Separate internal and LanguageTool data formats

This allows a bit simpler code and less clutter in the LanguageTool
communication.
This commit is contained in:
Simon Rozman 2024-02-01 09:43:18 +01:00
parent 279c815b15
commit 0d8d6e165a

View File

@ -24,7 +24,7 @@ async function besCheckText(el)
{
switch (el.nodeType) {
case Node.TEXT_NODE:
return [{ text: el.textContent, el: el }]
return [{ text: el.textContent, el: el, markup: false }]
case Node.ELEMENT_NODE:
let data = []
@ -36,15 +36,15 @@ async function besCheckText(el)
{
case 'inline':
case 'inline-block':
data.splice(0, 0, { markup: '<'+el.tagName+'>', el: el })
data.splice(data.length, 0, { markup: '</'+el.tagName+'>' })
data.splice(0, 0, { text: '<'+el.tagName+'>', el: el, markup: true })
data.splice(data.length, 0, { text: '</'+el.tagName+'>', markup: true })
return data;
default:
let regAllWhitespace = /^\s*$/
if (data.some(x => 'text' in x && !regAllWhitespace.test(x.text))) {
if (data.some(x => !x.markup && !regAllWhitespace.test(x.text))) {
const requestData = {
format: 'plain',
data: JSON.stringify({annotation: data}),
data: JSON.stringify({annotation: data.map(x => x.markup ? { markup: x.text } : { text: x.text })}),
language: 'sl',
level: 'picky'
}
@ -62,34 +62,27 @@ async function besCheckText(el)
return response.json()
})
.then(responseData => {
// responseData.matches.sort((a, b) => a.offset < b.offset ? -1 : a.offset > b.offset ? +1 : 0)
// Reverse sort grammar mistakes for easier markup insertion later.
// When we start inserting grammar mistakes at the back, indexes before that remain valid.
responseData.matches.sort((a, b) => a.offset < b.offset ? +1 : a.offset > b.offset ? -1 : 0);
responseData.matches.forEach(match => {
let range = document.createRange()
// Locate start of the grammar mistake.
for (let idx = 0, startingOffset = 0; ; ) {
if ('text' in data[idx]) {
if (startingOffset <= match.offset && match.offset < startingOffset + data[idx].text.length) {
range.setStart(data[idx].el, match.offset - startingOffset)
break
}
startingOffset += data[idx++].text.length
for (let idx = 0, startingOffset = 0; ; startingOffset += data[idx++].text.length) {
if (!data[idx].markup && /*startingOffset <= match.offset &&*/ match.offset < startingOffset + data[idx].text.length) {
range.setStart(data[idx].el, match.offset - startingOffset)
break
}
else startingOffset += data[idx++].markup.length
}
// Locate end of the grammar mistake.
let endOffset = match.offset + match.length
for (let idx = 0, startingOffset = 0; ; ) {
if ('text' in data[idx]) {
if (startingOffset <= endOffset && endOffset <= startingOffset + data[idx].text.length) {
range.setEnd(data[idx].el, endOffset - startingOffset)
break
}
startingOffset += data[idx++].text.length
for (let idx = 0, startingOffset = 0; ; startingOffset += data[idx++].text.length) {
if (!data[idx].markup && /*startingOffset <= endOffset &&*/ endOffset <= startingOffset + data[idx].text.length) {
range.setEnd(data[idx].el, endOffset - startingOffset)
break
}
else startingOffset += data[idx++].markup.length
}
const span = document.createElement('span')
@ -98,7 +91,7 @@ async function besCheckText(el)
// Dirty hack, copied from: https://stackoverflow.com/questions/67634286/invalidstateerror-failed-to-execute-surroundcontents-on-range-the-range-ha
span.appendChild(range.extractContents());
range.insertNode(span);
// This is a better way to apply span elements,
// but it doesnt work when the range has partially selected a non-Text node.
// range.surroundContents(span)
@ -106,14 +99,14 @@ async function besCheckText(el)
})
.catch(error => {
// TODO: Make parsing issues non-fatal.
throw new Error('Request to backend server failed: ' + error)
throw new Error('Parsing backend server response failed: ' + error)
})
}
return [{ markup: '<'+el.tagName+'/>', el: el }]
return [{ text: '<'+el.tagName+'/>', el: el, markup: true }]
}
default:
return [{ markup: '<?'+el.nodeType+'>', el: el }]
return [{ text: '<?'+el.nodeType+'>', el: el, markup: true }]
}
}