|
| 1 | +#!/usr/bin/env python3 |
| 2 | +"""Validate the bilingual PIG design-record contract.""" |
| 3 | + |
| 4 | +from __future__ import annotations |
| 5 | + |
| 6 | +import pathlib |
| 7 | +import re |
| 8 | +import sys |
| 9 | + |
| 10 | + |
| 11 | +DESIGN_DIR = pathlib.Path("content/blog/design") |
| 12 | +ALLOWED_TAGS = { |
| 13 | + "repo", |
| 14 | + "ext", |
| 15 | + "postgres", |
| 16 | + "patroni", |
| 17 | + "pgbackrest", |
| 18 | + "pitr", |
| 19 | + "sty", |
| 20 | + "build", |
| 21 | + "inventory", |
| 22 | + "catalog", |
| 23 | + "cli", |
| 24 | + "install", |
| 25 | +} |
| 26 | +REQUIRED_KEYS = { |
| 27 | + "title", |
| 28 | + "linkTitle", |
| 29 | + "date", |
| 30 | + "lastmod", |
| 31 | + "description", |
| 32 | + "tags", |
| 33 | + "weight", |
| 34 | + "authors", |
| 35 | + "draft", |
| 36 | +} |
| 37 | +REQUIRED_ANCHORS = { |
| 38 | + "decision", |
| 39 | + "context", |
| 40 | + "alternatives", |
| 41 | + "contract", |
| 42 | + "impact", |
| 43 | + "verification", |
| 44 | + "status", |
| 45 | +} |
| 46 | +FORBIDDEN = { |
| 47 | + "local absolute path": re.compile(r"/Users/|/home/[A-Za-z0-9_.-]+/"), |
| 48 | + "retired local document reference": re.compile(r"docs/(?:spec|refactor)/"), |
| 49 | + "temporary review reference": re.compile(r"(?:^|[\s`/])tmp/"), |
| 50 | + "unfinished marker": re.compile(r"\b(?:TODO|TBD|DRAFT)\b"), |
| 51 | + "raw agent-review detail": re.compile(r"Claude Code|claude-(?:sonnet|opus|fable)", re.I), |
| 52 | +} |
| 53 | +ANCHOR_RE = re.compile(r"\{#([a-z0-9-]+)\}") |
| 54 | + |
| 55 | + |
| 56 | +def parse_page(path: pathlib.Path) -> tuple[dict[str, str], str, list[str]]: |
| 57 | + errors: list[str] = [] |
| 58 | + text = path.read_text(encoding="utf-8") |
| 59 | + lines = text.splitlines() |
| 60 | + if not lines or lines[0] != "---": |
| 61 | + return {}, text, [f"{path}: missing YAML front matter"] |
| 62 | + try: |
| 63 | + end = lines.index("---", 1) |
| 64 | + except ValueError: |
| 65 | + return {}, text, [f"{path}: unclosed YAML front matter"] |
| 66 | + meta: dict[str, str] = {} |
| 67 | + for number, line in enumerate(lines[1:end], 2): |
| 68 | + if not line or line.startswith((" ", "\t", "#")): |
| 69 | + continue |
| 70 | + if ":" not in line: |
| 71 | + errors.append(f"{path}:{number}: unsupported front matter line") |
| 72 | + continue |
| 73 | + key, value = line.split(":", 1) |
| 74 | + meta[key.strip()] = value.strip() |
| 75 | + return meta, "\n".join(lines[end + 1 :]), errors |
| 76 | + |
| 77 | + |
| 78 | +def parse_list(value: str) -> list[str]: |
| 79 | + if not (value.startswith("[") and value.endswith("]")): |
| 80 | + return [] |
| 81 | + return [item.strip().strip("\"'") for item in value[1:-1].split(",") if item.strip()] |
| 82 | + |
| 83 | + |
| 84 | +def validate_index(errors: list[str]) -> None: |
| 85 | + for name in ("_index.md", "_index.zh.md"): |
| 86 | + path = DESIGN_DIR / name |
| 87 | + if not path.is_file(): |
| 88 | + errors.append(f"missing design index: {path}") |
| 89 | + continue |
| 90 | + meta, _, page_errors = parse_page(path) |
| 91 | + errors.extend(page_errors) |
| 92 | + for key in ("title", "linkTitle", "description", "weight", "module", "blog_index"): |
| 93 | + if not meta.get(key): |
| 94 | + errors.append(f"{path}: missing index front matter key {key}") |
| 95 | + |
| 96 | + |
| 97 | +def validate_pair(english: pathlib.Path, chinese: pathlib.Path, errors: list[str]) -> None: |
| 98 | + en_meta, en_body, en_errors = parse_page(english) |
| 99 | + zh_meta, zh_body, zh_errors = parse_page(chinese) |
| 100 | + errors.extend(en_errors) |
| 101 | + errors.extend(zh_errors) |
| 102 | + |
| 103 | + for path, meta in ((english, en_meta), (chinese, zh_meta)): |
| 104 | + missing = sorted(REQUIRED_KEYS - meta.keys()) |
