diff options
Diffstat (limited to 'bin/wiki/ImportarDesdeURL/node_modules/chrono-node/src/parsers/en/ENSlashDateFormatParser.js')
-rw-r--r-- | bin/wiki/ImportarDesdeURL/node_modules/chrono-node/src/parsers/en/ENSlashDateFormatParser.js | 133 |
1 files changed, 133 insertions, 0 deletions
diff --git a/bin/wiki/ImportarDesdeURL/node_modules/chrono-node/src/parsers/en/ENSlashDateFormatParser.js b/bin/wiki/ImportarDesdeURL/node_modules/chrono-node/src/parsers/en/ENSlashDateFormatParser.js new file mode 100644 index 00000000..c482ca1b --- /dev/null +++ b/bin/wiki/ImportarDesdeURL/node_modules/chrono-node/src/parsers/en/ENSlashDateFormatParser.js @@ -0,0 +1,133 @@ +/* + Date format with slash "/" (also "-" and ".") between numbers + - Tuesday 11/3/2015 + - 11/3/2015 + - 11/3 + + By default the paser us "middle-endien" format (US English), + then fallback to little-endian if failed. + - 11/3/2015 = November 3rd, 2015 + - 23/4/2015 = April 23th, 2015 + + If "littleEndian" config is set, the parser will try the little-endian first. + - 11/3/2015 = March 11th, 2015 +*/ +var moment = require('moment'); +var Parser = require('../parser').Parser; +var ParsedResult = require('../../result').ParsedResult; + +var PATTERN = new RegExp('(\\W|^)' + + '(?:' + + '(?:on\\s*?)?' + + '((?:sun|mon|tues?|wed(?:nes)?|thu(?:rs?)?|fri|sat(?:ur)?)(?:day)?)' + + '\\s*\\,?\\s*' + + ')?' + + '([0-3]{0,1}[0-9]{1})[\\/\\.\\-]([0-3]{0,1}[0-9]{1})' + + '(?:' + + '[\\/\\.\\-]' + + '([0-9]{4}\s*\,?\s*|[0-9]{2}\s*\,?\s*)' + + ')?' + + '(\\W|$)', 'i'); + +var DAYS_OFFSET = { 'sunday': 0, 'sun': 0, 'monday': 1, 'mon': 1,'tuesday': 2, 'wednesday': 3, 'wed': 3, + 'thursday': 4, 'thur': 4,'friday': 5, 'fri': 5,'saturday': 6, 'sat': 6,} + + +var OPENNING_GROUP = 1; +var ENDING_GROUP = 6; + +var WEEKDAY_GROUP = 2; + + +var FIRST_NUMBERS_GROUP = 3; +var SECOND_NUMBERS_GROUP = 4; + +var YEAR_GROUP = 5; + +exports.Parser = function ENSlashDateFormatParser(config) { + Parser.apply(this, arguments); + config = config || {}; + var littleEndian = config.littleEndian; + var MONTH_GROUP = littleEndian ? SECOND_NUMBERS_GROUP : FIRST_NUMBERS_GROUP; + var DAY_GROUP = littleEndian ? FIRST_NUMBERS_GROUP : SECOND_NUMBERS_GROUP; + + this.pattern = function () { return PATTERN; }; + this.extract = function(text, ref, match, opt){ + + if(match[OPENNING_GROUP] == '/' || match[ENDING_GROUP] == '/') { + // Long skip, if there is some overlapping like: + // XX[/YY/ZZ] + // [XX/YY/]ZZ + match.index += match[0].length + return; + } + + var index = match.index + match[OPENNING_GROUP].length; + var text = match[0].substr(match[OPENNING_GROUP].length, match[0].length - match[ENDING_GROUP].length); + + + var result = new ParsedResult({ + text: text, + index: index, + ref: ref, + }); + + if(text.match(/^\d\.\d$/)) return; + if(text.match(/^\d\.\d{1,2}\.\d{1,2}$/)) return; + + // MM/dd -> OK + // MM.dd -> NG + if(!match[YEAR_GROUP] && match[0].indexOf('/') < 0) return; + + var date = null; + var year = match[YEAR_GROUP] || moment(ref).year() + ''; + var month = match[MONTH_GROUP]; + var day = match[DAY_GROUP]; + + month = parseInt(month); + day = parseInt(day); + year = parseInt(year); + + if(month < 1 || month > 12) { + if(month > 12) { + // dd/mm/yyyy date format if day looks like a month, and month + // looks like a day. + if (day >= 1 && day <= 12 && month >= 13 && month <= 31) { + // unambiguous + var tday = month; + month = day; + day = tday; + } + else { + // both month and day are <= 12 + return null; + } + } + } + if(day < 1 || day > 31) return null; + + if(year < 100){ + if (year > 50) { + year = year + 1900; + } else { + year = year + 2000; + } + } + + result.start.assign('day', day); + result.start.assign('month', month); + if (match[YEAR_GROUP]) { + result.start.assign('year', year); + } else { + result.start.imply('year', year); + } + + //Day of week + if(match[WEEKDAY_GROUP]) { + result.start.assign('weekday', DAYS_OFFSET[match[WEEKDAY_GROUP].toLowerCase()]); + } + + result.tags['ENSlashDateFormatParser'] = true; + return result; + }; +}; |