Web Workbench

This browser-only workbench runs from GitHub Pages. It can extract starter MOSFET or diode parameters from PDF text, edit the project JSON, and generate starter SPICE files without installing Python.

The hosted page is useful for quick reviews and sharing. The local Python workbench remains the higher-fidelity path for screenshot evidence, calibrated raster plot digitization, and server-side PyMuPDF extraction.

Open the Web Workbench

Current Browser-Only Scope

• PDF text extraction through PDF.js.
• Heuristic MOSFET and diode parameter extraction.
• Diode series-part detection, explicit part selection, and all-variant ZIP export.
• Editable DeviceProject JSON.
• ABM, VDMOS, diode compact, and diode behavioral starter netlist generation.
• Dialects: common, LTspice, ngspice, PSpice, HSPICE, Xyce, and experimental QSPICE.
• ZIP download through JSZip.

Planned WASM Path

GitHub Pages can serve static WebAssembly artifacts. That means the heavier parts of the local workbench can later move into a Rust/WASM module:

• robust PDF text/table extraction,
• raster image preprocessing,
• curve tracing,
• fitting and quality scoring,
• deterministic netlist generation shared by CLI and browser.

The hosted workbench should remain a lightweight public entry point. Features that need local files, simulator execution, large OCR models, or long-running jobs should stay behind the local Python backend or a future optional remote API. See Deployment Modes.

Frontend source files live in web/; GitHub Pages copies are kept in docs/. Run python tools/sync_web_frontend.py after changing the source frontend. The browser code is split into workbench_app.js, workbench_runtime.js, pdf_extractors.js, and model_emitters.js so the hosted demo can evolve without turning the HTML page back into a monolith.