118792 - Save Complete Web Page screws up document.write() calls (and other dynamic DOM modifications)

Status: VERIFIED DUPLICATE of bug 115328

(Core Graveyard :: File Handling, defect)

Description

[Boris Zbarsky [:bzbarsky]]

See bug 118251, Comment #5 and bug 118251 comment #7.  Basically, passing the
current DOM to WebBrowserPersist is bad because it completely screws up on any
page that does dynamic DOM modification.  We need to pass in the original DOM
_before_ executing any top-level scripts or the like.

The problem with the current implementation is that content written with
document.write() will appear twice in the saved version -- once because the
script is saved, once because the content itself is saved.

Comment 1

[Boris Zbarsky [:bzbarsky]]

adding dataloss, since this can make _major_ modifications to the page source
(consider a script that removes some <script> nodes from the DOM after they have
run, for example).

Comment 2

[Jeremy M. Dolan]

Is it out of the question, for complete page save, to reget the source from
cache (or whereever the pristine copy is), and have a seperate parser reparse it
for links to rewrite?

I have a feeling we'd never be able to eliminiate all of the cases were
DOM->disk does bad stuff.

Comment 3

[Ben Goodger (use ben at mozilla dot org for email)]

->adamlock. This explains issues with saving sites like cnn.com

Comment 4

[Adam Lock]

I will dup this against bug 115328 seeing as I own both of them. While this bug
actually is actually an xpfe problem - to feed the persist object a pristine DOM
& not a used one - it shares the same issues as embedding faces and a solution
can probably be fashioned to address both instances (possibly).

*** This bug has been marked as a duplicate of 115328 ***

Comment 5

[sairuh (rarely reading bugmail)]

vrfy dup