451376 - IAccessibleText::attributes provides incorrect info after a mis-spelled word

Status: RESOLVED FIXED

(Core :: Disability Access APIs, defect)

Description

[Michael Curran]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.1a2pre) Gecko/20080819034536 Minefield/3.1a2pre
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.1a2pre) Gecko/20080819034536 Minefield/3.1a2pre ID:20080819034536

IAccessibleText::attributes returns the attribute run representing a mis-spelled word, if you give it an offset which is the offset after that attribute run. i.e. its the endOffset of the attribute run.

Reproducible: Always

Steps to Reproduce:
1. In a new thunderbird message, type "This iss a test".
2. call IAccessibleText::attributes for that node of text giving it an offset of 0, to get the "This " attribute run.
3. Give it an offset of 5, to get the "iss " attribute run [the mis-spelling] 
4. Give it an offset of 8, the start of the final attribute run [after the mis-spelling].
5. Give it an offset of 9.
Actual Results:  
*Giving 0 gives back startOffset: 0, endOffset: 5, and the appropriate attributes.
*Giving 5 gives back startOffset: 5, endOffset:8, and the attributes for that run.
*Giving 8 gives back StartOffset: 5, endOffset: 8, and attributes for that run.
*Giving 9 gives back startOffset: 8, endOffset: 15, and the attributes for that run.

Expected Results:  
*Giving 8 should give back startOffset: 8, endOffset: 15.

It seems that this bug is only related to the attribute runs of mis-spellings. The tests I have done seem to show that other attribute runs for bold, italic etc, do not cause this problem. 
This issue is very important as it can cause any AT using IAccessibleText::attributes to possibly go in to an infinite loop as the relationship between the given offset, and the resulting start and end offsets is not correct.

Comment 1

[alexander :surkov (:asurkov)]

What kind of intervals should we use [a, b) or [a, b]? Iirc, we talked with Marco and we bend to use closed intervals like [a,b]. But what do you thing?

Comment 2

[alexander :surkov (:asurkov)]

(In reply to comment #1)
> What kind of intervals should we use [a, b) or [a, b]? Iirc, we talked with
> Marco and we bend to use closed intervals like [a,b]. But what do you thing?
> 

by the way, ranges don't use closed intervals http://www.w3.org/TR/2000/REC-DOM-Level-2-Traversal-Range-20001113/ranges.html, possibly we should follow it too

Comment 3

[alexander :surkov (:asurkov)]

Attachment: patch

nsIDOMNSRange::ComparePoint should return 1 (the point is after range) if the given node and offset are equal to end node and end offset of the range because end bound isn't included into range

Comment 4

[Olli Pettay [:smaug][[e-mail]](vacation until July 14)]

Comment on attachment 334851 [details] [diff] [review]
patch

1.8 seems to have the same bug. Please write some mochitest to test range.comparePoint() and maybe also .isPointInRange(). If you do that, r=me :)

Comment 5

[Marco Zehe (:MarcoZ)]

Comment on attachment 334851 [details] [diff] [review]
patch

r=me for the test part. Thanks!

Comment 6

[alexander :surkov (:asurkov)]

Attachment: patch2

+ content mochitest

Comment 7

[alexander :surkov (:asurkov)]

http://hg.mozilla.org/mozilla-central/index.cgi/rev/3b707bbddf47