Merge pull request #849 from spencermountain/dev

Dev
spencermountain · Jun 21, 2021 · f94dff6 · f94dff6
2 parents 85aa1b0 + 836b174
commit f94dff6
Show file tree

Hide file tree

Showing 21 changed files with 2,658 additions and 2,024 deletions.
diff --git a/builds/compromise-tokenize.js b/builds/compromise-tokenize.js
diff --git a/builds/compromise.js b/builds/compromise.js
diff --git a/builds/compromise.min.js b/builds/compromise.min.js
diff --git a/builds/compromise.mjs b/builds/compromise.mjs
diff --git a/changelog.md b/changelog.md
@@ -12,6 +12,12 @@ While all _Major_ releases should be reviewed, our only two _large_ releases are
 
 -->
 
+#### 13.11.3 [June 2021]
+
+- **[fix]** - regex backtracing issue \#847 (thanks @srubin)
+- misc tagging fixes
+  update deps
+
 #### 13.11.2 [May 2021]
 
 - **[fix]** - verbphrase conjugation fixes

diff --git a/data/nouns/singulars.js b/data/nouns/singulars.js
@@ -72,6 +72,7 @@ module.exports = [
   'us dollar',
   'us state',
   'way',
+  'press release',
 
   //double-consonant rule
   'bottle',

diff --git a/data/nouns/uncountables.js b/data/nouns/uncountables.js
@@ -176,4 +176,5 @@ module.exports = [
   'wool',
   'monogamy',
   'polygamy',
+  'national security',
 ]
diff --git a/data/people/maleNames.js b/data/people/maleNames.js
@@ -862,7 +862,7 @@ module.exports = [
   'robert',
   'roberto',
   'rocco',
-  'rocky',
+  // 'rocky',
   'roderick',
   'rodger',
   'rodney',

diff --git a/data/verbs/infinitives.js b/data/verbs/infinitives.js
@@ -521,6 +521,7 @@ module.exports = [
   'talk',
   'tighten',
   'target',
+  'take part',
   'taste',
   'tend',
   'test',

diff --git a/package-lock.json b/package-lock.json
diff --git a/package.json b/package.json
@@ -2,7 +2,7 @@
   "author": "Spencer Kelly <[email protected]> (http://spencermounta.in)",
   "name": "compromise",
   "description": "modest natural language processing",
-  "version": "13.11.2",
+  "version": "13.11.3",
   "main": "./builds/compromise.js",
   "unpkg": "./builds/compromise.min.js",
   "module": "./builds/compromise.mjs",
@@ -77,22 +77,22 @@
     "efrt-unpack": "2.2.0"
   },
   "devDependencies": {
-    "@babel/core": "7.14.0",
-    "@babel/preset-env": "7.14.1",
+    "@babel/core": "7.14.6",
+    "@babel/preset-env": "7.14.5",
     "@rollup/plugin-alias": "3.1.2",
-    "@rollup/plugin-commonjs": "18.1.0",
+    "@rollup/plugin-commonjs": "19.0.0",
     "@rollup/plugin-json": "4.1.0",
-    "@rollup/plugin-node-resolve": "11.2.1",
+    "@rollup/plugin-node-resolve": "13.0.0",
     "amble": "1.3.0",
     "codecov": "3.8.2",
     "efrt": "2.2.2",
     "nyc": "^15.1.0",
-    "rollup": "2.47.0",
+    "rollup": "2.52.2",
     "rollup-plugin-babel": "4.4.0",
     "rollup-plugin-filesize-check": "0.0.1",
     "rollup-plugin-terser": "7.0.2",
     "shelljs": "0.8.4",
-    "tap-dancer": "0.3.1",
+    "tap-dancer": "0.3.2",
     "tape": "5.2.2"
   },
   "eslintIgnore": [

diff --git a/rollup.config.js b/rollup.config.js
@@ -60,7 +60,7 @@ export default [
         babelrc: false,
         presets: ['@babel/preset-env'],
       }),
-      sizeCheck({ expect: 351, warn: 10 }),
+      sizeCheck({ expect: 391, warn: 10 }),
     ],
   },
   {
@@ -75,7 +75,7 @@ export default [
         presets: ['@babel/preset-env'],
       }),
       terser(),
-      sizeCheck({ expect: 183, warn: 10 }),
+      sizeCheck({ expect: 190, warn: 10 }),
     ],
   },
 ]
diff --git a/scratch.js b/scratch.js
@@ -6,17 +6,56 @@ const nlp = require('./src/index')
 // nlp.extend(require('./plugins/sentences/src'))
 // const text =  require('/Users/spencer/mountain/compromise/scripts/perf/flame/_sotu-text.js')
 // const fmt = iso => (iso ? spacetime(iso).format('{day-short} {nice} {year}') : '-')
-// nlp.verbose(true)
+nlp.verbose(true)
 
