Changeset 1244 for trunk/gsdl/perllib/plugins/EMAILPlug.pm

Timestamp:

2000-06-27T17:10:07+12:00 (24 years ago)

Author:

sjboddie

Message:

Caught up most general plugins (that's the ones in gsdlhome/perllib/plugins)
with changes to BasPlug so that they can all now use the new general plugin
options. Those I didn't do were FoxPlug (as it's not actually used anywhere
and I don't know what it does) and WebPlug (as it's kind of a work in
progress and doesn't really work anyway). All plugins will still work
(including all the collection specific ones that are laying around), some
of them just won't have access to the general options.
I also wrote a short perl script (pluginfo.pl) that prints out all the
options available to a given plugin.

File:

• trunk/gsdl/perllib/plugins/EMAILPlug.pm (modified) (8 diffs)

trunk/gsdl/perllib/plugins/EMAILPlug.pm

@@ -70 +70 @@
 }
 
+use strict;
 
 # Create a new EMAILPlug object with which to parse a file.
@@ -77 +78 @@
 sub new {
     my ($class) = @_;
-    $self = new BasPlug ();
+    my $self = new BasPlug ("EMAILPlug", @_);
+
     return bless $self, $class;
 }
 
-
-# Is EMAILPlug recursive?  No.
-
-sub is_recursive {
-    return 0; 
-}
-
-
-# Read a file and store its contents in a new document object.
-# First, we check to see if it is an email message we're dealing
-# with, then we extract the text and metadata, then we store
-# all this information.
-#
-# Returns: number of files processed or undef if it can't process
-# a file.  This plugin only processes one file at a time.
-
-sub read {
+sub get_default_process_exp {
     my $self = shift (@_);
-    my ($pluginfo, $base_dir, $file, $metadata, $processor) = @_;
-
-    #
-    # Check that we're dealig with a valid mail file
-    #
-
-    # Make sure file exists 
-    my $filename = &util::filename_cat($base_dir, $file);
-    return undef unless (-e $filename);
-    return undef unless ($filename =~ /\d+(\.email)?$/);
-
-    # Read the text and make sure it is an email message
-    open (FILE, $filename) || die "EMAILPlug::read - can't open $filename\n";
-    my @text = <FILE>;
-    my $text = join("", @text);
-    return undef unless (($text =~ /From:/) || ($text =~ /To:/));
-
-    print STDERR "EMAILPlug: processing $filename\n" if $processor->{'verbosity'};
+
+    return q^\d+(\.email)?$^;
+}
+
+# do plugin specific processing of doc_obj
+sub process {
+    my $self = shift (@_);
+    my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj) = @_;
+    
+    # Check that we're dealing with a valid mail file
+    return undef unless (($$textref =~ /From:/) || ($$textref =~ /To:/));
+
+    print STDERR "EMAILPlug: processing $file\n" 
+    if $self->{'verbosity'} > 1;
+
+    my $cursection = $doc_obj->get_top_section();
 
     #
@@ -123 +107 @@
 
     # Separate header from body of message
-    my $Headers = $text;
+    my $Headers = $$textref;
     $Headers =~ s/\n\n.*//s;
-    $text = substr $text, (length $Headers);
+    $$textref = substr $$textref, (length $Headers);
 
     # Extract basic metadata from header
@@ -158 +142 @@
 
 
-    #
-    # Create a new document object
-    #
-
-    my $doc_obj = new doc ($file, "indexed_doc");
-    my $cursection = $doc_obj->get_top_section();
-
-    # Add specilised metadata
+    # Add extracted metadata to document object
     foreach my $name (keys %raw) {
     $value = $raw{$name};
@@ -173 +150 @@
         $value = "No $name field";
     }
-    $doc_obj->add_metadata ($cursection, $name, $value);
+    $doc_obj->add_utf8_metadata ($cursection, $name, $value);
     }
 
@@ -179 +156 @@
     $Headers = &text_into_html($Headers);
     $Headers = "No headers" unless ($Headers =~ /\w/);
-    $doc_obj->add_metadata ($cursection, "Headers", $Headers);
-
-    # Add document text
-    $text = &text_into_html($text);
-    $text = "No message" unless ($text =~ /\w/);
-    $doc_obj->add_text ($cursection, $text);
-    
-    # Add the OID - that is, the big HASH value used as a unique ID
-    $doc_obj->set_OID ();
-
-    # Process the document
-    $processor->process($doc_obj);
-
-    # Return the number of documents processed
-    return 1; 
-
+    $doc_obj->add_utf8_metadata ($cursection, "Headers", $Headers);
+
+    # Add text to document object
+    $$textref = &text_into_html($$textref);
+    $$textref = "No message" unless ($$textref =~ /\w/);
+    $doc_obj->add_utf8_text($cursection, $$textref);
+
+    return 1;
 }
 
@@ -213 +182 @@
     my ($text) = @_;
 
-    # Convert problem charaters into HTML symbols
+    # Convert problem characters into HTML symbols
     $text =~ s/&/&amp;/go;
     $text =~ s/</&lt;/go;
@@ -236 +205 @@
 # Perl packages have to return true if they are run.
 1;
-   
-
-
-
-
-
-