Dev

changelog.md

@@ -20,6 +20,12 @@ While all _Major_ releases should be reviewed, our only _large_ releases are **v
 - **[fix]** - .before() .after()
 -->
 
+#### 14.13.0 [Apr 2024]
+
+- **[new]** - .compute('freeze')
+- **[new]** - .debug('freeze')
+- **[change]** - allow 3-slashes in a word
+
 #### 14.12.0 [Feb 2024]
 
 - **[new]** - .payload() plugin

package.json

@@ -2,7 +2,7 @@
   "author": "Spencer Kelly <spencermountain@gmail.com> (http://spencermounta.in)",
   "name": "compromise",
   "description": "modest natural language processing",
-  "version": "14.12.0",
+  "version": "14.13.0",
   "module": "./src/three.js",
   "main": "./src/three.js",
   "unpkg": "./builds/compromise.js",

scratch.js

@@ -4,10 +4,21 @@ import nlp from './src/three.js'
 // nlp.plugin(plg)
 // nlp.verbose('tagger')
 
-let doc = nlp(`john jacob and john. foobar`)
-let m = doc.split('.')
-let res = m.joinIf('john', '.')
-res.debug()
+var doc = nlp('Poincare\u0301')
+console.log(doc.docs[0])
+// console.log(nlp(`IEEE/WIC`).debug().docs[0][0])
+
+// const prependingText = 'the patient will need an '
+// let doc = nlp('ECG')
+// doc.prepend(prependingText)
+// doc.debug('freeze')
+
+// nlp('i play dr who').debug()
+
+// let doc = nlp(`john jacob and john. foobar`)
+// let m = doc.split('.')
+// let res = m.joinIf('john', '.')
+// res.debug()
 // let doc = nlp('one foo two foo')
 // let m = doc.terms()
 // m = m.join()

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

@@ -3,17 +3,18 @@ import uuid from '../compute/uuid.js'
 // are we inserting inside a contraction?
 // expand it first
 const expand = function (m) {
-  if (m.has('@hasContraction') && typeof m.contractions === 'function') {//&& m.after('^.').has('@hasContraction')
+  if (m.has('@hasContraction') && typeof m.contractions === 'function') {
+    //&& m.after('^.').has('@hasContraction')
     let more = m.grow('@hasContraction')
     more.contractions().expand()
   }
 }
 
-const isArray = (arr) => Object.prototype.toString.call(arr) === '[object Array]'
+const isArray = arr => Object.prototype.toString.call(arr) === '[object Array]'
 
 // set new ids for each terms
 const addIds = function (terms) {
-  terms = terms.map((term) => {
+  terms = terms.map(term => {
     term.id = uuid(term)
     return term
   })
@@ -75,10 +76,11 @@ const insert = function (input, view, prepend) {
   // shift our self pointer, if necessary
   view.ptrs = selfPtrs
   // try to tag them, too
-  doc.compute(['id', 'index', 'lexicon'])
+  doc.compute(['id', 'index', 'freeze', 'lexicon'])
   if (doc.world.compute.preTagger) {
     doc.compute('preTagger')
   }
+  doc.compute('unfreeze')
   return doc
 }
 
@@ -89,7 +91,6 @@ const fns = {
   insertBefore: function (input) {
     return insert(input, this, true)
   },
-
 }
 fns.append = fns.insertAfter
 fns.prepend = fns.insertBefore

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

@@ -2,7 +2,7 @@ const dollarStub = /\$[0-9a-z]+/g
 const fns = {}
 
 const titleCase = function (str) {
-  return str.replace(/\w\S*/g, (txt) => txt.charAt(0).toUpperCase() + txt.substring(1).toLowerCase())
+  return str.replace(/\w\S*/g, txt => txt.charAt(0).toUpperCase() + txt.substring(1).toLowerCase())
 }
 
 // doc.replace('foo', (m)=>{})
@@ -20,7 +20,7 @@ const subDollarSign = function (input, main) {
     return input
   }
   let groups = main.groups()
-  input = input.replace(dollarStub, (a) => {
+  input = input.replace(dollarStub, a => {
     let num = a.replace(/\$/, '')
     if (groups.hasOwnProperty(num)) {
       return groups[num].text()
@@ -65,16 +65,17 @@ fns.replaceWith = function (input, keep = {}) {
     let tmp = main.docs[0]
     let term = tmp[tmp.length - 1]
     if (!term.tags.has('Possessive')) {
-      term.text += '\'s'
-      term.normal += '\'s'
+      term.text += "'s"
+      term.normal += "'s"
       term.tags.add('Possessive')
     }
   }
   // what should we return?
-  let m = main.toView(ptrs).compute(['index', 'lexicon'])
+  let m = main.toView(ptrs).compute(['index', 'freeze', 'lexicon'])
   if (m.world.compute.preTagger) {
     m.compute('preTagger')
   }
+  m.compute('unfreeze')
   // replace any old tags
   if (keep.tags) {
     m.terms().forEach((term, i) => {

