204322 - href's of url links contained within HTML message body are not scanned by Body:Contains filter

Status: RESOLVED WORKSFORME

(MailNews Core :: Filters, defect)

Description

[Christopher Tobar]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.3.1) Gecko/20030425
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.3.1) Gecko/20030425

Most of the junk email I receive has a generic sender address like @yahoo. or
@hotmail.com.  Most have an unsubscribe option contained within the body of the
email that say "click HERE to unsubscribe".  The "HERE" word is highlighted
because it contains a link.  the "BODY" filter option should be able to search
these links for text as well because the link is very consistant, whereas the
sender address is not.

Reproducible: Always

Steps to Reproduce:
1. In mail, click on tools, filters, and edit or add a new filter
2. The filter should be set to "match any of the following"
3. set the BODY to CONTAINS any link, like www.junkmail.com

Actual Results:  
The messages were not filtered out

Expected Results:  
It should have found the link text in the body of the email and performed the
appropriate MOVE action.

Comment 1

[Mike Cowperthwaite]

Original summary:  "when incoming mail is filtered, url links that are contained 
within the body of junk email are not searched."

duplicated with 1.3 Final, 1.4b-0430

Comment 2

[Mike Cowperthwaite]

*** Bug 204557 has been marked as a duplicate of this bug. ***

Comment 3

[(not reading, please use [e-mail] instead)]

mass re-assign.

Comment 4

[Mike Cowperthwaite]

This may qualify as a dupe of bug 144036, which is about generally filtering on 
the HTML source of a message.

Comment 5

[James Rome]

I think I see this bug in non-junk e-mails on build 2003072204. In the attached
message, the links at the top of the page do not work in Mozilla, but do work in IE.

Comment 6

[James Rome]

Attachment: example of bad linkjs

The links at the top do not work

Comment 7

[James Rome]

I think I now see this all html mail messages. For example, the Sun Core Java
Technologies newsletter. The article links which refer to links such as "/#2"
return "is not a registered protocol" message when clicked.

Comment 8

[Mike Cowperthwaite]

James Rome, I believe you have misunderstood this bug entirely.  This has 
nothing to do with whether links 'work'; it has to do with whether the href of 
the link is parsed by message filters.

Comment 9

[Christopher Tobar]

Attachment: Example of email with links that the filter did not find

Comment 10

[James Rome]

I keep getting accused of not searching the database before I file a new bug, so
I will transfer my comments to a new bug 215248.

Comment 11

[Marc DVer]

Still present in 1.6.  I tried to set a filter for a particular spam I've been
getting, but the filter won't recognize the url.

Comment 12

[Bill Turner]

This problem has been occuring ever since mail filtering was introducted.
and still exists in Mozilla 1.6
Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.6) Gecko/20040113

If I understand the reported bug correctly, then the filter will not
scan for "asd.asd" in the following example, but will find "dfgdfg."

		<A href="asd.asd"> dfgdfg </a> 

and yes, that is the exact problem that I am having.
My immediate question is why it appears to be scanning
the rendered html rather than the RAW message format. 

Comment 13

[John H. Miller]

I have tried this and agree that it is a bug and should be fixed.  It would be
very valuable in eliminating spam if I could use a URL link contained in the
body of an email to identify it as spam.

Comment 14

[Mike Cowperthwaite]

I've tested this in TB 1.5 and SM 1.1 -- it's working now for the general case, for POP.  (No body search for IMAP, of course... bug 67421.)

I noticed that if I download headers-only, the fetch of the actual message doesn't seem to be filtered, but I think that may be a general issue with filtering on the body when headers-only is enabled.