compromise-speech 0.1.0 → 0.2.0

package/src/syllables/split/find-cuts/_compound-prefixes.js

@@ -0,0 +1,22 @@
+// Match a compound's left component, including its silent e.
+// Both vowel counting and boundary placement use these same spelling rules.
+const compoundPrefixes = [
+  /^note(?=book)/, // note-book: silent e and a boundary after the complete base
+  /^side(?=kick|line|show|step|stream|track|walk|water|way|winder|wise)/,
+  /^some(?=bod|one|time|where)/,
+  /^none(?=such)/, // none-such retains the silent e of none
+  /^goose(?=berr)/, // goose-berry and goose-berries
+  /^above(?=board|ground)/,
+  /^a?fore(?=clos|taste|said|mentioned)/,
+  /^ware(?=house|housing)/,
+  /^space(?=port|ship)/,
+  /^race(?=horse)/,
+  /^lake(?=side)/,
+  /^white(?=side)/,
+  /^state(?=side)/,
+  /^juke(?=box)/,
+  /^police(?=woman|women|man|men)/,
+  /^(?:sale|spoke)s?(?=woman|women|man|men)/,
+]
+
+export default compoundPrefixes

package/src/syllables/split/find-cuts/index.js

@@ -0,0 +1,27 @@
+import checkExceptions from './00-exceptions.js'
+import vowelSpan from './01-vowel-span.js'
+import pickBoundaries from './02-boundaries.js'
+
+const findCuts = str => {
+  // Exact exceptions and their plural forms take priority over spelling rules.
+  const found = checkExceptions(str)
+  if (found !== null) {
+    return found
+  }
+  // Keep apostrophe suffixes intact so vowel rules can recognize an extra syllable.
+  const vowelSpans = vowelSpan(str)
+  const cuts = pickBoundaries(str, vowelSpans)
+  // An exception may supply base-word cuts without overriding a syllabic suffix.
+  const key = str.replace(/['’](?:s|d|ll|re|ve)$/, '')
+  // Without a suffix, the initial lookup already checked this word.
+  if (key === str) {
+    return cuts
+  }
+  const baseCuts = checkExceptions(key)
+  if (baseCuts !== null) {
+    return baseCuts.concat(cuts.filter(cut => cut >= key.length))
+  }
+  return cuts
+}
+
+export default findCuts

package/src/syllables/split/index.js

@@ -0,0 +1,22 @@
+import findCuts from './find-cuts/index.js'
+import { applyCuts, normalize, pushCutsBack } from './_lib.js'
+
+// Pass the original term text through; the tokenizer owns pre/post punctuation.
+const splitTerm = (term, options = {}) => {
+  if (!term.text) {
+    return []
+  }
+  let str = term.text
+  if (options.normalize) {
+    str = term.normal || term.text
+  }
+  // we can split on a normalized string
+  const { clean, prefix } = normalize(str)
+  let cuts = findCuts(clean)
+  if (prefix && prefix.length > 0) {
+    cuts = pushCutsBack(cuts, prefix)
+  }
+  return applyCuts(str, cuts)
+}
+
+export default splitTerm

package/types/index.d.cts

@@ -0,0 +1,9 @@
+import type esmPlugin from './index.d.ts' with { 'resolution-mode': 'import' }
+import type * as types from './index.d.ts' with { 'resolution-mode': 'import' }
+
+declare const plugin: typeof esmPlugin
+declare namespace plugin {
+  export type SpeechMethods = types.SpeechMethods
+}
+
+export = plugin

package/{index.d.ts → types/index.d.ts}

@@ -8,6 +8,6 @@ export interface SpeechMethods {
 }
 
 /** extended compromise lib **/
-declare const nlpSpeech: nlp.TypedPlugin<SpeechMethods>
+declare const nlpSpeech: nlp.TypedPlugin<SpeechMethods> & { version: string }
 
 export default nlpSpeech

package/demo/index.html

@@ -1,28 +0,0 @@
-<!DOCTYPE html>
-<html>
-
-<head>
-
-  <script src="https://unpkg.com/compromise"></script>
-  <!-- <script src="https://unpkg.com/compromise-speech"></script> -->
-  <script src="../builds/compromise-speech.min.js"></script>
-
-</head>
-
-<body>
-  <p>compromise-speech demo:</p>
-  <ul>
-    <kbd>chocolate microscopes </kbd>
-    <pre id="res">loading</pre>
-  </ul>
-
-  <script defer>
-    nlp.plugin(compromiseSpeech) // window.compromiseSpeech
-    let txt = document.querySelector('kbd').innerText
-    let doc = nlp(txt)
-    let json = doc.syllables()
-    document.querySelector('#res').innerHTML = JSON.stringify(json, null, 2)
-  </script>
-</body>
-
-</html>

