157299 - DOMParser's parseFromString limits text nodes to 4096 bytes

Status: RESOLVED DUPLICATE of bug 194231

(Core :: XML, defect)

Description

[dgym]

the above link gives the simplest demonstration I could think of (view the source).

any text element over 4096 bytes long in an xml string parsed with '(new
DOMParser()).parseFromString' is truncated to 4Kbytes in the corresponding text
node.

e.g.
doc = (new DOMParser()).parseFromString(
'<a>text...more...than...4096...bytes</a>', 'text/xml' );
doc.documentElement.firstChild.nodeValue.length == 4096;

Comment 1

[Kai Lahmann (is there, where MNG is)]

does this bug still exist in 1.1beta?

Comment 2

[Kenneth Herron]

Attachment: Another testcase

This builds a long string, passes it through DOMParser parseFromString, then
prints the length of the string, the length of the resulting dom object, and
the text of dom object. The object is being truncated to 4096 characters.

Comment 3

[Kenneth Herron]

Using Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3a) Gecko/20021207. I can
reproduce the behavior described. Both the reporter's testcase and the one I
just attached show the text element being truncated to 4096 characters.

Comment 4

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

This should happen for all XML documents you load, not just for the documents
you load using DOMParser; could you please test? Could you also please test if
there are more text nodes following the first one? I am pretty sure we break
longer than 4kb text blocks to several text nodes in the DOM for perf reasons.
If there are, then this is allowed per the DOM Recommendations, I believe.

Comment 5

[dgym]

Indeed, as Heikki Toivonen assumed, the text is being broken down into several
text nodes which is perfectly allowable. I have updated my test script to
include the length of all the text nodes and everything adds up just fine, so
this never was a bug - just a bit unexpected.
Thanks for all the help.