Skip to content

CLAUDE.md

Latest commit

 

History

History
117 lines (81 loc) · 4.63 KB

CLAUDE.md

File metadata and controls

117 lines (81 loc) · 4.63 KB

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

What is DashAI

DashAI is a desktop/web graphical toolbox for training, evaluating, and deploying ML models. It runs as a FastAPI backend serving a React frontend, optionally wrapped in PyWebView for a native desktop window.

Commands

Backend

# Install
pip install -e . -r requirements-dev.txt
pre-commit install

# Run dev server
python -m DashAI --no-browser --logging-level DEBUG

# Lint / format
ruff check --fix
ruff format

# Run all tests
pytest tests/

# Run a single test file or function
pytest tests/back/api/test_components_api.py -v
pytest tests/back/api/test_components_api.py::test_function_name -v

# Database migrations (auto-runs on startup, but also manually)
alembic upgrade head

Frontend

cd DashAI/front

yarn install        # requires Node LTS + Yarn 3.5.0
yarn start          # dev server on http://localhost:3000
yarn build
yarn lint
yarn test
yarn test FileName.test.tsx   # single test file

Architecture

Request flow

Browser / PyWebView
  → React (port 3000 dev / port 8000 prod)
  → FastAPI (/api/v1/...)
  → Service layer
  → SQLite (SQLAlchemy + Alembic)
  → Huey job queue (async tasks: training, conversion, exploration)

Long-running operations (training, data conversion, explanations) are dispatched as Huey tasks and tracked in the DB. The frontend polls /api/v1/jobs/{job_id} for status.

Backend layout

Path Purpose
DashAI/back/app.py FastAPI app factory (create_app)
DashAI/back/runner.py CLI entry point (starts uvicorn + huey)
DashAI/back/container.py Dependency injection (kink) — config, DB engine, registry, job queue
DashAI/back/dependencies/registry/ ComponentRegistry: central registry of all ML components
DashAI/back/dependencies/config_builder.py Registers initial components on startup
DashAI/back/api/api_v1/endpoints/ REST endpoints
DashAI/back/core/schema_fields/ Type system for component parameters (drives dynamic frontend forms)
DashAI/back/pipeline/ DAG pipeline nodes: DataSelector → Converter → Train → Prediction → Exploration
DashAI/back/plugins/ Plugin system (PyPI packages with dashai.plugins entry point)

Key patterns

ComponentRegistry — all ML components (models, metrics, converters, dataloaders, explorers, explainers, tasks) are registered at startup and resolved by name string. To add a new component: subclass the relevant base, define its schema, and add it to config_builder.py.

Schema / type system — every component declares its parameters using BaseSchema + field classes (IntField, FloatField, ComponentField, UnionType, etc.) with MultilingualString labels. The frontend uses these schemas to auto-generate configuration forms.

Dependency injection (kink) — singletons (config, DB session factory, registry, job queue) live in the di container. Use @inject to receive them; don't import globals directly.

Pipeline as DAG — PipelineValidator validates the graph before execution. Each node is a BaseJob subclass. Results flow between nodes via a shared context dict.

Plugin system — plugins are PyPI packages prefixed dashai-* that register components via entry_points group dashai.plugins. Install/uninstall via /api/v1/plugins.

Frontend layout

Path Purpose
DashAI/front/src/api/ Axios HTTP client for all backend calls
DashAI/front/src/components/configurableObject/ Auto-generates forms from backend component schemas
DashAI/front/src/components/pipelines/ Pipeline builder/editor
DashAI/front/src/i18n/ i18next translations

Data stored at runtime

Default path ~/.DashAI/ (overridable via DASHAI_LOCAL_PATH):

  • datasets/, runs/, explanations/, notebooks/, images/, sqlite.db

Testing notes

  • Backend tests use in-memory SQLite — no setup needed.
  • conftest.py fixtures: test_path, test_datasets_path, test_job_queue.
  • Job queue tests use test_job_queue.set_test_mode(immediate=True) to run tasks synchronously.
  • API tests use FastAPI TestClient defined in tests/back/api/conftest.py.

Adding new things

New ML model or converter: subclass BaseModel / BaseConverter, define schema, register in config_builder.py.

New API endpoint: add file in DashAI/back/api/api_v1/endpoints/, include its router in DashAI/back/api/api_v1/api.py.

New async job: subclass BaseJob, dispatch via job_queue.enqueue(...), track status through the jobs API.