package/rollup.config.js

@@ -1,23 +0,0 @@
-import { terser } from 'rollup-plugin-terser'
-import sizeCheck from 'rollup-plugin-filesize-check'
-import { nodeResolve } from '@rollup/plugin-node-resolve';
-
-const opts = { keep_classnames: true, module: true }
-
-export default [
-  {
-    input: 'src/plugin.js',
-    output: [{ file: 'builds/compromise-speech.cjs', format: 'umd', name: 'compromiseSpeech' }],
-    plugins: [nodeResolve(), sizeCheck({ expect: 11, warn: 15 })],
-  },
-  {
-    input: 'src/plugin.js',
-    output: [{ file: 'builds/compromise-speech.min.js', format: 'umd', name: 'compromiseSpeech' }],
-    plugins: [nodeResolve(), terser(opts), sizeCheck({ expect: 4, warn: 15 })],
-  },
-  {
-    input: 'src/plugin.js',
-    output: [{ file: 'builds/compromise-speech.mjs', format: 'esm' }],
-    plugins: [nodeResolve(), terser(opts), sizeCheck({ expect: 4, warn: 15 })],
-  }
-]

package/scratch.js

@@ -1,17 +0,0 @@
-/* eslint-disable no-console, no-unused-vars */
-import nlp from '../../src/three.js'
-
-import speechPlugin from './src/plugin.js'
-nlp.plugin(speechPlugin)
-// nlp.verbose(true)
-// nlp.verbose('date')
-
-let txt = ''
-txt = 'seventh millenium. white collar'
-
-// let doc = nlp(txt).compute(['soundsLike', 'syllables'])
-// console.dir(doc.json()[0], { depth: 5 })
-
-let doc = nlp('calgary')
-doc.compute('soundsLike')
-console.log(JSON.stringify(doc.json()[0], null, 2))

package/src/compute/index.js

@@ -1,7 +0,0 @@
-import soundsLike from './soundsLike/index.js'
-import syllables from './syllables/index.js'
-
-export default {
-  soundsLike,
-  syllables
-}

package/src/compute/syllables/index.js

@@ -1,13 +0,0 @@
-import getSyllables from './syllables.js'
-
-// const defaultObj = { normal: true, text: true, terms: false }
-
-const syllables = function (view) {
-  view.docs.forEach(terms => {
-    terms.forEach(term => {
-      term.syllables = getSyllables(term.normal || term.text)
-    })
-  })
-}
-
-export default syllables

