utf7.js

"use strict"

// UTF-7 codec, according to https://tools.ietf.org/html/rfc2152
// See also below a UTF-7-IMAP codec, according to http://tools.ietf.org/html/rfc3501#section-5.1.3

exports.utf7 = Utf7Codec;
exports.unicode11utf7 = 'utf7'; // Alias UNICODE-1-1-UTF-7
function Utf7Codec(codecOptions, iconv) {
    this.iconv = iconv;
};

Utf7Codec.prototype.encoder = Utf7Encoder;
Utf7Codec.prototype.decoder = Utf7Decoder;
Utf7Codec.prototype.bomAware = true;


// -- Encoding

var nonDirectChars = /[^A-Za-z0-9'\(\),-\.\/:\? \n\r\t]+/g;

function Utf7Encoder(options, codec) {
    this.iconv = codec.iconv;
}

Utf7Encoder.prototype.write = function(str) {
    // Naive implementation.
    // Non-direct chars are encoded as "+<base64>-"; single "+" char is encoded as "+-".
    return new Buffer(str.replace(nonDirectChars, function(chunk) {
        return "+" + (chunk === '+' ? '' : 
            this.iconv.encode(chunk, 'utf16-be').toString('base64').replace(/=+$/, '')) 
            + "-";
    }.bind(this)));
}

Utf7Encoder.prototype.end = function() {
}


// -- Decoding

function Utf7Decoder(options, codec) {
    this.iconv = codec.iconv;
    this.inBase64 = false;
    this.base64Accum = '';
}

var base64Regex = /[A-Za-z0-9\/+]/;
var base64Chars = [];
for (var i = 0; i < 256; i++)
    base64Chars[i] = base64Regex.test(String.fromCharCode(i));

var plusChar = '+'.charCodeAt(0), 
    minusChar = '-'.charCodeAt(0),
    andChar = '&'.charCodeAt(0);

Utf7Decoder.prototype.write = function(buf) {
    var res = "", lastI = 0,
        inBase64 = this.inBase64,
        base64Accum = this.base64Accum;

    // The decoder is more involved as we must handle chunks in stream.

    for (var i = 0; i < buf.length; i++) {
        if (!inBase64) { // We're in direct mode.
            // Write direct chars until '+'
            if (buf[i] == plusChar) {
                res += this.iconv.decode(buf.slice(lastI, i), "ascii"); // Write direct chars.
                lastI = i+1;
                inBase64 = true;
            }
        } else { // We decode base64.
            if (!base64Chars[buf[i]]) { // Base64 ended.
                if (i == lastI && buf[i] == minusChar) {// "+-" -> "+"
                    res += "+";
                } else {
                    var b64str = base64Accum + buf.slice(lastI, i).toString();
                    res += this.iconv.decode(new Buffer(b64str, 'base64'), "utf16-be");
                }

                if (buf[i] != minusChar) // Minus is absorbed after base64.
                    i--;

                lastI = i+1;
                inBase64 = false;
                base64Accum = '';
            }
        }
    }

    if (!inBase64) {
        res += this.iconv.decode(buf.slice(lastI), "ascii"); // Write direct chars.
    } else {
        var b64str = base64Accum + buf.slice(lastI).toString();

        var canBeDecoded = b64str.length - (b64str.length % 8); // Minimal chunk: 2 quads -> 2x3 bytes -> 3 chars.
        base64Accum = b64str.slice(canBeDecoded); // The rest will be decoded in future.
        b64str = b64str.slice(0, canBeDecoded);

        res += this.iconv.decode(new Buffer(b64str, 'base64'), "utf16-be");
    }

    this.inBase64 = inBase64;
    this.base64Accum = base64Accum;

    return res;
}

Utf7Decoder.prototype.end = function() {
    var res = "";
    if (this.inBase64 && this.base64Accum.length > 0)
        res = this.iconv.decode(new Buffer(this.base64Accum, 'base64'), "utf16-be");

    this.inBase64 = false;
    this.base64Accum = '';
    return res;
}


// UTF-7-IMAP codec.
// RFC3501 Sec. 5.1.3 Modified UTF-7 (http://tools.ietf.org/html/rfc3501#section-5.1.3)
// Differences:
//  * Base64 part is started by "&" instead of "+"
//  * Direct characters are 0x20-0x7E, except "&" (0x26)
//  * In Base64, "," is used instead of "/"
//  * Base64 must not be used to represent direct characters.
//  * No implicit shift back from Base64 (should always end with '-')
//  * String must end in non-shifted position.
//  * "-&" while in base64 is not allowed.


exports.utf7imap = Utf7IMAPCodec;
function Utf7IMAPCodec(codecOptions, iconv) {
    this.iconv = iconv;
};

Utf7IMAPCodec.prototype.encoder = Utf7IMAPEncoder;
Utf7IMAPCodec.prototype.decoder = Utf7IMAPDecoder;
Utf7IMAPCodec.prototype.bomAware = true;


