popfile-commit Mailing List for POPFile - Automatic Email Classification (Page 111)

SourceForge Headquarters 1320 Columbia Street Suite 310 San Diego, CA 92101 +1 (858) 422-6466

Update of /cvsroot/popfile/engine
In directory sc8-pr-cvs1:/tmp/cvs-serv11608

Modified Files:
	insert.pl 
Log Message:
Updated to work with new structure and use API function

Index: insert.pl
===================================================================
RCS file: /cvsroot/popfile/engine/insert.pl,v
retrieving revision 1.23
retrieving revision 1.24
diff -C2 -d -r1.23 -r1.24
*** insert.pl	31 Jul 2003 16:32:20 -0000	1.23
--- insert.pl	9 Oct 2003 18:22:11 -0000	1.24
***************
*** 25,134 ****

  use strict;
! use locale;
! use Classifier::MailParse;
! 
! my %words;
! 
! # ---------------------------------------------------------------------------------------------
! #
! # load_word_table
! #
! # $bucket    The name of the bucket we are loading words for
! #
! # Fills the words hash with the word frequencies for word loaded from the appropriate bucket
! #
! # ---------------------------------------------------------------------------------------------
! sub load_word_table
! {
!     my ($bucket) = @_;
! 
!     # Make sure that the bucket mentioned exists, if it doesn't the create an empty
!     # directory and word table
! 
!     mkdir("corpus");
!     mkdir("corpus/$bucket");
! 
!     print "Loading word table for bucket '$bucket'...\n";
! 
!     open WORDS, "<corpus/$bucket/table";
! 
!     # Each line in the word table is a word and a count
! 
!     while (<WORDS>) {
!         if ( /__CORPUS__ __VERSION__ (\d+)/ ) {
!             if ( $1 != 1 ) {
!                 print "Incompatible corpus version in $bucket\n";
!                 return;
!             }
! 
!             next;
!         }
! 
!         if ( /(.+) (.+)/ ) {
!             $words{$1} = $2;
!         }
!     }
! 
!     close WORDS;
! }
! 
! # ---------------------------------------------------------------------------------------------
! #
! # save_word_table
! #
! # $bucket    The name of the bucket we are loading words for
! #
! # Writes the words hash out to a bucket
! #
! # ---------------------------------------------------------------------------------------------
! 
! sub save_word_table
! {
!     my ($bucket) = @_;
! 
!     print "Saving word table for bucket '$bucket'...\n";
! 
!     open WORDS, ">corpus/$bucket/table";
!     print WORDS "__CORPUS__ __VERSION__ 1\n";
! 
!     # Each line in the word table is a word and a count

!     foreach my $word (keys %words) {
!         print WORDS "$word $words{$word}\n";
!     }

!     close WORDS;
! }

! # ---------------------------------------------------------------------------------------------
! #
! # split_mail_message
! #
! # $message    The name of the file containing the mail message
! #
! # Splits the mail message into valid words and updated the words hash
! #
! # ---------------------------------------------------------------------------------------------

! sub split_mail_message
! {
!     my ($message) = @_;
!     my $parser   = new Classifier::MailParse;
!     my $word;

!     print "Parsing message '$message'...\n";

!     $parser->parse_file($message);

!     foreach $word (keys %{$parser->{words__}}) {
!         $words{$word} += $parser->{words__}{$word};
!     }
! }

! # main

! if ( $#ARGV >= 1 )
! {
!     load_word_table($ARGV[0]);

      my @files;
--- 25,62 ----

  use strict;
! use Classifier::Bayes;
! use POPFile::Configuration;
! use POPFile::MQ;
! use POPFile::Logger;

! if ( $#ARGV > 0 ) {
!     my $c = new POPFile::Configuration;
!     my $mq = new POPFile::MQ;
!     my $l = new POPFile::Logger;
!     my $b = new Classifier::Bayes;

!     $c->configuration( $c );
!     $c->mq( $mq );
!     $c->logger( $l );

!     $l->configuration( $c );
!     $l->mq( $mq );
!     $l->logger( $l );

!     $l->initialize();

!     $mq->configuration( $c );
!     $mq->mq( $mq );
!     $mq->logger( $l );

!     $b->configuration( $c );
!     $b->mq( $mq );
!     $b->logger( $l );

!     $b->initialize();

!     $c->load_configuration();

!     $b->start();

      my @files;
***************
*** 140,150 ****
      }

!     foreach my $file (@files) {
!         split_mail_message($file);
      }
- 
-     save_word_table($ARGV[0]);
- 
-     print "done.\n";
  } else {
      print "insert.pl - insert mail messages into a specific bucket\n\n";
--- 68,76 ----
      }

!     if ( !$b->add_messages_to_bucket( $ARGV[0], @files ) ) {
!         print "Bucket $ARGV[0] does not exist\n";
!     } else {
!         print "Added ", $#files+1, " files to $ARGV[0]\n";
      }
  } else {
      print "insert.pl - insert mail messages into a specific bucket\n\n";
***************
*** 153,154 ****
--- 79,81 ----
      print "       <messages>         Filename of message(s) to insert\n";
  }
+ 

2003	Jan	Feb (160)	Mar (119)	Apr (111)	May (118)	Jun (101)	Jul (304)	Aug (113)	Sep (140)	Oct (137)	Nov (87)	Dec (122)
2004	Jan (78)	Feb (125)	Mar (131)	Apr (59)	May (121)	Jun (166)	Jul (150)	Aug (137)	Sep (73)	Oct (58)	Nov (27)	Dec (60)
2005	Jan (131)	Feb (84)	Mar (36)	Apr (8)	May (28)	Jun (20)	Jul (10)	Aug (72)	Sep (76)	Oct (34)	Nov (3)	Dec (29)
2006	Jan (13)	Feb (92)	Mar (7)	Apr (1)	May (1)	Jun (2)	Jul (4)	Aug (17)	Sep (5)	Oct (2)	Nov (8)	Dec (12)
2007	Jan (28)	Feb (15)	Mar	Apr	May (8)	Jun (4)	Jul (5)	Aug (8)	Sep (20)	Oct (38)	Nov (65)	Dec (92)
2008	Jan (21)	Feb (56)	Mar (27)	Apr (174)	May (25)	Jun	Jul	Aug	Sep	Oct	Nov	Dec

popfile-commit Mailing List for POPFile - Automatic Email Classification (Page 111)

popfile-commit — CVS commit list messages