node-nlp
Version:
Library for NLU (Natural Language Understanding) done in Node.js
396 lines (346 loc) • 14 kB
JavaScript
/*
*/
var moment = require('moment');
var Parser = require('../parser').Parser;
var ParsedResult = require('../../result').ParsedResult;
var ParsedComponents = require('../../result').ParsedComponents;
var util = require('../../utils/ZH-Hant.js');
var patternString1 = '(?:由|從|自)?' +
'(?:' +
'(今|明|聽|昨|尋|琴)(早|朝|晚)|' +
'(上(?:午|晝)|朝(?:早)|早(?:上)|下(?:午|晝)|晏(?:晝)|晚(?:上)|夜(?:晚)?|中(?:午)|凌(?:晨))|' +
'(今|明|聽|昨|尋|琴)(?:日|天)' +
'(?:[\\s,，]*)' +
'(?:(上(?:午|晝)|朝(?:早)|早(?:上)|下(?:午|晝)|晏(?:晝)|晚(?:上)|夜(?:晚)?|中(?:午)|凌(?:晨)))?' +
')?' +
'(?:[\\s,，]*)' +
'(?:(\\d+|[' + Object.keys(util.NUMBER).join('') + ']+)(?:\\s*)(?:點|時|:|：)' +
'(?:\\s*)' +
'(\\d+|半|正|整|[' + Object.keys(util.NUMBER).join('') + ']+)?(?:\\s*)(?:分|:|：)?' +
'(?:\\s*)' +
'(\\d+|[' + Object.keys(util.NUMBER).join('') + ']+)?(?:\\s*)(?:秒)?)' +
'(?:\\s*(A\.M\.|P\.M\.|AM?|PM?))?';
var patternString2 = '(?:\\s*(?:到|至|\\-|\\–|\\~|\\〜)\\s*)' +
'(?:' +
'(今|明|聽|昨|尋|琴)(早|朝|晚)|' +
'(上(?:午|晝)|朝(?:早)|早(?:上)|下(?:午|晝)|晏(?:晝)|晚(?:上)|夜(?:晚)?|中(?:午)|凌(?:晨))|' +
'(今|明|聽|昨|尋|琴)(?:日|天)' +
'(?:[\\s,，]*)' +
'(?:(上(?:午|晝)|朝(?:早)|早(?:上)|下(?:午|晝)|晏(?:晝)|晚(?:上)|夜(?:晚)?|中(?:午)|凌(?:晨)))?' +
')?' +
'(?:[\\s,，]*)' +
'(?:(\\d+|[' + Object.keys(util.NUMBER).join('') + ']+)(?:\\s*)(?:點|時|:|：)' +
'(?:\\s*)' +
'(\\d+|半|正|整|[' + Object.keys(util.NUMBER).join('') + ']+)?(?:\\s*)(?:分|:|：)?' +
'(?:\\s*)' +
'(\\d+|[' + Object.keys(util.NUMBER).join('') + ']+)?(?:\\s*)(?:秒)?)' +
'(?:\\s*(A\.M\.|P\.M\.|AM?|PM?))?';
var FIRST_REG_PATTERN = new RegExp(patternString1, 'i');
var SECOND_REG_PATTERN = new RegExp(patternString2, 'i');
var DAY_GROUP_1 = 1;
var ZH_AM_PM_HOUR_GROUP_1 = 2;
var ZH_AM_PM_HOUR_GROUP_2 = 3;
var DAY_GROUP_3 = 4;
var ZH_AM_PM_HOUR_GROUP_3 = 5;
var HOUR_GROUP = 6;
var MINUTE_GROUP = 7;
var SECOND_GROUP = 8;
var AM_PM_HOUR_GROUP = 9;
exports.Parser = function ZHHantTimeExpressionParser() {
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;
result.text = match[0];
result.tags.ZHTimeExpressionParser = true;
var startMoment = refMoment.clone();
// ----- Day
if (match[DAY_GROUP_1]) {
var day1 = match[DAY_GROUP_1];
if (day1 == '明' || day1 == '聽') {
// Check not "Tomorrow" on late night
if(refMoment.hour() > 1) {
startMoment.add(1, 'day');
}
} else if (day1 == '昨' || day1 == '尋' || day1 == '琴') {
startMoment.add(-1, 'day');
}
result.start.assign('day', startMoment.date());
result.start.assign('month', startMoment.month() + 1);
result.start.assign('year', startMoment.year());
} else if (match[DAY_GROUP_3]) {
var day3 = match[DAY_GROUP_3];
if (day3 == '明' || day3 == '聽') {
