transliterateBrahmic doesn't recognize multi-byte characters

[chchch]

Hi,

It looks like the transliterateBrahmic function was written before some scripts were added to Unicode, which are recognized by Javascript as multi-byte characters (e.g., grantha). This is probably why you're having trouble with transliterating from "superscripted" Tamil. To be able to transliterate from multi-byte or multi-character "Brahmic" scripts, you'll need to change the transliterateBrahmic function to mirror transliterateRoman. I'm using an older version of Sanscript.js, but it should look something like this:

var transliterateBrahmic = function(data, map, options) {
        var buf = [],
            consonants = map.consonants,
            hadRomanConsonant = false,
            letters = map.letters,
            marks = map.marks,
            dataLength = data.length,
            maxTokenLength = map.maxTokenLength,
            tempLetter,
            tokenBuffer = '',
            toRoman = map.toRoman,
            skippingTrans = false;

        for (var i = 0, L; (L = data.charAt(i)) || tokenBuffer; i++) {
            // Fill the token buffer, if possible.
            var difference = maxTokenLength - tokenBuffer.length;
            if (difference > 0 && i < dataLength) {
                tokenBuffer += L;
                if (difference > 1) {
                    continue;
                }
            }

            // Match all token substrings to our map.
            for (var j = 0; j < maxTokenLength; j++) {
                var token = tokenBuffer.substr(0,maxTokenLength-j);

                if((tempLetter = marks[token]) !== undefined && !skippingTrans) {
                    buf.push(tempLetter);
                    hadRomanConsonant = false;
                    tokenBuffer = tokenBuffer.substr(maxTokenLength-j);
                    break;
                } 
                else if((tempLetter = letters[token])) {
                    if (hadRomanConsonant) {
                        buf.push('a');
                        hadRomanConsonant = false;
                    }
                    buf.push(tempLetter);
                    hadRomanConsonant = toRoman && (token in consonants);
                    tokenBuffer = tokenBuffer.substr(maxTokenLength-j);
                    break;

                } else if (j === maxTokenLength - 1) {
                    if (hadRomanConsonant) {
                        buf.push('a');
                        hadRomanConsonant = false;
                    }
                    buf.push(token);
                    tokenBuffer = tokenBuffer.substr(1);
                }
            }
        }
        if (hadRomanConsonant) {
            buf.push('a');
        }
        return buf.join('');
    };

To make this work with your version of Sanscript.js, you'll need to change buf.push('a') to buf.push(map.toSchemeA). You'll probably also want to add in a check for the "#" character to skip transliteration.

It might also be nice to re-name the "transliterateBrahmic" function to something like "transliterateAbugida", since that's the issue here (it's about the inherent vowel) and "transliterateRoman" to something like "transliterateAlphabet" (no inherent vowel).

[vvasuki]

Why not send a pull request after ensuring that the tests pass?

[chchch]

I'm using an older version of the script that I've customized. I can send a pull request eventually, but it might take a while!