|
| 1 | +"""Require pyproject.toml to track the latest finalized HF data release.""" |
| 2 | + |
| 3 | +from __future__ import annotations |
| 4 | + |
| 5 | +import argparse |
| 6 | +import json |
| 7 | +import os |
| 8 | +from pathlib import Path |
| 9 | +import re |
| 10 | +import sys |
| 11 | +from urllib.error import URLError |
| 12 | +from urllib.request import urlopen |
| 13 | + |
| 14 | + |
| 15 | +REPO_ROOT = Path(__file__).resolve().parents[2] |
| 16 | +DEFAULT_VERSION_MANIFEST_URL = ( |
| 17 | + "https://huggingface.co/policyengine/policyengine-us-data/" |
| 18 | + "resolve/main/version_manifest.json" |
| 19 | +) |
| 20 | +VERSION_RE = re.compile(r'^version\s*=\s*"([^"]+)"', re.MULTILINE) |
| 21 | +SEMVER_RE = re.compile(r"^(\d+)\.(\d+)\.(\d+)(?:rc\d+)?$") |
| 22 | + |
| 23 | + |
| 24 | +def stable_version_tuple(version: str) -> tuple[int, int, int]: |
| 25 | + match = SEMVER_RE.match(version) |
| 26 | + if not match: |
| 27 | + raise ValueError(f"Unsupported version format: {version}") |
| 28 | + return tuple(int(part) for part in match.groups()) |
| 29 | + |
| 30 | + |
| 31 | +def pyproject_version(root: Path = REPO_ROOT) -> str: |
| 32 | + text = (root / "pyproject.toml").read_text() |
| 33 | + match = VERSION_RE.search(text) |
| 34 | + if not match: |
| 35 | + raise ValueError("Could not find project version in pyproject.toml") |
| 36 | + return match.group(1) |
| 37 | + |
| 38 | + |
| 39 | +def latest_hf_release_version( |
| 40 | + url: str = DEFAULT_VERSION_MANIFEST_URL, |
| 41 | +) -> str: |
| 42 | + with urlopen(url, timeout=30) as response: |
| 43 | + payload = json.load(response) |
| 44 | + current = payload.get("current") |
| 45 | + if isinstance(current, str) and current: |
| 46 | + return current |
| 47 | + versions = payload.get("versions") |
| 48 | + if not isinstance(versions, list) or not versions: |
| 49 | + raise ValueError("HF version_manifest.json has no current version") |
| 50 | + latest = versions[-1].get("version") |
| 51 | + if not isinstance(latest, str) or not latest: |
| 52 | + raise ValueError("HF version_manifest.json latest entry has no version") |
| 53 | + return latest |
| 54 | + |
| 55 | + |
| 56 | +def version_violations( |
| 57 | + *, |
| 58 | + package_version: str, |
| 59 | + finalized_release_version: str, |
| 60 | +) -> list[str]: |
| 61 | + if stable_version_tuple(package_version) >= stable_version_tuple( |
| 62 | + finalized_release_version |
| 63 | + ): |
| 64 | + return [] |
| 65 | + return [ |
| 66 | + "pyproject.toml version " |
| 67 | + f"{package_version} is behind finalized HF data release " |
| 68 | + f"{finalized_release_version}. Finalize the package version before " |
| 69 | + "creating another publication candidate." |
| 70 | + ] |
| 71 | + |
| 72 | + |
| 73 | +def check_repository( |
| 74 | + root: Path = REPO_ROOT, |
| 75 | + *, |
| 76 | + finalized_release_version: str | None = None, |
| 77 | + version_manifest_url: str = DEFAULT_VERSION_MANIFEST_URL, |
| 78 | +) -> list[str]: |
| 79 | + package_version = pyproject_version(root) |
| 80 | + finalized_release_version = finalized_release_version or latest_hf_release_version( |
| 81 | + version_manifest_url |
| 82 | + ) |
| 83 | + return version_violations( |
| 84 | + package_version=package_version, |
| 85 | + finalized_release_version=finalized_release_version, |
| 86 | + ) |
| 87 | + |
| 88 | + |
| 89 | +def main(argv: list[str] | None = None) -> int: |
| 90 | + parser = argparse.ArgumentParser(description=__doc__) |
| 91 | + parser.add_argument( |
| 92 | + "--mode", |
| 93 | + choices=("warn", "fail"), |
| 94 | + default="fail", |
| 95 | + help="Whether stale versions should fail the command.", |
| 96 | + ) |
| 97 | + parser.add_argument( |
| 98 | + "--version-manifest-url", |
| 99 | + default=os.environ.get( |
| 100 | + "US_DATA_VERSION_MANIFEST_URL", DEFAULT_VERSION_MANIFEST_URL |
| 101 | + ), |
| 102 | + ) |
| 103 | + args = parser.parse_args(argv) |
| 104 | + |
| 105 | + try: |
| 106 | + violations = check_repository( |
| 107 | + version_manifest_url=args.version_manifest_url, |
| 108 | + ) |
| 109 | + except (URLError, OSError, ValueError) as exc: |
| 110 | + print( |
| 111 | + f"Could not check finalized HF data release version: {exc}", file=sys.stderr |
| 112 | + ) |
| 113 | + return 1 if args.mode == "fail" else 0 |
| 114 | + |
| 115 | + if not violations: |
| 116 | + print("Data package version is current with the latest finalized HF release.") |
| 117 | + return 0 |
| 118 | + |
| 119 | + for violation in violations: |
| 120 | + print(violation, file=sys.stderr) |
| 121 | + return 1 if args.mode == "fail" else 0 |
| 122 | + |
| 123 | + |
| 124 | +if __name__ == "__main__": |
| 125 | + sys.exit(main()) |
0 commit comments