iso2022jp-encode-form-csiso2022jp.html (4307B)
1 <!DOCTYPE html> 2 <html> 3 <head> 4 <meta charset="csiso2022jp"> <!-- test breaks if the server overrides this --> 5 <title>csiso2022jp encoding (form)</title> 6 <meta name="timeout" content="long"> 7 <meta name="variant" content="?1-1000"> 8 <meta name="variant" content="?1001-2000"> 9 <meta name="variant" content="?2001-3000"> 10 <meta name="variant" content="?3001-4000"> 11 <meta name="variant" content="?4001-5000"> 12 <meta name="variant" content="?5001-6000"> 13 <meta name="variant" content="?6001-7000"> 14 <meta name="variant" content="?7001-last"> 15 <script src="/resources/testharness.js"></script> 16 <script src="/resources/testharnessreport.js"></script> 17 <script src="/common/subset-tests.js"></script> 18 <script src="jis0208_index.js"></script> 19 <script src="iso2022jp-encoder.js"></script> 20 <link rel="author" title="Richard Ishida" href="mailto:ishida@w3.org"> 21 <link rel="help" href="https://encoding.spec.whatwg.org/#names-and-labels"> 22 <meta name="assert" content="The browser produces the same encoding behavior for a document labeled 'csiso2022jp' as for a document labeled 'iso-2022-jp'."> 23 <style> 24 iframe { display:none } 25 form { display:none } 26 </style> 27 </head> 28 <body> 29 <div id="log"></div> 30 <script src="../../resources/ranges.js"></script> 31 <script> 32 var errors = false; 33 var separator = "\u0019"; 34 var ranges = rangesAll; 35 36 function expect(result, codepoint) { 37 return result; 38 } 39 40 function encoder(str) { 41 return getByteSequence(str.codePointAt(0)); 42 } 43 44 function getByteSequence(cp) { 45 // uses the Encoding spec algorithm to derive a sequence of bytes for a character that can be encoded 46 // the result is either a percent-encoded value or null (if the character can't be encoded) 47 // cp: integer, a code point number 48 49 var cps = [cp]; 50 var endofstream = 2000000; 51 cps.push(endofstream); 52 var out = ""; 53 var encState = "ascii"; 54 var finished = false; 55 while (!finished) { 56 cp = cps.shift(); 57 if (cp == endofstream && encState != "ascii") { 58 cps.unshift(cp); 59 encState = "ascii"; 60 out += "%1B%28%42"; 61 continue; 62 } 63 if (cp == endofstream && encState == "ascii") { 64 finished = true; 65 break; 66 } 67 if (encState == "ascii" && cp >= 0x00 && cp <= 0x7f) { 68 out += "%" + cp.toString(16).toUpperCase(); 69 continue; 70 } 71 if ( 72 encState == "roman" && 73 ((cp >= 0x00 && cp <= 0x7f && cp !== 0x5c && cp !== 0x7e) || 74 cp == 0xa5 || 75 cp == 0x203e) 76 ) { 77 if (cp >= 0x00 && cp <= 0x7f) { 78 // ASCII 79 out += "%" + cp.toString(16).toUpperCase(); 80 continue; 81 } 82 if (cp == 0xa5) { 83 out += "%5C"; 84 continue; 85 } 86 if (cp == 0x203e) { 87 out += "%7E"; 88 continue; 89 } 90 } 91 if (encState != "ascii" && cp >= 0x00 && cp <= 0x7f) { 92 cps.unshift(cp); 93 encState = "ascii"; 94 out += "%1B%28%42"; 95 continue; 96 } 97 if ((cp == 0xa5 || cp == 0x203e) && encState != "roman") { 98 cps.unshift(cp); 99 encState = "roman"; 100 out += "%1B%28%4A"; 101 continue; 102 } 103 if (cp == 0x2212) cp = 0xff0d; 104 ptr = indexcodepoints[cp]; 105 if (ptr == null) { 106 return null; 107 } 108 if (encState != "jis0208") { 109 cps.unshift(cp); 110 encState = "jis0208"; 111 out += "%1B%24%42"; 112 continue; 113 } 114 var lead = Math.floor(ptr / 94) + 0x21; 115 var trail = ptr % 94 + 0x21; 116 out += 117 "%" + 118 lead.toString(16).toUpperCase() + 119 "%" + 120 trail.toString(16).toUpperCase(); 121 } 122 return out.trim(); 123 } 124 125 // set up a sparse array of all unicode codepoints listed in the index 126 // this will be used for lookup in getByteSequence 127 var indexcodepoints = []; // index is unicode cp, value is pointer 128 for (p = 0; p < jis0208.length; p++) { 129 if (jis0208[p] != null && indexcodepoints[jis0208[p]] == null) { 130 indexcodepoints[jis0208[p]] = p; 131 } 132 } 133 </script> 134 <script src="../../resources/encode-form-common.js"></script> 135 </body> 136 </html>