Re: indexing Maildir mails

2005-07-21 Thread Joe Shaw
Hi,

On Wed, 2005-07-20 at 22:51 +0300, Aviram Jenik wrote:
> The problem I'm experiencing is quite different. The mail *is* rfc822, but it 
> does not show up in the results. In fact, when I add a first line to screw 
> the mime detection (make the mail file a 'text/plain') it *does* appear in 
> the results - so the problem is definitely in the FilterMail object.

Are you having a problem with body searches, or names of people?  I
noticed a problem where names were being indexed as keywords and not
regular text.  Which meant in a mail from me, "Joe" would match, but
"Shaw" wouldn't.  I just fixed this in CVS, but it will require you to
reindex the mail.

In my tests, I haven't had any problem searching for text in bodies of
emails.

Joe

___
Dashboard-hackers mailing list
Dashboard-hackers@gnome.org
http://mail.gnome.org/mailman/listinfo/dashboard-hackers


Re: indexing Maildir mails

2005-07-21 Thread Joe Shaw
Hi,

On Thu, 2005-07-21 at 10:30 +, Raghu GS wrote:
> i fixed a header and search in best for it, best showed that fixed header mail
> in same Files but this time it won't open it via a text editor but instead 
> shows
> this long error
> 
> Open URI: action:_tile_568!Open
> Cmd: kmail --view

Beagle is actually trying to run a program named "kmail --view" rather
than recognizing that "--view" is an argument.  I just checked a fix
into CVS.

Joe


___
Dashboard-hackers mailing list
Dashboard-hackers@gnome.org
http://mail.gnome.org/mailman/listinfo/dashboard-hackers


Re: indexing Maildir mails

2005-07-21 Thread Raghu GS
Hello Everybody

in my turn while trying to make Beagle index Maildir mail files produced by
Kmail with the great help from a person nicnamed dsd in #dashboard irc channal 
i had bit success,
first-of all for indexing Mails produced by KMail you must add
/.kde/shere/apps/kmail/mails/inbox/cur
/.kde/shere/apps/kmail/mails/sent-mails/cur to Beagle list of roots,
then it would be indexed, but if you use KMail it would be indexed by text
filter not by mail filter,
this is mishandling is not caused by beagle but by gnome-vfs's inability to
correct indentify Maildir Mail, i heard this is caused by manipulation of
headers by application after receiving it or by Mail server,

the result would be is in best when click a result it would call a text editor
to proccess it,
Note: even if after headers fixed, best shows results in Files section not in
mails section.

i fixed a header and search in best for it, best showed that fixed header mail
in same Files but this time it won't open it via a text editor but instead shows
this long error

Open URI: action:_tile_568!Open
Cmd: kmail --view
Arg: 
'file:///home/raghugs/.kde/share/apps/kmail
/mail/inbox/cur/1121582219.7423.q55ap'
Error in OpenFromMime: System.ComponentModel.Win32Exception: Cannot find the
specified file
in [0x00289] (at
/tmp/scratch/BUILD/mono-1.1.8.1/mcs/class/System
/System.Diagnostics/Process.cs:823)
System.Diagnostics.Process:Start_common (System.Diagnostics.ProcessStartInfo
startInfo, System.Diagnostics.Process process)
in [0x7] (at
/tmp/scratch/BUILD/mono-1.1.8.1/mcs/class/System
/System.Diagnostics/Process.cs:856)
System.Diagnostics.Process:Start ()
in (wrapper remoting-invoke-with-check) System.Diagnostics.Process:Start ()
in <0x0021c> Beagle.Tile.Tile:OpenFromMime (Beagle.Hit hit, System.String
command_fallback, System.String args_fallback, Boolean expects_uris_fallback)



___
Dashboard-hackers mailing list
Dashboard-hackers@gnome.org
http://mail.gnome.org/mailman/listinfo/dashboard-hackers


Re: indexing Maildir mails

2005-07-20 Thread Aviram Jenik
On Wednesday 20 July 2005 19:59, Ian Soboroff wrote:
> I have a similar problem... 
[...] 
> $ gnomevfs-info ~/Mail/mail/misc/8225
> MIME type : text/plain

The problem I'm experiencing is quite different. The mail *is* rfc822, but it 
does not show up in the results. In fact, when I add a first line to screw 
the mime detection (make the mail file a 'text/plain') it *does* appear in 
the results - so the problem is definitely in the FilterMail object.

- Aviram
___
Dashboard-hackers mailing list
Dashboard-hackers@gnome.org
http://mail.gnome.org/mailman/listinfo/dashboard-hackers


Re: indexing Maildir mails

2005-07-20 Thread Joe Shaw
Hi,

On Wed, 2005-07-20 at 13:15 -0400, Ian Soboroff wrote:
> $ head -4 ~/Mail/mail/misc/8225
> X-From-Line: imap Fri Jul  5 16:33:27 2002
> Received: from nist-i.nist.gov (nist1.nist.gov [129.6.94.1])
> by email.nist.gov (8.12.2/8.12.2) with SMTP id g65KV3FN011792
> for <[EMAIL PROTECTED]>; Fri, 5 Jul 2002 16:31:03 -0400 (EDT)

The "X-From-Line" is throwing off the MIME sniffer.  Most sniffers
(including gnome-vfs and Apache) expect one of "From", "Received",
"Relay-Version", or "Return-Path".  They're not technically required to
start the file, but it's generally a good, easy to detect convention.
Some KMail users are seeing something similar.

