Commit a78dfbb
Collapse internal whitespace in abstracts
Abstracts now have no internal line breaks and no multi-space runs. The
1,139 records that had OCR-introduced line wrapping (single \n between
words) are flattened to single-space-joined prose, which is how the
content was actually meant to read. Both the records.jsonl cache and the
bib reflect the change.
clean_abstract in pipeline/03_export_bib.py gains a final `\s+ → ` pass
after the line-anchored noise patterns have run, so future re-parses
stay clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>1 parent 545ccbf commit a78dfbb
3 files changed
Lines changed: 2340 additions & 14424 deletions
0 commit comments