Merge pull request #833 from spencermountain/dev

Dev

plugins/dates/changelog.md

@@ -1,7 +1,11 @@
 <!-- #### [Unreleased]
-
+- **[breaking]** - return array in .json().dates
 -->
 
+### 2.2.0 [April 2021]
+
+- **[new]** - start parsing or/and date combos
+
 ### 2.1.0 [April 2021]
 
 - **[change]** - null timezone is now the same as undefined timezone

plugins/dates/package.json

@@ -1,7 +1,7 @@
 {
   "name": "compromise-dates",
   "description": "plugin for nlp-compromise",
-  "version": "2.0.3",
+  "version": "2.2.0",
   "author": "Spencer Kelly <[email protected]> (http://spencermounta.in)",
   "main": "./builds/compromise-dates.js",
   "unpkg": "./builds/compromise-dates.min.js",

plugins/dates/scratch.js

@@ -3,24 +3,31 @@ const spacetime = require('spacetime')
 nlp.extend(require('../../plugins/numbers/src'))
 nlp.extend(require('../../plugins/dates/src'))
 // nlp.verbose(true)
-nlp.verbose('date')
+// nlp.verbose('date')
 
 const fmt = (iso) => (iso ? spacetime(iso).format('{day-short} {nice} {year}') : '-')
 
 const context = {
+  // today: '2021-04-17', //saturday
   // timezone: null, //'Asia/Shanghai',
-  today: '2021-04-01', //friday
+  // today: '2020-01-21',
+  // today: [2016, 10, 23], //wed nov 23rd
   // today: '2018-01-19T10:05:00',
   // dayStart: '8:00am',
   // dayEnd: '8:00pm',
 }
 
 // let doc = nlp('22-23 Feb 2016 at 7pm')
 // let doc = nlp('ten to 4 in the morning')
-// let doc = nlp('february 5th or 7th')
-let doc = nlp('2016-feb-01')
-let dates = doc.dates(context).debug()
+// let doc = nlp('february 5th, 6th or 7th') //.debug()
+let doc = nlp('during march or september').debug()
+// let doc = nlp('march 1')
+// let doc = nlp('next friday, this monday').debug()
+// let doc = nlp('wednesdays at 3') //.debug()
+let dates = doc.dates(context) //.debug()
+// console.log(dates.json(0))
 dates = dates.get()
+// console.log(dates)
 dates.forEach((date) => {
   console.log('start: ', fmt(date.start))
   console.log('  end: ', fmt(date.end))

plugins/dates/src/01-tagger/00-basic.js

@@ -33,6 +33,7 @@ const tagYearSafe = (m, reason) => {
 }
 
 const tagDates = function (doc) {
+  // doc.debug()
   // in the evening
   doc.match('in the (night|evening|morning|afternoon|day|daytime)').tag('Time', 'in-the-night')
   // 8 pm
@@ -45,6 +46,14 @@ const tagDates = function (doc) {
   // misc weekday words
   doc.match('(tue|thu)').tag('WeekDay', 'misc-weekday')
 
+  doc
+    .match('(march|april|may) (and|to|or|through|until)? (march|april|may)')
+    .tag('Date')
+    .match('(march|april|may)')
+    .tag('Month', 'march|april|may')
+  // april should almost-always be a date
+  // doc.match('[april] !#LastName?', 0).tag('Month', 'april')
+
   //months:
   let month = doc.if('#Month')
   if (month.found === true) {
@@ -55,9 +64,15 @@ const tagDates = function (doc) {
     //5 March
     month.match('#Cardinal #Month').tag('Date', 'cardinal-month')
     //march 5 to 7
-    month.match('#Month #Value to #Value').tag('Date', 'value-to-value')
+    month.match('#Month #Value (and|or|to)? #Value+').tag('Date', 'value-to-value')
     //march the 12th
     month.match('#Month the #Value').tag('Date', 'month-the-value')
+    // march to april
+    month.match('(march|may) to? #Date').tag('Date').match('^.').tag('Month', 'march-to')
+    // 'march'
+    month.match('^(march|may)$').tag('Month', 'single-march')
+    //March or June
+    month.match('#Month or #Month').tag('Date', 'month-or-month')
   }
 
