How jijavemi/partspec works
PartSpec’s documented product spans an iOS app and a backend; this repository contains only the FastAPI backend, so the native UI and its local predefined-part lookup are not represented as code components here. The backend exposes health and search endpoints. A search request retrieves DuckDuckGo results (retrying with a specifications query when needed), asks Groq to produce structured part data, normalizes it into the PartSpec response model, and returns it. The source shows this backend flow; other README-described capabilities are not evidenced in the available files.
Main entry points: main.py Search endpoint
Components 11
API
- Search endpoint
main.py - Health endpoint
main.py
Web retrieval
- Spec lookup
main.py - DuckDuckGo search
main.py - Result extraction
main.py
Spec generation
- Groq generation
main.py - Response normalization
main.py - PartSpec model
main.py
Other components
- API caller
- DuckDuckGo
- Groq API
Connections 11
- API caller → Search endpoint: queries
main.py - API caller → Health endpoint: checks status
main.py - Search endpoint → Spec lookup: calls
main.py - Spec lookup → DuckDuckGo search: searches; retries if sparse
main.py - DuckDuckGo search → DuckDuckGo: requests results
main.py - DuckDuckGo search → Result extraction: parses HTML
main.py - Spec lookup → Groq generation: submits context
main.py - Groq generation → Groq API: requests JSON
main.py - Groq generation → Response normalization: parses response
main.py - Response normalization → PartSpec model: constructs
main.py - Search endpoint → PartSpec model: returns specs
main.py