Changeset 15871 for gsdl/trunk/perllib/plugins/SplitTextFile.pm

Timestamp:

2008-06-05T09:26:56+12:00 (16 years ago)

Author:

kjdon

Message:

plugin overhaul: Split plug renamed to SplitTextFile, XMLPlug renamed to ReadXMLFile, ConvertToPlug renamed to ConvertBinaryFile. With the exception of BasePlugin, only 'real' plugins (top level ones) are named xxPlugin.

File:

• gsdl/trunk/perllib/plugins/SplitTextFile.pm (modified) (12 diffs)

gsdl/trunk/perllib/plugins/SplitTextFile.pm

@@ -1 +1 @@
 ###########################################################################
 #
-# SplitPlug.pm - a plugin for splitting input files into segments that
+# SplitTextFile.pm - a plugin for splitting input files into segments that
 #                will then be individually processed.
 #
@@ -29 +29 @@
 
 
-# SplitPlug is a plugin for splitting input files into segments that will
+# SplitTextFile is a plugin for splitting input files into segments that will
 # then be individually processed.  
 
@@ -35 +35 @@
 # process input files that contain several documents, you should write a
 # plugin with a process function that will handle one of those documents
-# and have it inherit from SplitPlug.  See ReferPlug for an example.
-
-
-package SplitPlug;
-
-use BasPlug;
+# and have it inherit from SplitTextFile.  See ReferPlug for an example.
+
+
+package SplitTextFile;
+
+use ReadTextFile;
 use gsprintf 'gsprintf';
 use util;
@@ -47 +47 @@
 no strict 'refs'; # allow filehandles to be variables and viceversa
 
-# SplitPlug is a sub-class of BasPlug.
+# SplitTextFile is a sub-class of BasPlug.
 sub BEGIN {
-    @SplitPlug::ISA = ('BasPlug');
+    @SplitTextFile::ISA = ('ReadTextFile');
 }
 
@@ -55 +55 @@
 my $arguments =
     [ { 'name' => "split_exp",
-    'desc' => "{SplitPlug.split_exp}",
+    'desc' => "{SplitTextFile.split_exp}",
     'type' => "regexp",
     #'deft' => &get_default_split_exp(),
@@ -61 +61 @@
     'reqd' => "no" } ];
 
-my $options = { 'name'     => "SplitPlug",
-        'desc'     => "{SplitPlug.desc}",
+my $options = { 'name'     => "SplitTextFile",
+        'desc'     => "{SplitTextFile.desc}",
         'abstract' => "yes",
         'inherits' => "yes",
@@ -73 +73 @@
     push(@$pluginlist, $class);
 
-    if(defined $arguments){ push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});}
-    if(defined $options) { push(@{$hashArgOptLists->{"OptList"}},$options)};
-
-    my $self = new BasPlug($pluginlist, $inputargs, $hashArgOptLists);
+    push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
+    push(@{$hashArgOptLists->{"OptList"}},$options);
+
+    my $self = new ReadTextFile($pluginlist, $inputargs, $hashArgOptLists);
 
     $self->{'textcat_store'} = {};
@@ -87 +87 @@
     my ($verbosity, $outhandle, $failhandle) = @_;
 
-    $self->BasPlug::init($verbosity, $outhandle, $failhandle);
-
+    $self->ReadTextFile::init($verbosity, $outhandle, $failhandle);
+
+    # why is this is init and not in new??
     if ((!defined $self->{'process_exp'}) || ($self->{'process_exp'} eq "")) {
 
@@ -119 +120 @@
     my ($pluginfo, $base_dir, $file, $metadata, $extrametakeys, $extrametadata, $processor, $maxdocs, $gli) = @_;
 
+    # returns 1 if matches process_exp, and has done blocking in the meantime
     my $matched = $self->SUPER::metadata_read($pluginfo, $base_dir, $file, 
                           $metadata, $extrametakeys, 
@@ -146 +148 @@
  
     if ($text !~ /\w/) {
-        gsprintf($outhandle, "$plugin_name: {BasPlug.file_has_no_text}\n",
+        gsprintf($outhandle, "$plugin_name: {ReadTextFile.file_has_no_text}\n",
              $file)
         if $self->{'verbosity'};
@@ -171 +173 @@
     }
 
-    print $outhandle "SplitPlug found " . (scalar @segments) . " documents in $filename\n" 
+    print $outhandle "SplitTextFile found " . (scalar @segments) . " documents in $filename\n" 
         if $self->{'verbosity'};
     
@@ -231 +233 @@
     $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Encoding", $encoding);
     my ($filemeta) = $file =~ /([^\\\/]+)$/;
-    $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Source", &ghtml::dmsafe($filemeta));
+    $self->set_Source_metadata($doc_obj, $filemeta, $encoding);
     $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "SourceSegment", "$segment");
     if ($self->{'cover_image'}) {