# ECE Compliance RAG — Developer Workflow
#
# Usage:
#   make check              Ollama running? Model loaded? Corpus & indexes OK?
#   make corpus             Build all domain compilations from sources
#   make index              Build all PageIndex trees (LLM-heavy)
#   make test               Staged pipeline validation
#   make app                Launch Streamlit UI
#   make help               Show all targets

.PHONY: help check corpus corpus-legislation corpus-licensing corpus-reform \
        index index-legislation index-licensing index-ero index-cross-regulator index-reform \
        test benchmark benchmark-quick app app-qwen app-gemma mlx-stop download rebuild

SHELL := /bin/bash
export PYTHONUNBUFFERED := 1
UV := /opt/homebrew/bin/uv run python

# ── Help ────────────────────────────────────────────────────────────────────

help:
	@echo ""
	@echo "ECE Compliance RAG — Developer Workflow"
	@echo "========================================"
	@echo ""
	@echo "Health checks (fast, no LLM):"
	@echo "  make check              Corpus, indexes, mlx-lm server"
	@echo "  make check-ollama       Just Ollama connectivity (legacy)"
	@echo ""
	@echo "Source acquisition (network, no LLM):"
	@echo "  make download           Fetch HTML/PDF sources"
	@echo ""
	@echo "Corpus building (local, no LLM):"
	@echo "  make corpus             Build all 5 domain compilations"
	@echo "  make corpus-legislation Just rebuild legislation.md from HTML"
	@echo "  make corpus-licensing   Just rebuild licensing-criteria.md"
	@echo "  make corpus-reform      Just rebuild reform-context.md"
	@echo ""
	@echo "Index building (LLM-heavy — slow):"
	@echo "  make index              Build all 5 PageIndex trees (auto-runs propagate-urls)"
	@echo "  make index-legislation  Just rebuild legislation index"
	@echo "  make index-licensing    Just rebuild licensing-criteria index"
	@echo "  make index-ero          Just rebuild ero index"
	@echo "  make index-cross-regulator"
	@echo "  make index-reform       Just rebuild reform-context index"
	@echo "  make propagate-urls     Propagate Source: URLs down index trees (no LLM)"
	@echo ""
	@echo "Testing:"
	@echo "  make test               Staged pipeline test (1 LLM call)"
	@echo "  make benchmark          Full 20-question benchmark (~30 min)"
	@echo "  make benchmark-quick    3 questions (~5 min)"
	@echo ""
	@echo "App:"
	@echo "  make app                Launch Streamlit UI (default: Qwen 2.5 14B)"
	@echo "  make app-qwen           Launch with Qwen 2.5 14B (~9 GB, local)"
	@echo "  make app-gemma          Launch with Gemma 4 26B MoE (~17 GB, local)"
	@echo "  make app-sonnet         Launch with Claude Sonnet (API, fast)"
	@echo "  make mlx-stop           Stop the mlx-lm server"
	@echo ""
	@echo "Full rebuild:"
	@echo "  make rebuild            corpus → index → test (end to end)"
	@echo ""

# ── Health checks ───────────────────────────────────────────────────────────

check:
	@$(UV) scripts/check.py

check-ollama:
	@$(UV) scripts/check.py --ollama

# ── Source acquisition ──────────────────────────────────────────────────────

download:
	@echo "Downloading legislation HTML sources..."
	@$(UV) scripts/convert_legislation_html.py
	@echo ""
	@echo "Converting PDF sources to Markdown..."
	@$(UV) scripts/convert_sources.py

# ── Corpus building ────────────────────────────────────────────────────────

corpus: corpus-legislation corpus-licensing corpus-reform corpus-ero
	@echo ""
	@echo "All corpus files built."

corpus-legislation:
	@echo "Building legislation corpus from HTML..."
	@$(UV) scripts/convert_legislation_html.py

corpus-licensing:
	@echo "Building licensing-criteria corpus from education.govt.nz..."
	@$(UV) scripts/build_licensing_html.py

corpus-reform:
	@echo "Building reform-context corpus..."
	@$(UV) scripts/build_reform_compilation.py

corpus-ero:
	@echo "Building ERO corpus from HTML sources..."
	@$(UV) scripts/build_ero_compilation.py