 // let doc = nlp('the dogs chased')
 // let doc = nlp('i do not really yell').debug()
 // let doc = nlp('i did not really yell').debug()
 
-let doc = nlp(`http:subdomain.cool.com/`).debug()
-// doc.verbs().toPresentTense()
-doc.verbs().toPastTense()
-console.log(doc.text())
+let arr = [
+  // `drinks and food fuel shopping at new Saks`,
+  // `litigation costs`,
+  // `the dog, whose skip was Frank`,
+  // `on non-Microsoft operating systems,`,
+  // `it’s a myth that uncovered wounds heal faster`,
+  // `a national security issue `,
+  // `I have no excuse, but surprise and fear `,
+  // `in the woods, sketching, boating, fishing`,
+  // `formal thought patterns `,
+  // `every parenting system`,
+  // `with Scotland winning 49 matches `,
+  // `come into the coach and [take] part`,
+  // `There’s a big plum tree growing on it close to the line fence .`,
+  // `become overly weakened`,
+  // `a completely beaten man`,
+  // `the said card`,
+  // `one super strong character`,
+  // `we charged back`,
+  // `for suspected terrorists`,
+  // `for discounted beauty items`,
+  // `the number of suspected terrorists`,
+  // `number of registered party members`,
+  // `rely on bottled water`,
+  // `will have waited until release`,
+  // `selling like hot cakes`,
+  // `have given up on reason`,
 
-// let doc = nlp('I will never do that')
-// doc.verbs().toInfinitive()
-// console.log(doc.text())
+  // `you have some valid points`,
+  // `for some reason`,
+  // `dirty tricks`,
+  // `press release`,
+  // `the same type of shouts`,
+  // `the same kind of shouts`,
+  // `they are essential to expand`,
+  // `had a rocky release`,
+  // `doing better for fights`,
+  // `might get better aim`,
+  `i think tipping blows`,
+]
+// `won’t take extra damage`,
+// `called the taylor rule`,
+// `naive and chubby cheeked`,
+// `he’s devastated`,
+
+let doc = nlp(arr[0]).debug()
+// nlp(`I have to say the value `).debug()
diff --git a/src/01-tokenizer/01-sentences.js b/src/01-tokenizer/01-sentences.js
@@ -51,14 +51,14 @@ const testHasEllipse = function (str, suffix) {
   return hasEllipse.test(str)
 }
 
-const testHasLetter = function (str, suffix, prefixHasLetter) {
+const testHasLetter = function (suffix, prefixHasLetter) {
   return prefixHasLetter || hasLetter.test(suffix)
 }
 
 /** does this look like a sentence? */
 const isSentence = function (str, suffix, abbrevs, prefixContext) {
   // must have a letter
-  prefixContext.hasLetter = testHasLetter(str, suffix, prefixContext.hasLetter)
+  prefixContext.hasLetter = testHasLetter(suffix, prefixContext.hasLetter)
   if (!prefixContext.hasLetter) {
     return false
   }

diff --git a/src/02-tagger/01-init/01-lexicon.js b/src/02-tagger/01-init/01-lexicon.js
@@ -1,4 +1,4 @@
-const underOver = /^(under|over)-?/
+const underOver = /^(under|over)-?.{3}/
 
 /** match a word-sequence, like 'super bowl' in the lexicon */
 const tryMultiple = function (terms, t, world) {
@@ -61,7 +61,7 @@ const checkLexicon = function (terms, world) {
     }
     // prefix strip: try to match 'take' for 'undertake'
     if (underOver.test(str) === true) {
-      let noPrefix = str.replace(underOver, '')
+      let noPrefix = str.replace(/^(under|over)-?/, '')
       if (lex.hasOwnProperty(noPrefix) === true) {
         terms[t].tag(lex[noPrefix], 'noprefix-lexicon', world)
       }

diff --git a/src/02-tagger/04-correction/matches/03-adjective.js b/src/02-tagger/04-correction/matches/03-adjective.js
@@ -50,6 +50,17 @@ let list = [
   { match: '#Copula #Adjective? [(out|in|through)]$', group: 0, tag: 'Adjective', reason: 'still-out' },
   // shut the door
   { match: '^[#Adjective] (the|your) #Noun', group: 0, tag: 'Infinitive', reason: 'shut-the' },
+  // the said card
+  { match: 'the [said] #Noun', group: 0, tag: 'Adjective', reason: 'the-said-card' },
+  // a myth that uncovered wounds heal
+  {
+    match: '#Noun (that|which|whose) [#PastTense && !#Copula] #Noun',
+    group: 0,
+    tag: 'Adjective',
+    reason: 'that-past-noun',
+  },
+  // the very fed character
+  // { match: `#Determiner #Adverb [#PastTense] #Noun`, group: 0, tag: 'Adjective', reason: 'very-x-noun' },
 ]
 
