open Bytesrw
(* Encoding detection example *)
let () =
Printf.printf "=== Encoding Detection ===\n\n";
(* Parse UTF-8 bytes with BOM *)
let utf8_bom = Bytes.of_string "\xEF\xBB\xBFUTF-8 with BOM" in
let result = Html5rw.parse_bytes utf8_bom in
(match Html5rw.encoding result with
| Some enc -> Printf.printf "Detected encoding: %s\n" (Html5rw.Encoding.encoding_to_string enc)
| None -> Printf.printf "No encoding detected\n");
Printf.printf "Text: %s\n\n" (Html5rw.to_text result);
(* Parse with meta charset *)
let meta_charset = Bytes.of_string {|
Encoding from meta tag
|} in
let result2 = Html5rw.parse_bytes meta_charset in
(match Html5rw.encoding result2 with
| Some enc -> Printf.printf "Detected encoding: %s\n" (Html5rw.Encoding.encoding_to_string enc)
| None -> Printf.printf "No encoding detected\n");
Printf.printf "Text: %s\n\n" (Html5rw.to_text result2);
(* Using low-level encoding functions *)
Printf.printf "=== Low-level Encoding API ===\n\n";
let bytes = Bytes.of_string "\xEF\xBB\xBFHello" in
(match Html5rw.Encoding.sniff_bom bytes with
| Some (enc, offset) ->
Printf.printf "BOM sniffing result: %s (skip %d bytes)\n"
(Html5rw.Encoding.encoding_to_string enc) offset
| None ->
Printf.printf "No BOM detected\n");
let html_bytes = Bytes.of_string {||} in
(match Html5rw.Encoding.prescan_for_meta_charset html_bytes with
| Some enc -> Printf.printf "Prescan found: %s\n" (Html5rw.Encoding.encoding_to_string enc)
| None -> Printf.printf "No charset in prescan\n")