   //months:
@@ -187,6 +202,8 @@ const tagDates = function (doc) {
   doc.match('(from|starting|until|by) now').tag('Date', 'for-now')
   // every night
   doc.match('(each|every) night').tag('Date', 'for-now')
+
+  // doc.debug()
   return doc
 }
 module.exports = tagDates

plugins/dates/src/01-tagger/01-values.js

@@ -24,6 +24,8 @@ const values = function (doc) {
     doc.match('#Value and a half (years|months|weeks|days)').tag('Date', here)
     //on the fifth
     doc.match('on the #Ordinal').tag('Date', here)
+    // 'jan 5 or 8'
+    doc.match('#Month #Value+ (and|or) #Value').tag('Date', 'date-or-date')
   }
   return doc
 }

plugins/dates/src/01-tagger/08-fixup.js

@@ -35,34 +35,34 @@ const fixUp = function (doc) {
       // d.match(`(this|last|next) #Date ${knownDate}$`).unTag('Date').lastTerm().tag('Date', 'this month yesterday')
     }
     //tomorrow on 5
-    d.match(`on #Cardinal$`).unTag('Date', here)
+    d.match(`on #Cardinal$`).unTag('Date', 'on 5')
     //this tomorrow
     d.match(`this tomorrow`).terms(0).unTag('Date', 'this-tomorrow')
     //q2 2019
     d.match(`(q1|q2|q3|q4) #Year`).tag('Date', here)
     //5 tuesday
     // d.match(`^#Value #WeekDay`).terms(0).unTag('Date');
     //5 next week
-    d.match(`^#Value (this|next|last)`).terms(0).unTag('Date', here)
+    d.match(`^#Value (this|next|last)`).terms(0).unTag('Date', '4 next')
 
     if (d.has('(last|this|next)')) {
       //this month 7
-      d.match(`(last|this|next) #Duration #Value`).terms(2).unTag('Date', here)
+      d.match(`(last|this|next) #Duration #Value`).terms(2).unTag('Date', 'this month 7')
       //7 this month
-      d.match(`!#Month #Value (last|this|next) #Date`).terms(0).unTag('Date', here)
+      d.match(`!#Month #Value (last|this|next) #Date`).terms(0).unTag('Date', '7 this month')
     }
     //january 5 5
-    if (d.has('(#Year|#Time|#TextValue|#NumberRange)') === false) {
-      d.match('(#Month|#WeekDay) #Value #Value').terms(2).unTag('Date', here)
-    }
+    // if (d.has('(#Year|#Time|#TextValue|#NumberRange)') === false) {
+    //   d.match('(#Month|#WeekDay) #Value #Value !(or|and)?').terms(2).unTag('Date', 'jan 5 5')
+    // }
     //between june
     if (d.has('^between') && !d.has('and .')) {
       d.unTag('Date', here)
     }
     //june june
-    if (d.has('#Month #Month') && !d.has('@hasHyphen') && !d.has('@hasComma')) {
-      d.match('#Month').lastTerm().unTag('Date', 'month-month')
-    }
+    // if (d.has('#Month #Month') && !d.has('@hasHyphen') && !d.has('@hasComma')) {
+    //   d.match('#Month').lastTerm().unTag('Date', 'month-month')
+    // }
     // over the years
     d.match('(in|over) the #Duration #Date+?').unTag('Date', 'over-the-duration')
     // log the hours

plugins/dates/src/02-find/index.js

@@ -0,0 +1,49 @@
+const split = require('./split')
+
+const findDate = function (doc) {
+  if (doc.world.isVerbose() === 'date') {
+    doc.debug()
+    console.log('          ---')
+  }
+  let dates = doc.match('#Date+')
+  // ignore only-durations like '20 minutes'
+  dates = dates.filter((m) => {
+    let isDuration = m.has('^#Duration+$') || m.has('^#Value #Duration+$')
+    // allow 'q4', etc
+    if (isDuration === true && m.has('(#FinancialQuarter|quarter)')) {
+      return true
+    }
+    return isDuration === false
+  })
+  // 30 minutes on tuesday
+  let m = dates.match('[#Cardinal #Duration (in|on|this|next|during|for)] #Date', 0)
+  if (m.found) {
+    dates = dates.not(m)
+  }
+  // 30 minutes tuesday
+  m = dates.match('[#Cardinal #Duration] #WeekDay', 0)
+  if (m.found) {
+    dates = dates.not(m)
+  }
+  // tuesday for 30 mins
+  m = dates.match('#Date [for #Value #Duration]$', 0)
+  if (m.found) {
+    dates = dates.not(m)
+  }
+  // '20 minutes june 5th'
+  m = dates.match('[#Cardinal #Duration] #Date', 0) //but allow '20 minutes ago'
+  if (m.found && !dates.has('#Cardinal #Duration] (ago|from|before|after|back)')) {
+    dates = dates.not(m)
+  }
+  // for 20 minutes
+  m = dates.match('for #Cardinal #Duration')
+  if (m.found) {
+    dates = dates.not(m)
+  }
+  // 'one saturday'
+  dates = dates.notIf('^one (#WeekDay|#Month)$')
+  // tokenize the dates
+  dates = split(dates)
+  return dates
+}
+module.exports = findDate

