src/third_party/WebKit/LayoutTests/fast/encoding/api/utf-round-trip.html

   1 <!DOCTYPE html>
   2 <title>Encoding API: UTF encoding round trips</title>
   3 <script src="../../../resources/testharness.js"></script>
   4 <script src="../../../resources/testharnessreport.js"></script>
   5 <script src="resources/shared.js"></script>
   6 <script>
   7
   8 var BATCH_SIZE = 0x1000; // Convert in batches spanning this many code points.
   9 var SKIP_SIZE = 0x77; // For efficiency, don't test every code point.
  10
  11 function fromCodePoint(cp) {
  12     if (0xD800 <= cp && cp <= 0xDFFF) throw new Error('Invalid code point');
  13
  14     if (cp <= 0xFFFF)
  15         return String.fromCharCode(cp);
  16
  17     // outside BMP - encode as surrogate pair
  18     return String.fromCharCode(0xD800 + ((cp >> 10) & 0x3FF), 0xDC00 + (cp & 0x3FF));
  19 }
  20
  21 function makeBatch(cp) {
  22     var string = '';
  23     for (var i = cp; i < cp + BATCH_SIZE && cp < 0x10FFFF; i += SKIP_SIZE) {
  24         if (0xD800 <= i && i <= 0xDFFF) {
  25             // surrogate half
  26             continue;
  27         }
  28         string += fromCodePoint(i);
  29     }
  30     return string;
  31  }
  32
  33 utf_encodings.forEach(function(encoding) {
  34     test(function() {
  35         for (var i = 0; i < 0x10FFFF; i += BATCH_SIZE) {
  36             var string = makeBatch(i);
  37             var encoded = new TextEncoder(encoding).encode(string);
  38             var decoded = new TextDecoder(encoding).decode(encoded);
  39             assert_equals(decoded, string);
  40         }
  41     }, encoding + ' - encode/decode round trip');
  42 });
  43
  44
  45 // Inspired by:
  46 // http://ecmanaut.blogspot.com/2006/07/encoding-decoding-utf8-in-javascript.html
  47 function encode_utf8(string) {
  48     var utf8 = unescape(encodeURIComponent(string));
  49     var octets = [];
  50     for (var i = 0; i < utf8.length; i += 1)
  51         octets.push(utf8.charCodeAt(i));
  52     return octets;
  53 }
  54
  55 function decode_utf8(octets) {
  56     var utf8 = String.fromCharCode.apply(null, octets);
  57     return decodeURIComponent(escape(utf8));
  58 }
  59
  60 test(function() {
  61     for (var i = 0; i < 0x10FFFF; i += BATCH_SIZE) {
  62         var string = makeBatch(i);
  63         var expected = encode_utf8(string);
  64         var actual = new TextEncoder('UTF-8').encode(string);
  65         assert_array_equals(actual, expected);
  66     }
  67 }, 'UTF-8 encoding (compare against unescape/encodeURIComponent)');
  68
  69 test(function() {
  70     for (var i = 0; i < 0x10FFFF; i += BATCH_SIZE) {
  71         var string = makeBatch(i);
  72         var encoded = encode_utf8(string);
  73         var expected = decode_utf8(encoded);
  74         var actual = new TextDecoder('UTF-8').decode(new Uint8Array(encoded));
  75         assert_equals(actual, expected);
  76     }
  77 }, 'UTF-8 decoding (compare against decodeURIComponent/escape)');
  78
  79 </script>