* fix(route): harden njxzc, gxmzu and jou routes
- parse all observed date formats (ISO, slash, Chinese with optional
time) instead of a fixed YYYY-MM-DD pattern
- fall back to list data when an article page fails or lacks content,
instead of crashing the whole feed on cheerio load(null)
- skip content fetching for off-site links (e.g. WeChat posts)
- only override title/pubDate when found on the detail page
- resolve relative links and images against the article URL
- deduplicate the list/detail scaffold into per-namespace utils
- remove dead response.status checks, unused ctx params and noise
comments; fix namespace URLs and antiCrawler flags
* refactor(route): address review feedback for gxmzu, jou and njxzc
- translate code comments to English (AGENTS.md rule 53)
- select the date cell via $(selector, context) instead of .find(selector)
to avoid unicorn/no-array-callback-reference false positives
* refactor(route): simplify gxmzu, jou and njxzc per review
- drop the unused type exports, unnecessary .first() calls and the
unreachable no-content early returns
- restore the original selector object order in gxmzu/jou call sites
- gxmzu/jou: drop try-catch in resolveArticles (no failing article
observed); keep the same-host guard for off-site list entries
- njxzc: drop the same-host guard (list links are same-host /_redirect
URLs); keep try-catch for redirects to intranet-only subdomains
- gxmzu/lib: read the link text directly (library rows have no title
attribute)
* refactor(route): drop jou host guard and njxzc intranet handling
- jou: remove the same-host guard — no off-site rows on the current
first pages, which are all the route reads
- njxzc: remove the intranet-notice branch and the try-catch — every
article on the current first pages loads fine from off-campus; drop
the now-stale route descriptions as well
---------
Co-authored-by: real-jiakai <real.jiakai@gmail.com>
39aa7af02c