[prev] [thread] [next] [lurker] [Date index for 2005/02/08]
Author: simon
Date: 2005-02-08 15:49:43 +0000 (Tue, 08 Feb 2005)
New Revision: 1856
Modified:
trunk/Email-Store-HTML/lib/Email/Store/HTML.pm
trunk/Email-Store-HTML/t/01basic.t
Log:
Update a test, make html return all html parts and update docs
Modified: trunk/Email-Store-HTML/lib/Email/Store/HTML.pm
===================================================================
--- trunk/Email-Store-HTML/lib/Email/Store/HTML.pm 2005-02-08 15:49:17 UTC (rev 1855)
+++ trunk/Email-Store-HTML/lib/Email/Store/HTML.pm 2005-02-08 15:49:43 UTC (rev 1856)
@@ -16,6 +16,7 @@
$VERSION = "0.1";
+
sub on_store_order { 2 }
sub on_store {
@@ -47,10 +48,86 @@
my $scrubbed = $scrubber->scrub($raw);
my $text = $f->parse($raw);
Email::Store::HTML->create( { mail => $mail->id, raw => $raw, scrubbed => $scrubbed, as_text => $text } );
- last;
}
}
+=head1 NAME
+
+Email::Store::HTML - various HTML related functions for Email::Store::Mail
+
+=head1 SYNOPSIS
+
+ my $mail = Email::Store::Mail->retrieve( $msgid );
+ exit unless $mail->html;
+
+ for ($mail->html) {
+ print $_->raw; # prints out the raw HTML version of the attachment
+ print $_->scrubbed; # prints out a scrubbed version of the mail which should be safe
+ print $_->as_text; # prints out a version of the HTML converted to plain text
+ }
+
+=head1 DESCRIPTION
+
+=head1 METHODS
+
+=head2 on_store <Email::Store::Mail>
+
+This finds every HTML attachment in the mail and performs various operations on them
+before storing them as a new C<Email::Store::HTML> object.
+
+=head2 raw
+
+The raw HTML, exactly as we found it.
+
+=head2 scrubbed
+
+A scrubbed version of the HTML with things like javascript removed.
+
+=head2 as_text
+
+The HTML run through C<HTML::FormatText::WithLinks>. Links are placed after the anchor
+word(a) in square brackets so that
+
+ <a href="http://thegestalt.org">HOME!</a>;
+
+becomes
+
+ HOME! [ http://thegestalt.org ]
+
+
+=head1 BUGS AND TODO
+
+No bugs known at the moment.
+
+It might be nice to give people access to to the scrubber and formatter so that they
+could change the options.
+
+=head1 SUPPORT
+
+This module is part of the Perl Email Project - http://pep.kwiki.org/
+
+There is a mailing list at pep@xxxx.xxx (subscribe at pep-subscribe@xxxx.xxx)
+and an archive available at http://nntp.perl.org/group/pep.php
+
+=head1 AUTHOR
+
+Simon Wistow <simon@xxxxxxxxxx.xxx>
+
+=head1 COPYRIGHT
+
+Copyright 2005, Simon Wistow
+
+This code is distributed under the same terms as Perl itself.
+
+
+=head1 SEE ALSO
+
+L<HTML::Scrubber>, L<HTML::FormatText::WithLinks>
+
+=cut
+
+
+
###
# Configuration for HTML::Scrubber
###
@@ -102,8 +179,7 @@
package Email::Store::Mail;
sub html {
my ($self) = @_;
- my ($html) = Email::Store::HTML->from_mail($self->message_id);
- return $html;
+ return Email::Store::HTML->from_mail($self->message_id);
}
package Email::Store::HTML;
Modified: trunk/Email-Store-HTML/t/01basic.t
===================================================================
--- trunk/Email-Store-HTML/t/01basic.t 2005-02-08 15:49:17 UTC (rev 1855)
+++ trunk/Email-Store-HTML/t/01basic.t 2005-02-08 15:49:43 UTC (rev 1856)
@@ -1,4 +1,4 @@
-use Test::More tests => 13;
+use Test::More tests => 14;
use File::Slurp;
BEGIN { unlink("t/test.db"); }
use Email::Store "dbi:SQLite:dbname=t/test.db";
@@ -11,8 +11,13 @@
ok($m, "Got the mail back");
-my ($html, $body, $raw, $scrubbed, $as_text);
-ok($html = $m->html, "Got html");
+
+my (@html, $html, $body, $raw, $scrubbed, $as_text);
+ok(@html = $m->html, "Got html");
+is(@html, 1, "Only one part");
+
+$html = shift @html;
+
ok($body = $m->simple->body, "Got body");
ok($raw = $html->raw, "Got raw");
ok($scrubbed = $html->scrubbed, "Got scrubbed");
Generated at 16:00 on 08 Feb 2005 by mariachi 0.52