Merge pull request #722 from spencermountain/dev

Dev

.github/workflows/build-and-test.yml

@@ -3,9 +3,12 @@ name: Build and test
 on: [pull_request]
 
 jobs:
-  build:
+  build-and-test:
     runs-on: ubuntu-latest
 
+    env:
+      CI: true
+
     strategy:
       matrix:
         node-version: [8.x, 10.x, 13.x]
@@ -27,18 +30,25 @@ jobs:
             ${{ runner.os }}-npm-${{ matrix.node-version }}-
             ${{ runner.os }}-npm-
 
-      - name: npm install, build, and test
+      - name: install
         run: |
           npm ci
           npm i --no-save eslint ts-node typescript
           npm run plugins:ci
+
+      - name: static checks
+        run: |
           npm run lint
+
+      - name: build
+        run: |
           npm run build
           npm run plugins:build
+
+      - name: test
+        run: |
           npm run test:smoke
           npm run testb
           npm run test:perf
           npm run test:types
           # npm run test:stress
-        env:
-          CI: true

.github/workflows/release.yml

@@ -0,0 +1,66 @@
+name: Release
+
+on:
+  release:
+    types: [created]
+
+jobs:
+  release:
+    runs-on: ubuntu-latest
+
+    env:
+      CI: true
+
+    # Note that these steps are *identical* to build-and-test (with the caveat
+    # that build-and-test uses several versions of Node, and Release only uses
+    # 10.x) at least until the actual publishing happens.  Ideally, we could
+    # delegate to the build- and-test workflow, but I haven't found a way to do
+    # that yet.
+    steps:
+      - uses: actions/checkout@v2
+
+      - uses: actions/setup-node@v1
+        with:
+          node-version: 10.x
+
+      - name: cache dependencies
+        uses: actions/cache@v1
+        with:
+          path: ~/.npm
+          key: ${{ runner.os }}-npm-10.x-${{ hashFiles('package-lock.json') }}
+          restore-keys: |
+            ${{ runner.os }}-npm-10.x-
+            ${{ runner.os }}-npm-
+
+      - name: install
+        run: |
+          npm ci
+          npm i --no-save eslint ts-node typescript
+          npm run plugins:ci
+
+      - name: static checks
+        run: |
+          npm run lint
+
+      - name: build
+        run: |
+          npm run build
+          npm run plugins:build
+
+      - name: test
+        run: |
+          npm run test:smoke
+          npm run testb
+          npm run test:perf
+          npm run test:types
+          # npm run test:stress
+
+      # And finally... publish it!  Note that we create the .npmrc file
+      # "just in time" so that `npm publish` can get the auth token from the
+      # environment
+      - name: publish
+        run: |
+          echo '//registry.npmjs.org/:_authToken=${NPM_TOKEN}' > .npmrc
+          npm publish --dry-run
+        env:
+          NPM_TOKEN: ${{ secrets.NPM_TOKEN }}

README.md

@@ -716,9 +716,10 @@ doc.world.hello = "Hello again!"
 
 - **[Geocoding Social Conversations with NLP and JavaScript](http://compromise.cool)** - by Microsoft
 - **[Microservice Recipe](https://eventn.com/recipes/text-parsing-with-nlp-compromise)** - by Eventn
+- **[Adventure Game Sentence Parsing with Compromise](https://killalldefects.com/2020/02/20/adventure-game-sentence-parsing-with-compromise/)
 - **[Building Text-Based Games](https://killalldefects.com/2019/09/24/building-text-based-games-with-compromise-nlp/)** - by Matt Eland
 - **[Fun with javascript in BigQuery](https://medium.com/@hoffa/new-in-bigquery-persistent-udfs-c9ea4100fd83#6e09)** - by Felipe Hoffa
-- - **[Natural Language Processing... in the Browser???](https://dev.to/charlesdlandau/natural-language-processing-in-the-browser-52hj)** - by Charles Landau
+- **[Natural Language Processing... in the Browser???](https://dev.to/charlesdlandau/natural-language-processing-in-the-browser-52hj)** - by Charles Landau
 
 ##### Talks:
 

changelog.md

@@ -8,6 +8,13 @@ compromise uses semver, and pushes to npm frequently
 
 While all _Major_ releases should be reviewed, our only two _large_ releases are **v6** in 2016 and and **v12** in 2019. Others have been mostly incremental, or niche.
 
