node-nlp
Version:
Library for NLU (Natural Language Understanding) done in Node.js
96 lines (74 loc) • 2.86 kB
JavaScript
/*
The parser for parsing month name and year.
EX.
- January
- January 2012
- January, 2012
*/
var moment = require('moment');
var Parser = require('../parser').Parser;
var ParsedResult = require('../../result').ParsedResult;
var util = require('../../utils/EN');
var PATTERN = new RegExp('(^|\\D\\s+|[^\\w\\s])' +
'(Jan\\.?|January|Feb\\.?|February|Mar\\.?|March|Apr\\.?|April|May\\.?|Jun\\.?|June|Jul\\.?|July|Aug\\.?|August|Sep\\.?|Sept\\.?|September|Oct\\.?|October|Nov\\.?|November|Dec\\.?|December)' +
'\\s*' +
'(?:' +
'[,-]?\\s*([0-9]{4})(\\s*BE|AD|BC)?' +
')?' +
'(?=[^\\s\\w]|\\s+[^0-9]|\\s+$|$)', 'i');
var MONTH_NAME_GROUP = 2;
var YEAR_GROUP = 3;
var YEAR_BE_GROUP = 4;
exports.Parser = function ENMonthNameParser(){
Parser.apply(this, arguments);
this.pattern = function() { return PATTERN; }
this.extract = function(text, ref, match, opt){
var result = new ParsedResult({
text: match[0].substr(match[1].length, match[0].length - match[1].length),
index: match.index + match[1].length,
ref: ref,
});
var month = match[MONTH_NAME_GROUP];
month = util.MONTH_OFFSET[month.toLowerCase()];
var day = 1;
var year = null;
if (match[YEAR_GROUP]) {
year = match[YEAR_GROUP];
year = parseInt(year);
if(match[YEAR_BE_GROUP]){
if (match[YEAR_BE_GROUP].match(/BE/)) {
// Buddhist Era
year = year - 543;
} else if (match[YEAR_BE_GROUP].match(/BC/)) {
// Before Christ
year = -year;
}
} else if (year < 100){
year = year + 2000;
}
}
if(year){
result.start.imply('day', day);
result.start.assign('month', month);
result.start.assign('year', year);
} else {
//Find the most appropriated year
var refMoment = moment(ref);
refMoment.month(month - 1);
refMoment.date(day);
var nextYear = refMoment.clone().add(1, 'y');
var lastYear = refMoment.clone().add(-1, 'y');
if( Math.abs(nextYear.diff(moment(ref))) < Math.abs(refMoment.diff(moment(ref))) ){
refMoment = nextYear;
}
else if( Math.abs(lastYear.diff(moment(ref))) < Math.abs(refMoment.diff(moment(ref))) ){
refMoment = lastYear;
}
result.start.imply('day', day);
result.start.assign('month', month);
result.start.imply('year', refMoment.year());
}
result.tags['ENMonthNameParser'] = true;
return result;
}
}