3851a543d8
Adds document extraction to the chat composer with a small footprint that reuses the existing RAG preview UI and is fast by default. Backend - Adaptive PDF extraction: born-digital pages produce layout-aware Markdown via pymupdf4llm and render no page images, so a text PDF issues no VLM calls. Only pages without a text layer are detected as scanned and rendered for OCR. - Scanned pages are transcribed (not summarized) through the already loaded vision model over /v1/chat/completions. No dedicated OCR model is loaded and the chat model is never swapped out. - Adaptive render DPI (120, env override) and bounded async caption concurrency (2 local, 3 gguf, env override). - /chat/document-support and /chat/extract-document endpoints with NDJSON streaming progress, cancellation, multipart size guards, and token-budget truncation. Frontend - Reuses the existing RAG DocumentPreviewSheet and MarkdownPreview to render an extracted document inline (a new markdown preview target), instead of a separate preview panel. - Extraction uses whatever model is loaded; there is no OCR model picker, cross-tab lock, or custom-code consent step. - Compact document chips in the composer and transcript; image data is stripped from persisted attachments. - Document settings expose a mode (fast text, auto, scanned), a caption toggle, a token budget, and concurrency. Unknown settings keys are ignored. Adds backend tests for the adaptive path, the support probe, NDJSON streaming, cancellation, error mapping, and the scanned-page dedup.
35 lines
1.1 KiB
Text
35 lines
1.1 KiB
Text
# Studio UI backend dependencies
|
|
typer
|
|
fastapi
|
|
uvicorn
|
|
pydantic
|
|
packaging
|
|
matplotlib==3.10.9
|
|
pandas
|
|
nest_asyncio
|
|
datasets==4.3.0
|
|
pyjwt
|
|
# gradio>=4.0.0 # 148 MB - Studio uses React + FastAPI, not Gradio
|
|
huggingface-hub==0.36.2
|
|
structlog>=24.1.0
|
|
diceware
|
|
ddgs
|
|
pypdf>=6.0.0,<7
|
|
python-multipart>=0.0.26
|
|
# Document extraction relies on pymupdf4llm 1.27+ (installed via
|
|
# data-designer-deps.txt), which pulls pymupdf-layout. The bundled ONNX
|
|
# models work fine on modern onnxruntime; we require >=1.19 because
|
|
# earlier wheels (e.g. 1.17.x) were built against NumPy 1.x and crash
|
|
# on import in venvs that have NumPy 2.x installed (pymupdf.layout ->
|
|
# onnxruntime -> numpy._multiarray_umath ABI mismatch). Verified
|
|
# end-to-end with onnxruntime 1.25.0 + numpy 2.4.x.
|
|
onnxruntime>=1.19
|
|
cryptography>=42.0.0
|
|
boto3>=1.34.0 # optional: S3 dataset loading
|
|
httpx>=0.27.0
|
|
fastmcp>=3.0.2
|
|
# RAG (knowledge bases, hybrid retrieval). sentence-transformers lives in
|
|
# extras-no-deps.txt; these add the lexical+dense store and document parsing.
|
|
sqlite-vec==0.1.9
|
|
pymupdf==1.27.2.3
|
|
python-docx==1.2.0
|