plugins/dates/src/02-find/split.js

@@ -0,0 +1,74 @@
+// chop things up into bite-size pieces
+const split = function (dates) {
+  let m = null
+  // don't split anything if it looks like a range
+  if (dates.has('^(between|within) #Date')) {
+    return dates
+  }
+
+  if (dates.has('#Month')) {
+    // 'june 5, june 10'
+    m = dates.match('[#Month #Value] and? #Month', 0).ifNo('@hasDash$')
+    if (m.found) {
+      dates = dates.splitAfter(m)
+    }
+
+    // '5 june, 10 june'
+    m = dates.match('[#Value #Month] and? #Value #Month', 0)
+    if (m.found) {
+      dates = dates.splitAfter(m)
+    }
+
+    // 'june, august'
+    m = dates.match('^[#Month] and? #Month #Ordinal?$', 0)
+    if (m.found) {
+      dates = dates.splitAfter(m)
+    }
+
+    // 'june 5th, june 10th'
+    m = dates.match('[#Month #Value] #Month', 0).ifNo('@hasDash$')
+    if (m.found) {
+      dates = dates.splitAfter(m)
+    }
+  }
+
+  if (dates.has('#WeekDay')) {
+    // 'tuesday, wednesday'
+    m = dates.match('^[#WeekDay] and? #WeekDay$', 0).ifNo('@hasDash$')
+    if (m.found) {
+      dates = dates.splitAfter(m)
+    }
+
+    // 'tuesday, wednesday, and friday'
+    m = dates.match('#WeekDay #WeekDay and? #WeekDay')
+    if (m.found) {
+      dates = dates.splitOn('#WeekDay')
+    }
+
+    // monday, wednesday
+    m = dates.match('[#WeekDay] (and|or|this|next)? #WeekDay', 0).ifNo('@hasDash$')
+    if (m.found) {
+      dates = dates.splitAfter('#WeekDay')
+    }
+  }
+
+  // next week tomorrow
+  m = dates.match('(this|next) #Duration [(today|tomorrow|yesterday)]', 0)
+  if (m.found) {
+    dates = dates.splitBefore(m)
+  }
+  // tomorrow 15 march
+  m = dates.match('[(today|tomorrow|yesterday)] #Value #Month', 0)
+  if (m.found) {
+    dates = dates.splitAfter(m)
+  }
+  // tomorrow yesterday
+  m = dates.match('[(today|tomorrow|yesterday)] (today|tomorrow|yesterday|#WeekDay)', 0).ifNo('@hasDash$')
+  if (m.found) {
+    dates = dates.splitAfter(m)
+  }
+  // cleanup any splits
+  dates = dates.not('^and')
+  return dates
+}
+module.exports = split

plugins/dates/src/02-ranges/01-two-times.js → plugins/dates/src/03-ranges/01-two-times.js

@@ -1,4 +1,4 @@
-const parseDate = require('../03-parse')
+const parseDate = require('../04-parse')
 const reverseMaybe = require('./_reverse')
 
 const moveToPM = function (obj) {
@@ -23,7 +23,7 @@ module.exports = [
       let end = parseDate(to, context)
       if (end) {
         let start = end.clone()
-        start.applyTime(from.text())
+        start.applyTime(from.text('reduced'))
         if (start) {
           let obj = {
             start: start,
@@ -51,7 +51,7 @@ module.exports = [
       from = parseDate(from, context)
       if (from) {
         let end = from.clone()
-        end.applyTime(to.text())
+        end.applyTime(to.text('reduced'))
         if (end) {
           let obj = {
             start: from,