How aianytime/mistral-ocr-app works
Mistral OCR App is a Streamlit interface for extracting text from PDFs and images supplied by URL or local upload. Users enter an API key, select document and source types, and process one or more documents. The app prepares each document, calls Mistral’s OCR endpoint, extracts page Markdown, and displays previews and OCR results with JSON and text download options. Components are mapped to main.py; some preparation and rendering details are supported by the README rather than the sampled excerpts.
Main entry points: main.py Session results
Components 14
User interface
- API key entry
main.py - Source selection
main.py - Document type
main.py
Document ingestion
- URL input
main.py - Local uploads
main.py - Document preparation
main.py
OCR processing
- OCR request
main.py - Result extraction
main.py
Results and export
- Session results
main.py - Document preview
main.py - OCR result views
main.py - Result downloads
main.py
Other components
- User
- Mistral OCR API
Connections 16
- User → API key entry: enters key
README.md - User → Source selection: chooses source
README.md - User → Document type: chooses type
README.md - User → URL input: provides URLs
README.md - User → Local uploads: uploads files
README.md - URL input → Document preparation: prepares input
README.md - Local uploads → Document preparation: encodes uploads
README.md - Document preparation → OCR request: submits document
main.py - API key entry → OCR request: authorizes request
main.py - OCR request → Mistral OCR API: posts OCR request
main.py - Mistral OCR API → Result extraction: returns pages
main.py - Result extraction → Session results: stores text
main.py - Document preparation → Session results: stores preview data
main.py - Session results → Document preview: renders preview
main.py - Session results → OCR result views: renders OCR text
main.py - Session results → Result downloads: exports results
README.md