Contract each match on its own instead of merging them into one
This commit is contained in:
1 parent
163edfec77
commit
cab0138668
2 files changed
+25
-5
No files matched your search
@@ -1,10 +1,6 @@
|
||||
const postPunct = /[,)"';:\-–—.…]/
|
||||
|
||||
const setContraction = function (m, suffix) {
|
||||
if (!m.found) {
|
||||
return
|
||||
}
|
||||
const terms = m.termList()
|
||||
const contractTerms = function (terms, suffix) {
|
||||
//avoid any problematic punctuation
|
||||
for (let i = 0; i < terms.length - 1; i++) {
|
||||
const t = terms[i]
|
||||
@@ -27,6 +23,11 @@ const setContraction = function (m, suffix) {
|
||||
}
|
||||
}
|
||||
|
||||
// contract each match on its own
|
||||
const setContraction = function (m, suffix) {
|
||||
m.docs.forEach(terms => contractTerms(terms, suffix))
|
||||
}
|
||||
|
||||
/** turn 'i am' into i'm */
|
||||
const contract = function () {
|
||||
const doc = this.not('@hasContraction')
|
||||
|
||||
@@ -74,3 +74,22 @@ test('multiple contraction in sentence', function (t) {
|
||||
t.equal(doc.match('he is dead').length, 2, here + 'multi-contraction-count')
|
||||
t.end()
|
||||
})
|
||||
|
||||
test('contract multiple matches', function (t) {
|
||||
let doc = nlp('he is here and she is there')
|
||||
doc.contract()
|
||||
t.equal(doc.text(), `he's here and she's there`, here + 'two in one sentence')
|
||||
|
||||
doc = nlp('we are tired and they are not')
|
||||
doc.contract()
|
||||
t.equal(doc.text(), `we're tired and they're not`, here + 'two of the same')
|
||||
|
||||
doc = nlp('he is here. she is there.')
|
||||
doc.contract()
|
||||
t.equal(doc.text(), `he's here. she's there.`, here + 'two sentences')
|
||||
|
||||
doc = nlp('Tony, is here and John is there')
|
||||
doc.contract()
|
||||
t.equal(doc.text(), `Tony, is here and John's there`, here + 'skip one with punctuation')
|
||||
t.end()
|
||||
})
|
||||
Reference in new issue
Block a user