v0.3.6 — PDF, Office, EXIF, Email Extraction + Full File Traversal

What's New

Binary File Extraction

• PDF: Full text extraction from PDF documents via pdf-extract
• DOCX: Text from Word documents (body, headers, footers) + author/title metadata
• PPTX: Text from PowerPoint slides and speaker notes
• XLSX: All cell text from Excel spreadsheets via shared strings table
• Email (.eml): Structured parsing — from, to, subject, date, body text (plain or stripped HTML)

Image EXIF Metadata

• GPS coordinates (latitude/longitude) from photos
• Date taken from camera timestamp
• Camera make/model and lens info
• Image dimensions and descriptions
• Creates timeline entries like "Photo taken: 2024-03-15, Location: 37.7749, -122.4194, Camera: iPhone 15 Pro"

Image Vision (Optional)

• If you have llava model in Ollama, images can be described by AI
• Extracts text from screenshots, describes scenes, identifies objects
• Async, won't block imports — runs as a post-processing step

Full File Traversal

• ZIP handler: switched from allowlist (14 types) to blocklist — now extracts everything except known binary media/archives
• Generic adapter: handles 40+ file extensions (RTF, XML, YAML, TOML, subtitles, etc.) with UTF-8 fallback for unknowns
• Folder watcher: monitors all non-binary file types, not just 7

Comprehensive Logging

• Every skipped file now logs a warning with the reason
• Every parsed file logs what was extracted
• ZIP extraction logs total extracted/skipped counts
• 12 adapters updated to log all silent skips

Task Progress Fix

• Folder watcher imports now properly show in the task banner
• Tasks complete correctly instead of getting stuck at "Complete" while still running

Full Changelog

6584f1f...6766163