134457 - Relative URLs in text output

Status: NEW

(Core :: DOM: Serializers, defect)

Description

[Ben Bucksch (:BenB)]

Reproduction:
1. Run "<a href="/foo/index.html">a link</a>" (in the document
<http://www.example.com/base/url/somedoc.html>) through the HTML->TXT converter.

Actual result:
"a link </foo/index.html>"

Expected result:
"a link <http://www.example.com/base/url/foo/index.html>"

Comment 1

[Felix Miata]

Expected result:

       'a link'

You're asking for a text file. The link in the HTML page is invisible unless
present in the status bar or you are viewing the HTML source, and so is not
expected from/in a text file. If you want a link, you save as html. If you need
the link elsewhere, you copy the link from the html page.

There is no need to redefine text. Text has been text a very long time. If you
want something different from what is currently defined as a text file, give it
a new name instead of emulating M$ and stealing an established name.

Comment 2

[Ben Bucksch (:BenB)]

> There is no need to redefine text.

That's what bug 131166 is about, please keep the discussion where it belongs.

Comment 3

[harishd]

--> Tanu

Comment 4

[Tanu Mutreja]

To me the most sensible solution is: while doing a copy & paste, behavior 
should be wysiwyg but while saving file as text, we should try to retain the 
important information to the maximum possible extent. So while pasting the 
stuff as plain text URL need not be retained though while saving as text it 
must be.
 
I know this is the kind of bug where we can always have counter arguments but 
I'm trying to fix this bug on the above basis.

Comment 5

[Tanu Mutreja]

Attachment: Patch to fix it ...

Akkana, I need to confirm my understanding about the situations where
PlainTextSerializer gets Parser nodes as input. I think it's only Mail-> sent
as plain text, in which case I do not see any sensible value as base URL of the
document. Are there other places from where PlainTextSerilaizer gets parser
nodes?
In the patch I have passed the value to BaseURI only from DocumentEncoder, I'm
not sure if this is required at some other places too.

Comment 6

[Daniel Bratell]

Does every document have an URL? 
Is that URL always meaningful? 
What if I copies from a html mail where someone has written a relative url? 
What if the html file is opened from the file system?

Comment 7

[Akkana Peck]

The plaintext serializer gets parser nodes as input whenever there isn't a
document created.  For some reason mail does this (serializes to html then
re-parses and then serializes to plaintext), but copy/paste will also go through
this path, because copy saves the html, and if the pasting application requests
plaintext then we parse the html to convert to plaintext.  Plus of course the
output tests.  There may be a few other places where this happens, but those are
all I can think of offhand, but you definitely can't assume that only mail does
this.

Comment 8

[Tanu Mutreja]

Once again we are close to "how" to do but considering comments #5 and #6, we 
need to decide "what" to do.

Comment 9

[Tanu Mutreja]

I need some inputs to move further...

Comment 10

[BugBot [:suhaib / :marco/ :calixte]]

The bug assignee didn't login in Bugzilla in the last 7 months.
:hsinyi, could you have a look please?
For more information, please visit auto_nag documentation.