mdscrub: decode character references, keep block boundaries, narrow the fence

parity/oracle/unescape.py

@@ -0,0 +1,23 @@
+"""Decode HTML character references the way Python does, one line at a time.
+
+Each line of stdin is one case; each line of stdout is that case decoded and
+JSON-encoded, because a decoded reference can itself be a newline.
+
+    python3 oracle/unescape.py < cases/unescape/entities.txt
+"""
+
+import html
+import json
+import sys
+
+
+def main() -> int:
+    text = sys.stdin.buffer.read().decode("utf-8")
+    out = sys.stdout
+    for line in text.split("\n"):
+        out.write(json.dumps(html.unescape(line), ensure_ascii=False) + "\n")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())