|
| 1 | +#!/usr/bin/env python3 |
| 2 | +# SPDX-FileCopyrightText: 2025 BayLibre, SAS |
| 3 | +# SPDX-FileCopyrightText: 2026 The RISE Project |
| 4 | +# SPDX-License-Identifier: MIT |
| 5 | +""" |
| 6 | +Extract metadata from a just-built riscv64 wheel, add or update the |
| 7 | +corresponding docs/packages/<name>.yaml entry with the new version, and open |
| 8 | +a pull request with the change. |
| 9 | +
|
| 10 | +docs/packages/generate_packages_doc.py renders this YAML into the published |
| 11 | +Markdown pages, so this script only needs to maintain the YAML source of |
| 12 | +truth; it never touches docs/packages/*.md or index.md directly. |
| 13 | +""" |
| 14 | + |
| 15 | +import os |
| 16 | +import re |
| 17 | +import string |
| 18 | +import subprocess |
| 19 | +import sys |
| 20 | +import zipfile |
| 21 | +from email.message import Message |
| 22 | +from email.parser import Parser |
| 23 | +from pathlib import Path |
| 24 | + |
| 25 | +import yaml |
| 26 | + |
| 27 | +REPO = "riseproject-dev/python-wheels" |
| 28 | +DOCS_DIR = Path("docs/packages") |
| 29 | +PACKAGES_FILE = Path("ci_scripts/packages.txt") |
| 30 | +ARTIFACTS_PATH = os.environ.get("ARTIFACTS_PATH", "dist") |
| 31 | + |
| 32 | + |
| 33 | +def find_wheel_file(path): |
| 34 | + for file in Path(path).glob("*.whl"): |
| 35 | + return file |
| 36 | + return None |
| 37 | + |
| 38 | + |
| 39 | +def normalize_name(name): |
| 40 | + """ |
| 41 | + https://packaging.python.org/en/latest/specifications/name-normalization/#name-normalization |
| 42 | + """ |
| 43 | + return re.sub(r"[-_.]+", "-", name).lower() |
| 44 | + |
| 45 | + |
| 46 | +def normalize_label(label): |
| 47 | + """ |
| 48 | + https://packaging.python.org/en/latest/specifications/well-known-project-urls/#label-normalization |
| 49 | + """ |
| 50 | + chars_to_remove = string.punctuation + string.whitespace |
| 51 | + removal_map = str.maketrans("", "", chars_to_remove) |
| 52 | + return label.translate(removal_map).lower() |
| 53 | + |
| 54 | + |
| 55 | +def extract_license(message): |
| 56 | + license = message.get("License-Expression") |
| 57 | + if not license: |
| 58 | + license = message.get("License", "Unknown") |
| 59 | + return license |
| 60 | + |
| 61 | + |
| 62 | +def extract_source_code_url(message): |
| 63 | + # Collect all "Project-URL" lines |
| 64 | + project_urls = message.get_all("Project-URL", []) |
| 65 | + well_known_labels = ["source", "repository", "sourcecode", "github"] |
| 66 | + |
| 67 | + for entry in project_urls: |
| 68 | + try: |
| 69 | + label, url = map(str.strip, entry.split(",", 1)) |
| 70 | + if normalize_label(label) in well_known_labels: |
| 71 | + return url |
| 72 | + except ValueError: |
| 73 | + continue # skip malformed lines |
| 74 | + |
| 75 | + # A lot of projects use homepage as source code url. Done in a second |
| 76 | + # loop so a homepage entry appearing before a well-known source label |
| 77 | + # doesn't win by accident. |
| 78 | + for entry in project_urls: |
| 79 | + try: |
| 80 | + label, url = map(str.strip, entry.split(",", 1)) |
| 81 | + if normalize_label(label) == "homepage": |
| 82 | + return url |
| 83 | + except ValueError: |
| 84 | + continue |
| 85 | + |
| 86 | + return message.get("Home-page") # deprecated fallback, may be None |
| 87 | + |
| 88 | + |
| 89 | +def extract_metadata_from_whl(whl_path): |
| 90 | + """ |
