Changeset 24487

Timestamp:

2011-08-26T09:39:01+12:00 (13 years ago)

Author:

sjm84

Message:

Added two new options in EmbeddedMetadataPlugin for filtering the embedded sets that are returned

Location:

main/trunk/greenstone2/perllib

Files:

• plugins/EmbeddedMetadataPlugin.pm (modified) (9 diffs)
• strings.properties (modified) (1 diff)

main/trunk/greenstone2/perllib/plugins/EmbeddedMetadataPlugin.pm

@@ -75 +75 @@
     'list' => [{'name' => "true", 'desc' => "{common.true}"}, {'name' => "false", 'desc' => "{common.false}"}],
     'deft' => "true"
+    },{
+    'name' => "set_filter_list",
+    'desc' => "{EmbeddedMetadataPlugin.set_filter_list}",
+    'type' => "string"
+    },{
+    'name' => "set_filter_regexp",
+    'desc' => "{EmbeddedMetadataPlugin.set_filter_regexp}",
+    'type' => "string",
+    'deft' => ".*" #If changing this default, also need to update the constructor
     }];
 
@@ -102 +111 @@
     $exifTool->Options('Verbose');
     $self->{'exiftool'} = $exifTool;
+    
+    my $setFilterList = $self->{'set_filter_list'};
+    my $setFilterRegexp = $self->{'set_filter_regexp'};
+    if ((defined $setFilterList) && ($setFilterList ne ""))
+    {
+        if ((defined $setFilterRegexp) && ($setFilterRegexp ne ".*") && ($setFilterRegexp ne ""))
+        {
+            my $outhandle = $self->{'outhandle'};
+            print $outhandle "Warning: can only specify 'set_filter_list' or 'set_filter_regexp'\n";
+            print $outhandle "         defaulting to 'set_filter_list'\n";
+        }
+
+        my @sets = split(/,/,$setFilterList);
+        my @sets_bracketed;
+        foreach my $s (@sets)
+        {
+            $s =~ s/^(ex\.)?(.*)$/(ex.$2)/;
+            push (@sets_bracketed, $s);
+        }
+
+        my $setFilterRegexp = join("|",@sets_bracketed);
+        $self->{'set_filter_regexp'} = $setFilterRegexp;
+    }
 
     return bless $self, $class;
@@ -145 +177 @@
 }
 
+sub checkAgainstFilters
+{
+    my $self = shift(@_);
+    my $name = shift(@_);
+    
+    my $setFilterRegexp = $self->{'set_filter_regexp'};
+    if((defined $setFilterRegexp) && ($setFilterRegexp ne ""))
+    {
+        return ($name =~ m/($setFilterRegexp)/i);
+    }
+    else
+    {
+        return 1;
+    }
+}
 
 sub extractEmbeddedMetadata()
@@ -215 +262 @@
                                 foreach my $val (@vs) {
                                     if ($val =~ /\S/) {
-                                        push (@{$exif_metadata{$field}}, $self->gsSafe($val));
+                                        push (@{$exif_metadata{$field}}, $self->gsSafe($val)) if $self->checkAgainstFilters($field);
                                         ++$metadata_count;
                                     }
@@ -222 +269 @@
                             else
                             {
-                                push (@{$exif_metadata{$field}}, $self->gsSafe($v));
+                                push (@{$exif_metadata{$field}}, $self->gsSafe($v)) if $self->checkAgainstFilters($field);
                                 ++$metadata_count;
                             }
@@ -239 +286 @@
                             foreach my $val (@vs) {
                                 if ($val =~ /\S/) {
-                                    push (@{$exif_metadata{$field}}, $self->gsSafe($val));
+                                    push (@{$exif_metadata{$field}}, $self->gsSafe($val)) if $self->checkAgainstFilters($field);
                                     ++$metadata_count;
                                 }
@@ -246 +293 @@
                         else
                         {
-                            push (@{$exif_metadata{$field}}, $self->gsSafe($allvals));
+                            push (@{$exif_metadata{$field}}, $self->gsSafe($allvals)) if $self->checkAgainstFilters($field);
                             ++$metadata_count;
                         }
@@ -258 +305 @@
                         foreach my $v (@vs) {
                             if ($v =~ /\S/) {
-                                push (@{$exif_metadata{$field}}, $self->gsSafe($v));
+                                push (@{$exif_metadata{$field}}, $self->gsSafe($v)) if $self->checkAgainstFilters($field);
                                 ++$metadata_count; 
                             }
@@ -265 +312 @@
                 }
                 if (!$metadata_done) {
-                    push (@{$exif_metadata{$field}}, $self->gsSafe($value));
+                    push (@{$exif_metadata{$field}}, $self->gsSafe($value)) if $self->checkAgainstFilters($field);
                     ++$metadata_count;
                 }

main/trunk/greenstone2/perllib/strings.properties

@@ -866 +866 @@
 EmbeddedMetadataPlugin.join_character:The character to use with join_before_split (default is a single space).
 EmbeddedMetadataPlugin.trim_whitespace:Trim whitespace from start and end of any extracted metadata values (Note: this also applies to any values generated through joining with join_before_split or splitting through metadata_field_separator).
+EmbeddedMetadataPlugin.set_filter_list:A comma-separated list of the metadata sets we would like to retrieve.
+EmbeddedMetadataPlugin.set_filter_regexp:A regular expression that filters out the metadata we would like to retrieve.
 
 ExcelPlugin.desc:A plugin for importing Microsoft Excel files (versions 95 and 97).