| 1 | # Document Query Plugin Configuration |
| 2 | # All timeout values in seconds |
| 3 | |
| 4 | # --- Timeouts --- |
| 5 | fetch_timeout: 30 # HTTP fetch connect/read timeout |
| 6 | fetch_retries: 3 # HTTP retry attempts |
| 7 | fetch_retry_backoff: 1.0 # delay between HTTP retry attempts |
| 8 | per_document_timeout: 60 # max time for a single document parse |
| 9 | gather_timeout: 120 # max time for all documents combined in one call |
| 10 | |
| 11 | # --- Parser settings --- |
| 12 | parser_concurrency: 1 # max parser jobs running across all chats in this process |
| 13 | context_intro_chunks: 2 # always include leading chunks per document for title/abstract grounding |
| 14 | chunk_size: 1000 |
| 15 | chunk_overlap: 100 |
| 16 | max_index_chunks: 1200 # adapt chunk size above this many indexed chunks |
| 17 | search_threshold: 0.5 |
| 18 | search_limit: 100 |
| 19 | max_remote_bytes: 52428800 # 50 MB |
| 20 | |
| 21 | # --- Feature flags --- |
| 22 | liteparse_enabled: true # prefer LiteParse before legacy parser fallbacks |
| 23 | liteparse_ocr_enabled: true |
| 24 | liteparse_ocr_language: eng |
| 25 | liteparse_ocr_server_url: |
| 26 | liteparse_tessdata_path: |
| 27 | liteparse_max_pages: 1000 |
| 28 | liteparse_target_pages: |
| 29 | liteparse_dpi: 150 |
| 30 | liteparse_preserve_very_small_text: false |
| 31 | liteparse_output_format: text |
| 32 | liteparse_num_workers: 2 # balanced default for OCR speed without overloading shared Web UI runtime |
| 33 | liteparse_ocr_auto_disable: true # disable OCR automatically for long PDFs |
| 34 | liteparse_ocr_auto_disable_pages: 30 # OCR-on runtime climbs sharply around this page count |
| 35 | liteparse_ocr_auto_sample_pages: 5 |
| 36 | pdf_ocr_fallback: true # enable legacy Tesseract fallback after PyMuPDF |
| 37 | thread_offload: true # offload sync parsers to thread pool |