| 91 | + Extract metadata according to https://packaging.python.org/en/latest/specifications/core-metadata/ |
| 92 | + """ |
| 93 | + with zipfile.ZipFile(whl_path, "r") as z: |
| 94 | + metadata_file = next(f for f in z.namelist() if f.endswith("METADATA")) |
| 95 | + content = z.read(metadata_file).decode() |
| 96 | + message: Message = Parser().parsestr(content) |
| 97 | + return { |
| 98 | + "name": message.get("Name"), |
| 99 | + "version": message.get("Version"), |
| 100 | + "license": extract_license(message), |
| 101 | + "source_code": extract_source_code_url(message), |
| 102 | + } |
| 103 | + |
| 104 | + |
| 105 | +def find_patch_dir(slug, version): |
| 106 | + """ |
| 107 | + Look for a `patches/<slug>/<version_tag>` directory as described in |
| 108 | + docs/development.md, trying both a `v`-prefixed and bare version tag. |
| 109 | + """ |
| 110 | + for tag in (f"v{version}", version): |
| 111 | + candidate = Path("patches") / slug / tag |
| 112 | + if candidate.exists(): |
| 113 | + return candidate |
| 114 | + return None |
| 115 | + |
| 116 | + |
| 117 | +def yaml_line(key, value): |
| 118 | + """Render a single `key: value` YAML mapping line, quoted as needed.""" |
| 119 | + return yaml.safe_dump( |
| 120 | + {key: value}, default_flow_style=False, allow_unicode=True |
| 121 | + ).rstrip("\n") |
| 122 | + |
| 123 | + |
| 124 | +def render_new_yaml(slug, source_code, license, version, patch_dir): |
| 125 | + """Render a brand-new docs/packages/<slug>.yaml for a package's first version.""" |
| 126 | + lines = [yaml_line("package-name", slug)] |
| 127 | + if source_code: |
| 128 | + lines.append(yaml_line("source-code", source_code)) |
| 129 | + lines.append(yaml_line("license", license)) |
| 130 | + lines.append("versions:") |
| 131 | + lines.append(f" - {yaml_line('version', version)}") |
| 132 | + if patch_dir is not None: |
| 133 | + lines.append(" patched:") |
| 134 | + return "\n".join(lines) + "\n" |
| 135 | + |
| 136 | + |
| 137 | +def append_version(content, package_data, version, license, patch_dir): |
| 138 | + """ |
| 139 | + Append a new version entry to the end of an existing package YAML file's |
| 140 | + `versions:` list, preserving the rest of the file byte-for-byte. |
| 141 | +
|
| 142 | + Returns None if this exact version is already documented. |
| 143 | + """ |
| 144 | + existing_versions = { |
| 145 | + str(v.get("version")) for v in (package_data.get("versions") or []) |
| 146 | + } |
| 147 | + if str(version) in existing_versions: |
| 148 | + return None |
| 149 | + |
| 150 | + top_level_license = package_data.get("license") |
| 151 | + lines = [f" - {yaml_line('version', version)}"] |
| 152 | + if patch_dir is not None: |
| 153 | + lines.append(" patched:") |
| 154 | + if license and license != top_level_license: |
| 155 | + lines.append(f" {yaml_line('license', license)}") |
| 156 | + |
| 157 | + return content.rstrip("\n") + "\n" + "\n".join(lines) + "\n" |
| 158 | + |
| 159 | + |
| 160 | +def add_to_packages_file(slug): |
| 161 | + lines = PACKAGES_FILE.read_text().splitlines() |
| 162 | + header_end = next(i for i, line in enumerate(lines) if line and not line.startswith("#")) |
| 163 | + header, entries = lines[:header_end], [line for line in lines[header_end:] if line] |
| 164 | + entries = sorted(set(entries) | {slug}, key=str.casefold) |
| 165 | + PACKAGES_FILE.write_text("\n".join(header + entries) + "\n") |
