thomwiggers / onebot / 24400164567

67%

push

github

Update URL handler to extract og:title metadata (#245)

* update devcontainer

* fix: extract og:title for Instagram link previews

Instagram pages have a generic <title> ("Instagram") but put the full
post description in the og:title meta tag. Add a dedicated
_process_url_instagram handler that detects instagram.com URLs and
extracts og:title instead of falling through to the default <title>
extraction.

Add a fixture HTML file and unit test that verify the og:title is used
and the result includes the actual post content.

https://claude.ai/code/session_01HpfD4oaZHyZS9gePtLqsHC

* refactor: prefer og:title over <title> for all URLs

Rather than a dedicated Instagram processor, simply update
_extract_title_from_content to check og:title first and fall back to
<title>. This fixes Instagram (and any other site where the <title> is
generic) in one place.

Remove the now-redundant _process_url_instagram method. Update the old
test_too_long_title_text test to assert og:title preference and add a
separate test_too_long_title_truncated for the truncation logic.

https://claude.ai/code/session_01HpfD4oaZHyZS9gePtLqsHC

* refactor: deduplicate truncation/formatting in _extract_title_from_content

Select the title string first (og:title then <title>), then apply the
truncation and curly-quote wrapping once instead of repeating it in
both branches.

https://claude.ai/code/session_01HpfD4oaZHyZS9gePtLqsHC

* test: replace invented Instagram fixture with real Betamax cassette

Replace the hand-crafted instagram-reel-example.html and its mock-based
test with a Betamax cassette recorded against the live Instagram URL.
Set Accept-Encoding: gzip, deflate in the session to avoid zstd, which
Betamax cannot round-trip through JSON (urllib3 advertises zstd but the
cassette serialiser corrupts the binary frame, causing ZstdError on playback).

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* test: document Betamax zstd workaround in test_instagram

Co-Authored-... (continued)

6 of 7 new or added lines in 1 file covered. (85.71%)

954 of 1414 relevant lines covered (67.47%)

3.37 hits per line