Changeset 17727 for gsdl/trunk/perllib/plugins/HTMLImagePlugin.pm

Timestamp:

2008-11-06T11:35:21+13:00 (15 years ago)

Author:

kjdon

Message:

W3ImagePlugin renamed to HTMLImagePlugin

File:

• gsdl/trunk/perllib/plugins/HTMLImagePlugin.pm (moved) (moved from gsdl/trunk/perllib/plugins/W3ImagePlugin.pm ) (19 diffs)

gsdl/trunk/perllib/plugins/HTMLImagePlugin.pm

@@ -1 +1 @@
 ###########################################################################
 #
-# W3ImagePlugin.pm -- Context-based image indexing plugin for HTML documents
+# HTMLImagePlugin.pm -- Context-based image indexing plugin for HTML documents
 #
 # A component of the Greenstone digital library software
@@ -39 +39 @@
 #  collection builds at the import stage.
 #
-#  W3ImagePlugin is a subclass of HTMLPlug (i.e. it will index pages also 
+#  HTMLImagePlugin is a subclass of HTMLPlug (i.e. it will index pages also 
 #  if required). It can be used in place of HTMLPlugin to index both
 #  pages and their images.
@@ -55 +55 @@
 #    ImageMagick can be downloaded from the website above.
 #    Make sure the system path includes the ImageMagick binaries
-#    before using W3ImagePlugin.
+#    before using HTMLImagePlugin.
 #
 #    NOTE: NT/2000/XP contain a filesystem utility 'convert.exe' 
@@ -97 +97 @@
 #   ...
 #
-#   plugin W3ImagePlugin -index_pages -aggressiveness 6
+#   plugin HTMLImagePlugin -index_pages -aggressiveness 6
 #
 #   ...
@@ -110 +110 @@
 #
  
-package W3ImagePlugin;
+package HTMLImagePlugin;
 
 use HTMLPlugin;
@@ -120 +120 @@
 
 sub BEGIN {
-    @W3ImagePlugin::ISA = qw( HTMLPlugin );
+    @HTMLImagePlugin::ISA = qw( HTMLPlugin );
 }
 
@@ -218 +218 @@
     'reqd' => "no" } ];
 
-my $options = { 'name'     => "W3ImagePlugin",
+my $options = { 'name'     => "HTMLImagePlugin",
         'desc'     => "{W3ImagePlugin.desc}",
         'abstract' => "no",
@@ -281 +281 @@
 
 # get complex configuration options from configuration files
-# -- $GSDLCOLLECTION/etc/W3ImagePlugin.cfg (tag sets for aggr 2+)
+# -- $GSDLCOLLECTION/etc/HTMLImagePlugin.cfg (tag sets for aggr 2+)
 # -- $GSDLHOME/etc/packages/phind/stopword/en/brown.sw (stopwords for aggr 5+)
 
-# If there's no W3ImagePlugin.cfg file we'll use the following default values
+# If there's no HTMLImagePlugin.cfg file we'll use the following default values
 my $defaultcfg = '
 <delimitertagset>
@@ -323 +323 @@
     my ($filepath);
 
-    print {$self->{'outhandle'}} "W3ImagePlugin: Initialising\n"
+    print {$self->{'outhandle'}} "HTMLImagePlugin: Initialising\n"
     if $self->{'verbosity'} > 1;
-    # etc/W3ImagePlugin.cfg (XML)
+    # etc/HTMLImagePlugin.cfg (XML)
     # tag sets for captions and neartext
     if ( $self->{'aggressiveness'} > 1 && $self->{'aggressiveness'} != 9 ) {
@@ -332 +332 @@
     my ($cfg, @tagsets, $tagset, $type, @delims);
 
-    $filepath = "$collpath/etc/W3ImagePlugin.cfg";
+    $filepath = "$collpath/etc/HTMLImagePlugin.cfg";
     if ( open CFG, "<$filepath" ) {
         while (<CFG>) { $cfg .= $_ }
@@ -354 +354 @@
     # output a warning if there seem to be no delimiters
     if ( scalar(@{$self->{'cdelims'}} == 0)) {
-        print {$self->{'outhandle'}} "W3ImagePlugin: Warning: no caption delimiters found in $filepath\n";
+        print {$self->{'outhandle'}} "HTMLImagePlugin: Warning: no caption delimiters found in $filepath\n";
     }
     if ( scalar(@{$self->{'delims'}} == 0)) {
-        print {$self->{'outhandle'}} "W3ImagePlugin: Warning: no neartext delimiters found in $filepath\n";
+        print {$self->{'outhandle'}} "HTMLImagePlugin: Warning: no neartext delimiters found in $filepath\n";
     }
     }
@@ -373 +373 @@
         close STOPWORDS;
     } else {
-        print {$self->{'outhandle'}} "W3ImagePlugin: Warning: couldn't open stopwords file at $filepath ($!)\n";
+        print {$self->{'outhandle'}} "HTMLImagePlugin: Warning: couldn't open stopwords file at $filepath ($!)\n";
     }
     
