CSV parser

CSVParser.js

/**
 * CSV Parser.  Takes a string as input and returns
 * an array of arrays (for each row).
 * 
 * @param input String, CSV input
 * @param options.delimiter String, single character used to separate fields.
 *        Defaults to null, if null, tries to guess it.
 * @param doptions.quote String, single character used to quote non-simple fields.
 *        Defaults to "\"".
 */

class CSVParser {
  constructor(data, options = {delimiter: null, quote: '"'}) {
    this.delimiter = options.delimiter || this._guessDelimiter(data);
    this.quote = options.quote;

    if(!data) {
      this.data = [['']];
      return;
    }

    this.data = this._parse(data.split(''));
  }
  _addToStream(stream, c) {
     if(c === this.quote && stream[stream.length - 1] === c) { return; }
     stream.push(c)
  }
  _addStream(row, stream) {
    row.push(stream.join(''));
    stream.length = 0;
  }
  _addRow(rows, row, stream) {
    this._addStream(row, stream);
    rows.push(row); 
    return [];
  }
  _extractContent(data, stream, lastColumn) {
    var c;
    while((c = data.shift()) !== undefined) {
      if(c === '\r' && data[0] === '\n') { continue; }
      if(c === this.quote && (data[0] === undefined || (lastColumn === false && data[0] === this.delimiter) || (lastColumn === true && data[0] === '\n') || (lastColumn === undefined && (data[0] === this.delimiter || data[0] === '\n')))) {
        return;
      }
      this._addToStream(stream, c);
    }
  }
  _parse(data) {
    var c,
        row = [],
        rows = [],
        stream = [];
    while((c = data.shift()) !== undefined) {
      if(c === '\r' && data[0] === '\n') { continue; }
      if(c === this.delimiter) {
        this._addStream(row, stream);
      } else if(c === '\n') {
        row = this._addRow(rows, row, stream);
      } else if(c === this.quote) {
        this._extractContent(data, stream, rows[0] ? rows[0].length === row.length : undefined);
      } else {
        this._addToStream(stream, c);
      }
    }
    this._addRow(rows, row, stream);
    return rows;
  }
  _guessDelimiter(data) {
    //guess delimiters using string occurences.
    var map = {},
        count = [],
        possibleDelimiters = [',', ';', '^', '$', '.', '|', '\t'];
    data.split(/\n|\r/).every(function(line, index, array) {
      possibleDelimiters.forEach(function(possibleDelimiter) {
        map[possibleDelimiter] = (map[possibleDelimiter] || 0) + line.split(possibleDelimiter).length;
        count[map[possibleDelimiter]] = possibleDelimiter;
      });

      //only analyse 10% of the lines. It should be enough
      if(index >= array.length * 0.1) {
        return false;
      }
      return true;
    });
    
    return count.pop();
  }
}