v2.0.31
2.0.31 (2026-08-01)
Bug Fixes
-
indexing: skip pages with no indexable content instead of failing the crawl (5f276e6)
A page that extracted to nothing made the processor throw, and the workflow
turned any processing error into a failed run - so one blank page discarded
every other page that had already been crawled and embedded.- Return a document with no chunks for pages with nothing to index, and skip
those pages instead of abandoning the crawl - Treat content that parses to no sections the same way, so a front-matter or
chrome-only page no longer fails the run - Rethrow from the content processors rather than returning undefined, so a
genuine parse failure cannot be mistaken for an empty page, and reject an
unknown content format outright - Refuse to store when most pages were skipped: extraction failing site-wide
would otherwise replace a healthy index with a gutted one and report success - Count only pages that produced chunks as processed, and report the skip
count in the progress notification - Replace the README troubleshooting entry for a parse error that no longer
exists with one describing skipped pages
- Return a document with no chunks for pages with nothing to index, and skip