teleo-infrastructure/scripts/backfill-synthetic-recovery-prs.py

#!/usr/bin/env python3
"""Reconstruct synthetic `prs` rows for historical GitHub PRs lost pre-mirror-wiring.

Two PRs merged on GitHub before our sync-mirror.sh tracked `github_pr`:
  - GitHub PR #68: alexastrum — 6 claims, merged 2026-03-09 via GitHub squash,
    recovered to Forgejo via commit dba00a79 (Apr 16, after mirror erased files)
  - GitHub PR #88: Cameron-S1 — 1 claim, recovered via commit da64f805

The recovery commits wrote the files directly to main, so our `prs` table has
no row to attach originator events to — the backfill-events.py strategies all
return NULL. We reconstruct one synthetic `prs` row per historical GitHub PR so
the events pipeline (and `github_pr` strategy in backfill-events) can credit
Alex and Cameron properly.

Numbers 900000+ are clearly synthetic and won't collide with real Forgejo PRs.

Idempotent via INSERT OR IGNORE.

Usage:
  python3 scripts/backfill-synthetic-recovery-prs.py --dry-run
  python3 scripts/backfill-synthetic-recovery-prs.py
"""
import argparse
import os
import sqlite3
import sys
from pathlib import Path

DB_PATH = os.environ.get("PIPELINE_DB", "/opt/teleo-eval/pipeline/pipeline.db")

# Historical GitHub PRs recovered via direct-to-main commits.
# Original GitHub merge dates come from the recovery commit messages.
RECOVERY_PRS = [
    {
        "number": 900068,
        "github_pr": 68,
        "branch": "gh-pr-68",
        "status": "merged",
        "domain": "ai-alignment",
        "commit_type": "knowledge",
        "tier": "STANDARD",
        "leo_verdict": "approve",
        "domain_verdict": "approve",
        "submitted_by": "alexastrum",
        "source_channel": "github",
        # origin='human' matches lib/merge.py convention for external contributors
        # (default is 'pipeline' which misclassifies us as machine-authored).
        "origin": "human",
        "priority": "high",
        "description": "Multi-agent git workflows production maturity | Cryptographic agent trust ratings | Defense in depth for AI agent oversight | Deterministic policy engines below LLM layer | Knowledge validation four-layer architecture | Structurally separating proposer and reviewer agents",
        "merged_at": "2026-03-09 00:00:00",
        "created_at": "2026-03-08 00:00:00",
        "last_error": "synthetic_recovery: GitHub PR #68 pre-mirror-wiring reconstruction (commit dba00a79)",
    },
    {
        "number": 900088,
        "github_pr": 88,
        "branch": "gh-pr-88",
        "status": "merged",
        "domain": "ai-alignment",
        "commit_type": "knowledge",
        "tier": "STANDARD",
        "leo_verdict": "approve",
        "domain_verdict": "approve",
        "submitted_by": "cameron-s1",
        "source_channel": "github",
        "origin": "human",
        "priority": "high",
        "description": "Orthogonality is an artefact of specification architectures not a property of intelligence itself",
        "merged_at": "2026-04-01 00:00:00",
        "created_at": "2026-04-01 00:00:00",
        "last_error": "synthetic_recovery: GitHub PR #88 pre-mirror-wiring reconstruction (commit da64f805)",
    },
]


def main():
    parser = argparse.ArgumentParser()
    parser.add_argument("--dry-run", action="store_true")
    args = parser.parse_args()

    if not Path(DB_PATH).exists():
        print(f"ERROR: DB not found at {DB_PATH}", file=sys.stderr)
        sys.exit(1)

    conn = sqlite3.connect(DB_PATH, timeout=30)
    conn.row_factory = sqlite3.Row

    # Guard against synthetic-range colonization (Ganymede review): check for
    # any row in the synthetic range that isn't one of ours. INSERT OR IGNORE on
    # the specific numbers is the real collision defense; this is belt-and-suspenders.
    max_real = conn.execute(
        "SELECT MAX(number) FROM prs WHERE number < 900000"
    ).fetchone()[0] or 0
    print(f"Max real Forgejo PR number: {max_real}")
    synth_conflict = conn.execute(
        "SELECT number FROM prs WHERE number >= 900000 AND number NOT IN (900068, 900088) LIMIT 1"
    ).fetchone()
    if synth_conflict:
        print(f"ERROR: PR #{synth_conflict[0]} already exists in synthetic range. "
              f"Pick a new range before running.", file=sys.stderr)
        sys.exit(2)

    inserted = 0
    skipped = 0
    for row in RECOVERY_PRS:
        existing = conn.execute(
            "SELECT number FROM prs WHERE number = ? OR github_pr = ?",
            (row["number"], row["github_pr"]),
        ).fetchone()
        if existing:
            print(f"  PR #{row['number']} (github_pr={row['github_pr']}): already exists — skip")
            skipped += 1
            continue
        print(f"  {'(dry-run) ' if args.dry_run else ''}INSERT synthetic PR #{row['number']} "
              f"(github_pr={row['github_pr']}, submitted_by={row['submitted_by']}, "
              f"merged_at={row['merged_at']})")
        if not args.dry_run:
            conn.execute(
                """INSERT INTO prs (
                    number, github_pr, branch, status, domain, commit_type, tier,
                    leo_verdict, domain_verdict, submitted_by, source_channel,
                    origin, priority,
                    description, merged_at, created_at, last_error
                ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)""",
                (
                    row["number"], row["github_pr"], row["branch"], row["status"],
                    row["domain"], row["commit_type"], row["tier"],
                    row["leo_verdict"], row["domain_verdict"],
                    row["submitted_by"], row["source_channel"],
                    row["origin"], row["priority"],
                    row["description"], row["merged_at"], row["created_at"],
                    row["last_error"],
                ),
            )
            inserted += 1

    if not args.dry_run:
        conn.commit()

    print(f"\nInserted {inserted}, skipped {skipped}")
    if not args.dry_run and inserted:
        print("\nNext step: re-run backfill-events.py to attach originator events")
        print("  python3 ops/backfill-events.py")


if __name__ == "__main__":
    main()