@@ -380 +380 @@
     if ( $self->{'neartext_length'} > $self->{'max_near_text'} ) {
     $self->{'max_near_text'} = $self->{'neartext_length'} * 1.33;
-    print {$self->{'outhandle'}} "W3ImagePlugin: Warning: adjusted max_text to $self->{'max_near_text'}\n";
+    print {$self->{'outhandle'}} "HTMLImagePlugin: Warning: adjusted max_text to $self->{'max_near_text'}\n";
     } 
     if ( $self->{'caption_length'} > $self->{'max_near_text'} ) {
     $self->{'max_near_text'} = $self->{'caption_length'} * 1.33;
-    print {$self->{'outhandle'}} "W3ImagePlugin: Warning: adjusted max_text to $self->{'max_near_text'}\n";
+    print {$self->{'outhandle'}} "HTMLImagePlugin: Warning: adjusted max_text to $self->{'max_near_text'}\n";
     }
 
@@ -420 +420 @@
         ($imgtag) = ($context =~ /(<(?:img|a|body)\s[^>]*$filepath[^>]*>)/is );
         if (! defined($imgtag)) { $imgtag = $filepath }
-        print $outhandle "W3ImagePlugin: extracting $filepath\n"
+        print $outhandle "HTMLImagePlugin: extracting $filepath\n"
         if ( $self->{'verbosity'} > 1 );
         $doc_obj = new doc ("", "indexed_doc");
@@ -434 +434 @@
     return $numdocs;
     } else {
-    print $outhandle "W3ImagePlugin: No images from $file indexed\n"
+    print $outhandle "HTMLImagePlugin: No images from $file indexed\n"
         if ( $self->{'verbosity'} > 2 );
     return 1;
@@ -473 +473 @@
     `convert -flatten -filter Hanning $self->{'convert_params'} -geometry "$self->{'thumb_size'}x$self->{'thumb_size'}>" $filepath $thumbfp` unless -e $thumbfp;
     if ( ! (-e $thumbfp) ) {
-    print STDERR "W3ImagePlugin: 'convert' failed. Check ImageMagicK binaries are installed and working correctly\n"; return 0; 
+    print STDERR "HTMLImagePlugin: 'convert' failed. Check ImageMagicK binaries are installed and working correctly\n"; return 0; 
     }
     
@@ -854 +854 @@
     } elsif ( $bestlen[$best1] < $mintext ) {
     # use plain text extraction if tags failed (e.g. usable tag outside context)
-    print {$self->{'outhandle'}} "W3ImagePlugin: Fallback to plain-text extraction for $tag\n" 
+    print {$self->{'outhandle'}} "HTMLImagePlugin: Fallback to plain-text extraction for $tag\n" 
         if $self->{'verbosity'} > 2;
     $neartext[0] = "<tr><td>RawNeartext</td><td>" . $self->extract_raw_neartext($tag, $textref) . "</td></tr>";
@@ -986 +986 @@
         `identify $abspath -ping -format "%wx%h"` =~ /^(\d*)x(\d*)$/m;
     if (! ($width && $height)) { 
-        print STDERR "W3ImagePlugin: ($abspath) 'identify' failed. Check ImageMagicK binaries are installed and working correctly\n"; next;
+        print STDERR "HTMLImagePlugin: ($abspath) 'identify' failed. Check ImageMagicK binaries are installed and working correctly\n"; next;
     }
     $filesize = (-s $abspath);
@@ -999 +999 @@
        $imgs->{$filepath}{'filesize'} = $filesize;
        } else {
-       print {$self->{'outhandle'}} "W3ImagePlugin: skipping $self->{'base_path'}/$relpath: $filesize, $width x $height\n" 
+       print {$self->{'outhandle'}} "HTMLImagePlugin: skipping $self->{'base_path'}/$relpath: $filesize, $width x $height\n" 
            if $self->{'verbosity'} > 2;
        }