csv/lib/parser.js

CSV parser with simple api, full of options and tested against large datasets.

// Generated by CoffeeScript 1.6.2
var EventEmitter, Parser;

EventEmitter = require('events').EventEmitter;

/*

Parsing
=======

The library extend the [EventEmitter][event] and emit the following events:

*   *row*   
  Emitted by the parser on each line with the line content as an array of fields.
*   *end*   
  Emitted when no more data will be parsed.
*   *error*   
  Emitted when an error occured.
*/


Parser = function(csv) {
  this.csv = csv;
  this.options = csv.options.from;
  this.state = csv.state;
  this.lines = 0;
  this.buf = '';
  this.quoting = false;
  this.field = '';
  this.lastC = '';
  this.nextChar = null;
  this.closingQuote = 0;
  this.line = [];
  return this;
};

Parser.prototype.__proto__ = EventEmitter.prototype;

/*

`write(chars)`
--------------

Parse a string which may hold multiple lines.
Private state object is enriched on each character until 
transform is called on a new line.
*/


Parser.prototype.write = function(chars, end) {
  var areNextCharsRowDelimiters, char, csv, delimLength, escapeIsQuote, i, isDelimiter, isEscape, isQuote, isReallyEscaped, isRowDelimiter, l, ltrim, rtrim, _results;

  csv = this.csv;
  ltrim = this.options.trim || this.options.ltrim;
  rtrim = this.options.trim || this.options.rtrim;
  chars = this.buf + chars;
  l = chars.length;
  delimLength = this.options.rowDelimiter ? this.options.rowDelimiter.length : 0;
  i = 0;
  if (this.lines === 0 && csv.options.from.encoding === 'utf8' && 0xFEFF === chars.charCodeAt(0)) {
    i++;
  }
  while (i < l) {
    if ((i + delimLength >= l && chars.substr(i, this.options.rowDelimiter.length) !== this.options.rowDelimiter) && !end) {
      break;
    }
    if ((i + this.options.escape.length >= l && chars.substr(i, this.options.escape.length) === this.options.escape) && !end) {
      break;
    }
    char = this.nextChar ? this.nextChar : chars.charAt(i);
    this.lastC = char;
    this.nextChar = chars.charAt(i + 1);
    if ((this.options.rowDelimiter == null) && (this.nextChar === '\n' || this.nextChar === '\r')) {
      this.options.rowDelimiter = this.nextChar;
      if (this.nextChar === '\r' && chars.charAt(i + 2) === '\n') {
        this.options.rowDelimiter += '\n';
      }
      delimLength = this.options.rowDelimiter.length;
    }
    isReallyEscaped = false;
    if (char === this.options.escape) {
      escapeIsQuote = this.options.escape === this.options.quote;
      isEscape = this.nextChar === this.options.escape;
      isQuote = this.nextChar === this.options.quote;
      if (!(escapeIsQuote && !this.field && !this.quoting) && (isEscape || isQuote)) {
        i++;
        char = this.nextChar;
        this.nextChar = chars.charAt(i + 1);
        this.field += char;
        i++;
        continue;
      }
    }
    if (!isReallyEscaped && char === this.options.quote) {
      if (this.quoting) {
        areNextCharsRowDelimiters = this.options.rowDelimiter && chars.substr(i + 1, this.options.rowDelimiter.length) === this.options.rowDelimiter;
        if (this.nextChar && !areNextCharsRowDelimiters && this.nextChar !== this.options.delimiter) {
          return this.error(new Error("Invalid closing quote at line " + (this.lines + 1) + "; found " + (JSON.stringify(this.nextChar)) + " instead of delimiter " + (JSON.stringify(this.options.delimiter))));
        }
        this.quoting = false;
        this.closingQuote = i;
        i++;
        continue;
      } else if (!this.field) {
        this.quoting = true;
        i++;
        continue;
      }
    }
    isDelimiter = char === this.options.delimiter;
    isRowDelimiter = this.options.rowDelimiter && chars.substr(i, this.options.rowDelimiter.length) === this.options.rowDelimiter;
    if (!this.quoting && (isDelimiter || isRowDelimiter)) {
      if (rtrim) {
        if (this.closingQuote) {
          this.field = this.field.substr(0, this.closingQuote);
        } else {
          this.field = this.field.trimRight();
        }
      }
      this.line.push(this.field);
      this.closingQuote = 0;
      this.field = '';
      if (isRowDelimiter) {
        this.emit('row', this.line);
        this.line = [];
        i += this.options.rowDelimiter.length;
        this.nextChar = chars.charAt(i);
        continue;
      }
    } else if (!this.quoting && (char === ' ' || char === '\t')) {
      if (!(ltrim && !this.field)) {
        this.field += char;
      }
    } else {
      this.field += char;
    }
    i++;
  }
  this.buf = '';
  _results = [];
  while (i < l) {
    this.buf += chars.charAt(i);
    _results.push(i++);
  }
  return _results;
};

Parser.prototype.end = function() {
  this.write('', true);
  if (this.quoting) {
    return this.error(new Error("Quoted field not terminated at line " + (this.lines + 1)));
  }
  if (this.field || this.lastC === this.options.delimiter || this.lastC === this.options.quote) {
    if (this.options.trim || this.options.rtrim) {
      this.field = this.field.trimRight();
    }
    this.line.push(this.field);
    this.field = '';
  }
  if (this.line.length > 0) {
    this.emit('row', this.line);
  }
  return this.emit('end', null);
};

Parser.prototype.error = function(e) {
  return this.emit('error', e);
};

module.exports = function(csv) {
  return new Parser(csv);
};

module.exports.Parser = Parser;

/*
[event]: http://nodejs.org/api/events.html
*/