Merge pull request #28128 from overleaf/mj-client-word-count-magic-comments

[web] Support some texcount magic comments in client side word count

GitOrigin-RevId: 12749876edbf7940ca9adfc63fcc7e787894e727

services/web/frontend/js/features/word-count-modal/utils/count-words-in-file.ts

@@ -139,6 +139,10 @@ export const countWordsInFile = (
   const headMatcher = NodeType.match<
     (nodeRef: SyntaxNodeRef) => boolean | void
   >({
+    Comment(nodeRef) {
+      handleComment(nodeRef)
+      return false
+    },
     Title(nodeRef) {
       data.headers++
       iterateNode(nodeRef, 'header')
@@ -149,6 +153,10 @@ export const countWordsInFile = (
   const bodyMatcher = NodeType.match<
     (nodeRef: SyntaxNodeRef) => boolean | void
   >({
+    Comment(nodeRef) {
+      handleComment(nodeRef)
+      return false
+    },
     Normal(nodeRef) {
       textNodes.push({
         from: nodeRef.from,
@@ -265,17 +273,48 @@ export const countWordsInFile = (
 
   const preambleExtent = findPreambleExtent(tree)
 
+  const state = {
+    skipping: false,
+  }
+
+  const TC_REGEX = /^%+TC:\s*(\w+)\s*/i
+
+  const handleComment = (nodeRef: SyntaxNodeRef) => {
+    const comment = content.slice(nodeRef.from, nodeRef.to)
+
+    // look for TeXcount instructions
+    const match = TC_REGEX.exec(comment)
+    if (match) {
+      switch (match[1].toLowerCase()) {
+        case 'ignore':
+          state.skipping = true
+          break
+        case 'endignore':
+          state.skipping = false
+          break
+        default:
+          break
+      }
+    }
+  }
+
   tree.iterate({
     from: 0,
     to: preambleExtent.to,
-    enter(nodeRef) {
+    enter(nodeRef: SyntaxNodeRef) {
+      if (state.skipping && !nodeRef.type.is('Comment')) {
+        return false
+      }
       return headMatcher(nodeRef.type)?.(nodeRef)
     },
   })
 
   tree.iterate({
     from: preambleExtent.to,
-    enter(nodeRef) {
+    enter(nodeRef: SyntaxNodeRef) {
+      if (state.skipping && !nodeRef.type.is('Comment')) {
+        return false
+      }
       return bodyMatcher(nodeRef.type)?.(nodeRef)
     },
   })

services/web/test/frontend/features/word-count-modal/utils/count-words-in-file.test.ts

@@ -5,7 +5,7 @@ import { WordCountData } from '@/features/word-count-modal/components/word-count
 import { createSegmenters } from '@/features/word-count-modal/utils/segmenters'
 import { expect } from 'chai'
 
-describe('word count', function () {
+describe('word-count', function () {
   beforeEach(async function () {
     this.data = {
       encode: '',
@@ -35,6 +35,10 @@ describe('word count', function () {
         path.join(__dirname, 'word-count.tex'),
         'utf-8'
       ),
+      'word-count-with-ignored-sections.tex': await readFile(
+        path.join(__dirname, 'word-count-with-ignored-sections.tex'),
+        'utf-8'
+      ),
     }
 
     this.projectSnapshot = {
@@ -69,4 +73,28 @@ describe('word count', function () {
       textWords: 42,
     })
   })
+
+  it('skips ignored sections', function () {
+    countWordsInFile(
+      this.data,
+      this.projectSnapshot,
+      'word-count-with-ignored-sections.tex',
+      this.segmenters
+    )
+
+    expect(this.data).to.deep.include({
+      abstractCharacters: 0,
+      abstractWords: 0,
+      captionCharacters: 0,
+      captionWords: 0,
+      footnoteCharacters: 0,
+      footnoteWords: 0,
+      headCharacters: 0,
+      headWords: 0,
+      otherCharacters: 0,
+      otherWords: 0,
+      textCharacters: 10,
+      textWords: 3,
+    })
+  })
 })

services/web/test/frontend/features/word-count-modal/utils/word-count-with-ignored-sections.tex

@@ -0,0 +1,9 @@
+foo
+%TC:ignore
+bar
+%TC:endignore
+baz
+%TC: ignore
+qux
+%%TC: endignore
+quux