oxjs/source/Ox/js/Encoding.js

'use strict';

(function() {

    function cap(width, height) {
        // returns maximum encoding capacity of an image
        return parseInt(width * height * 3/8) - 4;
    }

    function seek(data, px) {
        // returns this, or the next, opaque pixel
        while (data[px * 4 + 3] < 255) {
            if (++px * 4 == data.length) {
                throwPNGError('de');
            }
        }
        return px;
    }

    function xor(byte) {
        // returns "1"-bits-in-byte % 2
        // use: num.toString(2).replace(/0/g, '').length % 2
        var xor = 0;
        Ox.range(8).forEach(function(i) {
            xor ^= byte >> i & 1;
        });
        return xor;
    }

    function throwPNGError(str) {
        throw new RangeError(
            'PNG codec can\'t ' +
            (str == 'en' ? 'encode data' : 'decode image')
        );
    }

    function throwUTF8Error(byte, pos) {
        throw new RangeError(
            'UTF-8 codec can\'t decode byte 0x' +
            byte.toString(16).toUpperCase() + ' at position ' + pos
        );
    }

    /*@
    Ox.encodeBase26 <b> Encode a number as bijective base26
        See <a href="http://en.wikipedia.org/wiki/Bijective_numeration">
        Bijective numeration</a>.
        > Ox.encodeBase26(4461)
        'FOO'
    @*/
    Ox.encodeBase26 = function(num) {
        var ret = '';
        while (num) {
            ret = String.fromCharCode(64 + num % 26) + ret;
            num = parseInt(num / 26);
        }
        return ret;
    };

    /*@
    Ox.decodeBase26 <f> Decodes a bijective base26-encoded number
        See <a href="http://en.wikipedia.org/wiki/Bijective_numeration">
        Bijective numeration</a>.
        > Ox.decodeBase26('foo')
        4461
    @*/
    Ox.decodeBase26 = function(str) {
        return str.toUpperCase().split('').reverse().reduce(function(p, v, i) {
            return p + (v.charCodeAt(0) - 64) * Math.pow(26, i);
        }, 0);
    };

    /*@
    Ox.encodeBase32 <b> Encode a number as base32
        See <a href="http://www.crockford.com/wrmg/base32.html">Base 32</a>.
        > Ox.encodeBase32(15360)
        'F00'
        > Ox.encodeBase32(33819)
        '110V'
    @*/
    Ox.encodeBase32 = function(num) {
        return Ox.map(num.toString(32), function(char) {
            return Ox.BASE_32_DIGITS[parseInt(char, 32)];
        });
    };

    /*@
    Ox.decodeBase32 <f> Decodes a base32-encoded number
        See <a href="http://www.crockford.com/wrmg/base32.html">Base 32</a>.
        > Ox.decodeBase32('foo')
        15360
        > Ox.decodeBase32('ILOU')
        33819
        > Ox.decodeBase32('?').toString()
        'NaN'
    @*/
    Ox.decodeBase32 = function(str) {
        return parseInt(Ox.map(str.toUpperCase(), function(char) {
            var index = Ox.BASE_32_DIGITS.indexOf(
                Ox.BASE_32_ALIASES[char] || char
            );
            return (index == -1 ? ' ' : index).toString(32);
        }), 32);
    };

    /*@
    Ox.encodeBase64 <f> Encode a number as base64
        > Ox.encodeBase64(32394)
        'foo'
    @*/
    Ox.encodeBase64 = function(num) {
        return btoa(Ox.encodeBase256(num)).replace(/=/g, '');
    };

    /*@
    Ox.decodeBase64 <f> Decodes a base64-encoded number
        > Ox.decodeBase64('foo')
        32394
    @*/
    Ox.decodeBase64 = function(str) {
        return Ox.decodeBase256(atob(str));
    };

    /*@
    Ox.encodeBase128 <f> Encode a number as base128
        > Ox.encodeBase128(1685487)
        'foo'
    @*/
    Ox.encodeBase128 = function(num) {
        var str = '';
        while (num) {
            str = Ox.char(num & 127) + str;
            num >>= 7;
        }
        return str;
    };

    /*@
    Ox.decodeBase128 <f> Decode a base128-encoded number
        > Ox.decodeBase128('foo')
        1685487
    @*/
    Ox.decodeBase128 = function(str) {
        return str.split('').reverse().reduce(function(p, v, i) {
            return p + (v.charCodeAt(0) << i * 7);
        }, 0);
    };

    /*@
    Ox.encodeBase256 <f> Encode a number as base256
        > Ox.encodeBase256(6713199)
        'foo'
    @*/
    Ox.encodeBase256 = function(num) {
        var str = '';
        while (num) {
            str = Ox.char(num & 255) + str;
            num >>= 8;
        }
        return str;
    };