You could argue that the bug is in freedesktop.org's mime magic
database, but realistically it's impossible to detect every possible
first line that would be a compliant message.  It's generally good
behavior for mailers to append their headers rather than prepend them.
For this reason you may want to file an interoperability bug against
Gnus.

See http://bugzilla.gnome.org/show_bug.cgi?id=310462

Joe

___
Dashboard-hackers mailing list
Dashboard-hackers@gnome.org
http://mail.gnome.org/mailman/listinfo/dashboard-hackers


Re: indexing Maildir mails

2005-07-20 Thread Ian Soboroff
Joe Shaw <[EMAIL PROTECTED]> writes:

> Hi,
>
> On Wed, 2005-07-20 at 12:59 -0400, Ian Soboroff wrote:
>> I have a similar problem... I use Emacs Gnus for mail, which stores a
>> single email per file, but since gnomevfs-info doesn't correctly
>> identify the file type, it doesn't come up as mail:
>
> What are the first 4 lines of the file?

$ head -4 ~/Mail/mail/misc/8225
X-From-Line: imap Fri Jul  5 16:33:27 2002
Received: from nist-i.nist.gov (nist1.nist.gov [129.6.94.1])
by email.nist.gov (8.12.2/8.12.2) with SMTP id g65KV3FN011792
for <[EMAIL PROTECTED]>; Fri, 5 Jul 2002 16:31:03 -0400 (EDT)

Ian

___
Dashboard-hackers mailing list
Dashboard-hackers@gnome.org
http://mail.gnome.org/mailman/listinfo/dashboard-hackers


Re: indexing Maildir mails

2005-07-20 Thread Joe Shaw
Hi,

On Wed, 2005-07-20 at 12:59 -0400, Ian Soboroff wrote:
> I have a similar problem... I use Emacs Gnus for mail, which stores a
> single email per file, but since gnomevfs-info doesn't correctly
> identify the file type, it doesn't come up as mail:

What are the first 4 lines of the file?

Joe


___
Dashboard-hackers mailing list
Dashboard-hackers@gnome.org
http://mail.gnome.org/mailman/listinfo/dashboard-hackers


Re: indexing Maildir mails

2005-07-20 Thread Ian Soboroff

I have a similar problem... I use Emacs Gnus for mail, which stores a
single email per file, but since gnomevfs-info doesn't correctly
identify the file type, it doesn't come up as mail:


$ gnomevfs-info ~/Mail/mail/misc/8225
Name  : 8225
Type  : Regular
MIME type : text/plain
Size  : 1383
Blocks: 8
I/O block size: 32768
Local : YES
SUID  : NO
SGID  : NO
Sticky: NO
Permissions   : 600600
Link count: 1
UID   : 34544
GID   : 4254
Access time   : Wed Jul 20 09:09:02 2005
Modification time : Fri Jul  5 16:33:27 2002
Change time   : Mon Mar 14 14:49:36 2005
Device #  : 31
Inode #   : 1023412966
Readable  : YES
Writable  : YES
Executable: NO

Ian

___
Dashboard-hackers mailing list
Dashboard-hackers@gnome.org
http://mail.gnome.org/mailman/listinfo/dashboard-hackers


Re: indexing Maildir mails

2005-07-19 Thread Noam Rathaus
Hi,

The only issue in Maildir case (Kmail at least) is the fact that most
if not all the directories start with a "." (the name of the folder),
which by default, as far as I remember are regarded by Beagle has
hidden and shouldn't be indexed.

I am not sure whether they have addressed this or not.

On 7/20/05, Aviram Jenik <[EMAIL PROTECTED]> wrote:
> On Wednesday 20 July 2005 05:23, D Bera wrote:
> > > I think it might still be possible to recusrively scan for cur/ new/
> > tmp/ directories in the maildir directory and find rfc822 files in
> > them. At least, that is what I think
> > http://www.rpcurnow.force9.co.uk/mairix/ does.
> >
> I'm pretty sure this is what the FilterMail filter in beagle *currently* does.
> 
> > I will see if I can recursively scan directories and pick up mail
> > files for indexing.
> >
> 
> From what I've seen, it seems beagle uses the standard File Indexing backend
> to recursively scan directories. If the file is an rfc822 file, it filters it
> through the mail filter which indexes it - all that works fine for me, except
> for the fact the mail doesn't show up later in the search results.
> 
> > - Bera
> >
> - Aviram
> ___
> Dashboard-hackers mailing list
> Dashboard-hackers@gnome.org
> http://mail.gnome.org/mailman/listinfo/dashboard-hackers
> 


-- 
Thanks
Noam Rathaus
CTO
Beyond Security Ltd.