| 166 | + |
| 167 | + |
| 168 | +def git_run(*args): |
| 169 | + subprocess.run(["git", *args], check=True) |
| 170 | + |
| 171 | + |
| 172 | +def configure_git_identity(): |
| 173 | + git_run("config", "user.name", "github-actions[bot]") |
| 174 | + git_run("config", "user.email", "41898282+github-actions[bot]@users.noreply.github.com") |
| 175 | + |
| 176 | + |
| 177 | +def extract_pr_url(stdout): |
| 178 | + for line in stdout.split("\n"): |
| 179 | + line = line.strip() |
| 180 | + if "github.com" in line and "/pull/" in line: |
| 181 | + return line |
| 182 | + return None |
| 183 | + |
| 184 | + |
| 185 | +def main(): |
| 186 | + whl_file = find_wheel_file(ARTIFACTS_PATH) |
| 187 | + if not whl_file: |
| 188 | + print(f"No .whl file found in {ARTIFACTS_PATH}") |
| 189 | + sys.exit(1) |
| 190 | + |
| 191 | + metadata = extract_metadata_from_whl(whl_file) |
| 192 | + display_name = metadata["name"] |
| 193 | + version = metadata["version"] |
| 194 | + license = metadata["license"] |
| 195 | + source_code = metadata["source_code"] |
| 196 | + |
| 197 | + if not display_name or not version: |
| 198 | + print("Name or version could not be extracted") |
| 199 | + sys.exit(1) |
| 200 | + |
| 201 | + slug = normalize_name(display_name) |
| 202 | + patch_dir = find_patch_dir(slug, version) |
| 203 | + yaml_path = DOCS_DIR / f"{slug}.yaml" |
| 204 | + is_new = not yaml_path.exists() |
| 205 | + |
| 206 | + if is_new: |
| 207 | + yaml_path.write_text( |
| 208 | + render_new_yaml(slug, source_code, license, version, patch_dir) |
| 209 | + ) |
| 210 | + else: |
| 211 | + content = yaml_path.read_text() |
| 212 | + package_data = yaml.safe_load(content) or {} |
| 213 | + updated = append_version(content, package_data, version, license, patch_dir) |
| 214 | + if updated is None: |
| 215 | + print(f"{slug} {version} is already documented; nothing to do") |
| 216 | + return |
| 217 | + yaml_path.write_text(updated) |
| 218 | + |
| 219 | + configure_git_identity() |
| 220 | + |
| 221 | + branch = f"github-actions/{'add' if is_new else 'update'}-doc-for-{slug}" |
| 222 | + git_run("switch", "-c", branch) |
| 223 | + git_run("add", str(yaml_path)) |
| 224 | + |
| 225 | + if is_new: |
| 226 | + add_to_packages_file(slug) |
| 227 | + git_run("add", str(PACKAGES_FILE)) |
| 228 | + git_run("commit", "-s", "-m", f"docs: add {slug}\n\nAdd version {version}") |
| 229 | + else: |
| 230 | + git_run("commit", "-s", "-m", f"docs: update {slug}\n\nAdd version {version}") |
| 231 | + |
| 232 | + git_run("push", "origin", branch) |
| 233 | + |
| 234 | + result = subprocess.run( |
| 235 | + [ |
| 236 | + "gh", "pr", "create", "--draft", |
| 237 | + "--repo", REPO, |
| 238 | + "--base", "main", |
| 239 | + "--head", branch, |
| 240 | + "--reviewer", "threexc,justeph", |
| 241 | + "--title", f"docs: {'add' if is_new else 'update'} {slug}", |
| 242 | + "--body", |
| 243 | + "Automatically generated PR to document a newly published wheel. " |
| 244 | + "Please review it carefully before merging.\n\n" |
| 245 | + "If necessary, force-push this branch.", |
| 246 | + ], |
| 247 | + capture_output=True, text=True, check=True, |
| 248 | + ) |
| 249 | + pr_url = extract_pr_url(result.stdout) |
| 250 | + print(f"[+] Opened PR: {pr_url or '(URL not found in output)'}") |
| 251 | + |
| 252 | + |
| 253 | +if __name__ == "__main__": |
| 254 | + main() |
0 commit comments