+#### 13.1.1
+
+- fix `verbs.json()` runtime-error
+- improve empty `.lists()` methods
+- allow custom tag colors
+- test new github action workflow
+
 #### 13.1.0
 
 - significant (~30%) speed up of parsing

plugins/scan/builds/compromise-scan.mjs

@@ -1,4 +1,4 @@
-/* compromise-scan 0.0.1 MIT */
+/* compromise-scan 0.1.0 MIT */
 // edited by Spencer Kelly
 // credit to https://github.com/BrunoRB/ahocorasick by Bruno Roberto Búrigo.
 // object v. array

plugins/scan/package.json

@@ -1,7 +1,7 @@
 {
   "name": "compromise-scan",
   "description": "plugin for nlp-compromise",
-  "version": "0.0.1",
+  "version": "0.1.0",
   "author": "Spencer Kelly <[email protected]> (http://spencermounta.in)",
   "main": "./builds/compromise-scan.js",
   "unpkg": "./builds/compromise-scan.min.js",

scratch.js

@@ -4,5 +4,14 @@ let txt = require('./scripts/test/speed/_sotu-text.js')
 // nlp.extend(require('./plugins/numbers/src'))
 // nlp.extend(require('./plugins/dates/src'))
 
-let doc = nlp(txt)
-// console.log(doc.dates().json())
+// let doc = nlp('i am a modern major general').replace('modern', 'old')
+// let doc = nlp('i am a modern major general')
+// doc.match('modern').tag('Person')
+// // console.log(doc._cache)
+// console.log(doc.has('#Person'))
+
+let doc = nlp('i saw red, blue, and green.')
+// doc.lists().addOxfordComma()
+doc.lists().removeOxfordComma()
+console.log(doc.text())
+// console.log(arr)

src/02-tagger/index.js

@@ -16,6 +16,8 @@ const tagger = function(doc) {
   doc.cache()
   // wiggle-around the results, so they make more sense
   doc = corrections(doc)
+  // remove our cache, as it's invalidated now
+  doc.uncache()
   // run any user-given tagger functions
   doc.world.taggers.forEach(fn => {
     fn(doc)

src/Doc/match/checkCache.js

@@ -0,0 +1,37 @@
+// cache the easier conditions up-front
+const cacheRequired = function(reg) {
+  let needTags = []
+  let needWords = []
+  reg.forEach(obj => {
+    if (obj.optional === true || obj.negative === true) {
+      return
+    }
+    if (obj.tag !== undefined) {
+      needTags.push(obj.tag)
+    }
+    if (obj.word !== undefined) {
+      needWords.push(obj.word)
+    }
+  })
+  return { tags: needTags, words: needWords }
+}
+
+const failFast = function(doc, regs) {
+  if (doc._cache && doc._cache.set === true) {
+    let { words, tags } = cacheRequired(regs)
+    //check required words
+    for (let i = 0; i < words.length; i++) {
+      if (doc._cache.words[words[i]] === undefined) {
+        return false
+      }
+    }
+    //check required tags
+    for (let i = 0; i < tags.length; i++) {
+      if (doc._cache.tags[tags[i]] === undefined) {
+        return false
+      }
+    }
+  }
+  return true
+}
+module.exports = failFast

src/Doc/methods/03-match.js

@@ -1,4 +1,5 @@
 const parseSyntax = require('../match/syntax')
+const checkCache = require('../match/checkCache')
 
