keep a match's quotes and brackets out of m.text() in wrap()

This commit is contained in:
dualfroz committed 2026-10-14 19:36:07 +02:00
1 parent 219f140a9d
commit c3666c98d6
2 files changed
+35 -1

No files matched your search

+13 -1
View File
@@ -1,9 +1,21 @@
import fmts from './_fmts.js'
const toText = function (term) {
const pre = term.pre || ''
const post = term.post || ''
return pre + term.text + post
}
// the match's pre and post are added around fn(m), so m.text() should leave them out
const withoutEnds = function (m) {
const text = m.text
m.text = fmt => {
const opts = typeof fmt === 'string' ? fmts[fmt] : fmt
return text.call(m, { keepEndPunct: false, ...opts })
}
return m
}
const findStarts = function (doc, obj) {
const starts = {}
Object.keys(obj).forEach(reg => {
@@ -25,7 +37,7 @@ const wrap = function (doc, obj) {
// do a span tag
if (Object.hasOwn(starts, t.id)) {
const { fn, end } = starts[t.id]
const m = doc.update([[n, i, end]])
const m = withoutEnds(doc.update([[n, i, end]]))
text += terms[i].pre || ''
text += fn(m)
i = end - 1
+22
View File
@@ -42,3 +42,25 @@ test('out-wrap', function (t) {
t.equal(out, `before (few more words) after`, here + 'pre+post')
t.end()
})
test('out-wrap-punctuation', function (t) {
const arr = [
[`"hello world" he said`, `"[hello world]" he said`],
[`he said "hello world" today`, `he said "[hello world]" today`],
[`he said "hello world"`, `he said "[hello world]"`],
[`(hello world) he said`, `([hello world]) he said`],
[`he said (hello world) today`, `he said ([hello world]) today`],
[`he said (hello world).`, `he said ([hello world]).`],
[`hello world, he said`, `[hello world], he said`],
[`he left. "hello world" she said`, `he left. "[hello world]" she said`],
[`"hello world."`, `"[hello world]."`],
]
arr.forEach(function (a) {
const doc = nlp(a[0])
t.equal(doc.out({ 'hello world': m => `[${m.text()}]` }), a[1], here + 'text: ' + a[0])
t.equal(doc.wrap({ 'hello world': m => `[${m.out('normal')}]` }), a[1], here + 'normal: ' + a[0])
})
const out = nlp(`"Hello World" he said`).wrap({ 'hello world': m => `[${m.text('normal')}]` })
t.equal(out, `"[hello world]" he said`, here + 'keeps text format')
t.end()
})