# 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

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