    /*@
    Ox.decodeBase256 <f> Decode a base256-encoded number
        > Ox.decodeBase256('foo')
        6713199
    @*/
    Ox.decodeBase256 = function(str) {
        return str.split('').reverse().reduce(function(p, v, i) {
            return p + (v.charCodeAt(0) << i * 8);
        }, 0);
    };

    /*@
    Ox.encodeDeflate <f> Encodes a string, using deflate
        Since PNGs are deflate-encoded, the <code>canvas</code> object's
        <code>toDataURL</code> method provides an efficient implementation.
        The string is encoded as UTF-8 and written to the RGB channels of a
        canvas element, then the PNG dataURL is decoded from base64, and some
        head, tail and chunk names are removed.
        (str) -> <s> The encoded string
            str <s> The string to be encoded
        # Test with: Ox.decodeDeflate(Ox.encodeDeflate('foo'), alert)
    @*/
    
    Ox.encodeDeflate = function(str, callback) {
        // Make sure we can encode the full unicode range of characters.
        str = Ox.encodeUTF8(str);
        // We can only safely write to RGB, so we need 1 pixel for 3 bytes.
        // The string length may not be a multiple of 3, so we need to encode
        // the number of padding bytes (1 byte), the string, and non-0-bytes
        // as padding, so that the combined length becomes a multiple of 3.
        var len = 1 + str.length, c = Ox.canvas(Math.ceil(len / 3), 1),
            data, idat, pad = (3 - len % 3) % 3;
        str = Ox.char(pad) + str + Ox.repeat('\u00FF', pad);
        Ox.loop(c.data.length, function(i) {
            // Write character codes into RGB, and 255 into ALPHA
            c.data[i] = i % 4 < 3 ? str.charCodeAt(i - parseInt(i / 4)) : 255;
        });
        c.context.putImageData(c.imageData, 0, 0);
        // Get the PNG data from the data URL and decode it from base64.
        str = atob(c.canvas.toDataURL().split(',')[1]);
        // Discard bytes 0 to 15 (8 bytes PNG signature, 4 bytes IHDR length, 4
        // bytes IHDR name), keep bytes 16 to 19 (width), discard bytes 20 to 29
        // (4 bytes height, 5 bytes flags), keep bytes 29 to 32 (IHDR checksum),
        // keep the rest (IDAT chunks), discard the last 12 bytes (IEND chunk).
        data = str.substr(16, 4) + str.substr(29, 4);
        idat = str.substr(33, str.length - 45);
        while (idat) {
            // Each IDAT chunk is 4 bytes length, 4 bytes name, length bytes
            // data and 4 bytes checksum. We can discard the name parts.
            len = idat.substr(0, 4);
            data += len + idat.substr(8, 4 + (len = Ox.decodeBase256(len)));
            idat = idat.substr(12 + len);
        }
        callback && callback(data);
        return data;
    };

    /*@
    Ox.decodeDeflate <f> Decodes an deflate-encoded string
        Since PNGs are deflate-encoded, the <code>canvas</code> object's
        <code>drawImage</code> method provides an efficient implementation. The
        string will be wrapped as a PNG dataURL, encoded as base64, and drawn
        onto a canvas element, then the RGB channels will be read, and the
        result will be decoded from UTF8.
        (str) -> <u> undefined
            str <s> The string to be decoded
            callback <f> Callback function
                str <s> The decoded string
    @*/

    Ox.decodeDeflate = function(str, callback) {
        var image = new Image(),
            // PNG file signature and IHDR chunk
            data = '\u0089PNG\r\n\u001A\n\u0000\u0000\u0000\u000DIHDR'
                + str.substr(0, 4) + '\u0000\u0000\u0000\u0001'
                + '\u0008\u0006\u0000\u0000\u0000' + str.substr(4, 4),
            // IDAT chunks
            idat = str.substr(8), len;
        function error() {
            throw new RangeError('Deflate codec can\'t decode data.');
        }
        while (idat) {
            // Reinsert the IDAT chunk names
            len = idat.substr(0, 4);
            data += len + 'IDAT' + idat.substr(4, 4 + (len = Ox.decodeBase256(len)));
            idat = idat.substr(8 + len);
        }
        // IEND chunk
        data += '\u0000\u0000\u0000\u0000IEND\u00AE\u0042\u0060\u0082';
        // Unfortunately, we can't synchronously set the source of an image,
        // draw it onto a canvas, and read its data.
        image.onload = function() {
            str = Ox.toArray(Ox.canvas(image).data).map(function(v, i) {
                // Read one character per RGB byte, ignore ALPHA.
                return i % 4 < 3 ? Ox.char(v) : '';
            }).join('');
            try {
                // Parse the first byte as number of bytes to chop at the end,
                // and the rest, without these bytes, as an UTF8-encoded string.
                str = Ox.decodeUTF8(str.substr(1, str.length - 1 - str.charCodeAt(0)))
            } catch (e) {
                error();
            }
            callback(str);
        }
        image.onerror = error;
        image.src = 'data:image/png;base64,' + btoa(data);
    };