| 105 | + if missing: |
| 106 | + errors.append(f"{path}: missing front matter keys: {', '.join(missing)}") |
| 107 | + if meta.get("authors") != "[Vonng]": |
| 108 | + errors.append(f"{path}: authors must be [Vonng]") |
| 109 | + if meta.get("draft") != "false": |
| 110 | + errors.append(f"{path}: draft must be false") |
| 111 | + tags = parse_list(meta.get("tags", "")) |
| 112 | + if not tags or len(tags) > 4: |
| 113 | + errors.append(f"{path}: tags must contain one to four entries") |
| 114 | + unknown = sorted(set(tags) - ALLOWED_TAGS) |
| 115 | + if unknown: |
| 116 | + errors.append(f"{path}: unknown tags: {', '.join(unknown)}") |
| 117 | + |
| 118 | + for key in ("date", "lastmod", "weight", "authors", "draft"): |
| 119 | + if en_meta.get(key) != zh_meta.get(key): |
| 120 | + errors.append(f"{english} / {chinese}: {key} differs") |
| 121 | + |
| 122 | + en_anchors = ANCHOR_RE.findall(en_body) |
| 123 | + zh_anchors = ANCHOR_RE.findall(zh_body) |
| 124 | + if set(en_anchors) != REQUIRED_ANCHORS: |
| 125 | + errors.append(f"{english}: design anchors differ from the required contract") |
| 126 | + if en_anchors != zh_anchors: |
| 127 | + errors.append(f"{english} / {chinese}: heading anchors or order differ") |
| 128 | + |
| 129 | + for path, body, marker in ( |
| 130 | + (english, en_body, "**Decision date:**"), |
| 131 | + (chinese, zh_body, "**决策日期：**"), |
| 132 | + ): |
| 133 | + if marker not in body: |
| 134 | + errors.append(f"{path}: missing visible decision metadata") |
| 135 | + if "]( /" in body: |
| 136 | + errors.append(f"{path}: malformed internal link spacing") |
| 137 | + if "](/" not in body: |
| 138 | + errors.append(f"{path}: missing current-site reference") |
| 139 | + if "https://github.com/pgsty/pig/" not in body: |
| 140 | + errors.append(f"{path}: missing pinned source or release evidence") |
| 141 | + for label, pattern in FORBIDDEN.items(): |
| 142 | + match = pattern.search(body) |
| 143 | + if match: |
| 144 | + errors.append(f"{path}: contains {label} near {match.group(0)!r}") |
| 145 | + |
| 146 | + |
| 147 | +def main() -> int: |
| 148 | + errors: list[str] = [] |
| 149 | + if not DESIGN_DIR.is_dir(): |
| 150 | + print(f"missing design directory: {DESIGN_DIR}", file=sys.stderr) |
| 151 | + return 1 |
| 152 | + |
| 153 | + validate_index(errors) |
| 154 | + english_pages = sorted( |
| 155 | + path |
| 156 | + for path in DESIGN_DIR.glob("*.md") |
| 157 | + if path.name != "_index.md" and not path.name.endswith(".zh.md") |
| 158 | + ) |
| 159 | + if not english_pages: |
| 160 | + errors.append("design section has no English records") |
| 161 | + expected_chinese = {path.with_name(f"{path.stem}.zh.md") for path in english_pages} |
| 162 | + actual_chinese = { |
| 163 | + path for path in DESIGN_DIR.glob("*.zh.md") if path.name != "_index.zh.md" |
| 164 | + } |
| 165 | + for missing in sorted(expected_chinese - actual_chinese): |
| 166 | + errors.append(f"missing Chinese design record: {missing}") |
| 167 | + for orphan in sorted(actual_chinese - expected_chinese): |
| 168 | + errors.append(f"orphan Chinese design record: {orphan}") |
| 169 | + for english in english_pages: |
| 170 | + chinese = english.with_name(f"{english.stem}.zh.md") |
| 171 | + if chinese.is_file(): |
| 172 | + validate_pair(english, chinese, errors) |
| 173 | + |
| 174 | + if errors: |
| 175 | + print(f"design check failed: {len(errors)} issues", file=sys.stderr) |
| 176 | + for error in errors: |
| 177 | + print(f"- {error}", file=sys.stderr) |
| 178 | + return 1 |
| 179 | + |
| 180 | + print(f"design check passed: {len(english_pages)} bilingual records") |
| 181 | + return 0 |
| 182 | + |
| 183 | + |
| 184 | +if __name__ == "__main__": |
| 185 | + raise SystemExit(main()) |
0 commit comments