Join the SecuriTeam community on Orkut:
http://www.orkut.com/Community.aspx?cmm=1
___
Dashboard-hackers mailing list
Dashboard-hackers@gnome.org
http://mail.gnome.org/mailman/listinfo/dashboard-hackers


Re: indexing Maildir mails

2005-07-19 Thread Aviram Jenik
On Wednesday 20 July 2005 05:23, D Bera wrote:
> > I think it might still be possible to recusrively scan for cur/ new/ 
> tmp/ directories in the maildir directory and find rfc822 files in
> them. At least, that is what I think
> http://www.rpcurnow.force9.co.uk/mairix/ does.
>
I'm pretty sure this is what the FilterMail filter in beagle *currently* does.

> I will see if I can recursively scan directories and pick up mail
> files for indexing.
>

From what I've seen, it seems beagle uses the standard File Indexing backend 
to recursively scan directories. If the file is an rfc822 file, it filters it 
through the mail filter which indexes it - all that works fine for me, except 
for the fact the mail doesn't show up later in the search results.

> - Bera
>
- Aviram
___
Dashboard-hackers mailing list
Dashboard-hackers@gnome.org
http://mail.gnome.org/mailman/listinfo/dashboard-hackers


Re: indexing Maildir mails

2005-07-19 Thread D Bera
It seems that the directory structure for maildir is not standardised.
This page talks about two possible formats:
http://wiki.mutt.org/index.cgi?action=browse&diff=2&id=MuttFaq/Maildir

Ubuntu seems to be coming up with yet another naming standard (more at
https://wiki.ubuntu.com//MailStorageStandardisation ). And to talk
about kmail, their structure is given in
http://wiki.mutt.org/index.cgi?ExtendedMaildirFormat and it doesnt
seem they have any plans of changing it soon
(http://mail.kde.org/pipermail/kmail-devel/2005-January/014614.html ).

However, all these are different conventions for directory structures.
I think it might still be possible to recusrively scan for cur/ new/
tmp/ directories in the maildir directory and find rfc822 files in
them. At least, that is what I think
http://www.rpcurnow.force9.co.uk/mairix/ does.

I will see if I can recursively scan directories and pick up mail
files for indexing.

- Bera

> > The changelog of beagle-0.12 says it can index maildir mails. Does
> > that mean mails in maildir format from any mail application ? I tried
> > to change local_path, imap_path etc. in  EvolutionMailQueryable to
> > point to a maildir directory I have but there was no response in the
> > debug output.
> 
> Filtering maildir mail is done via a filter on regular file indexing. If the
> file is identified to be message/rfc822 (you can check this with
> gnomevfs-info) it passed through the FilterMail filter and indexed.
___
Dashboard-hackers mailing list
Dashboard-hackers@gnome.org
http://mail.gnome.org/mailman/listinfo/dashboard-hackers


Re: indexing Maildir mails

2005-07-19 Thread Aviram Jenik
Hi,

On Wednesday 20 July 2005 01:07, D Bera wrote:
> Hi,
> The changelog of beagle-0.12 says it can index maildir mails. Does
> that mean mails in maildir format from any mail application ? I tried
> to change local_path, imap_path etc. in  EvolutionMailQueryable to
> point to a maildir directory I have but there was no response in the
> debug output.

Filtering maildir mail is done via a filter on regular file indexing. If the 
file is identified to be message/rfc822 (you can check this with 
gnomevfs-info) it passed through the FilterMail filter and indexed.

My experience has been as follows: so far I haven't been able to see the 
indexed files in the results (either best or beagle-query) but I haven't 
investigated enough to see where the problem is - according to the logs the 
file is indeed indexed, and according to beagle-index-info the indexed file 
count grows (notice Maildir is marked by the backend as "file" and not 
"mail).

If you can check this out and let me know what you find that would be great - 
Maildir support is extremely important to me and I would like to give the 
beagle developers as much accurate information as I can regarding bugs in 
this feature.

- Aviram
___
Dashboard-hackers mailing list
Dashboard-hackers@gnome.org
http://mail.gnome.org/mailman/listinfo/dashboard-hackers


indexing Maildir mails

2005-07-19 Thread D Bera
Hi,
The changelog of beagle-0.12 says it can index maildir mails. Does
that mean mails in maildir format from any mail application ? I tried
to change local_path, imap_path etc. in  EvolutionMailQueryable to
point to a maildir directory I have but there was no response in the
debug output.
From the code it seems the crawler tries to find files named
"summary" or *.ev-summary and quits otherwise. Are these evolution
specific files ?
   If indexing general maildir mails is not yet supported, is anybody
working on it ?

Thanks,
- Bera
___
Dashboard-hackers mailing list
Dashboard-hackers@gnome.org
http://mail.gnome.org/mailman/listinfo/dashboard-hackers