// -- Encoding

function Utf7IMAPEncoder(options, codec) {
    this.iconv = codec.iconv;
    this.inBase64 = false;
    this.base64Accum = new Buffer(6);
    this.base64AccumIdx = 0;
}

Utf7IMAPEncoder.prototype.write = function(str) {
    var inBase64 = this.inBase64,
        base64Accum = this.base64Accum,
        base64AccumIdx = this.base64AccumIdx,
        buf = new Buffer(str.length*5 + 10), bufIdx = 0;

    for (var i = 0; i < str.length; i++) {
        var uChar = str.charCodeAt(i);
        if (0x20 <= uChar && uChar <= 0x7E) { // Direct character or '&'.
            if (inBase64) {
                if (base64AccumIdx > 0) {
                    bufIdx += buf.write(base64Accum.slice(0, base64AccumIdx).toString('base64').replace(/\//g, ',').replace(/=+$/, ''), bufIdx);
                    base64AccumIdx = 0;
                }

                buf[bufIdx++] = minusChar; // Write '-', then go to direct mode.
                inBase64 = false;
            }

            if (!inBase64) {
                buf[bufIdx++] = uChar; // Write direct character

                if (uChar === andChar)  // Ampersand -> '&-'
                    buf[bufIdx++] = minusChar;
            }

        } else { // Non-direct character
            if (!inBase64) {
                buf[bufIdx++] = andChar; // Write '&', then go to base64 mode.
                inBase64 = true;
            }
            if (inBase64) {
                base64Accum[base64AccumIdx++] = uChar >> 8;
                base64Accum[base64AccumIdx++] = uChar & 0xFF;

                if (base64AccumIdx == base64Accum.length) {
                    bufIdx += buf.write(base64Accum.toString('base64').replace(/\//g, ','), bufIdx);
                    base64AccumIdx = 0;
                }
            }
        }
    }

    this.inBase64 = inBase64;
    this.base64AccumIdx = base64AccumIdx;

    return buf.slice(0, bufIdx);
}

Utf7IMAPEncoder.prototype.end = function() {
    var buf = new Buffer(10), bufIdx = 0;
    if (this.inBase64) {
        if (this.base64AccumIdx > 0) {
            bufIdx += buf.write(this.base64Accum.slice(0, this.base64AccumIdx).toString('base64').replace(/\//g, ',').replace(/=+$/, ''), bufIdx);
            this.base64AccumIdx = 0;
        }

        buf[bufIdx++] = minusChar; // Write '-', then go to direct mode.
        this.inBase64 = false;
    }

    return buf.slice(0, bufIdx);
}


// -- Decoding

function Utf7IMAPDecoder(options, codec) {
    this.iconv = codec.iconv;
    this.inBase64 = false;
    this.base64Accum = '';
}

var base64IMAPChars = base64Chars.slice();
base64IMAPChars[','.charCodeAt(0)] = true;

Utf7IMAPDecoder.prototype.write = function(buf) {
    var res = "", lastI = 0,
        inBase64 = this.inBase64,
        base64Accum = this.base64Accum;

    // The decoder is more involved as we must handle chunks in stream.
    // It is forgiving, closer to standard UTF-7 (for example, '-' is optional at the end).

    for (var i = 0; i < buf.length; i++) {
        if (!inBase64) { // We're in direct mode.
            // Write direct chars until '&'
            if (buf[i] == andChar) {
                res += this.iconv.decode(buf.slice(lastI, i), "ascii"); // Write direct chars.
                lastI = i+1;
                inBase64 = true;
            }
        } else { // We decode base64.
            if (!base64IMAPChars[buf[i]]) { // Base64 ended.
                if (i == lastI && buf[i] == minusChar) { // "&-" -> "&"
                    res += "&";
                } else {
                    var b64str = base64Accum + buf.slice(lastI, i).toString().replace(/,/g, '/');
                    res += this.iconv.decode(new Buffer(b64str, 'base64'), "utf16-be");
                }

                if (buf[i] != minusChar) // Minus may be absorbed after base64.
                    i--;

                lastI = i+1;
                inBase64 = false;
                base64Accum = '';
            }
        }
    }

    if (!inBase64) {
        res += this.iconv.decode(buf.slice(lastI), "ascii"); // Write direct chars.
    } else {
        var b64str = base64Accum + buf.slice(lastI).toString().replace(/,/g, '/');

        var canBeDecoded = b64str.length - (b64str.length % 8); // Minimal chunk: 2 quads -> 2x3 bytes -> 3 chars.
        base64Accum = b64str.slice(canBeDecoded); // The rest will be decoded in future.
        b64str = b64str.slice(0, canBeDecoded);

        res += this.iconv.decode(new Buffer(b64str, 'base64'), "utf16-be");
    }

    this.inBase64 = inBase64;
    this.base64Accum = base64Accum;

    return res;
}

Utf7IMAPDecoder.prototype.end = function() {
    var res = "";
    if (this.inBase64 && this.base64Accum.length > 0)
        res = this.iconv.decode(new Buffer(this.base64Accum, 'base64'), "utf16-be");

    this.inBase64 = false;
    this.base64Accum = '';
    return res;
}