Subject heuristics

operasoftware · Sep 5, 2011 · 911d4fc · 911d4fc
1 parent 064f9b1
commit 911d4fc
Show file tree

Hide file tree

Showing 7 changed files with 41 additions and 37 deletions.
diff --git a/TODO b/TODO
@@ -0,0 +1,4 @@
+When we have too many subclasses - then checking each of them to mach
+each of the messages will be too slow.  Probably we'll need a hash
+translating the email origin domain to the subclass.
+
diff --git a/lib/Courriel/MMS.pm b/lib/Courriel/MMS.pm
@@ -35,8 +35,36 @@ sub rebless {
     return bless $email, $class;
 }
 
+sub bad_subject {}
+
 # --- Instance methods ---
 
+around subject => sub {
+    my $orig = shift;
+    my $self = shift;
+    my $subject = $self->$orig( @_ );
+    if( $self->bad_subject( $subject ) ){
+        $subject = '';
+    }
+    if( !length( $subject ) ) {
+        my $plain_content = $self->plain_content;
+        if( length( $plain_content ) ) { 
+            ( $subject ) = $plain_content =~ /^([^\.]+\.)/g; # use the first sentence.
+            if( !$subject ) {
+                $subject = substr( $plain_content, 0, 25 ); # if still not subject, use some of the text
+            }
+        }
+        else{
+            my @images = $self->get_mms_images;
+            if( @images ) { 
+                $subject = $images[0][0]; # set subject to image filename
+            }
+        }
+    }
+    return $subject;
+};
+
+
 sub plain_content { 
     my $self = shift;
     my $part = $self->plain_body_part;

diff --git a/lib/Courriel/MMS/Plugin/O2Ie.pm b/lib/Courriel/MMS/Plugin/O2Ie.pm
@@ -17,17 +17,10 @@ sub match {
     return;
 }
 
+sub bad_subject { $_[1] eq 'Multimedia message' }
 
 # --- Instance methods ---
 
-around subject => sub {
-    my $orig = shift;
-    my $self = shift;
-    my $subject = $self->$orig;
-    return undef if $subject eq 'Multimedia message';
-    return $subject;
-};
-
 __PACKAGE__->meta()->make_immutable();
 
 1;

diff --git a/lib/Courriel/MMS/Plugin/Tele2LT.pm b/lib/Courriel/MMS/Plugin/Tele2LT.pm
@@ -17,17 +17,10 @@ sub match {
     return;
 }
 
+sub bad_subject { $_[1] =~ /MMS via e-mail/ }
 
 # --- Instance methods ---
 
-around subject => sub {
-    my $orig = shift;
-    my $self = shift;
-    my $subject = $self->$orig;
-    return undef if $subject =~ /MMS via e-mail/;
-    return $subject;
-};
-
 __PACKAGE__->meta()->make_immutable();
 
 1;

diff --git a/lib/Courriel/MMS/Plugin/VodafoneDE.pm b/lib/Courriel/MMS/Plugin/VodafoneDE.pm
@@ -17,17 +17,10 @@ sub match {
     return;
 }
 
+sub bad_subject { $_[1] eq 'Sie haben eine MMS erhalt' }
 
 # --- Instance methods ---
 
-around subject => sub {
-    my $orig = shift;
-    my $self = shift;
-    my $subject = $self->$orig;
-    return undef if $subject eq 'Sie haben eine MMS erhalt';
-    return $subject;
-};
-
 override plain_content => sub {
     my $self = shift;
     my $part = $self->first_part_matching(

diff --git a/lib/Courriel/MMS/Plugin/VodafoneNZ.pm b/lib/Courriel/MMS/Plugin/VodafoneNZ.pm
@@ -17,17 +17,10 @@ sub match {
     return;
 }
 
+sub bad_subject { $_[1] =~ /^You have a PXT from/ }
 
 # --- Instance methods ---
 
-around subject => sub {
-    my $orig = shift;
-    my $self = shift;
-    my $subject = $self->$orig;
-    return undef if $subject =~ /^You have a PXT from/;
-    return $subject;
-};
-
 __PACKAGE__->meta()->make_immutable();
 
 1;

diff --git a/t/test.t b/t/test.t
@@ -48,11 +48,11 @@ use File::Slurp 'slurp';
         subject('Multimedia message'),
         from( 'aaa@mms.o2.ie' ),
         to( 'example@example.com' ),
-        plain_body( 'test' ),
+        plain_body( '' ),
     );
     my $email = Courriel::MMS->parse( text => $c_email->as_string );
     isa_ok( $email, 'Courriel::MMS::Plugin::O2Ie', 'MMS from O2 Ireland' );
-    is( $email->subject, undef, 'subject cleared for O2Ie' );
+    is( $email->subject, '', 'subject cleared for O2Ie' );
 }
 
 {
@@ -66,7 +66,7 @@ use File::Slurp 'slurp';
     );
     my $email = Courriel::MMS->parse( text => $c_email->as_string );
     isa_ok( $email, 'Courriel::MMS::Plugin::VodafoneDE', 'MMS from Vodafone DE' );
-    is( $email->subject, undef, 'subject cleared for VodafoneDE' );
+    is( $email->subject, 'some text', 'subject cleared for VodafoneDE' );
     is( $email->plain_content, 'some text', 'plain_content ignored http://www.vodafone.de' );
 }
 
@@ -93,7 +93,7 @@ use File::Slurp 'slurp';
     );
     my $email = Courriel::MMS->parse( text => $c_email->as_string );
     isa_ok( $email, 'Courriel::MMS::Plugin::VodafoneNZ', 'MMS from Vodafone New Zealand' );
-    is( $email->subject, undef, 'subject cleared for VodafoneNZ' );
+    is( $email->subject, 'test', 'subject cleared for VodafoneNZ' );
 }
 
 {
@@ -135,7 +135,7 @@ use File::Slurp 'slurp';
     );
     my $email = Courriel::MMS->parse( text => $c_email->as_string );
     isa_ok( $email, 'Courriel::MMS::Plugin::Tele2LT', 'MMS from Tele2 Lithuania' );
-    is( $email->subject, undef, 'subject cleared for Tele2LT' );
+    is( $email->subject, 'test', 'subject cleared for Tele2LT' );
 }
 
 {