Files
jobtrackingapp/scripts/compare-cv-approved-fixtures.py

172 lines
7.2 KiB
Python

#!/usr/bin/env python3
import argparse
import importlib.util
import json
import pathlib
import sys
from typing import Any
HELPER_PATH = pathlib.Path(__file__).with_name("prepare-cv-approved-fixtures.py")
HELPER_SPEC = importlib.util.spec_from_file_location("prepare_cv_approved_fixtures", HELPER_PATH)
if HELPER_SPEC is None or HELPER_SPEC.loader is None:
raise RuntimeError(f"Unable to load helper script: {HELPER_PATH}")
HELPER_MODULE = importlib.util.module_from_spec(HELPER_SPEC)
HELPER_SPEC.loader.exec_module(HELPER_MODULE)
normalize_fixture = HELPER_MODULE.normalize_fixture
def load_json(path: pathlib.Path) -> Any:
return json.loads(path.read_text(encoding="utf-8", errors="replace"))
def diff_values(expected: Any, actual: Any, path: str = "$") -> list[dict[str, Any]]:
diffs: list[dict[str, Any]] = []
if type(expected) is not type(actual):
diffs.append({"path": path, "expected": expected, "actual": actual, "kind": "type-mismatch"})
return diffs
if isinstance(expected, dict):
expected_keys = set(expected.keys())
actual_keys = set(actual.keys())
for key in sorted(expected_keys | actual_keys):
child_path = f"{path}.{key}"
if key not in expected:
diffs.append({"path": child_path, "expected": None, "actual": actual[key], "kind": "unexpected-key"})
continue
if key not in actual:
diffs.append({"path": child_path, "expected": expected[key], "actual": None, "kind": "missing-key"})
continue
diffs.extend(diff_values(expected[key], actual[key], child_path))
return diffs
if isinstance(expected, list):
if len(expected) != len(actual):
diffs.append({"path": path, "expected": f"len={len(expected)}", "actual": f"len={len(actual)}", "kind": "length-mismatch"})
for index, (expected_item, actual_item) in enumerate(zip(expected, actual)):
diffs.extend(diff_values(expected_item, actual_item, f"{path}[{index}]"))
return diffs
if expected != actual:
diffs.append({"path": path, "expected": expected, "actual": actual, "kind": "value-mismatch"})
return diffs
def summarize_sections(diffs: list[dict[str, Any]]) -> dict[str, int]:
sections: dict[str, int] = {}
for diff in diffs:
path = diff["path"]
remainder = path[2:] if path.startswith('$.') else path
head = remainder.split('.', 1)[0].split('[', 1)[0]
sections[head] = sections.get(head, 0) + 1
return dict(sorted(sections.items(), key=lambda item: (-item[1], item[0])))
def render_markdown(results: list[dict[str, Any]], missing_expected: list[str], missing_actual: list[str], approved_dir: pathlib.Path, actual_dir: pathlib.Path) -> str:
total = len(results)
matched = sum(1 for result in results if result["diffCount"] == 0)
lines = [
"# CV fixture comparison report",
"",
f"- Approved fixtures: `{approved_dir}`",
f"- Actual outputs: `{actual_dir}`",
f"- Compared files: {total}",
f"- Exact matches: {matched}",
f"- Files with differences: {total - matched}",
"",
]
if missing_expected:
lines.append("## Missing approved fixtures")
lines.extend(f"- {name}" for name in missing_expected)
lines.append("")
if missing_actual:
lines.append("## Missing actual outputs")
lines.extend(f"- {name}" for name in missing_actual)
lines.append("")
lines.append("## File results")
lines.append("")
lines.append("| File | Diffs | Top sections |")
lines.append("|---|---:|---|")
for result in results:
section_summary = ", ".join(f"{key}:{value}" for key, value in result["sectionCounts"].items()) or ""
lines.append(f"| {result['file']} | {result['diffCount']} | {section_summary} |")
lines.append("")
lines.append("## Detailed mismatches")
lines.append("")
for result in results:
if result["diffCount"] == 0:
continue
lines.append(f"### {result['file']}")
for diff in result["diffs"][:50]:
lines.append(f"- `{diff['path']}` ({diff['kind']})")
lines.append(f" - expected: `{json.dumps(diff['expected'], ensure_ascii=False)}`")
lines.append(f" - actual: `{json.dumps(diff['actual'], ensure_ascii=False)}`")
if result["diffCount"] > 50:
lines.append(f"- … {result['diffCount'] - 50} more differences omitted")
lines.append("")
return "\n".join(lines)
def main() -> int:
parser = argparse.ArgumentParser(description="Compare actual CV extractor outputs against approved fixtures.")
parser.add_argument("--approved", default="/home/pi/cvs/approved-jsons", help="Directory containing approved fixture JSON files")
parser.add_argument("--actual", required=True, help="Directory containing actual extractor output JSON files")
parser.add_argument("--output", help="Directory to write comparison artifacts. Defaults to <actual>/../comparison")
args = parser.parse_args()
approved_dir = pathlib.Path(args.approved)
actual_dir = pathlib.Path(args.actual)
output_dir = pathlib.Path(args.output) if args.output else actual_dir.parent / "comparison"
if not approved_dir.is_dir():
print(f"Approved fixture directory not found: {approved_dir}", file=sys.stderr)
return 1
if not actual_dir.is_dir():
print(f"Actual output directory not found: {actual_dir}", file=sys.stderr)
return 1
approved_files = {path.name: path for path in approved_dir.glob("*.json")}
actual_files = {path.name: path for path in actual_dir.glob("*.json")}
common_names = sorted(set(approved_files) & set(actual_files))
missing_expected = sorted(set(actual_files) - set(approved_files))
missing_actual = sorted(set(approved_files) - set(actual_files))
results: list[dict[str, Any]] = []
for name in common_names:
expected = normalize_fixture(load_json(approved_files[name]))
actual = normalize_fixture(load_json(actual_files[name]))
diffs = diff_values(expected, actual)
results.append({
"file": name,
"diffCount": len(diffs),
"sectionCounts": summarize_sections(diffs),
"diffs": diffs,
})
output_dir.mkdir(parents=True, exist_ok=True)
(output_dir / "comparison.json").write_text(json.dumps({
"approvedDir": str(approved_dir),
"actualDir": str(actual_dir),
"comparedFiles": len(results),
"missingApprovedFixtures": missing_expected,
"missingActualOutputs": missing_actual,
"results": results,
}, indent=2, ensure_ascii=False) + "\n", encoding="utf-8")
(output_dir / "comparison.md").write_text(render_markdown(results, missing_expected, missing_actual, approved_dir, actual_dir) + "\n", encoding="utf-8")
print(f"Comparison JSON: {output_dir / 'comparison.json'}")
print(f"Comparison markdown: {output_dir / 'comparison.md'}")
print(f"Compared files: {len(results)}")
print(f"Files with differences: {sum(1 for result in results if result['diffCount'] > 0)}")
return 0
if __name__ == "__main__":
raise SystemExit(main())