# ── Index building ──────────────────────────────────────────────────────────

dry-run:
	@$(UV) scripts/build_indexes.py --dry-run

index:
	@$(UV) scripts/build_indexes.py
	@$(MAKE) --no-print-directory propagate-urls

index-legislation:
	@$(UV) scripts/build_indexes.py legislation
	@$(UV) scripts/propagate_source_urls.py --apply legislation

index-licensing:
	@$(UV) scripts/build_indexes.py licensing-criteria
	@$(UV) scripts/propagate_source_urls.py --apply licensing-criteria

index-ero:
	@$(UV) scripts/build_indexes.py ero
	@$(UV) scripts/propagate_source_urls.py --apply ero

index-cross-regulator:
	@$(UV) scripts/build_indexes.py cross-regulator
	@$(UV) scripts/propagate_source_urls.py --apply cross-regulator

index-reform:
	@$(UV) scripts/build_indexes.py reform-context
	@$(UV) scripts/propagate_source_urls.py --apply reform-context

# Propagate Source: URLs down every index tree so deep nodes inherit
# the nearest ancestor's URL. Zero API cost. Run automatically after
# `make index` but can be re-run standalone on existing indexes.
propagate-urls:
	@$(UV) scripts/propagate_source_urls.py --apply

# ── Testing ─────────────────────────────────────────────────────────────────

test:
	@$(UV) test_pipeline.py

benchmark:
	@$(UV) benchmark/run_benchmark.py

benchmark-quick:
	@$(UV) benchmark/run_benchmark.py --ids 1,4,5

benchmark-compare:
	@$(UV) benchmark/run_benchmark.py --compare qwen,sonnet

# ── App ─────────────────────────────────────────────────────────────────────

# Model presets — override with: make app MLX_MODEL=your-model-here
MODEL_QWEN := mlx-community/Qwen2.5-14B-Instruct-4bit
MODEL_GEMMA := mlx-community/gemma-4-26b-a4b-it-4bit
MLX_MODEL ?= $(MODEL_QWEN)
MLX_PORT := 8080

# Stop any running mlx-lm server (useful when swapping models)
mlx-stop:
	@echo "Stopping mlx-lm server..."
	@pkill -f "mlx_lm.*server" 2>/dev/null && echo "Server stopped." || echo "No server running."

# Start mlx-lm server for a given model, restarting if a different model is running
define start-mlx-server
	@RUNNING_MODEL=$$(curl -s --max-time 2 http://localhost:$(MLX_PORT)/v1/models 2>/dev/null | python3 -c "import sys,json; d=json.load(sys.stdin); print(d['data'][0]['id'])" 2>/dev/null || echo ""); \
	if [ "$$RUNNING_MODEL" = "$(1)" ]; then \
		echo "mlx-lm server already running $(1)"; \
	else \
		if [ -n "$$RUNNING_MODEL" ]; then \
			echo "Stopping server (was running $$RUNNING_MODEL)..."; \
			pkill -f "mlx_lm.*server" 2>/dev/null; sleep 2; \
		fi; \
		echo "Starting mlx-lm server ($(1))..."; \
		ECE_MLX_MODEL=$(1) /opt/homebrew/bin/uv run python -m mlx_lm server --model $(1) --port $(MLX_PORT) > /dev/null 2>&1 & \
		echo "Waiting for server..."; sleep 10; \
		echo "Server ready."; \
	fi
endef

app: ## Launch with current MLX_MODEL (default: Qwen)
	$(call start-mlx-server,$(MLX_MODEL))
	@echo "Launching Streamlit UI ($(MLX_MODEL))..."
	@ECE_MLX_MODEL=$(MLX_MODEL) /opt/homebrew/bin/uv run streamlit run app.py --server.headless true

app-qwen: ## Launch with Qwen 2.5 14B
	$(call start-mlx-server,$(MODEL_QWEN))
	@echo "Launching Streamlit UI (Qwen 2.5 14B)..."
	@ECE_MLX_MODEL=$(MODEL_QWEN) /opt/homebrew/bin/uv run streamlit run app.py --server.headless true

app-gemma: ## Launch with Gemma 4 26B MoE
	$(call start-mlx-server,$(MODEL_GEMMA))
	@echo "Launching Streamlit UI (Gemma 4 26B MoE)..."
	@ECE_MLX_MODEL=$(MODEL_GEMMA) /opt/homebrew/bin/uv run streamlit run app.py --server.headless true

app-sonnet: ## Launch with Claude Sonnet (API — no local server needed)
	@echo "Launching Streamlit UI (Claude Sonnet — API)..."
	@ECE_MODEL=sonnet /opt/homebrew/bin/uv run streamlit run app.py --server.headless true

# ── Full rebuild ────────────────────────────────────────────────────────────

rebuild: corpus index test
	@echo ""
	@echo "Full rebuild complete."