package/src/compute/syllables/postProcess.js

@@ -1,107 +0,0 @@
-const starts_with_single_vowel_combos = /^(eu)/i
-const joining_consonant_vowel = /^[^aeiou]e([^d]|$)/
-const cvcv_same_consonant = /^([^aeiouy])[aeiouy]\1[aeiouy]/
-const cvcv_same_vowel = /^[^aeiouy]([aeiouy])[^aeiouy]\1/
-const cvcv_known_consonants = /^([tg][aeiouy]){2}/
-const only_one_or_more_c = /^[^aeiouy]+$/
-
-const ends_with_vowel = /[aeiouy]$/
-const starts_with_consonant_vowel = /^[^aeiouy]h?[aeiouy]/
-
-const ones = [
-  /^[^aeiou]?ion/,
-  /^[^aeiou]?ised/,
-  /^[^aeiou]?iled/,
-
-  // -ing, -ent
-  /[aeiou]n[gt]$/,
-
-  // -ate, -age
-  /\wa[gt]e$/,
-]
-
-//suffix fixes
-const postprocess = function (arr) {
-  //trim whitespace
-  arr = arr.map(function (w) {
-    return w.trim()
-  })
-  arr = arr.filter(function (w) {
-    return w !== ''
-  })
-  // if (arr.length > 2) {
-  //   return arr;
-  // }
-  let l = arr.length
-  if (l > 1) {
-    let suffix = arr[l - 2] + arr[l - 1]
-    for (let i = 0; i < ones.length; i++) {
-      if (suffix.match(ones[i])) {
-        arr[l - 2] = arr[l - 2] + arr[l - 1]
-        arr.pop()
-      }
-    }
-  }
-
-  // since the open syllable detection is overzealous,
-  // sometimes need to rejoin incorrect splits
-  if (arr.length > 1) {
-    let first_is_open =
-      (arr[0].length === 1 || arr[0].match(starts_with_consonant_vowel)) &&
-      arr[0].match(ends_with_vowel)
-    let second_is_joining = arr[1].match(joining_consonant_vowel)
-
-    if (first_is_open && second_is_joining) {
-      let possible_combination = arr[0] + arr[1]
-      let probably_separate_syllables =
-        possible_combination.match(cvcv_same_consonant) ||
-        possible_combination.match(cvcv_same_vowel) ||
-        possible_combination.match(cvcv_known_consonants)
-
-      if (!probably_separate_syllables) {
-        arr[0] = arr[0] + arr[1]
-        arr.splice(1, 1)
-      }
-    }
-  }
-
-  if (arr.length > 1) {
-    let second_to_last_is_open =
-      arr[arr.length - 2].match(starts_with_consonant_vowel) &&
-      arr[arr.length - 2].match(ends_with_vowel)
-    let last_is_joining =
-      arr[arr.length - 1].match(joining_consonant_vowel) &&
-      ones.every(re => !arr[arr.length - 1].match(re))
-
-    if (second_to_last_is_open && last_is_joining) {
-      let possible_combination = arr[arr.length - 2] + arr[arr.length - 1]
-      let probably_separate_syllables =
-        possible_combination.match(cvcv_same_consonant) ||
-        possible_combination.match(cvcv_same_vowel) ||
-        possible_combination.match(cvcv_known_consonants)
-
-      if (!probably_separate_syllables) {
-        arr[arr.length - 2] = arr[arr.length - 2] + arr[arr.length - 1]
-        arr.splice(arr.length - 1, 1)
-      }
-    }
-  }
-
-  if (arr.length > 1) {
-    let single = arr[0] + arr[1]
-    if (single.match(starts_with_single_vowel_combos)) {
-      arr[0] = single
-      arr.splice(1, 1)
-    }
-  }
-
-  if (arr.length > 1) {
-    if (arr[arr.length - 1].match(only_one_or_more_c)) {
-      arr[arr.length - 2] = arr[arr.length - 2] + arr[arr.length - 1]
-      arr.splice(arr.length - 1, 1)
-    }
-  }
-
-  return arr
-}
-export default postprocess

package/src/compute/syllables/syllables.js

@@ -1,90 +0,0 @@
-//chop a string into pronounced syllables
-import postProcess from './postProcess.js'
-
-const all_spaces = / +/g
-const ends_with_vowel = /[aeiouy]$/
-const starts_with_consonant_vowel = /^[^aeiouy]h?[aeiouy]/
-const starts_with_e_then_specials = /^e[sm]/
-const starts_with_e = /^e/
-const ends_with_noisy_vowel_combos = /(eo|eu|ia|oa|ua|ui)$/i
-const aiouy = /[aiouy]/
-const ends_with_ee = /ee$/
-// const whitespace_dash = /\s\-/
-
-//method is nested because it's called recursively
-const doWord = function (w) {
-  let all = []
-  let chars = w.split('')
-  let before = ''
-  let after = ''
-  let current = ''
-  for (let i = 0; i < chars.length; i++) {
-    before = chars.slice(0, i).join('')
-    current = chars[i]
-    after = chars.slice(i + 1, chars.length).join('')
-    let candidate = before + chars[i]
-
-    //it's a consonant that comes after a vowel
-    if (before.match(ends_with_vowel) && !current.match(ends_with_vowel)) {
-      if (after.match(starts_with_e_then_specials)) {
-        candidate += 'e'
-        after = after.replace(starts_with_e, '')
-      }
-      all.push(candidate)
-      return all.concat(doWord(after))
-    }
-
-    //unblended vowels ('noisy' vowel combinations)
-    if (candidate.match(ends_with_noisy_vowel_combos)) {
-      //'io' is noisy, not in 'ion'
-      all.push(before)
-      all.push(current)
-      return all.concat(doWord(after)) //recursion
-    }
-
-    // if candidate is followed by a CV, assume consecutive open syllables
-    if (candidate.match(ends_with_vowel) && after.match(starts_with_consonant_vowel)) {
-      all.push(candidate)
-      return all.concat(doWord(after))
-    }
-  }
-  //if still running, end last syllable
-  if (w.match(aiouy) || w.match(ends_with_ee)) {
-    //allow silent trailing e
-    all.push(w)
-  } else if (w) {
-    let last = all.length - 1
-    if (last < 0) {
-      last = 0
-    }
-    all[last] = (all[last] || '') + w //append it to the last one
-  }
-  return all
-}
-
-let syllables = function (str) {
-  let all = []
-  if (!str) {
-    return all
-  }
-  str = str.replace(/[.,?]/g, '')
-  str.split(all_spaces).map(s => {
-    all = all.concat(doWord(s))
-  })
-
-  // str.split(whitespace_dash).forEach(doWord)
-  all = postProcess(all)
-
-  //for words like 'tree' and 'free'
-  if (all.length === 0) {
-    all = [str]
-  }
-  //filter blanks
-  all = all.filter(s => s)
-
-  return all
-}
-
-// console.log(syllables('civilised'))
-
-export default syllables

