122681 - XML DOM Javascript .nodeValue Null

Status: VERIFIED INVALID

(Core :: DOM: HTML Parser, defect)

Description

[marc]

I have "walked the DOM" with Javascript.  When I get to a text node and use 
nodeValue to get the contents, it returns null when it should return the 
contents.  I have doulbe checked with javascript to do this when the nodeType 
==3. I have tried to use .text with .innerHTML. That doesn't work either.

Comment 1

[Heikki Toivonen (remove -bugzilla when emailing directly)]

Please attach a test case.

Comment 2

[marc]

Attachment: HTML files with Javascript

This is the HTML file w/ javascript

Comment 3

[marc]

Attachment: xml file used with instr.html

Comment 4

[marc]

Attachment: ** 67146 Obsolete ** HTML file with javascript

Comment 5

[marc]

Attachment: 2nd Test Case

This uses a different XML file, same exact structure and contents. 
Compare Instructor.xml to Instructor1.xml then view instr.html vs instr2.html. 

There are Parsing problems.  This probably goes beyond the scope of BUG 122681

Comment 6

[marc]

Attachment: goes with instr2.html

Read comments on attachment 67155 [details]

Comment 7

[Heikki Toivonen (remove -bugzilla when emailing directly)]

Attachment: 2nd Test Case

Comment 8

[Heikki Toivonen (remove -bugzilla when emailing directly)]

Attachment: 1st Test Case

Comment 9

[Heikki Toivonen (remove -bugzilla when emailing directly)]

Attachment: Fixed (?) 1st Test Case

Comment 10

[Heikki Toivonen (remove -bugzilla when emailing directly)]

Attachment: Fixed 1st Test Case

Comment 11

[Heikki Toivonen (remove -bugzilla when emailing directly)]

Attachment: Fixed 1st Test Case, please...

Argh!

Comment 12

[Heikki Toivonen (remove -bugzilla when emailing directly)]

Finally got it right. Ok, there were several problems in the testcases. In
short, Mozilla is doing the right thing and your script needs to be modified.
Read on:

1. Mozilla's document.load() executes asynchronously (in IE you can choose).
Therefore, you MUST set an onload handler to be called before you initiate the
load(). Otherwise you have no idea when the document is loaded and when you can
use it.

2. You were going past the length of the ndes array, and using the property
every time in the loop is a (small) perf hit.

3. In XML whitespace between elements is generally not removed. Therefore, the
empty text nodes you were seeing between each element were the newline text nodes.

4. The nodeValue of element name was empty, as expected (I think, I should
really check the spec whether or not it should be empty or contain the element
name). You probably wanted all of the nodeValues of the element's child text
nodes concatenated. In the last testcase I attached, I show how you get the
firstChild's nodeValue (which is a text node). You can probably elaborate from
there...

Also, in the future when you attach test cases, it would be nice if you could
first attach the XML files that are referenced by other files, and then attach
the "driver" file last using the correct URLs to the earlier attachments. That
way, I can test just by clicking on the last attachment link without needing to
download all of them into my computer.
Anyway, thanks for your interest in Mozilla. It's great you took the effort to
file a bug and provide testcases!

Comment 13

[marc]

Well, I thought it was resolved, because I found an answer to my problem, but..
<INFO>some information</INFO> is actually 2 or 3 nodes, however you look at it.  
if I get to "some information" it has a nodeType of 3, and so it can then be
used with .nodeValue.  YOu should not have to say Element.firstChild.nodeValue. 
It should just be Node.NodeValue.  

http://www.mozilla.org/docs/dom/domref/dom_el_ref17.html#1018663
also from a previous email, the nodeValue of an Element is null, but not for a
text node. 
Marc
Certified XML Developer and Related Technologies

Comment 14

[Boris Zbarsky [:bzbarsky]]

The documentation you linked to is misleading at best.  The only useful thing
there is that there is a link to the specification near the bottom
(http://www.w3.org/TR/2000/REC-DOM-Level-2-Core-20001113/core.html#ID-F68D080).

The specification says:

Interface               nodeValue
----------------------------------------------
Attr                    value of attribute
CDATASection            content of the CDATA Section
Comment                 content of the comment
Document                null
DocumentFragment        null
DocumentType            null
Element                 null 
NamedNodeMap            null
EntityReference         null 
Notation                null 
ProcessingInstruction   entire content excluding the target
Text                    content of the text node

So returning null for an Element node is in fact correct.

Bug 122741 filed on the documentation people.

Comment 15

[Moied]

Verified Invalid