 module.exports = list
diff --git a/src/02-tagger/04-correction/matches/04-noun.js b/src/02-tagger/04-correction/matches/04-noun.js
@@ -161,6 +161,32 @@ module.exports = [
   { match: `to #PresentTense #Noun [#PresentTense] #Preposition`, group: 0, tag: 'Noun', reason: 'gas-exchange' },
   // a comdominium, or simply condo
   { match: `a #Noun+ or #Adverb+? [#Verb]`, group: 0, tag: 'Noun', reason: 'noun-or-noun' },
+  // operating system
+  { match: `[#Gerund] system`, group: 0, tag: 'Noun', reason: 'operating-system' },
+  // waited until release
+  {
+    match: `#PastTense (until|as|through|without) [#PresentTense]`,
+    group: 0,
+    tag: 'Noun',
+    reason: 'waited-until-release',
+  },
+  // selling like hot cakes
+  { match: `#Gerund like #Adjective? [#PresentTense]`, group: 0, tag: 'Plural', reason: 'like-hot-cakes' },
+
+  // some valid reason
+  { match: `some #Adjective [#PresentTense]`, group: 0, tag: 'Noun', reason: 'some-reason' },
+  // for some reason
+  { match: `for some [#PresentTense]`, group: 0, tag: 'Noun', reason: 'for-some-reason' },
+  // same kind of shouts
+  { match: `(same|some|the|that|a) kind of [#PresentTense]`, group: 0, tag: 'Noun', reason: 'some-kind-of' },
+  // a type of shout
+  { match: `(same|some|the|that|a) type of [#PresentTense]`, group: 0, tag: 'Noun', reason: 'some-type-of' },
+  // doing better for fights
+  { match: `#Gerund #Adjective #Preposition [#PresentTense]`, group: 0, tag: 'Noun', reason: 'doing-better-for-x' },
+  // get better aim
+  { match: `(get|got|have|had) #Comparative [#PresentTense]`, group: 0, tag: 'Noun', reason: 'got-better-aim' },
+  // i think tipping sucks
+  { match: `#Pronoun #Infinitive [#Gerund] #PresentTense`, group: 0, tag: 'Noun', reason: 'tipping-sucks' },
 
   // in various sensory functions
   // {

diff --git a/src/02-tagger/04-correction/matches/05-adverb.js b/src/02-tagger/04-correction/matches/05-adverb.js
@@ -40,4 +40,6 @@ module.exports = [
   // kinda sparkly
   { match: `#Adverb [#Adverb]$`, group: 0, tag: 'Adjective', reason: 'kinda-sparkly' },
   { match: `#Adverb [#Adverb] (and|or|then)`, group: 0, tag: 'Adjective', reason: 'kinda-sparkly-and' },
+  // super strong
+  { match: `[super] #Adjective #Noun`, group: 0, tag: 'Adverb', reason: 'super-strong' },
 ]
diff --git a/src/02-tagger/04-correction/matches/07-verbs.js b/src/02-tagger/04-correction/matches/07-verbs.js
@@ -127,7 +127,7 @@ let list = [
   { match: '#Pronoun [#Adjective] #Determiner #Adjective? #Noun', group: 0, tag: 'Verb', reason: 'he-adj-the' },
 
   //is eager to go
-  { match: '#Copula [#Adjective to] #Verb', group: 0, tag: 'Verb', reason: 'adj-to' },
+  { match: '#Copula [#Adjective] to #Verb', group: 0, tag: 'Verb', reason: 'adj-to' },
   // open the door
   { match: '[open] #Determiner', group: 0, tag: 'Infinitive', reason: 'open-the' },
   // compromises are possible
@@ -149,6 +149,14 @@ let list = [
   { match: '[shit] (#Determiner|#Possessive|them)', group: 0, tag: 'Verb', reason: 'swear1-verb' },
   { match: '[damn] (#Determiner|#Possessive|them)', group: 0, tag: 'Verb', reason: 'swear2-verb' },
   { match: '[fuck] (#Determiner|#Possessive|them)', group: 0, tag: 'Verb', reason: 'swear3-verb' },
+  // become overly weakened
+  { match: '(become|fall|grow) #Adverb? [#PastTense]', group: 0, tag: 'Adjective', reason: 'overly-weakened' },
+  // a completely beaten man
+  { match: '(a|an) #Adverb [#Participle] #Noun', group: 0, tag: 'Adjective', reason: 'completely-beaten' },
+  // whose name was
+  { match: 'whose [#PresentTense] #Copula', group: 0, tag: 'Noun', reason: 'whos-name-was' },
+  // give up on reason
+  { match: `#PhrasalVerb #PhrasalVerb #Preposition [#PresentTense]`, group: 0, tag: 'Noun', reason: 'given-up-on-x' }, // wants to be #Particle
 ]
 
 module.exports = list