package/tests/_lib.js

@@ -1,15 +0,0 @@
-import build from '../../../builds/one/compromise-one.mjs'
-import src from '../../../src/one.js'
-import plgBuild from '../builds/compromise-speech.mjs'
-import plg from '../src/plugin.js'
-let nlp;
-
-if (process.env.TESTENV === 'prod') {
-  console.warn('== production build test 🚀 ==')  // eslint-disable-line
-  nlp = build
-  nlp.plugin(plgBuild)
-} else {
-  nlp = src
-  nlp.plugin(plg)
-}
-export default nlp

package/tests/soundsLike.test.js

@@ -1,21 +0,0 @@
-import test from 'tape'
-import nlp from './_lib.js'
-
-test('soundsLike-tests', function (t) {
-  let arr = [
-    ['phil collins', 'fil kolins'],
-    ['Philadelphia freedom', 'filatelfia fretom'],
-    ['Shine on me', 'shine on me'],
-    ['peace of mind', 'pease of mint'],
-    ['Yes I do', 'yes i to'],
-    ['Zapped me', 'sapet me'],
-    ['Right between the eyes', 'rit betwen the eyes'],
-    // ['But the times have changed', 'but the times hafe kshanjet'],
-  ]
-  arr.forEach((a) => {
-    let doc = nlp(a[0])
-    let str = doc.soundsLike({})[0].join(' ')
-    t.equal(str, a[1], a[0])
-  })
-  t.end()
-})

package/tests/syllables.test.js

@@ -1,73 +0,0 @@
-import test from 'tape'
-import nlp from './_lib.js'
-
-const words = [
-  'sud den ly',
-  'con sti pa tion',
-  'di a bo lic',
-  'fa ted',
-  'ge ne tic',
-  'hy giene',
-  'o ma ha',
-  'i mi ta ted',
-  'tree',
-  'ci vi lised',
-  'went',
-  'to ge ther',
-  'tog gle',
-  'move ment',
-  'mo ment',
-  'do nate',
-  'vi king',
-  'wat tage',
-  'con gre gate',
-  'some thing',
-  'sales man',
-  're sour ces',
-  'eve ry thing',
-  'eve ry bo dy',
-  'hole',
-  'ho ly',
-  'sec ret',
-  'cause',
-  'fate',
-  'fates',
-  'eu lo gy',
-  'e on',
-  'rea sons',
-  'and',
-  'hoc key',
-  'wild',
-  'fun',
-  'fun ny',
-  'hor ses',
-  'cal ga ry',
-  'tor na do',
-  'pen',
-  'stu dy',
-  'ja mai ca',
-  'tri ni dad',
-  'ca na da',
-  'i ran',
-  'i raq',
-  'sau di a ra bi a',
-  'clin ton',
-  'hub ris',
-  'tur key',
-  'cor si ca',
-  'dev iled',
-  // 'horse',
-  // 'chance',
-  // 'peo ple'
-]
-
-test('test length', function (t) {
-  words.forEach(sep => {
-    let str = sep.replace(/ /g, '')
-    let doc = nlp(str)
-    let have = doc.syllables()[0]
-    let want = sep.split(/ /g)
-    t.equal(have.length, want.length, str)
-  })
-  t.end()
-})