big5-decoder.js (2072B)
1 function dec2char(n) { 2 // converts a decimal number to a Unicode character 3 // n: the dec codepoint value to be converted 4 if (n <= 0xffff) { 5 out = String.fromCharCode(n); 6 } else if (n <= 0x10ffff) { 7 n -= 0x10000; 8 out = 9 String.fromCharCode(0xd800 | (n >> 10)) + 10 String.fromCharCode(0xdc00 | (n & 0x3ff)); 11 } else out = "dec2char error: Code point out of range: " + n; 12 return out; 13 } 14 15 function big5Decoder(stream) { 16 stream = stream.replace(/%/g, " "); 17 stream = stream.replace(/[\s]+/g, " ").trim(); 18 var bytes = stream.split(" "); 19 for (var i = 0; i < bytes.length; i++) bytes[i] = parseInt(bytes[i], 16); 20 var out = ""; 21 var lead, byte, offset, ptr, cp; 22 var big5lead = 0x00; 23 var endofstream = 2000000; 24 var finished = false; 25 26 while (!finished) { 27 if (bytes.length == 0) byte = endofstream; 28 else byte = bytes.shift(); 29 30 if (byte == endofstream && big5lead != 0x00) { 31 big5lead = 0x00; 32 out += "�"; 33 continue; 34 } 35 if (byte == endofstream && big5lead == 0x00) { 36 finished = true; 37 continue; 38 } 39 40 if (big5lead != 0x00) { 41 lead = big5lead; 42 ptr = null; 43 big5lead = 0x00; 44 if (byte < 0x7f) offset = 0x40; 45 else offset = 0x62; 46 if ((byte >= 0x40 && byte <= 0x7e) || (byte >= 0xa1 && byte <= 0xfe)) 47 ptr = (lead - 0x81) * 157 + (byte - offset); 48 // "If there is a row in the table below whose first column is pointer, return the two code points listed in its second column" 49 switch (ptr) { 50 case 1133: 51 out += "Ê̄"; 52 continue; 53 case 1135: 54 out += "Ê̌"; 55 continue; 56 case 1164: 57 out += "ê̄"; 58 continue; 59 case 1166: 60 out += "ê̌"; 61 continue; 62 } 63 if (ptr == null) cp = null; 64 else cp = big5[ptr]; 65 if (cp == null && byte >= 0x00 && byte <= 0x7f) { 66 bytes.unshift(byte); 67 } 68 if (cp == null) { 69 out += "�"; 70 continue; 71 } 72 out += dec2char(cp); 73 continue; 74 } 75 if (byte >= 0x00 && byte <= 0x7f) { 76 out += dec2char(byte); 77 continue; 78 } 79 if (byte >= 0x81 && byte <= 0xfe) { 80 big5lead = byte; 81 continue; 82 } 83 out += "�"; 84 } 85 return out; 86 }