startMoment.add(1, 'day');
} else if (day3 == '昨' || day3 == '尋' || day3 == '琴') {
startMoment.add(-1, 'day');
}
result.start.assign('day', startMoment.date());
result.start.assign('month', startMoment.month() + 1);
result.start.assign('year', startMoment.year());
} else {
result.start.imply('day', startMoment.date());
result.start.imply('month', startMoment.month() + 1);
result.start.imply('year', startMoment.year());
}
var hour = 0;
var minute = 0;
var meridiem = -1;
// ----- Second
if (match[SECOND_GROUP]) {
var second = parseInt(match[SECOND_GROUP]);
if (isNaN(second)) {
second = util.zhStringToNumber(match[SECOND_GROUP]);
}
if (second >= 60) return null;
result.start.assign('second', second);
}
hour = parseInt(match[HOUR_GROUP]);
if (isNaN(hour)) {
hour = util.zhStringToNumber(match[HOUR_GROUP]);
}
// ----- Minutes
if (match[MINUTE_GROUP]) {
if (match[MINUTE_GROUP] == '半') {
minute = 30;
} else if (match[MINUTE_GROUP] == '正' || match[MINUTE_GROUP] == '整') {
minute = 0;
} else {
minute = parseInt(match[MINUTE_GROUP]);
if (isNaN(minute)) {
minute = util.zhStringToNumber(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]) {
if (hour > 12) return null;
var ampm = match[AM_PM_HOUR_GROUP][0].toLowerCase();
if (ampm == "a") {
meridiem = 0;
if (hour == 12) hour = 0;
}
if (ampm == "p") {
meridiem = 1;
if (hour != 12) hour += 12;
}
} else if (match[ZH_AM_PM_HOUR_GROUP_1]) {
var zhAMPMString1 = match[ZH_AM_PM_HOUR_GROUP_1];
var zhAMPM1 = zhAMPMString1[0];
if (zhAMPM1 == '朝' || zhAMPM1 == '早') {
meridiem = 0;
if (hour == 12) hour = 0;
} else if (zhAMPM1 == '晚') {
meridiem = 1;
if (hour != 12) hour += 12;
}
} else if (match[ZH_AM_PM_HOUR_GROUP_2]) {
var zhAMPMString2 = match[ZH_AM_PM_HOUR_GROUP_2];
var zhAMPM2 = zhAMPMString2[0];
if (zhAMPM2 == '上' || zhAMPM2 == '朝' || zhAMPM2 == '早' || zhAMPM2 == '凌') {
meridiem = 0;
if (hour == 12) hour = 0;
} else if (zhAMPM2 == '下' || zhAMPM2 == '晏' || zhAMPM2 == '晚') {
meridiem = 1;
if (hour != 12) hour += 12;
}
} else if (match[ZH_AM_PM_HOUR_GROUP_3]) {
var zhAMPMString3 = match[ZH_AM_PM_HOUR_GROUP_3];
var zhAMPM3 = zhAMPMString3[0];
if (zhAMPM3 == '上' || zhAMPM3 == '朝' || zhAMPM3 == '早' || zhAMPM3 == '凌') {
meridiem = 0;
if (hour == 12) hour = 0;
} else if (zhAMPM3 == '下' || zhAMPM3 == '晏' || zhAMPM3 == '晚') {
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;
}
var endMoment = startMoment.clone();
result.end = new ParsedComponents(null, null);
// ----- Day
if (match[DAY_GROUP_1]) {
var day1 = match[DAY_GROUP_1];
if (day1 == '明' || day1 == '聽') {
// Check not "Tomorrow" on late night
if(refMoment.hour() > 1) {
endMoment.add(1, 'day');
}
