How ossteam5/sagijeone works
사기전에 is a web service for reviewing second-hand-market chats and links, returning risk signals with quoted evidence and recommended actions rather than declaring a seller fraudulent. The main flow is browser input → FastAPI analysis endpoint → chat parsing → registered detectors → context filtering → risk scoring and advice → structured response. URL checks use local domain lists; the README also describes screenshot OCR as a roadmap capability, but the only OCR code is explicitly an experiment, so it is not shown as a production stage. The input extractors are present, but their runtime wiring is not evidenced in the excerpts.
Main entry points: pipeline.py Analysis pipeline · models.py Request and result types · main.py Analysis API · url.py URL detector
Components 14
Web interface
- Web page
frontend/index.html - Analysis API
backend/app/main.py
Input handling
- Chat parser
backend/app/input/chat_parser.py - Input extractors
backend/app/input/extractors.py
Risk analysis
- Analysis pipeline
backend/app/pipeline.py - Payment detectors
backend/app/detectors/payment.py - Messenger detector
backend/app/detectors/messenger.py - URL detector
backend/app/detectors/url.py - Detector registry
backend/app/detectors/registry.py - Context filter
backend/app/context.py - Risk engine
backend/app/risk_engine.py - Domain lists
backend/data
API and data
- Request and result types
backend/app/models.py
Other components
- User
Connections 19
- User → Web page: enters input
README.md - Web page → Analysis API: submits analysis
README.md - Analysis API → Analysis pipeline: invokes analysis
backend/app/main.py - Analysis API → Request and result types: validates request
backend/app/main.py - Chat parser → Request and result types: produces messages
backend/app/input/chat_parser.py - Analysis pipeline → Chat parser: parses chat
backend/app/pipeline.py - Analysis pipeline → Payment detectors: invokes detectors
backend/app/pipeline.py - Analysis pipeline → Messenger detector: invokes detectors
backend/app/pipeline.py - Analysis pipeline → URL detector: invokes detectors
backend/app/pipeline.py - Payment detectors → Detector registry: registers detectors
backend/app/detectors/payment.py - Messenger detector → Detector registry: registers detector
backend/app/detectors/messenger.py - URL detector → Detector registry: registers detector
backend/app/detectors/url.py - URL detector → Domain lists: reads domain lists
backend/app/detectors/url.py - Analysis pipeline → Context filter: filters signals
backend/app/pipeline.py - Context filter → Request and result types: uses message types
backend/app/context.py - Analysis pipeline → Risk engine: builds result
backend/app/pipeline.py - Risk engine → Request and result types: uses signal type
backend/app/risk_engine.py - Analysis pipeline → Request and result types: returns analysis data
backend/app/pipeline.py - Analysis API → Web page: serves page
backend/app/main.py