Use OffsetConverter for OffsetTextInfo find (#16657)

Fixes #16317

Summary of the issue:
Find in browse mode (Virtual Buffers) can be inaccurate when a python character spans two character offsets in the underlying TextInfo, such as when emojis come in to play.

Description of user facing changes
Find functionality is now accurate when a page contains emoji characters and similar.

Description of development approach
Use an OffsetConverter to correct offsets.

source/textInfos/offsets.py

@@ -665,10 +665,11 @@ def find(self,text,caseSensitive=False,reverse=False):
 		m=re.search(re.escape(text),inText,(0 if caseSensitive else re.IGNORECASE)|re.UNICODE)
 		if not m:
 			return False
+		converter = textUtils.getOffsetConverter(self.encoding)(inText)
 		if reverse:
-			offset=self._startOffset-m.end()
+			offset = self._startOffset - converter.strToEncodedOffsets(m.end())
 		else:
-			offset=self._startOffset+1+m.start()
+			offset = self._startOffset + 1 + converter.strToEncodedOffsets(m.start())
 		self._startOffset=self._endOffset=offset
 		return True
 

user_docs/en/changes.md

@@ -40,6 +40,7 @@ Unicode CLDR has been updated.
   * Error messages referenced with `aria-errormessage` are now reported in Google Chrome and Mozilla Firefox. (#8318)
   * If present, NVDA will now use `aria-labelledby` to provide accessible names for tables in Mozilla Firefox. (#5183)
   * NVDA will correctly announce radio and checkbox menuitems when first entering submenus in Google Chrome and Mozilla Firefox. (#14550)
+  * NVDA's browse mode find functionality is now more accurate when the page contains emojis. (#16317, @LeonarddeR)
 * NVDA will announce correctly the autocomplete suggestions in Eclipse and other Eclipse-based environments on Windows 11. (#16416, @thgcode)
 * Improved reliability of automatic text readout, particularly in terminal applications. (#15850, #16027, @Danstiv)
 * NVDA will correctly announce selection changes when editing a cell's text in Microsoft Excel. (#15843)