    /*@
    Ox.encodeHTML <f> HTML-encodes a string
        > Ox.encodeHTML('\'<"&">\'')
        '&apos;&lt;&quot;&amp;&quot;&gt;&apos;'
        > Ox.encodeHTML('äbçdê')
        '&#x00E4;b&#x00E7;d&#x00EA;'
    @*/
    Ox.encodeHTML = function(str) {
        return Ox.map(str.toString(), function(v) {
            var code = v.charCodeAt(0);
            return code < 128
                ? (v in Ox.HTML_ENTITIES ? Ox.HTML_ENTITIES[v] : v)
                : '&#x' + Ox.pad(code.toString(16).toUpperCase(), 4) + ';';
        });
    };

    /*@
    Ox.decodeHTML <f> Decodes an HTML-encoded string
        > Ox.decodeHTML('&apos;&lt;&quot;&amp;&quot;&gt;&apos;')
        '\'<"&">\''
        > Ox.decodeHTML('&#x0027;&#x003C;&#x0022;&#x0026;&#x0022;&#x003E;&#x0027;')
        '\'<"&">\''
        > Ox.decodeHTML('&auml;b&ccedil;d&ecirc;')
        'äbçdê'
        > Ox.decodeHTML('&#x00E4;b&#x00E7;d&#x00EA;')
        'äbçdê'
        > Ox.decodeHTML('<b>bold</b>')
        '<b>bold</b>'
    @*/
    Ox.decodeHTML = function(str) {
        // relies on dom, but shorter than using this:
        // http://www.w3.org/TR/html5/named-character-references.html 
        return Ox.decodeHTMLEntities(Ox.element('<div>').html(str).html());
    };

    Ox.encodeHTMLEntities = function(str) {
        return str.replace(
            new RegExp('(' + Object.keys(Ox.HTML_ENTITIES).join('|') + ')', 'g'),
            function(match) {
                return Ox.HTML_ENTITIES[match];
            }
        );
    };

    Ox.decodeHTMLEntities = function(str) {
        return str.replace(
            new RegExp('(' + Ox.values(Ox.HTML_ENTITIES).join('|') + ')', 'g'),
            function(match) {
                return Ox.keyOf(Ox.HTML_ENTITIES, match);
            }
        );
    };

    /*@
    Ox.encodeUTF8 <f> Encodes a string as UTF-8
        see http://en.wikipedia.org/wiki/UTF-8
        (string) -> <s> UTF-8 encoded string
        string <s> Any string
        > Ox.encodeUTF8("YES")
        "YES"
        > Ox.encodeUTF8("¥€$")
        "\u00C2\u00A5\u00E2\u0082\u00AC\u0024"
    @*/
    Ox.encodeUTF8 = function(str) {
        return Ox.map(str, function(chr) {
            var code = chr.charCodeAt(0),
                str = '';
            if (code < 128) {
                str = chr;
            } else if (code < 2048) {
                str = String.fromCharCode(code >> 6 | 192)
                    + String.fromCharCode(code & 63 | 128);
            } else {
                str = String.fromCharCode(code >> 12 | 224)
                    + String.fromCharCode(code >> 6 & 63 | 128)
                    + String.fromCharCode(code & 63 | 128);
            }
            return str;
        });
    };

    /*@
    Ox.decodeUTF8 <f> Decodes an UTF-8-encoded string
        see http://en.wikipedia.org/wiki/UTF-8
        (utf8) -> <s> string
        utf8 <s> Any UTF-8-encoded string
        > Ox.decodeUTF8('YES')
        'YES'
        > Ox.decodeUTF8('\u00C2\u00A5\u00E2\u0082\u00AC\u0024')
        '¥€$'
    @*/
    Ox.decodeUTF8 = function(str) {
        var code,
            i = 0,
            len = str.length,
            ret = '';
        while (i < len) {
            code = Ox.range(3).map(function(o) {
                return str.charCodeAt(i + o);
            });
            if (code[0] <= 128) {
                ret += str[i];
                i++;
            } else if (
                code[0] >= 192 && code[0] < 240
                && i < len - (code[0] < 224 ? 1 : 2)
            ) {
                if (code[1] >= 128 && code[1] < 192) {
                    if (code[0] < 224) {
                        ret += String.fromCharCode(
                            (code[0] & 31) << 6 | code[1] & 63
                        );
                        i += 2;
                    } else if (code[2] >= 128 && code[2] < 192) {
                        ret += String.fromCharCode(
                            (code[0] & 15) << 12 | (code[1] & 63) << 6
                            | code[2] & 63
                        );
                        i += 3;
                    } else {
                        throwUTF8Error(code[2], i + 2);
                    }
                } else {
                    throwUTF8Error(code[1], i + 1);
                }
            } else {
                throwUTF8Error(code[0], i);
            }
        }
        return ret;
    };

})();