#!/usr/bin/env python3
"""Add decision_refs provenance frontmatter to workshop outcomes (pql migration, Phase 5).

Workshop -> decision provenance was prose-only. This injects a `decision_refs:` YAML
frontmatter list (the confirmed D-records each workshop-outcomes.md touches) so pql can
answer "which workshop produced D-NNN" via
  pql query "SELECT name, fm.workshop WHERE fm.decision_refs CONTAINS 'D-238'"

decision_refs = distinct valid confirmed-D ids mentioned in the file body (filtered
against the governance decision set, so typos/multi-refs/stale ids are dropped). Files
without frontmatter get a minimal block. Idempotent: re-running rewrites the same line.

Default is a dry run. Pass --apply to write.
"""
import glob
import os
import re
import sqlite3
import subprocess
import sys

APPLY = "--apply" in sys.argv
# Derive the repo root from git so the script targets the checkout it's run from.
REPO = subprocess.check_output(["git", "rev-parse", "--show-toplevel"], text=True).strip()
DB = f"{REPO}/.pql/pql.db"

D_IDS = set(r[0] for r in sqlite3.connect(DB).execute(
    "SELECT id FROM decisions WHERE type='confirmed'"))

D_REF = re.compile(r"\bD-\d{3}\b")
FILES = sorted(glob.glob(f"{REPO}/docs/workshops/*/workshop-outcomes.md"))


def split_frontmatter(text):
    """Return (fm_lines, body) where fm_lines excludes the --- fences; fm is None if absent."""
    if text.startswith("---\n"):
        end = text.find("\n---\n", 4)
        if end != -1:
            return text[4:end].splitlines(), text[end + 5:]
    return None, text


def refs_in(body):
    return sorted({m for m in D_REF.findall(body) if m in D_IDS}, key=lambda d: int(d[2:]))


def main():
    print(f"=== add_workshop_provenance.py ({'APPLY' if APPLY else 'DRY-RUN'}) ===\n")
    for f in FILES:
        text = open(f, encoding="utf-8").read()
        fm, body = split_frontmatter(text)
        refs = refs_in(body if fm is not None else text)
        rel = os.path.relpath(f, REPO)
        wname = os.path.basename(os.path.dirname(f))
        refs_yaml = "[" + ", ".join(refs) + "]"

        if fm is not None:
            fm = [ln for ln in fm if not ln.startswith("decision_refs:")]
            fm.append(f"decision_refs: {refs_yaml}")
            new = "---\n" + "\n".join(fm) + "\n---\n" + body
            action = f"inject decision_refs ({len(refs)})"
        else:
            new = (f"---\ntitle: \"Workshop Outcomes: {wname}\"\ntype: workshop\n"
                   f"workshop: {wname}\nstatus: archived\ndecision_refs: {refs_yaml}\n---\n\n" + text)
            action = f"ADD frontmatter + decision_refs ({len(refs)})"

        print(f"{rel}: {action}")
        if refs:
            print(f"    {refs_yaml}")
        if APPLY:
            open(f, "w", encoding="utf-8").write(new)
    print("\nDone." + ("" if APPLY else "  (dry run — pass --apply)"))


if __name__ == "__main__":
    main()