First commit

1 files changed, 94 insertions, 0 deletions

diff --git a/tools/html_audit.py b/tools/html_audit.py
new file mode 100644
index 0000000..4e85c36
--- /dev/null
+++ b/tools/html_audit.py
@@ -0,0 +1,94 @@
+#!/usr/bin/env python3
+"""Lightweight HTML audit for common a11y/markup issues."""
+
+from __future__ import annotations
+
+from html.parser import HTMLParser
+from pathlib import Path
+
+ROOT = Path(__file__).resolve().parents[1]
+
+SKIP_DIRS = {"partials", ".git"}
+SKIP_FILES = {
+    "test.html",
+    "test_jp.html",
+    "test_zh.html",
+    "startpage/test.html",
+}
+
+
+class AuditParser(HTMLParser):
+    def __init__(self) -> None:
+        super().__init__()
+        self.ids: dict[str, int] = {}
+        self.duplicate_ids: set[str] = set()
+        self.missing_alt: list[str] = []
+        self.missing_iframe_title: list[str] = []
+        self.blank_rel: list[str] = []
+
+    def handle_starttag(self, tag: str, attrs: list[tuple[str, str | None]]) -> None:
+        attr_map = {k.lower(): (v or "") for k, v in attrs}
+
+        if "id" in attr_map:
+            ident = attr_map["id"]
+            if ident:
+                if ident in self.ids:
+                    self.duplicate_ids.add(ident)
+                self.ids[ident] = self.ids.get(ident, 0) + 1
+
+        if tag == "img":
+            if "alt" not in attr_map:
+                src = attr_map.get("src", "")
+                self.missing_alt.append(src)
+
+        if tag == "iframe":
+            if not attr_map.get("title", ""):
+                src = attr_map.get("src", "")
+                self.missing_iframe_title.append(src)
+
+        if tag == "a":
+            if attr_map.get("target", "") == "_blank":
+                rel = attr_map.get("rel", "")
+                if "noopener" not in rel:
+                    href = attr_map.get("href", "")
+                    self.blank_rel.append(href)
+
+
+def main() -> int:
+    issues = []
+
+    for html in ROOT.rglob("*.html"):
+        if any(part in SKIP_DIRS for part in html.parts):
+            continue
+        rel = html.relative_to(ROOT).as_posix()
+        if rel in SKIP_FILES:
+            continue
+
+        parser = AuditParser()
+        parser.feed(html.read_text(encoding="utf-8", errors="ignore"))
+
+        if parser.duplicate_ids:
+            issues.append((rel, "duplicate-ids", sorted(parser.duplicate_ids)))
+        if parser.missing_alt:
+            issues.append((rel, "img-missing-alt", parser.missing_alt))
+        if parser.missing_iframe_title:
+            issues.append((rel, "iframe-missing-title", parser.missing_iframe_title))
+        if parser.blank_rel:
+            issues.append((rel, "target-blank-missing-noopener", parser.blank_rel))
+
+    if not issues:
+        print("OK: no audit issues found")
+        return 0
+
+    print("HTML audit issues:")
+    for rel, kind, items in issues:
+        print(f"- {rel}: {kind}")
+        for item in items[:10]:
+            print(f"  - {item}")
+        if len(items) > 10:
+            print(f"  - ... ({len(items) - 10} more)")
+    return 1
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())