Update magnifier Point at start

[Boumtchack]

Link to issue number:

fixes #19746

Summary of the issue:

Text caret follow would fail in some text editor (python console of nvda or notepad), if it would be the last character

Description of user facing changes:

text caret will now focus properly end character

Description of developer facing changes:

Description of development approach:

Update _get_pointAtStart now falls back to the previous character's position when the caret is at end-of-text and both _getBoundingRectFromOffset and _getPointFromOffset fail for the current offset.

Testing strategy:

Manual

Known issues with pull request:

Code Review Checklist:

• Documentation:
  • Change log entry
  • User Documentation
  • Developer / Technical Documentation
  • Context sensitive help for GUI changes
• Testing:
  • Unit tests
  • System (end to end) tests
  • Manual testing
• UX of all users considered:
  • Speech
  • Braille
  • Low Vision
  • Different web browsers
  • Localization in other languages / culture than English
• API is compatible with existing add-ons.
• Security precautions taken.

[Copilot]

Pull request overview

Fixes magnifier caret-follow failures when the caret is positioned at end-of-text by improving how OffsetsTextInfo.pointAtStart is computed when the current offset can’t be mapped to screen coordinates.

Changes:

• Extend _get_pointAtStart to handle LookupError as well as NotImplementedError when resolving a point for the current offset.
• Add a fallback for collapsed ranges to use the previous character’s offset when the caret is at end-of-text and current-offset location lookup fails.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[CyrilleB79]

I am not an expert of this code.

But I wonder if it would not be better, if possible, to implement the fallback in magnifier files rather than in offset.py. Indeed an offset function used elsewhere returning an erroneous information could have side effect; couldn't it?

[LeonarddeR]

I think I agree with @CyrilleB79 here.
Note that the end offset is exclusive in text controls. So when you have a text control containing one character, that one character is at offset 0. The story length is 1 and collapsing at end is offset 1, yet offset 1 is exclusive to the text.
What would make sense to me though is returning the top right of the rectangle at the previous offset.

[seanbudd]

Is this ready for re-review?

[seanbudd]

I think it would be cleaner to just indent this into the except block. It will make reading the logs when it's raised more obvious too

[seanbudd]

can we debug log this error?

[seanbudd]

Suggested change

	try:
	return textInfo._getPointFromOffset(prevOffset)
	except (NotImplementedError, LookupError, AttributeError):
	pass
	raise originalExc
	log.debug(...)
	try:
	return textInfo._getPointFromOffset(prevOffset)
	except (NotImplementedError, LookupError, AttributeError) as e:
	log.debug(...)
	raise e

[Boumtchack]

I kept both fallbacks for better precision and added debug logging to each. The final raise re-raises the original pointAtStart error, which is the most useful for debugging.