| 1969 | // `cache` (optional) memoizes the result for the run: with several batches each |
| 1970 | // failing 422, the inventory would otherwise be refetched per batch. |
| 1971 | async function getPrDiffHunks({ github, owner, repo, prNumber, commitSha, log, cache }) { |
| 1972 | const logFn = typeof log === "function" ? log : () => {}; |
| 1973 | if (cache && cache.diff !== undefined) return cache.diff; |
| 1974 | |
| 1975 | const files = new Map(); |
| 1976 | const known = new Set(); |
| 1977 | let complete = true; |
| 1978 | |
| 1979 | const PER_PAGE = 100; |
| 1980 | const MAX_PAGES = 30; |
| 1981 | let page = 1; |
| 1982 | while (page <= MAX_PAGES) { |
| 1983 | const res = await readWithPacing( |
| 1984 | `listFiles (page ${page})`, |
| 1985 | () => github.rest.pulls.listFiles({ owner, repo, pull_number: prNumber, per_page: PER_PAGE, page }), |
| 1986 | logFn |
| 1987 | ); |
| 1988 | const batch = (res && res.data) || []; |
| 1989 | for (const file of batch) { |
| 1990 | if (!file || !file.filename) continue; |
| 1991 | known.add(file.filename); |
| 1992 | if (file.patch) { |
| 1993 | const parsed = parseDiffHunkInventory(file.patch); |
| 1994 | if (parsed.complete) { |
| 1995 | files.set(file.filename, parsed.ranges); |
| 1996 | } else { |
| 1997 | logFn( |
| 1998 | `[422-fallback] Patch data for ${file.filename} is incomplete or malformed; ` + |
| 1999 | `comments on that file will be treated as unknown rather than out-of-diff.` |
| 2000 | ); |
| 2001 | } |
| 2002 | } |
| 2003 | } |
| 2004 | if (batch.length < PER_PAGE) break; |
| 2005 | page++; |
| 2006 | } |
| 2007 | if (page > MAX_PAGES) { |
| 2008 | complete = false; |
| 2009 | logFn( |
| 2010 | `[422-fallback] PR changed-file list exceeded ${MAX_PAGES * PER_PAGE} files; ` + |
| 2011 | `diff inventory is incomplete, so no comment will be discarded as out-of-diff.` |
| 2012 | ); |
| 2013 | } |
| 2014 | |
| 2015 | // An empty inventory proves nothing. A PR that produced review comments |
| 2016 | // necessarily has changed files, so an empty listFiles response is an anomaly |
| 2017 | // (diff not yet materialized server-side, or a malformed/empty response body) |
| 2018 | // rather than evidence that every commented path sits outside the diff. |
| 2019 | // Trusting it would classify EVERY comment "invalid" and discard the whole |
| 2020 | // batch without a single posting attempt — the exact outcome the tri-state |
| 2021 | // classification exists to prevent. Note this is the mirror of the truncation |
| 2022 | // case above: too many files and zero files are both "cannot judge". |
| 2023 | if (known.size === 0) { |
| 2024 | complete = false; |
| 2025 | logFn( |
| 2026 | `[422-fallback] PR changed-file list came back empty, which cannot be right for a PR ` + |
| 2027 | `under review; treating the diff inventory as incomplete, so no comment will be ` + |
| 2028 | `discarded as out-of-diff.` |