#!/usr/bin/env python3
"""Generated regression tests from fix commits — Compounding Intelligence #87."""

import argparse, re, subprocess, sys
from pathlib import Path

HERE = Path(__file__).parent
ROOT = HERE.parent
TESTS_DIR = ROOT / "tests"
OUT_FILE = TESTS_DIR / "test_regression_generated.py"

def run_git(args, cwd):
    r = subprocess.run(["git"] + args, capture_output=True, text=True, cwd=str(cwd))
    if r.returncode != 0:
        raise RuntimeError(r.stderr.strip() or "git error")
    return r.stdout.strip()

def get_fix_commits(since=None):
    args = ["log", "--all", "--grep=fix", "--format=%H"]
    if since:
        args.append(f"--since={since}")
    out = run_git(args, ROOT)
    return [l.strip() for l in out.splitlines() if l.strip()]

def get_commit_info(sha):
    """Return message, full diff, and list of changed file paths."""
    msg = run_git(["show", "--no-patch", "--format=%s", sha], ROOT)
    diff = run_git(["show", "--format=full", sha], ROOT)
    files_out = run_git(["diff-tree", "--no-commit-id", "--name-only", "-r", sha], ROOT)
    files = [p for p in files_out.splitlines() if p.strip()]
    return {"sha": sha, "msg": msg, "diff": diff, "files": files}

# ── Test templates ───────────────────────────────────────────────────────
REGEX_TEST = """
class TestRegression_{prefix}(unittest.TestCase):
    \"\"\"Regression: regex syntax fix - commit {commit}.\"\"\"
    def test_regex_compiles(self):
        import re
        pattern = r"open\\\\([^)]*)[\\x27\\x22]w[\\x27\\x22]"
        try:
            regex = re.compile(pattern)
        except SyntaxError as e:
            self.fail(f"Regex still invalid after fix: {e}")
        self.assertRegex("open(test_file, 'w')", regex)
        self.assertRegex('open(test_file, "w")', regex)
        self.assertNotRegex("open(test_file, 'r')", regex)
"""

GENERIC_TEST = """
class TestRegression_{prefix}(unittest.TestCase):
    \"\"\"Regression guard: {first_line} - commit {sha}.\"\"\"
    def test_fixed_file_exists(self):
        from pathlib import Path
        p = Path("{file_path}")
        self.assertTrue(p.exists(), f"Fixed file missing: {file_path}")
"""

# ── Generation ───────────────────────────────────────────────────────────
def generate(commits):
    cases = []
    for sha in commits:
        try:
            info = get_commit_info(sha)
            # Keep only existing files (skip ones deleted/removed later)
            existing = [p for p in info["files"] if (ROOT / p).exists()]
            if not existing:
                continue
            first_file = existing[0]
            # Heuristic: regex-related fix if message or diff mentions open( with write mode pattern
            content = info["msg"] + "n" + info["diff"]
            if re.search(r"open\\\\([^)]*)[\"']w[\"']", content, re.IGNORECASE):
                cases.append(REGEX_TEST.format(prefix=sha[:8], commit=sha))
            else:
                first_line = info["msg"].replace('"', '\\"')[:80]
                cases.append(GENERIC_TEST.format(
                    prefix=sha[:8],
                    file_path=first_file,
                    first_line=first_line,
                    sha=sha))
        except Exception as e:
            print(f"[WARN] {sha[:8]}: {e}", file=sys.stderr)

    OUT_FILE.parent.mkdir(parents=True, exist_ok=True)
    OUT_FILE.write_text(
        f"""# AUTO-GENERATED — DO NOT EDIT
import unittest
from pathlib import Path

{"".join(cases)}

if __name__ == "__main__":
    unittest.main()
""",
        encoding="utf-8"
    )
    print(f"Wrote {OUT_FILE} — {len(cases)} test cases")

def main():
    parser = argparse.ArgumentParser()
    parser.add_argument("--commit", help="specific commit SHA")
    parser.add_argument("--since", help="e.g. 2025-01-01")
    args = parser.parse_args()
    shas = [args.commit] if args.commit else get_fix_commits(args.since)
    print(f"Scanning {len(shas)} fix commits…")
    generate(shas)

if __name__ == "__main__":
    main()