open Bytesrw
(* Error handling and malformed HTML example *)
let malformed_html = {|
Unclosed title
<meta charset="utf-8">
</head>
<body>
<div>
<p>Unclosed paragraph
<p>Another paragraph (implicitly closes the previous one)
<span><div>Misnested tags</span></div>
</div>
<table>
<tr><td>Cell 1<td>Cell 2</td>
</table>
<!-- Unclosed comment
</body>
</html>
|}
let () =
Printf.printf "=== Parsing Malformed HTML ===\n\n";
(* Parse with error collection enabled *)
let result = Html5rw.parse ~collect_errors:true (Bytes.Reader.of_string malformed_html) in
(* Get parse errors *)
let errs = Html5rw.errors result in
Printf.printf "Parse errors: %d\n\n" (List.length errs);
List.iter (fun err ->
Printf.printf " Line %d, Col %d: %s\n"
(Html5rw.error_line err)
(Html5rw.error_column err)
(Html5rw.Parse_error_code.to_string (Html5rw.error_code err))
) errs;
(* The parser still produces a valid DOM tree *)
Printf.printf "\n=== Recovered DOM Tree ===\n";
let html = Html5rw.to_string ~pretty:true ~indent_size:2 result in
Printf.printf "%s\n" html;
(* Query the recovered tree *)
Printf.printf "\n=== Query Results ===\n";
let paragraphs = Html5rw.query result "p" in
Printf.printf "Found %d paragraphs\n" (List.length paragraphs);
let cells = Html5rw.query result "td" in
Printf.printf "Found %d table cells\n" (List.length cells)