src/1-one/freeze/compute.js

@@ -0,0 +1,46 @@
+const freeze = function (view) {
+  const world = view.world
+  const { model, methods } = view.world
+  const setTag = methods.one.setTag
+  const { frozenLex } = model.one
+  const multi = model.one._multiCache || {}
+
+  view.docs.forEach(terms => {
+    for (let i = 0; i < terms.length; i += 1) {
+      // basic lexicon lookup
+      let t = terms[i]
+      let word = t.machine || t.normal
+
+      // test a multi-word
+      if (multi[word] !== undefined && terms[i + 1]) {
+        let end = i + multi[word] - 1
+        for (let k = end; k > i; k -= 1) {
+          let words = terms.slice(i, k + 1)
+          let str = words.map(term => term.machine || term.normal).join(' ')
+          // lookup frozen lexicon
+          if (frozenLex.hasOwnProperty(str) === true) {
+            setTag(words, frozenLex[str], world, false, '1-frozen-multi-lexicon')
+            words.forEach(term => (term.frozen = true))
+            continue
+          }
+        }
+      }
+      // test single word
+      if (frozenLex[word] !== undefined && frozenLex.hasOwnProperty(word)) {
+        setTag([t], frozenLex[word], world, false, '1-freeze-lexicon')
+        t.frozen = true
+        continue
+      }
+    }
+  })
+}
+
+const unfreeze = function (view) {
+  view.docs.forEach(ts => {
+    ts.forEach(term => {
+      delete term.frozen
+    })
+  })
+  return view
+}
+export default { frozen: freeze, freeze, unfreeze }

src/1-one/freeze/debug.js

@@ -0,0 +1,20 @@
+/* eslint-disable no-console */
+const blue = str => '\x1b[34m' + str + '\x1b[0m'
+const dim = str => '\x1b[3m\x1b[2m' + str + '\x1b[0m'
+
+const debug = function (view) {
+  view.docs.forEach(terms => {
+    console.log(blue('\n  ┌─────────'))
+    terms.forEach(t => {
+      let str = `  ${dim('│')}  `
+      let txt = t.implicit || t.text || '-'
+      if (t.frozen === true) {
+        str += `${blue(txt)} ❄️`
+      } else {
+        str += dim(txt)
+      }
+      console.log(str)
+    })
+  })
+}
+export default debug

src/1-one/freeze/plugin.js

@@ -1,13 +1,17 @@
+import compute from './compute.js'
+import debug from './debug.js'
+
 export default {
-  // lib: {
-  //   freeze: function (obj) {
-  //     this.world().model.two.freeze = obj
-  //   },
-  // },
+  // add .compute('freeze')
+  compute,
 
   mutate: world => {
+    const methods = world.methods.one
     // add @isFrozen method
-    world.methods.one.termMethods.isFrozen = term => term.frozen === true
+    methods.termMethods.isFrozen = term => term.frozen === true
+    // adds `.debug('frozen')`
+    methods.debug.freeze = debug
+    methods.debug.frozen = debug
   },
 
   api: function (View) {
@@ -22,16 +26,13 @@ export default {
     }
     // reset all terms to allow  any desctructive tags
     View.prototype.unfreeze = function () {
-      this.docs.forEach(ts => {
-        ts.forEach(term => {
-          delete term.frozen
-        })
-      })
-      return this
+      this.compute('unfreeze')
     }
     // return all frozen terms
     View.prototype.isFrozen = function () {
       return this.match('@isFrozen+')
     }
   },
+  // run it in init
+  hooks: ['freeze'],
 }

src/1-one/lexicon/compute/index.js

@@ -5,6 +5,7 @@ import singleWord from './single-word.js'
 // rest of pre-tagger is in ./two/preTagger
 const lexicon = function (view) {
   const world = view.world
+  // loop through our terms
   view.docs.forEach(terms => {
     for (let i = 0; i < terms.length; i += 1) {
       if (terms[i].tags.size === 0) {

src/1-one/lexicon/compute/multi-word.js

@@ -3,7 +3,7 @@ const multiWord = function (terms, start_i, world) {
   const { model, methods } = world
   const setTag = methods.one.setTag
   const multi = model.one._multiCache || {}
-  const { lexicon, frozenLex } = model.one || {}
+  const { lexicon } = model.one || {}
   let t = terms[start_i]
   let word = t.machine || t.normal
 
@@ -16,12 +16,6 @@ const multiWord = function (terms, start_i, world) {
         return false
       }
       let str = words.map(term => term.machine || term.normal).join(' ')
-      // lookup frozen lexicon
-      if (frozenLex.hasOwnProperty(str) === true) {
-        setTag(words, frozenLex[str], world, false, '1-frozen-multi-lexicon')
-        words.forEach(term => (term.frozen = true))
-        return true
-      }
       // lookup regular lexicon
       if (lexicon.hasOwnProperty(str) === true) {
         let tag = lexicon[str]

src/1-one/lexicon/compute/single-word.js

@@ -7,17 +7,11 @@ const checkLexicon = function (terms, i, world) {
   const { model, methods } = world
   // const fastTag = methods.one.fastTag
   const setTag = methods.one.setTag
-  const { lexicon, frozenLex } = model.one
+  const { lexicon } = model.one
 
