node-nlp
Version:
Library for NLU (Natural Language Understanding) done in Node.js
253 lines (201 loc) • 7.36 kB
JavaScript
/*
*/
var moment = require('moment');
var Parser = require('../parser').Parser;
var ParsedResult = require('../../result').ParsedResult;
var ParsedComponents = require('../../result').ParsedComponents;
var FIRST_REG_PATTERN = new RegExp("(^|\\s|T)" +
"(?:(?:um|von)\\s*)?" +
"(\\d{1,4}|mittags?|mitternachts?)" +
"(?:" +
"(?:\\.|\\:|\\：)(\\d{1,2})" +
"(?:" +
"(?:\\:|\\：)(\\d{2})" +
")?" +
")?" +
"(?:\\s*uhr)?" +
"(?:\\s*(morgens|vormittags|mittags|nachmittags|abends|nachts))?" +
"(?=\\W|$)", 'i');
var SECOND_REG_PATTERN = new RegExp("^\\s*" +
"(\\-|\\–|\\~|\\〜|bis|\\?)\\s*" +
"(\\d{1,4})" +
"(?:" +
"(?:\\.|\\:|\\：)(\\d{1,2})" +
"(?:" +
"(?:\\.|\\:|\\：)(\\d{1,2})" +
")?" +
")?" +
"(?:\\s*(morgens|vormittags|mittags|nachmittags|abends|nachts))?" +
"(?=\\W|$)", 'i');
var HOUR_GROUP = 2;
var MINUTE_GROUP = 3;
var SECOND_GROUP = 4;
var AM_PM_HOUR_GROUP = 5;
exports.Parser = function DETimeExpressionParser() {
Parser.apply(this, arguments);
this.pattern = function() { return FIRST_REG_PATTERN; }
this.extract = function(text, ref, match, opt){
// This pattern can be overlaped Ex. [12] AM, 1[2] AM
if (match.index > 0 && text[match.index-1].match(/\w/)) return null;
var refMoment = moment(ref);
var result = new ParsedResult();
result.ref = ref;
result.index = match.index + match[1].length;
result.text = match[0].substring(match[1].length);
result.tags['DETimeExpressionParser'] = true;
result.start.imply('day', refMoment.date());
result.start.imply('month', refMoment.month()+1);
result.start.imply('year', refMoment.year());
var hour = 0;
var minute = 0;
var meridiem = -1;
// ----- Second
if(match[SECOND_GROUP] != null){
var second = parseInt(match[SECOND_GROUP]);
if(second >= 60) return null;
result.start.assign('second', second);
}
// ----- Hours
if (/mittags?/i.test(match[HOUR_GROUP])) {
meridiem = 1;
hour = 12;
} else if (/mitternachts?/i.test(match[HOUR_GROUP])) {
meridiem = 0;
hour = 0;
} else {
hour = parseInt(match[HOUR_GROUP]);
}
// ----- Minutes
if(match[MINUTE_GROUP] != null){
minute = parseInt(match[MINUTE_GROUP]);
} else if(hour > 100) {
minute = hour%100;
hour = parseInt(hour/100);
}
if(minute >= 60) {
return null;
}
if(hour > 24) {
return null;
}
if (hour >= 12) {
meridiem = 1;
}
// ----- AM & PM
if (match[AM_PM_HOUR_GROUP] != null) {
if (hour > 12) return null;
var ampm = match[AM_PM_HOUR_GROUP][0].toLowerCase();
if (ampm === 'morgens' || ampm === 'vormittags') {
meridiem = 0;
if(hour == 12) hour = 0;
} else {
meridiem = 1;
if(hour != 12) hour += 12;
}
}
result.start.assign('hour', hour);
result.start.assign('minute', minute);
if (meridiem >= 0) {
result.start.assign('meridiem', meridiem);
} else {
if (hour < 12) {
result.start.imply('meridiem', 0);
} else {
result.start.imply('meridiem', 1);
}
}
// ==============================================================
// Extracting the 'to' chunk
// ==============================================================
match = SECOND_REG_PATTERN.exec(text.substring(result.index + result.text.length));
if (!match) {
// Not accept number only result
if (result.text.match(/^\d+$/)) {
return null;
}
return result;
}
// Pattern "YY.YY -XXXX" is more like timezone offset
if (match[0].match(/^\s*(\+|\-)\s*\d{3,4}$/)) {
return result;
}
if(result.end == null){
result.end = new ParsedComponents(null, result.start.date());
}
var hour = 0;
var minute = 0;
var meridiem = -1;
// ----- Second
if(match[SECOND_GROUP] != null){
var second = parseInt(match[SECOND_GROUP]);
if(second >= 60) return null;
result.end.assign('second', second);
}
hour = parseInt(match[2]);
// ----- Minute
if (match[MINUTE_GROUP]!= null) {
minute = parseInt(match[MINUTE_GROUP]);
if(minute >= 60) return result;
} else if (hour > 100) {
minute = hour%100;
hour = parseInt(hour/100);
}
if(minute >= 60) {
return null;
}
if(hour > 24) {
return null;
}
if (hour >= 12) {
meridiem = 1;
}
// ----- AM & PM
if (match[AM_PM_HOUR_GROUP] != null) {
if (hour > 12) return null;
var ampm = match[AM_PM_HOUR_GROUP][0].toLowerCase();
if (ampm === 'morgens' || ampm === 'vormittags') {
meridiem = 0;
if(hour == 12) {
hour = 0;
if (!result.end.isCertain('day')) {
result.end.imply('day', result.end.get('day') + 1);
}
}
} else {
meridiem = 1;
if(hour != 12) hour += 12;
}
if (!result.start.isCertain('meridiem')) {
if (meridiem == 0) {
result.start.imply('meridiem', 0);
if (result.start.get('hour') == 12) {
result.start.assign('hour', 0);
}
} else {
result.start.imply('meridiem', 1);
if (result.start.get('hour') != 12) {
result.start.assign('hour', result.start.get('hour') + 12);
}
}
}
}
result.text = result.text + match[0];
result.end.assign('hour', hour);
result.end.assign('minute', minute);
if (meridiem >= 0) {
result.end.assign('meridiem', meridiem);
} else {
var startAtPM = result.start.isCertain('meridiem') && result.start.get('meridiem') == 1;
if (startAtPM && result.start.get('hour') > hour) {
// 10pm - 1 (am)
result.end.imply('meridiem', 0);
} else if (hour > 12) {
result.end.imply('meridiem', 1);
}
}
if (result.end.date().getTime() < result.start.date().getTime()) {
result.end.imply('day', result.end.get('day') + 1)
}
return result;
}
}