 100 comments take the single-fetch path (no extra GraphQL calls). The commentsTruncated stat is true if ANY issue hit the per-issue page cap.

Params

• { owner, repo, client, cursorPath = DEFAULT_CURSOR_PATH, pageSize = DEFAULT_PAGE_SIZE, maxPages = DEFAULT_MAX_PAGES, reset = false, now = () => new Date(), }

Returns

Promise<object>

Examples

None.

extractJsDoc

extractJsDoc({ path, source })

Params

• { path, source }

Returns

None.

Examples

None.

extractReadmes

async extractReadmes(rootDir, opts): Promise<unknown>

Walk rootDir and collect every README.md in any subdirectory.

Params

• rootDir
• opts = {}

Returns

Promise<unknown>

Examples

None.

notifyParseFailure

notifyParseFailure({ path, message }): void

Params

• { path, message }

Returns

void

Examples

None.

parseSource

parseSource({ path, source }): object

Params

• { path, source }

Returns

object

Examples

None.

readCursor

readCursor(filePath): object

Read the cursor file. Returns safe defaults on any failure path.

Validation rules:

• Missing file, unreadable file, malformed JSON → defaults.
• last_updated_at: must be a string, must parse via Date.parse, and must not be in the future. On any failure → coerce to null and warn.
• end_cursor: must match CURSOR_REGEX. On failure → coerce to null.

Params

• filePath

Returns

object

Examples

None.

validatePath

validatePath(rawPath, root, kind)

Validate that rawPath resolves to a location inside root. On any failure, write a generic error to stderr and call process.exit(2).

Params

• rawPath
• root
• kind

Returns

None.

Examples

None.

writeCursor

writeCursor(filePath, cursor): void

Atomically persist the cursor. Always overwrites updated_at with the current ISO timestamp. Creates the parent directory if missing.

Params

• filePath
• cursor

Returns

void

Examples

None.