py/ai-job-search
1
0
Fork 0
mirror of https://github.com/MadsLorentzen/ai-job-search.git synced 2026-09-17 00:26:26 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(jobindex-search): reject non-jobindex detail URLs instead of fetching them verbatim (#447) (#448)

Browse Source 
detail fetched any http(s) input verbatim with no host check and, when
the path didn't match, silently used the whole input URL as the job id -
a non-posting page came back as a well-formed fake posting with exit 0.
buildUrl now requires a jobindex.dk host and a /jobannonce/<id> path,
rebuilds the fetch URL from the extracted id, and exits 1 with the
stderr-JSON BAD_ID contract otherwise; bare ids stay permissive
slash-free tokens per the jobnet precedent. Eight cases in the new
detail-input.test.ts; the five rejection/canonicalization cases fail
against the verbatim unguarded extraction.
This commit is contained in:
Ayobami Adegoke
2026-09-10 20:36:54 +02:00
committed by GitHub
parent c776e3f2b6
commit c7bd494f11
3 changed files with 110 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files
CHANGELOG.md
+19
View File
@@ -47,6 +47,25 @@ per-file diff commands.
### Fixed
- **`jobindex-search detail` no longer fetches arbitrary URLs or invents posting-shaped
output** (#447) - the command fetched any `http(s)` input verbatim (no host check) and,
when the path didn't match its one pattern, silently used the whole input URL as the job
id; the only net was "the fetched page has a title", so a non-posting page came back as
a well-formed fake posting with exit 0 (demonstrated with jobindex's own homepage:
`id` = the URL, `title` = the site's tagline, `description` = navigation chrome). Every
other portal CLI rejects unparseable detail input with `BAD_ID` and constructs its fetch
URL from the extracted id; jobindex was the one CLI trusting the raw string - and
`/scrape`/`/rank` agents feed it stored URLs, so a ghost or redirected URL (the #331
class) yielded plausible garbage instead of an error. `buildUrl` now requires a
jobindex.dk host (apex or subdomain - look-alike and userinfo tricks rejected via real
URL parsing) plus a `/jobannonce/<id>` path, rebuilds the fetch URL from the extracted
id (the canonical short form the bare-id path always used), and exits 1 with the
stderr-JSON `BAD_ID` contract otherwise; bare ids stay permissive scheme- and
slash-free tokens (the jobnet precedent - the server 404s unknowns loudly). Pinned by
eight cases in the new `detail-input.test.ts`; the five rejection/canonicalization
cases fail against the verbatim unguarded extraction. Complementary to the `/apply`
host-check rule proposed in #431, which stays with its proposer.
- **`/outcome` and `/interview` no longer confuse two roles at the same company** (#443)
(`.claude/commands/outcome.md`, `.claude/commands/interview.md`,
`tests/test_apply_records_application.py`) - when a tracker row's `cv_file` /