Remove Portico archive artefacts from featured references

Merged
LU
5d ago
+78-1

[!NOTE]
šŸ¤– GPT-6.1 Sol responding on behalf of Lukas Wallrich

Publication cards display Portico, an archive provider, as a publisher because the Publications sheet's APA text includes it. The generator now removes only . Portico. immediately before https://doi.org/, using the exact regex introduced in #984. Other mentions of Portico and all other citation formatting stay intact.

Cleanup happens before the shared card is copied into pub_cards.json and featured_resources.json. No generated outputs are changed; the next daily data-processing run on main will regenerate them.

Verification:

  • Three new pure-Python tests cover the known artefact, preservation of other Portico mentions, and an offline run of the generator that verifies both outputs and the short reference. All CSV input is mocked and outputs go to a temporary directory.
  • python3 -m unittest discover -s scripts/tests: 42 tests passed.
  • python3 -m unittest discover -s scripts -p test_parse_disciplines_gdoc.py: 11 tests passed.
  • Full hugo --gc --minify build and check_seo_html.py passed all ten hard checks.
  • Applied the cleanup in memory to the existing 820 card citations: exactly 42 have this archive-provider artefact.
  • git diff --check passed. The redesign-only scripts/rebrand/check_source_rules.py is absent from main; the diff contains only the generator and its test, with no generated paths.

Other entry points: scripts/parse_clusters_to_sheet.py receives the same artefact through doi.org APA lookup and already removes it in tidy_lookup_citation (#984). The featured generator reads APA from the Publications sheet; the Recommendations form supplies DOI selections and commentary, rather than a second APA field. Its shared card also feeds featured resources, so both destinations receive this fix. A scan of committed scripts, content, data, static files, and the BibTeX converter found no additional occurrences outside the affected cards and the existing clusters cleanup/test.

Fixes #1006

šŸ¤– Generated by GPT-6.1 Sol (Codex); review and UI decisions by Claude Code

Summary by cubic

Fixes publication cards showing Portico as the publisher by stripping the archive-provider artefact from APA references before writing pub_cards.json and featured_resources.json. Only . Portico. immediately before a DOI URL is removed; other Portico mentions and citation formatting stay intact.

  • Applies the same narrow cleanup used in parse_clusters_to_sheet (#984).
  • Adds three regression tests covering the known artefact, preserved Portico mentions, and cleaned output in both destinations.

Fixes #1006