   // basic lexicon lookup
   let t = terms[i]
   let word = t.machine || t.normal
-  // freeze lex
-  if (frozenLex[word] !== undefined && frozenLex.hasOwnProperty(word)) {
-    setTag([t], frozenLex[word], world, false, '1-freeze-lexicon')
-    t.frozen = true
-    return true
-  }
   // normal lexicon lookup
   if (lexicon[word] !== undefined && lexicon.hasOwnProperty(word)) {
     setTag([t], lexicon[word], world, false, '1-lexicon')

src/1-one/tokenize/compute/alias.js

@@ -16,7 +16,7 @@ const addAliases = function (term, world) {
   if (hasSlash.test(str) && !hasDomain.test(str) && !isMath.test(str)) {
     let arr = str.split(hasSlash)
     // don't split urls and things
-    if (arr.length <= 2) {
+    if (arr.length <= 3) {
       arr.forEach(word => {
         word = word.trim()
         if (word !== '') {

src/2-two/contraction-two/compute/index.js

@@ -28,7 +28,7 @@ const reTag = function (terms, view, start, len) {
     end += 1
   }
   tmp.ptrs = [[0, start, end]]
-  tmp.compute(['lexicon', 'preTagger'])
+  tmp.compute(['freeze', 'lexicon', 'preTagger', 'unfreeze'])
   // don't for a reindex of the whole document
   reIndex(terms)
 }

src/2-two/postTagger/compute/index.js

@@ -20,6 +20,6 @@ const postTagger = function (view) {
 }
 
 // helper function for compute('tagger')
-const tagger = view => view.compute(['lexicon', 'preTagger', 'postTagger'])
+const tagger = view => view.compute(['freeze', 'lexicon', 'preTagger', 'postTagger', 'unfreeze'])
 
 export default { postTagger, tagger }

src/2-two/preTagger/model/lexicon/frozenLex.js

@@ -28,6 +28,7 @@ export default {
 
   'bill gates': 'Person',
   'doctor who': 'Person',
+  'dr who': 'Person',
   'he man': 'Person',
   'iron man': 'Person',
   'kid cudi': 'Person',

src/3-three/numbers/plugin.js

@@ -10,12 +10,12 @@ export default {
   api,
 
   // add @greaterThan, @lessThan
-  mutate: world => {
-    let termMethods = world.methods.one.termMethods
+  // mutate: world => {
+  //   let termMethods = world.methods.one.termMethods
 
-    termMethods.lessThan = function (term) {
-      console.log(term)
-      // return /[aeiou]/.test(term.text)
-    }
-  },
+  //   termMethods.lessThan = function (term) {
+  //     return false //TODO: implement
+  //     // return /[aeiou]/.test(term.text)
+  //   }
+  // },
 }

src/3-three/verbs/api/conjugate/toInfinitive.js

@@ -24,7 +24,7 @@ const toInf = function (vb, parsed) {
     let does = doDoes(vb, parsed)
     vb.prepend(does)
   }
-  vb.fullSentence().compute(['lexicon', 'preTagger', 'postTagger', 'chunks'])
+  vb.fullSentence().compute(['freeze', 'lexicon', 'preTagger', 'postTagger', 'unfreeze', 'chunks'])
   return vb
 }
 export default toInf

src/API/View.js

@@ -4,13 +4,14 @@ import api from './methods/index.js'
 class View {
   constructor(document, pointer, groups = {}) {
     // invisible props
-    [
+    let props = [
       ['document', document],
       ['world', world],
       ['_groups', groups],
       ['_cache', null],
-      ['viewType', 'View']
-    ].forEach(a => {
+      ['viewType', 'View'],
+    ]
+    props.forEach(a => {
       Object.defineProperty(this, a[0], {
         value: a[1],
         writable: true,
@@ -101,15 +102,16 @@ class View {
     let document = methods.one.tokenize.fromString(input, this.world)
     let doc = new View(document)
     doc.world = this.world
-    doc.compute(['normal', 'lexicon'])
+    doc.compute(['normal', 'freeze', 'lexicon'])
     if (this.world.compute.preTagger) {
       doc.compute('preTagger')
     }
+    doc.compute('unfreeze')
     return doc
   }
   clone() {
     // clone the whole document
-    let document = this.document.slice(0)    //node 17: structuredClone(document);
+    let document = this.document.slice(0) //node 17: structuredClone(document);
     document = document.map(terms => {
       return terms.map(term => {
         term = Object.assign({}, term)

src/_version.js

@@ -1 +1 @@
-export default '14.12.0'
+export default '14.13.0'