chardet/index.js

88 lines
2.1 KiB
JavaScript

var fs = require('fs');
var utf8 = require('./encoding/utf8'),
unicode = require('./encoding/unicode'),
mbcs = require('./encoding/mbcs'),
sbcs = require('./encoding/sbcs'),
iso2022 = require('./encoding/iso2022');
var self = this;
var recognisers = [
new utf8,
new unicode.UTF_16BE,
new unicode.UTF_16LE,
new unicode.UTF_32BE,
new unicode.UTF_32LE,
new mbcs.sjis,
new mbcs.big5,
new mbcs.euc_jp,
new mbcs.euc_kr,
new mbcs.gb_18030,
new iso2022.ISO_2022_JP,
new iso2022.ISO_2022_KR,
new iso2022.ISO_2022_CN,
new sbcs.ISO_8859_1,
new sbcs.ISO_8859_2,
new sbcs.ISO_8859_5,
new sbcs.ISO_8859_6,
new sbcs.ISO_8859_7,
new sbcs.ISO_8859_8,
new sbcs.ISO_8859_9,
new sbcs.windows_1251,
new sbcs.windows_1256,
new sbcs.KOI8_R
];
module.exports.detect = function(buffer) {
// Tally up the byte occurence statistics.
var fByteStats = [];
for (var i = 0; i < 256; i++)
fByteStats[i] = 0;
for (var i = buffer.length - 1; i >= 0; i--)
fByteStats[buffer[i] & 0x00ff]++;
var fC1Bytes = false;
for (var i = 0x80; i <= 0x9F; i += 1) {
if (fByteStats[i] != 0) {
fC1Bytes = true;
break;
}
}
var context = {
fByteStats: fByteStats,
fC1Bytes: fC1Bytes,
fRawInput: buffer,
fRawLength: buffer.length,
fInputBytes: buffer,
fInputLen: buffer.length
};
var matches = recognisers.map(function(rec) {
return rec.match(context);
}).filter(function(match) {
return !!match;
});
matches.sort(function(a, b) {
return a.confidence - b.confidence;
});
return matches.length ? matches.pop().name : null;
};
module.exports.detectFile = function(filepath, fn) {
fs.readFile(filepath, function(err, res) {
if (err)
return fn(err, null);
fn(null, self.detect(res));
});
};
module.exports.detectFileSync = function(filepath) {
return self.detect(fs.readFileSync(filepath));
};