py/ai-job-search
1
0 Fork 0
mirror of https://github.com/MadsLorentzen/ai-job-search.git synced 2026-11-02 13:12:50 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(freehire-search): decode HTML entities and trim whitespace in titles (#505)

Browse Source 
Titles are indexed as the source ATS wrote them and toResult passed them
through untouched, so `detail intern-fullstack-amp-ai-innovation-noise-edaawqq2`
printed `Intern - Fullstack & AI Innovation ` and the same title reached
`search` output. In a 93-result /scrape batch, 1 title carried `&` and
3 ended in whitespace; /scrape Step 4.75 reads an undecoded entity in a
title as a half-working parser.

Run the title through the existing decodeHtmlEntities (already used for
detail descriptions) and trim it; an empty or whitespace-only title still
falls back to "(untitled)". Three new parsing.test.ts cases pin it; all
three fail on the old pass-through.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Sahil Kumar SahuandClaude Opus 5.5 authored and GitHub committed 2026-09-29 20:24:56 +02:00
1 parent 9187f894a5
commit 4f5520841f
3 files changed
+37 -2
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

CHANGELOG.md
+13
View File
@@ -105,6 +105,19 @@ per-file diff commands.
### Fixed
- **`freehire-search` titles no longer carry HTML entities or stray whitespace**
(`.agents/skills/freehire-search/cli/src/helpers.ts`) - titles are indexed as the source ATS
wrote them and `toResult` passed them through untouched, so `detail
intern-fullstack-amp-ai-innovation-noise-edaawqq2` printed `Intern - Fullstack &amp; AI
Innovation ` (trailing space included), and the same title reached `search` output. In a
93-result `/scrape` batch (`--country IN --seniority intern`), 1 title carried `&amp;` and 3
ended in whitespace. `/scrape` Step 4.75 reads undecoded entities in titles as a
half-working parser, so by that rule a single such posting marks a healthy portal
degraded. The title now goes through the CLI's existing `decodeHtmlEntities` (already used
for `detail` descriptions) and is trimmed; an empty or whitespace-only title still falls
back to `(untitled)`. Pinned by three new `parsing.test.ts` cases, which fail on the old
pass-through.
- **`freehire-search detail` keeps the salary period, so a yearly figure no longer reads as
monthly** (`.agents/skills/freehire-search/cli/src/helpers.ts`, `SKILL.md`) - `formatSalary`
printed only the currency and amounts, although freehire's enrichment records the period: