Merge pull request #945 from spencermountain/dev

Dev

changelog.md

@@ -8,6 +8,11 @@ While all _Major_ releases should be reviewed, our only _large_ releases are **v
 
 <!-- #### 14.5.0 [Unreleased]
 -->
+#### 14.4.5 [August 2022]
+- **[fix]** - fix logic for greedy-negative matches - #936 
+- **[fix]** - fix tagging for 3-digit year iso dates - #868
+- **[update]** - dependencies
+
 #### 14.4.4 [August 2022]
 - **[fix]** - support {root} matches without compromise/two
 

package.json

@@ -2,7 +2,7 @@
   "author": "Spencer Kelly <[email protected]> (http://spencermounta.in)",
   "name": "compromise",
   "description": "modest natural language processing",
-  "version": "14.4.4",
+  "version": "14.4.5",
   "main": "./src/three.js",
   "unpkg": "./builds/compromise.js",
   "type": "module",

plugins/dates/src/model/regex.js

@@ -10,7 +10,7 @@ export default [
   // @4pm
   [/^@[1-9]+(am|pm)$/, 'Time', '@5pm'],
   // 03/02
-  [/^[0-9]{2}\/[0-9]{2}/, 'Date', '03/02'],
+  [/^[0-9]{2}\/[0-9]{2}$/, 'Date', '03/02'],
   // iso-time
   // [/^[0-9]{4}[:-][0-9]{2}[:-][0-9]{2}T[0-9]/i, 'Time', 'iso-time-tag']
 

scratch.js

@@ -3,7 +3,7 @@ import nlp from './src/three.js'
 import plg from './plugins/dates/src/plugin.js'
 nlp.plugin(plg)
 
-// nlp.verbose('tagger')
+nlp.verbose('tagger')
 
 let txt = ''
 let doc
@@ -13,8 +13,7 @@ let m
 // doc.nouns(0).toSingular()
 // console.log(doc.text())
 
-let str = nlp('I will never do that').verbs().toInfinitive().all().out()
-console.log(str)
+doc = nlp('30/01/194').debug()
 
 // console.log(doc.world)
 // console.log(nlp.world())

src/1-one/change/api/insert.js

@@ -3,7 +3,7 @@ import uuid from '../compute/uuid.js'
 // are we inserting inside a contraction?
 // expand it first
 const expand = function (m) {
-  if (m.has('@hasContraction')) {//&& m.after('^.').has('@hasContraction')
+  if (m.has('@hasContraction') && typeof m.contractions === 'function') {//&& m.after('^.').has('@hasContraction')
     let more = m.grow('@hasContraction')
     more.contractions().expand()
   }

src/1-one/match/methods/match/steps/logic/negative-greedy.js

@@ -0,0 +1,37 @@
+import doesMatch from '../../term/doesMatch.js'
+
+const negGreedy = function (state, reg, nextReg) {
+  let skip = 0
+  for (let t = state.t; t < state.terms.length; t += 1) {
+    let found = doesMatch(state.terms[t], reg, state.start_i + state.t, state.phrase_length)
+    // we don't want a match, here
+    if (found) {
+      break//stop going
+    }
+    // are we doing 'greedy-to'?
+    // - "!foo+ after"  should stop at 'after'
+    if (nextReg) {
+      found = doesMatch(state.terms[t], nextReg, state.start_i + state.t, state.phrase_length)
+      if (found) {
+        break
+      }
+    }
+    skip += 1
+    // is it max-length now?
+    if (reg.max !== undefined && skip === reg.max) {
+      break
+    }
+  }
+  if (skip === 0) {
+    return false //dead
+  }
+  // did we satisfy min for !foo{min,max}
+  if (reg.min && reg.min > skip) {
+    return false//dead
+  }
+  state.t += skip
+  // state.r += 1
+  return true
+}
+
+export default negGreedy

src/1-one/match/methods/match/steps/negative.js

@@ -1,25 +1,24 @@
 import matchTerm from '../term/doesMatch.js'
+import negGreedy from './logic/negative-greedy.js'
 
 // '!foo' should match anything that isn't 'foo'
 // if it matches, return false
 const doNegative = function (state) {
   const { regs } = state
   let reg = regs[state.r]
 
-
-
   // match *anything* but this term
   let tmpReg = Object.assign({}, reg)
   tmpReg.negative = false // try removing it
+
+  // found it? if so, we die here
   let found = matchTerm(state.terms[state.t], tmpReg, state.start_i + state.t, state.phrase_length)
   if (found) {
-    return false//die
+    return false//bye
   }
-
   // should we skip the term too?
-  // "before after"
-  //  match("before !foo? after")
   if (reg.optional) {
+    // "before after" - "before !foo? after"
     // does the next reg match the this term?
     let nextReg = regs[state.r + 1]
     if (nextReg) {
@@ -37,7 +36,10 @@ const doNegative = function (state) {
       }
     }
   }
-
+  // negative greedy - !foo+  - super hard!
+  if (reg.greedy) {
+    return negGreedy(state, tmpReg, regs[state.r + 1])
+  }
   state.t += 1
   return true
 }

src/2-two/preTagger/model/regex/regex-numbers.js

@@ -11,7 +11,7 @@ export default [
   // iso-dates
   [/^[0-9]{4}-[0-9]{2}-[0-9]{2}T[0-9]{2}:[0-9]{2}/i, 'Date', 'iso-date'],
   [/^[0-9]{1,4}-[0-9]{1,2}-[0-9]{1,4}$/, 'Date', 'iso-dash'],
-  [/^[0-9]{1,4}\/[0-9]{1,2}\/[0-9]{1,4}$/, 'Date', 'iso-slash'],
+  [/^[0-9]{1,4}\/[0-9]{1,2}\/([0-9]{4}|[0-9]{2})$/, 'Date', 'iso-slash'],
   [/^[0-9]{1,4}\.[0-9]{1,2}\.[0-9]{1,4}$/, 'Date', 'iso-dot'],
   [/^[0-9]{1,4}-[a-z]{2,9}-[0-9]{1,4}$/i, 'Date', '12-dec-2019'],
 

src/_version.js

@@ -1 +1 @@
-export default '14.4.4'
+export default '14.4.5'

tests/one/match/negative.test.js

@@ -52,7 +52,6 @@ test('! negative match syntax :', function (t) {
   // doc = nlp("I will send her.")
   // t.equal(doc.has(m), true, here + 'multi-optional')
 
-
   t.end()
 })
 
@@ -77,3 +76,75 @@ test('negative optional logic', function (t) {
 
   t.end()
 })
+
+test('negative greedy-max', function (t) {
+  const check = (doc) => doc.has('before !(not|no|maybe|perhaps){0,3} after')
+
+  let doc = nlp('before after')
+  t.equal(check(doc), true, 'no middle')
+
+  doc = nlp('before one after')
+  t.equal(check(doc), true, 'one middle')
+
+  doc = nlp('before one two after')
+  t.equal(check(doc), true, 'two middle')
+
+  doc = nlp('before one two three after')
+  t.equal(check(doc), true, 'three middle')
+
+  doc = nlp('before one two three four after')
+  t.equal(check(doc), false, 'four middle')
+
+  doc = nlp('before one two three maybe')
+  let m = doc.match('before !maybe{1}')
+  t.equal(m.text(), 'before one', 'greedy-max-one')
+
+  doc = nlp('before one two three maybe')
+  m = doc.match('before !maybe{1,2}')
+  t.equal(m.text(), 'before one two', 'greedy-max-two')
+
+  doc = nlp('before one two three maybe')
+  m = doc.match('before !maybe{4,5}')
+  t.equal(m.text(), '', 'greedy-unmet-min')
+
+  t.end()
+})
+
+test('negative greedy-end', function (t) {
+  // simpler version
+  let doc = nlp('before one after')
+  t.equal(doc.has('before !maybe+ after'), true, 'greedy-not-one')
+
+  doc = nlp('before one two after')
+  t.equal(doc.has('before !maybe+ after'), true, 'greedy-not-two')
+
+  // greedyEnd
+  doc = nlp('before one two')
+  let m = doc.match('before !maybe+')
+  t.equal(m.text(), 'before one two', 'greedy-end')
+
+  doc = nlp('before maybe')
+  m = doc.match('before !maybe+')
+  t.equal(m.text(), '', 'greedy-zero-fail')
+
+  doc = nlp('before one maybe')
+  m = doc.match('before !maybe+')
+  t.equal(m.text(), 'before one', 'greedy-one-stop')
+
+  doc = nlp('before one two maybe')
+  m = doc.match('before !maybe+')
+  t.equal(m.text(), 'before one two', 'greedy-two-stop')
+  t.end()
+})
+
+
+test('negative greedy-to', function (t) {
+  let doc = nlp('before one after end')
+  let m = doc.match('before !maybe+ after')
+  t.equal(m.text(), 'before one after', 'greedy-to-pos')
+
+  doc = nlp('before one after end')
+  m = doc.match('before !maybe+ none')
+  t.equal(m.text(), '', 'greedy-no-after')
+  t.end()
+})

tests/two/match.test.js

@@ -659,6 +659,7 @@ let arr = [
   ["I was an expert", 'i was an #Noun'],
   ["definitely worth a rental.", '#Adverb #Verb a #Noun'],
   ["keeping the matter a secret", '#Gerund the #Noun a #Noun'],
+  ["30/01/194", '!#Date'],
 ]
 test('match:', function (t) {
   let res = []