 /** return a new Doc, with this one as a parent */
 exports.match = function(reg, name) {
@@ -7,6 +8,10 @@ exports.match = function(reg, name) {
   if (regs.length === 0) {
     return this.buildFrom([])
   }
+  //check our cache, if it exists
+  if (checkCache(this, regs) === false) {
+    return this.buildFrom([])
+  }
   //try expression on each phrase
   let matches = this.list.reduce((arr, p) => {
     return arr.concat(p.match(regs))
@@ -23,7 +28,7 @@ exports.not = function(reg) {
   //parse-up the input expression
   let regs = parseSyntax(reg)
   //if it's empty, return them all!
-  if (regs.length === 0) {
+  if (regs.length === 0 || checkCache(this, regs) === false) {
     return this
   }
   //try expression on each phrase
@@ -36,6 +41,10 @@ exports.not = function(reg) {
 /** return only the first match */
 exports.matchOne = function(reg) {
   let regs = parseSyntax(reg)
+  //check our cache, if it exists
+  if (checkCache(this, regs) === false) {
+    return this.buildFrom([])
+  }
   for (let i = 0; i < this.list.length; i++) {
     let match = this.list[i].match(regs, true)
     return this.buildFrom(match)
@@ -46,6 +55,10 @@ exports.matchOne = function(reg) {
 /** return each current phrase, only if it contains this match */
 exports.if = function(reg) {
   let regs = parseSyntax(reg)
+  //consult our cache, if it exists
+  if (checkCache(this, regs) === false) {
+    return this.buildFrom([])
+  }
   let found = this.list.filter(p => p.has(regs) === true)
   return this.buildFrom(found)
 }
@@ -60,6 +73,10 @@ exports.ifNo = function(reg) {
 /**Return a boolean if this match exists */
 exports.has = function(reg) {
   let regs = parseSyntax(reg)
+  //consult our cache, if it exists
+  if (checkCache(this, regs) === false) {
+    return false
+  }
   return this.list.some(p => p.has(regs) === true)
 }
 

src/Doc/methods/07-cache.js

@@ -8,6 +8,7 @@ exports.cache = function(options) {
   this._cache.set = true
   this.list.forEach((p, i) => {
     p.cache = p.cache || {}
+    //p.terms get cached automatically
     let terms = p.terms()
     // cache all the terms
     terms.forEach(t => {
@@ -21,26 +22,12 @@ exports.cache = function(options) {
         tags[tag] = tags[tag] || []
         tags[tag].push(i)
       })
-      // Object.assign(this._cache.tags, t.tags)
 
-      //   words[t.clean] = true
-      //   words[t.text.toLowerCase()] = true
-      //   if (t.implicit) {
-      //     words[t.implicit] = true
-      //   }
-      //   if (t.root) {
-      //     words[t.root] = true
-      //   }
-      //   if (t.alias !== undefined) {
-      //     words = Object.assign(words, t.alias)
-      //   }
+      // cache root-form on Term, too
       if (options.root) {
         t.setRoot(this.world)
         words[t.root] = true
       }
-      // })
-      // delete words['']
-      // p.cache.words = words
     })
   })
   return this

src/Doc/methods/output/_debug.js

@@ -1,5 +1,3 @@
-const tagset = require('../../../World/tags')
-
 // https://stackoverflow.com/questions/9781218/how-to-change-node-jss-console-font-color
 const reset = '\x1b[0m'
 
@@ -27,6 +25,7 @@ const css = {
 }
 
 const logClientSide = function(doc) {
+  let tagset = doc.world.tags
   doc.list.forEach(p => {
     console.log('\n%c"' + p.text() + '"', 'color: #e6d7b3;')
     let terms = p.terms()
@@ -74,12 +73,12 @@ const cli = {
   },
 }
 
-const tagString = function(tags) {
+const tagString = function(tags, world) {
   tags = tags.map(tag => {
-    if (!tagset.hasOwnProperty(tag)) {
+    if (!world.tags.hasOwnProperty(tag)) {
       return tag
     }
-    const c = tagset[tag].color || 'blue'
+    const c = world.tags[tag].color || 'blue'
     return cli[c](tag)
   })
   return tags.join(', ')
@@ -106,7 +105,7 @@ const debug = function(doc) {
       }
       let word = "'" + text + "'"
       word = padEnd(word, 18)
-      let str = cli.blue('  ｜ ') + word + '  - ' + tagString(tags)
+      let str = cli.blue('  ｜ ') + word + '  - ' + tagString(tags, doc.world)
       console.log(str)
     })
   })