# ResearchOwl — developer tasks.
#
# SEO engine vendoring: src/seo/rules.py is a byte-for-byte copy of the canonical
# seo_rules.py (repo git.chemavx.xyz/chemavx/chemavx-seo-tools; local working copy
# ~/seo-tools). These targets keep the vendored copy in sync; CI enforces it too.

SEO_TOOLS_DIR ?= $(HOME)/seo-tools
CANON         := $(SEO_TOOLS_DIR)/seo_rules.py
VENDORED      := src/seo/rules.py
HEADER        := src/seo/_vendor_header.py
MARKER        := BEGIN VENDORED seo_rules.py

.PHONY: sync-seo check-seo-sync test golden-db golden shortsmith-live

test:              ## Suite completa (los tests vivos se saltan solos)
	python3 -m pytest tests/ -q

# --- Shorts: pruebas contra el servicio vivo ---------------------------------
# El renderizador no se expone fuera del cluster, así que desde el nodo se le
# habla por la ClusterIP. Los dos targets de abajo la resuelven solos.

SHORTSMITH_IP = $(shell kubectl get svc shortsmith-svc -n shortsmith \
                  -o jsonpath='{.spec.clusterIP}' 2>/dev/null)
GOLDEN_SESSION ?= 153
GOLDEN_DB      ?= /tmp/researchowl-golden.db

golden-db:         ## Extrae UNA sesión de la DB de producción a un fichero pequeño
	@test -n "$(SHORTSMITH_IP)" || echo "aviso: shortsmith-svc no encontrado"
	kubectl exec -n researchowl deployment/researchowl -- python3 -c "\
import sqlite3, os; \
src = sqlite3.connect('file:/data/researchowl.db?immutable=1', uri=True); \
out = '/tmp/golden.db'; os.path.exists(out) and os.remove(out); \
src.execute('ATTACH DATABASE ? AS g', (out,)); \
src.executescript('''CREATE TABLE g.research_sessions AS SELECT * FROM research_sessions WHERE 0; \
CREATE TABLE g.sources AS SELECT * FROM sources WHERE 0; \
CREATE TABLE g.chunks AS SELECT * FROM chunks WHERE 0; \
CREATE TABLE g.outputs AS SELECT * FROM outputs WHERE 0; \
CREATE TABLE g.api_usage AS SELECT * FROM api_usage WHERE 0;'''); \
[src.execute(f'INSERT INTO g.{t} SELECT * FROM {t} WHERE ' + ('id=$(GOLDEN_SESSION)' if t=='research_sessions' else 'session_id=$(GOLDEN_SESSION)')) \
 for t in ('research_sessions','sources','chunks','outputs')]; \
src.commit(); print('sesión $(GOLDEN_SESSION) ->', os.path.getsize(out), 'bytes')"
	kubectl cp -n researchowl \
	  $$(kubectl get pod -n researchowl -o name | head -1 | cut -d/ -f2):/tmp/golden.db \
	  $(GOLDEN_DB)
	@echo "escrito en $(GOLDEN_DB)"

golden:            ## Eval dorada (GASTA dinero: una llamada a Claude + un render)
	@test -f $(GOLDEN_DB) || { echo "falta $(GOLDEN_DB) — ejecuta 'make golden-db'"; exit 1; }
	RESEARCHOWL_GOLDEN_DB=$(GOLDEN_DB) \
	RESEARCHOWL_GOLDEN_SESSION=$(GOLDEN_SESSION) \
	SHORTSMITH_LIVE_URL=http://$(SHORTSMITH_IP):8080 \
	SHORTS_DIR=/tmp/researchowl-shorts \
	python3 -m pytest tests/test_short_golden.py -v -s

shortsmith-live:   ## Fontanería contra el shortsmith vivo (renderiza el ejemplo)
	SHORTSMITH_LIVE_URL=http://$(SHORTSMITH_IP):8080 \
	python3 -m pytest tests/test_shortsmith_live.py -v -s


sync-seo:          ## Re-copy canonical seo_rules.py into the vendored file + record hash
	@test -f "$(CANON)" || { echo "canonical not found at $(CANON)"; exit 1; }
	@cat "$(HEADER)" "$(CANON)" > "$(VENDORED)"
	@sha256sum "$(CANON)" | cut -d' ' -f1 > src/seo/.rules.sha256
	@echo "synced $(VENDORED) from $(CANON) (sha $$(cat src/seo/.rules.sha256))"

check-seo-sync:    ## Fail if the vendored copy diverges from the local canonical
	@test -f "$(CANON)" || { echo "canonical not found at $(CANON) — skipping (run on a machine with seo-tools)"; exit 0; }
	@N=$$(grep -n "$(MARKER)" "$(VENDORED)" | head -1 | cut -d: -f1); \
	 tail -n +$$((N+1)) "$(VENDORED)" > /tmp/_vendored_body.py; \
	 if cmp -s /tmp/_vendored_body.py "$(CANON)"; then \
	   echo "seo engine in sync ($(VENDORED) == $(CANON))"; \
	 else \
	   echo "SEO ENGINE DRIFT: $(VENDORED) != $(CANON). Run 'make sync-seo' and commit."; \
	   diff -u "$(CANON)" /tmp/_vendored_body.py | head -40; exit 1; \
	 fi