Repository navigation
v0.1.0
Added
- Automatic chunked processing for large PDFs that exceed Gemini's token limit
- Try-and-split algorithm: tries the full PDF first, splits in half on token limit error, retries
- Rolling context across chunks preserves findings from earlier sections
cached_urisarray in responses for re-analyzing previously chunked documentspdf_sourcenow acceptsstring[](array of cached Gemini URIs) for chunk re-analysis- New dependency:
pdf-libfor PDF page splitting
Changed
- Renamed response field
file_uritocached_uris(now always an array) pdf_sourceinput accepts file paths, URLs, single Gemini URIs, or arrays of Gemini URIs
Installation
See README for installation instructions.
Checksums
See checksums.txt for SHA256 checksums.