} else if (day1 == '昨' || day1 == '尋' || day1 == '琴') {
endMoment.add(-1, 'day');
}
result.end.assign('day', endMoment.date());
result.end.assign('month', endMoment.month() + 1);
result.end.assign('year', endMoment.year());
} else if (match[DAY_GROUP_3]) {
var day3 = match[DAY_GROUP_3];
if (day3 == '明' || day3 == '聽') {
endMoment.add(1, 'day');
} else if (day3 == '昨' || day3 == '尋' || day3 == '琴') {
endMoment.add(-1, 'day');
}
result.end.assign('day', endMoment.date());
result.end.assign('month', endMoment.month() + 1);
result.end.assign('year', endMoment.year());
} else {
result.end.imply('day', endMoment.date());
result.end.imply('month', endMoment.month() + 1);
result.end.imply('year', endMoment.year());
}
hour = 0;
minute = 0;
meridiem = -1;
// ----- Second
if (match[SECOND_GROUP]) {
var second = parseInt(match[SECOND_GROUP]);
if (isNaN(second)) {
second = util.zhStringToNumber(match[SECOND_GROUP]);
}
if (second >= 60) return null;
result.end.assign('second', second);
}
hour = parseInt(match[HOUR_GROUP]);
if (isNaN(hour)) {
hour = util.zhStringToNumber(match[HOUR_GROUP]);
}
// ----- Minutes
if (match[MINUTE_GROUP]) {
if (match[MINUTE_GROUP] == '半') {
minute = 30;
} else if (match[MINUTE_GROUP] == '正' || match[MINUTE_GROUP] == '整') {
minute = 0;
} else {
minute = parseInt(match[MINUTE_GROUP]);
if (isNaN(minute)) {
minute = util.zhStringToNumber(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]) {
if (hour > 12) return null;
var ampm = match[AM_PM_HOUR_GROUP][0].toLowerCase();
if (ampm == "a") {
meridiem = 0;
if (hour == 12) hour = 0;
}
if (ampm == "p") {
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);
}
}
}
} else if (match[ZH_AM_PM_HOUR_GROUP_1]) {
var zhAMPMString1 = match[ZH_AM_PM_HOUR_GROUP_1];
var zhAMPM1 = zhAMPMString1[0];
if (zhAMPM1 == '朝' || zhAMPM1 == '早') {
meridiem = 0;
if (hour == 12) hour = 0;
} else if (zhAMPM1 == '晚') {
meridiem = 1;
if (hour != 12) hour += 12;
}
} else if (match[ZH_AM_PM_HOUR_GROUP_2]) {
var zhAMPMString2 = match[ZH_AM_PM_HOUR_GROUP_2];
var zhAMPM2 = zhAMPMString2[0];
if (zhAMPM2 == '上' || zhAMPM2 == '朝' || zhAMPM2 == '早' || zhAMPM2 == '凌') {
meridiem = 0;
if (hour == 12) hour = 0;
} else if (zhAMPM2 == '下' || zhAMPM2 == '晏' || zhAMPM2 == '晚') {
meridiem = 1;
if (hour != 12) hour += 12;
}
} else if (match[ZH_AM_PM_HOUR_GROUP_3]) {
var zhAMPMString3 = match[ZH_AM_PM_HOUR_GROUP_3];
var zhAMPM3 = zhAMPMString3[0];
if (zhAMPM3 == '上' || zhAMPM3 == '朝' || zhAMPM3 == '早' || zhAMPM3 == '凌') {
meridiem = 0;
if (hour == 12) hour = 0;
} else if (zhAMPM3 == '下' || zhAMPM3 == '晏' || zhAMPM3 == '晚') {
meridiem = 1;
if (hour != 12) 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;
};
};