fix(editor): preserve combining marks in whole-word replacement (#34957)

* fix(editor): preserve combining marks in whole-word replacement

* fix(editor): keep whole-word boundaries after emoji presentation sequences

Resolve a leading combining mark to its base character so VS16 after an
emoji does not count as a word character, treat enclosing marks (keycaps)
as symbols, and drop the redundant hook-level undo/redo test.

---------

Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>

src/renderer/src/components/editor/markdown-preview-search-word-boundaries.test.ts

@@ -0,0 +1,36 @@
+import { describe, expect, it } from 'vitest'
+import { findTextMatchRanges } from './markdown-preview-search'
+
+describe('whole-word search around combining marks', () => {
+  it.each(['́', 'ि'])('keeps the trailing combining mark %s attached to its word', (mark) => {
+    const text = `cat${mark} cat ${mark}cat`
+    expect(findTextMatchRanges(text, 'cat', { wholeWord: true })).toEqual([
+      { start: 4 + mark.length, end: 7 + mark.length },
+      { start: 8 + 2 * mark.length, end: 11 + 2 * mark.length }
+    ])
+  })
+
+  it('does not match a Devanagari word inside a longer word ending in a vowel sign', () => {
+    expect(findTextMatchRanges('किताबें किताब', 'किताब', { wholeWord: true })).toEqual([
+      { start: 8, end: 13 }
+    ])
+  })
+
+  it('finds a complete decomposed accented word and preserves source code-unit offsets', () => {
+    const text = 'café cafe caféine'
+    expect(findTextMatchRanges(text, 'café', { wholeWord: true })).toEqual([{ start: 0, end: 5 }])
+    expect(findTextMatchRanges(text, 'cafe', { wholeWord: true })).toEqual([{ start: 6, end: 10 }])
+    expect(findTextMatchRanges(text, 'cafe')).toHaveLength(3)
+  })
+
+  it.each([
+    ['⚠️Warning here', 'Warning'],
+    ['✔️Done', 'Done'],
+    ['1️⃣Step', 'Step']
+  ])('matches a whole word right after an emoji presentation sequence in %s', (text, query) => {
+    const start = text.indexOf(query)
+    expect(findTextMatchRanges(text, query, { wholeWord: true })).toEqual([
+      { start, end: start + query.length }
+    ])
+  })
+})

src/renderer/src/lib/text-match-ranges.ts

@@ -82,8 +82,10 @@ function findCaseInsensitiveMatchRanges(
   return matches
 }
 
-// Treat Unicode letters/digits/underscore as word characters, matching editor whole-word find boundaries.
-const WORD_CHARACTER = /[\p{L}\p{N}_]/u
+// Combining marks stay attached to their letters when checking whole-word boundaries.
+const WORD_CHARACTER = /[\p{L}\p{M}\p{N}_]/u
+const COMBINING_MARK = /\p{M}/u
+const ENCLOSING_MARK = /\p{Me}/u
 
 function isWordCharacter(char: string | undefined): boolean {
   return char !== undefined && WORD_CHARACTER.test(char)
@@ -113,10 +115,22 @@ function codePointAt(text: string, index: number): string | undefined {
   return codePoint === undefined ? undefined : String.fromCodePoint(codePoint)
 }
 
+// A leading mark belongs to the preceding base (emoji+VS16 stays a boundary); enclosing marks make a symbol (keycaps).
+function isWordGraphemeBefore(text: string, index: number): boolean {
+  let offset = index
+  let codePoint = codePointBefore(text, offset)
+  while (codePoint !== undefined && COMBINING_MARK.test(codePoint)) {
+    if (ENCLOSING_MARK.test(codePoint)) {
+      return false
+    }
+    offset -= codePoint.length
+    codePoint = codePointBefore(text, offset)
+  }
+  return isWordCharacter(codePoint)
+}
+
 function isWholeWordMatch(text: string, start: number, end: number): boolean {
-  const before = codePointBefore(text, start)
-  const after = codePointAt(text, end)
-  return !isWordCharacter(before) && !isWordCharacter(after)
+  return !isWordGraphemeBefore(text, start) && !isWordCharacter(codePointAt(text, end))
 }
 
 function buildLocaleLowercaseIndex(text: string): {