424264 - Same URI exposed for all link children associated with imagemaps

Status: RESOLVED FIXED

(Core :: Disability Access APIs, defect)

Description

[Joanmarie Diggs]

Steps to reproduce:

1. Examine the contents of the URL

2. Query the accessible Hypertext interface for the table cell which contains the imagemap.

Expected results:  Unique URIs would be exposed

Actual results: The URI associated with the first link is exposed as the URI for all of the links.

Other comments:  When I originally tested the patch for bug 491786 (using safeway.com) prior to it being checked in, I saw different URIs.  However, this bug seems to exist since the patch was checked into trunk.  I haven't yet worked out what has changed. :-(

Comment 1

[alexander :surkov (:asurkov)]

Joanie, could you show piece of code to see the problem, also please comment what accessible you operate with like acc.GetURI(0) // imagemap accessible

Comment 2

[Joanmarie Diggs]

Attachment: screenshot of hierarchy in Accerciser

In the spirit (and hope) that a picture is worth a thousand words I've attached a screenshot of Accerciser showing the hierarchy and the URIs for the test case.  If this doesn't help, I'll try words. :-)

Thanks for looking at this so quickly Alexander!

Comment 3

[alexander :surkov (:asurkov)]

Ok, on imagemap links are exposed correctly, DOMi test is:

accessible.QueryInterface(Components.interfaces.nsIAccessibleHyperLink);
var len = accessible.anchors;
for(var i=0; i < len; i++) {
  output(accessible.getURI(i).spec + "\n");
}

on table cell there is only one link accessible, DOMi test:
accessible.QueryInterface(Components.interfaces.nsIAccessibleHyperText);
output(accessible.links);

Joanie, your picture is very good and it shows there is a bug :) but unfortunately I don't know how to reproduce it. Possibly you'll look at my examples and say me what I do wrong or put your example. Thank you.

Comment 4

[Evan Yan]

Attachment: patch

The bug is on atk part. We only cached one uri for hyperlink

Comment 5

[Joanmarie Diggs]

Phew!  :-)  Thanks Evan!!

I just tried your patch, I'm now seeing the unique URIs.  I double checked this time to be sure. :-) I also used your patch during some Orca regression testing and did not see anything change/break as a result of your patch. 

I really appreciate it!

Comment 6

[Ginn Chen]

Comment on attachment 310977 [details] [diff] [review]
patch

I think we'd better not cache uri and anchor counts in MaiAtkHyperlinkClass.

Just remove the code would be fine.

Comment 7

[Evan Yan]

Attachment: just not cache uri

Comment 8

[Mike Beltzner [:beltzner, not reading bugmail]]

Comment on attachment 311333 [details] [diff] [review]
just not cache uri

a=beltzner

Comment 9

[alexander :surkov (:asurkov)]

/cvsroot/mozilla/accessible/src/atk/nsMaiHyperlink.cpp,v  <--  nsMaiHyperlink.cpp
new revision: 1.9; previous revision: 1.8
done