Changeset 10338

Timestamp:

2005-07-28T11:31:20+12:00 (19 years ago)

Author:

kjdon

Message:

tidied this up: uses new parse2, use strict

File:

• trunk/gsdl/bin/script/explode_metadata_database.pl (modified) (13 diffs)

trunk/gsdl/bin/script/explode_metadata_database.pl

@@ -8 +8 @@
 }
 
-
-use parsargv;
+use strict;
+no strict 'subs'; # allow barewords (eg STDERR) as function arguments
+no strict 'refs'; # allow filehandles to be variables and vice versa
 use printusage;
+use parse2;
+
 my $unicode_list =
     [ { 'name' => "auto",
@@ -22 +25 @@
 
 my $arguments = 
-    [ { 'name' => "input_encoding",
+    [ 
+      { 'name' => "plugin",
+    'desc' => "{explode.plugin}",
+    'type' => "string",
+    'reqd' => "yes",
+    'hiddengli' => "yes"},
+      { 'name' => "input_encoding",
     'desc' => "{explode.encoding}",
     'type' => "enum",
@@ -33 +42 @@
     'reqd' => "no",
     'hiddengli' => "yes"},
-      { 'name' => "plugin",
-    'desc' => "{explode.plugin}",
-    'type' => "string",
-    'reqd' => "yes",
-    'hiddengli' => "yes"},
       { 'name' => "document_field",
     'desc' => "{explode.document_field}",
@@ -53 +57 @@
     'desc' => "{explode.filename_field}",
     'type' => "string",
-    'reqd' => "no"}     
+    'reqd' => "no"},
+      { 'name' => "verbosity",
+    'desc' => "{import.verbosity}",
+    'type' => "int",
+    'range' => "0,",
+    'deft' => "1",
+    'reqd' => "no",
+    'modegli' => "4" },
+      { 'name' => "xml",
+    'desc' => "",
+    'type' => "flag",
+    'reqd' => "no",
+    'hiddengli' => "yes" }
       ];
     
@@ -63 +79 @@
 sub main
 {
-    my ($encoding, $metadata_set, $plugin, $filename_field, 
-    $document_field, $document_prefix, $document_suffix);
+    my ($language, $input_encoding, $metadata_set, $plugin, $filename_field, 
+    $document_field, $document_prefix, $document_suffix, $verbosity);
 
     my $xml = 0;
-    # Parse command line arguments
-    if (!parsargv::parse(\@ARGV, 
-             'language/.*/', \$language,
-             'input_encoding/.*/auto', \$encoding,
-             'metadata_set/.*/', \$metadata_set,
-             'plugin/.*/', \$plugin,
-             'filename_field/.*/', \$filename_field,
-             'document_field/.*/', \$document_field,
-             'document_prefix/.*/', \$document_prefix,
-             'document_suffix/.*/', \$document_suffix,
-             q^xml^, \$xml)) {
+
+    my $hashParsingResult = {};
+    my $blnParseFailed = "false";
+    # parse the options
+    my $intArgLeftinAfterParsing = parse2::parse(\@ARGV,$arguments,$hashParsingResult,"allow_extra_options");
+    # There should one arg left after parsing
+    if($intArgLeftinAfterParsing > 1)
+    {
     &PrintUsage::print_txt_usage($options, "{explode.params}");
     die "\n";
     }
-    
+
+    foreach my $strVariable (keys %$hashParsingResult)
+    {
+    eval "\$$strVariable = \$hashParsingResult->{\"\$strVariable\"}";
+    }
+        
     # If $language has been specified, load the appropriate resource bundle
     # (Otherwise, the default resource bundle will be loaded automatically)
-    if ($language) {
+    if ($language && $language =~ /\S/) {
     &gsprintf::load_language_specific_resource_bundle($language);
     }
@@ -121 +139 @@
 
     #check filename field
-    if (defined $fileanme_field && $filenmae_field eq "") {
+    if (defined $filename_field && $filename_field eq "") {
     undef $filename_field;
     }
@@ -132 +150 @@
     $plugobj->init(1, "STDERR", "STDERR");
     
-    if ($encoding eq "auto") {
-    $plugobj->{'input_encoding'} = $encoding;   
-    (my $language, $encoding) = $plugobj->textcat_get_language_encoding ($filename);
-    }
+    if ($input_encoding eq "auto") {
+    $plugobj->{'input_encoding'} = $input_encoding;   
+    ($language, $input_encoding) = $plugobj->textcat_get_language_encoding ($filename);
+    }
+    my $text = "";
     # Use the plugin's read_file function to avoid duplicating code
-    $plugobj->read_file($filename, $encoding, undef, \$text);
+    $plugobj->read_file($filename, $input_encoding, undef, \$text);
 
     # Create a directory to store the document files...
@@ -147 +166 @@
 
     # ...and a metadata.xml file for the document metadata (extracted from the database)
-    $documents_metadata_xml_file = &util::filename_cat($documents_directory, "metadata.xml");
+    my $documents_metadata_xml_file = &util::filename_cat($documents_directory, "metadata.xml");
     if (-e $documents_metadata_xml_file) {
     die "Error: document metadata.xml file $documents_metadata_xml_file already exists (bailing).\n";
@@ -166 +185 @@
     # Write the metadata from each record to the metadata.xml file
     my $record_number = 0;
-    foreach $record_text (@metadata_records) {
+    foreach my $record_text (@metadata_records) {
     
     # Use the plugin's process function to avoid duplicating code
@@ -183 +202 @@
         if ($field eq $document_field) {
             my $document_file_full = $document_prefix . $value . $document_suffix;
-            $document_file = &obtain_document($self, $document_file_full, $documents_directory);
+            $document_file = &obtain_document($document_file_full, $documents_directory, $verbosity);
         }
         }
@@ -191 +210 @@
         # try to get a file name
         if (defined $filename_field) {
+        
         my $meta = $doc_obj->get_metadata_element($doc_obj->get_top_section(), $filename_field);
         if (defined $meta) {
@@ -284 +304 @@
 sub obtain_document
 {
-    my $self = shift(@_);
     my $document_file_full = shift(@_);
     my $documents_directory = shift(@_);
-
-    my $outhandle = $self->{'outhandle'};
-    print $outhandle "Obtaining document file $document_file_full...\n"
-    if ($self->{'verbosity'} > 1);
+    my $verbosity = shift(@_);
+    
+    print STDERR "Obtaining document file $document_file_full...\n" if ($verbosity > 1);
 
     my $document_file_name;
@@ -302 +320 @@
 
     my $wget_options = "--quiet";
-    $wget_options = "--verbose" if ($self->{'verbosity'} > 2);
+    $wget_options = "--verbose" if ($verbosity > 2);
     $wget_options .= " --timestamping";  # Only re-download files if they're newer
     `wget $wget_options $document_file_full --output-document $local_document_file`;