Makefile

.PHONY: up down seed pipeline demo test logs rebuild index-code eval

# Bring up Postgres(pgvector) + API. First run downloads the embedding model.
up:
	docker compose up -d --build
	@echo "API:  http://localhost:8000  (dashboard at /)"

down:
	docker compose down

# Rebuild the API image without cache (after dependency changes)
rebuild:
	docker compose build --no-cache api

# Load sample issues + case library and embed them
seed:
	docker compose exec api python -m scripts.seed

# Run the full pipeline: ingest -> embed -> classify -> RAG analyze
pipeline:
	docker compose exec api python -m scripts.run_pipeline

# Index this repo's own source (+ sample Java) for semantic code search
index-code:
	docker compose exec api python -m scripts.index_code

# Evaluate classification accuracy against ground-truth labels
eval:
	docker compose exec api python -m scripts.eval_classifier

# End-to-end demo (seed + pipeline + a couple of queries)
demo: seed pipeline
	@echo "--- analysis for issue #1 ---"
	curl -s localhost:8000/issues/1/analysis | python -m json.tool || true
	@echo "--- semantic search: 'login timeout' ---"
	curl -s 'localhost:8000/search?q=login%20timeout&k=5' | python -m json.tool || true

# Run tests (no LLM key required: rule-based fallback path)
test:
	docker compose exec api python -m pytest

logs:
	docker compose logs -f api