Changeset 1244 for trunk/gsdl/perllib/plugins/TEXTPlug.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/TEXTPlug.pm (modified) (2 diffs)

trunk/gsdl/perllib/plugins/TEXTPlug.pm

@@ -24 +24 @@
 ###########################################################################
 
-# creates simple single-level document from .txt or .text files 
-# (case-insensitive match on filenames). Adds Title metadata 
-# of first 100 characters found.
+# creates simple single-level document. Adds Title metadata 
+# of first line of text (up to 100 characters long).
 
 package TEXTPlug;
 
 use BasPlug;
-use sorttools;
 
 sub BEGIN {
@@ -37 +35 @@
 }
 
+use strict;
+
 sub new {
     my ($class) = @_;
-    $self = new BasPlug ();
+    my $self = new BasPlug ("TEXTPlug", @_);
 
     return bless $self, $class;
 }
 
-sub is_recursive {
+sub get_default_process_exp {
     my $self = shift (@_);
 
-    return 0; # this is not a recursive plugin
+    return q^(?i)\.te?xt$^;
 }
 
-
-# return number of files processed, undef if can't process
-# Note that $base_dir might be "" and that $file might 
-# include directories
-sub read {
+# do plugin specific processing of doc_obj
+sub process {
     my $self = shift (@_);
-    my ($pluginfo, $base_dir, $file, $metadata, $processor) = @_;
-
-    my $filename = &util::filename_cat($base_dir, $file);
-
-    return undef unless ($filename =~ /\.(te?xt(\.gz)?)$/i && (-e $filename));
-
-    my $gz = 0;
-    if (defined $2) {
-    $gz = $2;
-    $gz = 1 if ($gz =~ /\.gz/i);
-    }
-
-    print STDERR "TEXTPlug: processing $filename\n" if $processor->{'verbosity'};
-
-    # create a new document
-    my $doc_obj = new doc ($file, "indexed_doc");
-
-    if ($gz) {
-    open (FILE, "zcat $filename |") || die "TEXTPlug::read - zcat can't open $filename\n";
-    } else {
-    open (FILE, $filename) || die "TEXTPlug::read - can't open $filename\n";
-    }
+    my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj) = @_;
+    
+    print STDERR "TEXTPlug: processing $file\n" 
+    if $self->{'verbosity'} > 1;
+    
     my $cursection = $doc_obj->get_top_section();
-
-    my $text = "";
-    my $line = "";
-    my $foundtitle = 0;
-    # don't need to get title if it has been passed
-    # in from another plugin
-    if (defined $metadata->{'Title'}) {
-    $foundtitle = 1;
-    }
-    while (defined ($line = <FILE>)) {
-    # use first line as title (or first 100 characters if it's long)
-    if (!$foundtitle && length($line) > 5) {
-        my $title = "";
-        if (length($line) > 100) {
-        $title = substr ($line, 0, 100);
-        } else {
-        $title = $line;
-        }
-        $doc_obj->add_metadata ($cursection, "Title", $title);
-        $foundtitle = 1;
+    
+    # get title metadata
+    # (don't need to get title if it has been passed
+    # in from another plugin)
+    if (!defined $metadata->{'Title'}) {
+    my ($title) = $$textref =~ /^([^\n]*)/;
+    if (length($title) > 100) {
+        $title = substr ($title, 0, 100);
     }
-    $text .= $line;
+    $doc_obj->add_utf8_metadata ($cursection, "Title", $title);
     }
     
-    $doc_obj->add_text ($cursection, "<pre>\n$text\n</pre>");
+    # insert preformat tags and add text to document object
+    $doc_obj->add_utf8_text($cursection, "<pre>\n$$textref\n</pre>");
 
-
-    foreach $field (keys(%$metadata)) {
-    # $metadata->{$field} may be an array reference
-    if (ref ($metadata->{$field}) eq "ARRAY") {
-        map { 
-        $doc_obj->add_metadata ($cursection, $field, $_); 
-        } @{$metadata->{$field}};
-    } else {
-        $doc_obj->add_metadata ($cursection, $field, $metadata->{$field}); 
-    }
-    }
-
-    # add OID
-    $doc_obj->set_OID ();
-
-    # process the document
-    $processor->process($doc_obj);
-
-    return 1; # processed the file
+    return 1;
 }