[popfile-commit] engine insert.pl,1.23,1.24

[<jgr...@us...>]

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";
  }
+