Fix leaked Google Docs comment markers on the clusters page
Merged
LU
4mo ago+10+10+01234567890123456789-10-10-01234567890123456789
Summary
- Six cluster/sub-cluster descriptions on the live
/clusters/page showed literal bracket artifacts likeprocess:[ag][ah]andfeedback[ai][aj], have receivedinstead of clean prose. - Root cause:
scripts/parse_clusters_to_sheet.pyfetches the source Google Doc via its plain-text export (export?format=txt). Google Docs' plain-text export represents unresolved inline comments as bracketed reference codes (a,b, ...aa,ab, ...) inserted directly into the body text at the comment's anchor point. - The script already had an
extract_annotations()helper built for exactly this pattern (used for citation text and sub-cluster ToC names), but the code paths that build cluster- and sub-cluster-level description text appended the raw, uncleaned line instead ā 5 call sites acrossparse_cluster_toc(),parse_subclusters_standard(), andparse_cluster_11().
Changes
- Fixed all 5 call sites to strip annotation markers, so future re-syncs from the Google Doc won't reintroduce this even if comments are left open.
- Directly cleaned the 6 already-leaked instances in
data/clusters_v4.jsonwith a minimal, scoped fix targeting only the affected fields ā not a full re-sync from the live doc, to avoid pulling in unrelated content drift.
Test plan
- Rebuilt the site locally and confirmed 0 occurrences of the leaked markers in the rendered
/clusters/page - Confirmed the diff only touches the 5 affected description fields, nothing else
š¤ Generated with Claude Code