Upstream spencermountain/compromise state of the touched files before d4dd232f45

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

@@ -0,0 +1,87 @@
+import splice from './_splice.js'
+import apostropheS from './apostrophe-s.js'
+import apostropheD from './apostrophe-d.js'
+import apostropheT from './apostrophe-t.js'
+import isPossessive from './isPossessive.js'
+
+const byApostrophe = /'/
+
+// poor-mans reindexing of this sentence only
+const reIndex = function (terms) {
+  terms.forEach((t, i) => {
+    if (t.index) {
+      t.index[1] = i
+    }
+  })
+}
+
+// run tagger on our new implicit terms
+const reTag = function (terms, view, start, len) {
+  const tmp = view.update()
+  tmp.document = [terms]
+  // offer to re-tag neighbours, too
+  let end = start + len
+  if (start > 0) {
+    start -= 1
+  }
+  if (terms[end]) {
+    end += 1
+  }
+  tmp.ptrs = [[0, start, end]]
+  tmp.compute(['freeze', 'lexicon', 'preTagger', 'unfreeze'])
+  // don't for a reindex of the whole document
+  reIndex(terms)
+}
+
+const byEnd = {
+  // how'd
+  d: (terms, i) => apostropheD(terms, i),
+  // we ain't
+  t: (terms, i) => apostropheT(terms, i),
+  // bob's
+  s: (terms, i, world) => {
+    // [bob's house] vs [bob's cool]
+    if (isPossessive(terms, i)) {
+      return world.methods.one.setTag([terms[i]], 'Possessive', world, null, '2-contraction')
+    }
+    return apostropheS(terms, i)
+  },
+}
+
+const toDocs = function (words, view) {
+  const doc = view.fromText(words.join(' '))
+  doc.compute('id')
+  return doc.docs[0]
+}
+
+//really easy ones
+const contractionTwo = view => {
+  const { world, document } = view
+  // each sentence
+  document.forEach((terms, n) => {
+    // loop through terms backwards
+    for (let i = terms.length - 1; i >= 0; i -= 1) {
+      // is it already a contraction
+      if (terms[i].implicit) {
+        continue
+      }
+      let after = null
+      if (byApostrophe.test(terms[i].normal) === true) {
+        after = terms[i].normal.split(byApostrophe)[1]
+      }
+      let words = null
+      // any known-ones, like 'dunno'?
+      if (Object.hasOwn(byEnd, after)) {
+        words = byEnd[after](terms, i, world)
+      }
+      // actually insert the new terms
+      if (words) {
+        words = toDocs(words, view)
+        splice(document, [n, i], words)
+        reTag(document[n], view, i, words.length)
+        continue
+      }
+    }
+  })
+}
+export default { contractionTwo }

tests/two/contractions/expand.test.js

@@ -0,0 +1,88 @@
+import test from 'tape'
+import nlp from '../_lib.js'
+const here = '[two/expand] '
+
+test('basic is contractions', function (t) {
+  let r = nlp(`he is cool.`)
+  r.contractions().expand()
+  t.equal(r.out('text'), `he is cool.`, here + 'expanded-expand')
+
+  r = nlp(`he's cool.`)
+  r.contractions().expand()
+  t.equal(r.out('text'), `he is cool.`, here + 'contracted-expand')
+
+  r = nlp(`he ain't going`)
+  r.contractions().expand()
+  t.equal(r.out('text'), `he is not going`, here + 'aint')
+
+  r = nlp(`that's really great.`)
+  r.contractions().expand()
+  t.equal(r.out('text'), `that is really great.`, here + 'contracted-expand')
+
+  r = nlp(`she'll, eat icecream`)
+  r.contractions().expand()
+  t.equal(r.out('text'), `she will, eat icecream`, here + 'with-punctuation')
+
+  r = nlp("we're not gonna take it, no we're not gonna take it")
+  r.contractions().expand()
+  t.equal(r.out('text'), `we are not going to take it, no we are not going to take it`, here + 'expand gonna twice')
+
+  r = nlp("we're we're gonna gonna")
+  r.contractions().expand()
+  t.equal(r.out('text'), `we are we are going to going to`, here + 'expand consecutive')
+  t.end()
+})
+
+test('do-not contractions', function (t) {
+  let r = nlp(`please do not eat the marshmellow`)
+  r.contractions().expand()
+  t.equal(r.out('text'), `please do not eat the marshmellow`, here + 'expanded-expand')
+
+  r = nlp(`please don't eat the marshmellow`)
+  r.contractions().expand()
+  t.equal(r.out('text'), `please do not eat the marshmellow`, here + 'contracted-expand')
+
+  t.end()
+})
+
+test('have contractions', function (t) {
+  let r = nlp(`i have stood`)
+  r.contractions().expand()
+  t.equal(r.out('text'), `i have stood`, here + 'expanded-expand')
+
+  r = nlp(`i've stood`)
+  r.contractions().expand()
+  t.equal(r.out('text'), `i have stood`, here + 'contracted-expand')
+
+  t.end()
+})
+
+test('repeated contract-expand', function (t) {
+  const r = nlp(`i'm good`)
+  r.contractions().expand()
+
+  r.contractions().expand().expand().expand()
+  t.equal(r.out('text'), `i am good`, here + 'expand-n')
+  t.end()
+})
+
+test('contracted', function (t) {
+  const r = nlp(`I'll go to Toronto. I will see.`)
+  const m = r.contractions()
+  const str = m.out('text')
+  t.equal(str, `I'll`, here + 'contracted')
+
+  t.equal(m.length, 1, here + 'no-expanded')
+  t.end()
+})
+
+test('would-or-did', function (t) {
+  let r = nlp(`i'd contemplate`)
+  let str = r.contractions().expand().all().out('text')
+  t.equal(str, `i would contemplate`, here + 'i-would')
+
+  r = nlp(`i'd contemplated`)
+  str = r.contractions().expand().all().out('text')
+  t.equal(str, `i had contemplated`, here + 'i-had')
+  t.end()
+})