Changeset 17318 for gsdl/trunk/bin


Ignore:
Timestamp:
2008-09-18T10:02:41+12:00 (16 years ago)
Author:
kjdon
Message:

tidied this up a bit. Modified the bit using metadata_read so that it works with my fixed OAIPlugin. Also allows multiple records to come back from metadata_read, but this hasn't been tested yet as its not currently used.

File:
1 edited

Legend:

Unmodified
Added
Removed
  • gsdl/trunk/bin/script/explode_metadata_database.pl

    r17301 r17318  
    227227
    228228    my @metadata_records = split(/$split_exp/, $text);
    229     print STDERR "Number of records: " . scalar(@metadata_records) . "\n";
     229    my $total_num_records = scalar(@metadata_records);
     230    print STDERR "Number of records: $total_num_records\n";
    230231   
    231232    # Write the metadata from each record to the metadata.xml file
     
    235236       
    236237        # Check if we need to start a new directory for these records
    237         check_need_new_directory($exploded_base_dir,$record_number,$records_per_folder,
    238                      \@metadata_records,\$documents_directory);
     238        check_need_new_directory($exploded_base_dir,$record_number,
     239                     $records_per_folder,$total_num_records,
     240                     \$documents_directory);
    239241        # Use the plugin's process function to avoid duplicating code
    240242        my $doc_obj = new doc($filename, "nonindexed_doc");
     
    249251               
    250252       
    251         check_close_directory($record_number,$records_per_folder,\@metadata_records);
     253        check_close_directory($record_number,$records_per_folder,$total_num_records);
    252254       
    253255        $record_number = $record_number + 1;
     
    255257    }
    256258    else {
    257     # Call metadata_read to sets up associated metadata
     259    # Call metadata_read to set up associated metadata
    258260
    259261    my $pluginfo = undef;
     
    264266    my $gli = undef;
    265267
    266     my $extrametakeys = {};
     268    my $extrametakeys = [];
    267269    my $extrametadata = {};
    268270
     
    271273                $extrametakeys, $extrametadata, $processor, $maxdocs, $gli);
    272274
    273 
    274     my $documents_directory = need_new_directory($exploded_base_dir);
    275 
    276     # Attach metadata to object
    277     # => use the plugin's extra_metadata function to avoid duplicating code
    278     my $doc_obj = new doc($filename, "nonindexed_doc");
    279 
    280     $plugobj->extra_metadata ($doc_obj, $doc_obj->get_top_section(), $extrametadata);
    281 
    282     # Try to get a doc to attach the metadata to
    283     # If no match found, create a dummy .nul file
    284     attach_metadata_or_make_nul_doc($document_field, $doc_obj, undef,
    285                     $documents_directory, $orig_base_dir,
    286                     $document_prefix, $document_suffix, $metadata_set, $verbosity);
    287 
    288 
    289     close_directory();
    290     }
    291 
     275    my $total_num_records = scalar (@$extrametakeys);
     276    print STDERR "Number of records: $total_num_records\n";
     277    my $record_number = 1;
     278    my $documents_directory;
     279    foreach my $record (@$extrametakeys) {
     280        &check_need_new_directory($exploded_base_dir, $record_number, $records_per_folder, $total_num_records, \$documents_directory);
     281       
     282        # Attach metadata to object
     283        # => use the plugin's extra_metadata function to avoid duplicating code
     284        my $doc_obj = new doc($filename, "nonindexed_doc");
     285        # all the metadata has been extracted into extrametadata
     286        $plugobj->extra_metadata ($doc_obj, $doc_obj->get_top_section(), $extrametadata->{$record});
     287
     288        # Try to get a doc to attach the metadata to
     289        # If no match found, create a dummy .nul file
     290        attach_metadata_or_make_nul_doc($document_field, $doc_obj, $record_number, $documents_directory, $orig_base_dir, $document_prefix, $document_suffix, $metadata_set, $verbosity);
     291       
     292        &check_close_directory($record_number,$records_per_folder,$total_num_records);
     293       
     294        $record_number = $record_number + 1;
     295
     296    }
     297    }
    292298
    293299    # Explode means just that: the original file is deleted
     
    326332sub check_need_new_directory
    327333{
    328     my ($exploded_base_dir,$record_number, $records_per_folder,$metadata_records,
    329     $documents_dir_ref) = @_;
     334    my ($exploded_base_dir,$record_number, $records_per_folder,
     335    $total_num_records, $documents_dir_ref) = @_;
    330336   
    331337
     
    334340    my $documents_directory = $exploded_base_dir;
    335341
    336     if (scalar(@$metadata_records) > $records_per_folder) {
     342    if ($total_num_records > $records_per_folder) {
    337343        $documents_directory .= "." . sprintf("%8.8d", $record_number);
    338344    }
     
    366372        my $document_file_full = $document_prefix . $value . $document_suffix;
    367373
     374        # this either downloads/copies the document, or creates a nul file for it.
    368375        $document_file = &obtain_document($document_file_full, $documents_directory, $orig_base_dir, $verbosity);
    369376        &write_metadata_xml_file_entry(METADATA_XML_FILE, $document_file, $record_metadata, $metadata_set);
    370377        }
    371378    }
    372     }
    373    
    374     # Create a dummy .nul file if we haven't obtained any documents for this record
     379    }
     380   
     381    # Create a dummy .nul file if we haven't obtained a document (or null file) for this record
    375382    if (not defined $document_file) {
    376383
     
    399406sub check_close_directory
    400407{
    401     my ($record_number,$records_per_folder,$metadata_records) = @_;
    402 
    403     if (($record_number % $records_per_folder) == 0 || $record_number == scalar(@$metadata_records)) {
     408    my ($record_number,$records_per_folder,$total_num_records) = @_;
     409
     410    if (($record_number % $records_per_folder) == 0 || $record_number == $total_num_records) {
    404411    # Finish and close the metadata.xml file
    405412    close_directory();
     
    511518    $local_document_file = &util::filename_cat($documents_directory, $document_file_name);
    512519
    513     &util::cp($document_file_full, $documents_directory);
    514 
     520    if (-e $document_file_full) {
     521        &util::cp($document_file_full, $documents_directory);
     522    }
     523   
    515524    # Check the document was obtained successfully
    516525    if (!-e $local_document_file) {
     
    531540    open(NULL_FILE, ">$local_document_file.nul");
    532541    close(NULL_FILE);
     542    print STDERR "Creating a nul document $document_file_name\n";
    533543    }
    534544
Note: See TracChangeset for help on using the changeset viewer.