LumeWeb
/
chardet
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

cleaning up the code

This commit is contained in:
Dmitry Shirokov 2013-08-22 09:39:36 +10:00
parent d084074d0c
commit 7bd721b1d7
3 changed files with 13 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
encoding/unicode.js
View File

@ -12,7 +12,7 @@ module.exports.UTF_16BE = function() {
this.match = function(det) { this.match = function(det) {
var input = det.fRawInput; var input = det.fRawInput;
if (input.length >= 2 && ((input[0] & 0xFF) == 0xFE && (input[1] & 0xFF) == 0xFF)) if (input.length >= 2 && ((input[0] & 0xff) == 0xfe && (input[1] & 0xff) == 0xff))
return new Match(det, this, confidence = 100); return new Match(det, this, confidence = 100);
// TODO: Do some statistics to check for unsigned UTF-16BE // TODO: Do some statistics to check for unsigned UTF-16BE
@ -27,7 +27,7 @@ module.exports.UTF_16LE = function() {
this.match = function(det) { this.match = function(det) {
var input = det.fRawInput; var input = det.fRawInput;
if (input.length >= 2 && ((input[0] & 0xFF) == 0xFF && (input[1] & 0xFF) == 0xFE)) { if (input.length >= 2 && ((input[0] & 0xff) == 0xff && (input[1] & 0xff) == 0xfe)) {
// An LE BOM is present. // An LE BOM is present.
if (input.length >= 4 && input[2] == 0x00 && input[3] == 0x00) if (input.length >= 4 && input[2] == 0x00 && input[3] == 0x00)
// It is probably UTF-32 LE, not UTF-16 // It is probably UTF-32 LE, not UTF-16
@ -89,8 +89,8 @@ module.exports.UTF_32BE = function() {
return 'UTF-32BE'; return 'UTF-32BE';
}; };
this.getChar = function(input, index) { this.getChar = function(input, index) {
return (input[index + 0] & 0xFF) << 24 | (input[index + 1] & 0xFF) << 16 | return (input[index + 0] & 0xff) << 24 | (input[index + 1] & 0xff) << 16 |
(input[index + 2] & 0xFF) << 8 | (input[index + 3] & 0xFF); (input[index + 2] & 0xff) << 8 | (input[index + 3] & 0xff);
}; };
}; };
util.inherits(module.exports.UTF_32BE, UTF_32); util.inherits(module.exports.UTF_32BE, UTF_32);
@ -100,8 +100,8 @@ module.exports.UTF_32LE = function() {
return 'UTF-32LE'; return 'UTF-32LE';
}; };
this.getChar = function(input, index) { this.getChar = function(input, index) {
return (input[index + 3] & 0xFF) << 24 | (input[index + 2] & 0xFF) << 16 | return (input[index + 3] & 0xff) << 24 | (input[index + 2] & 0xff) << 16 |
(input[index + 1] & 0xFF) << 8 | (input[index + 0] & 0xFF); (input[index + 1] & 0xff) << 8 | (input[index + 0] & 0xff);
}; };
}; };
util.inherits(module.exports.UTF_32LE, UTF_32); util.inherits(module.exports.UTF_32LE, UTF_32);

13
encoding/utf8.js
View File

@ -18,7 +18,7 @@ module.exports = function() {
confidence; confidence;
if (det.fRawLength >= 3 && if (det.fRawLength >= 3 &&
(input[0] & 0xFF) == 0xef && (input[1] & 0xFF) == 0xbb && (input[2] & 0xFF) == 0xbf) { (input[0] & 0xff) == 0xef && (input[1] & 0xff) == 0xbb && (input[2] & 0xff) == 0xbf) {
hasBOM = true; hasBOM = true;
} }
@ -28,7 +28,6 @@ module.exports = function() {
if ((b & 0x80) == 0) if ((b & 0x80) == 0)
continue; // ASCII continue; // ASCII
// Hi bit on char found. Figure out how long the sequence should be // Hi bit on char found. Figure out how long the sequence should be
if ((b & 0x0e0) == 0x0c0) { if ((b & 0x0e0) == 0x0c0) {
trailBytes = 1; trailBytes = 1;
@ -38,20 +37,18 @@ module.exports = function() {
trailBytes = 3; trailBytes = 3;
} else { } else {
numInvalid++; numInvalid++;
if (numInvalid > 5) { if (numInvalid > 5)
break; break;
}
trailBytes = 0; trailBytes = 0;
} }
// Verify that we've got the right number of trail bytes in the sequence // Verify that we've got the right number of trail bytes in the sequence
for (;;) { for (;;) {
i++; i++;
if (i >= det.fRawLength) { if (i >= det.fRawLength)
break; break;
}
b = input[i]; if ((input[i] & 0xc0) != 0x080) {
if ((b & 0xc0) != 0x080) {
numInvalid++; numInvalid++;
break; break;
} }

3
match.js
View File

@ -2,4 +2,5 @@
module.exports = function(det, rec, confidence, name, lang) { module.exports = function(det, rec, confidence, name, lang) {
this.confidence = confidence; this.confidence = confidence;
this.name = name || rec.name(det); this.name = name || rec.name(det);
}; this.lang = lang;
};