Changeset 16392 for gsdl/trunk/perllib/plugins/SplitTextFile.pm

Timestamp:

2008-07-14T14:57:38+12:00 (16 years ago)

Author:

kjdon

Message:

global block pass: read_block is no more, use can_process_this_file to see whether a file is for us or not. extra arg (block_hash) to read, read_into_doc_obj, metadata_read etc

File:

• gsdl/trunk/perllib/plugins/SplitTextFile.pm (modified) (4 diffs)

gsdl/trunk/perllib/plugins/SplitTextFile.pm

@@ -118 +118 @@
 sub metadata_read {
     my $self = shift (@_);  
-    my ($pluginfo, $base_dir, $file, $metadata, $extrametakeys, $extrametadata, $processor, $maxdocs, $gli) = @_;
+    my ($pluginfo, $base_dir, $file, $block_hash, $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, 
+    my $matched = $self->SUPER::metadata_read($pluginfo, $base_dir, $file,
+                          $block_hash, 
                           $metadata, $extrametakeys, 
                           $extrametadata, $processor, 
@@ -192 +193 @@
 sub read {
     my $self = shift (@_);
-    my ($pluginfo, $base_dir, $file, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
+    my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
     my $outhandle = $self->{'outhandle'};
     my $verbosity = $self->{'verbosity'};
 
-    #check process and block exps, smart block, etc
-    my ($block_status,$filename) = $self->read_block(@_);    
-    return $block_status if ((!defined $block_status) || ($block_status==0));
+    # can we process this file??
+    my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
+    return undef unless $self->can_process_this_file($filename_full_path);
 
     $file =~ s/^[\/\\]+//; # $file often begins with / so we'll tidy it up
@@ -232 +233 @@
 
     # create a new document
-    my $doc_obj = new doc ($filename, "indexed_doc");
+    my $doc_obj = new doc ($filename_full_path, "indexed_doc");
     $doc_obj->set_OIDtype ($processor->{'OIDtype'}, $processor->{'OIDmetadata'});
     $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Language", $language);
@@ -240 +241 @@
     $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "SourceSegment", "$segment");
     if ($self->{'cover_image'}) {
-        $self->associate_cover_image($doc_obj, $filename);
+        $self->associate_cover_image($doc_obj, $filename_full_path);
     }
     $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Plugin", "$self->{'plugin_type'}");