Changeset 13198 for trunk/gsdl/perllib/plugins/MARCPlug.pm

Timestamp:

2006-10-30T16:18:23+13:00 (17 years ago)

Author:

shaoqun

Message:

now this plugin can handle ascii marc files

File:

• trunk/gsdl/perllib/plugins/MARCPlug.pm (modified) (7 diffs)

trunk/gsdl/perllib/plugins/MARCPlug.pm

@@ -80 +80 @@
     my $self = new SplitPlug($pluginlist, $inputargs, $hashArgOptLists);
 
+    $self->{'type'} = "";
+    
     return bless $self, $class;
 }
@@ -150 +152 @@
 sub get_default_split_exp {
     # \r\n for msdos eol, \n for unix
-    return q^\r?\n\s*\r?\n^;
+    return q^\r?\n\s*\r?\n|\[\w+\]Record type: USmarc^;
 }
 
@@ -181 +183 @@
     $self->{'readfile_encoding'}->{$filename} = $encoding;
 
-    my @marc_entries = ();
-
+       
     if (!-r $filename)
     {
@@ -190 +191 @@
     }
 
+    ##handle ascii marc
+    #test whether this is ascii marc file 
+    if (open (FILE, $filename)) {
+    while (defined (my $line = <FILE>)) {
+        $$textref .= $line;
+       if ($line =~ /\[\w+\]Record type:/){
+           undef $/;
+           $$textref .= <FILE>;
+           $/ = "\n";
+           $self->{'type'} = "ascii";
+           close FILE;
+           return;
+       }
+    }
+    close FILE; 
+    }
+
+      
+    $$textref = "";
+    my @marc_entries = ();
+ 
     my $batch = new MARC::Batch( 'USMARC', $filename );
     while ( my $marc = $batch->next ) 
     {
-    push(@marc_entries,$marc);
+        push(@marc_entries,$marc);
     $$textref .= $marc->as_formatted();
     $$textref .= "\n\n"; # for SplitPlug - see default_split_exp above...
@@ -231 +253 @@
 
     my $encoding = $self->{'readfile_encoding'}->{$filename};
-    $self->extract_metadata ($marc, $metadata, $encoding, $doc_obj, $cursection);
+
+    if ($self->{'type'} ne "ascii" ){   
+        $self->extract_metadata ($marc, $metadata, $encoding, $doc_obj, $cursection);
+    }
+    else{
+    $self->extract_ascii_metadata ($$textref,$metadata,$doc_obj, $cursection);
+    }
 
     # add spaces after the sub-field markers, for word boundaries
@@ -290 +318 @@
     my $outhandle = $self->{'outhandle'};
 
+    if (!defined $marc){
+    return;
+    }
+
     my $metadata_mapping = $self->{'metadata_mapping'};
     my $mm;
+
     foreach $mm ( @$metadata_mapping )
     {
     my $marc_field = $mm->{'marc'};
+
     my @metavalues = $marc->field($marc_field);
 
@@ -309 +343 @@
     }
 }
+
+sub extract_ascii_metadata 
+{
+    my $self = shift (@_);
+    my ($text, $metadata,$doc_obj, $section) = @_;
+    my $outhandle = $self->{'outhandle'};
+    my $metadata_mapping = $self->{'metadata_mapping'};
+    ## get fields
+    my @fields = split(/[\n\r]+/,$text);
+    my $marc_mapping ={};
+
+    foreach my $field (@fields){
+    if ($field ne ""){
+        $field =~ /^(\d\d\d)\s/;
+        my $code = $1;
+        $field = $';
+        ##get subfields
+        my @subfields = split(/\$/,$field);
+        my $i=0;
+        $marc_mapping->{$code} = [];  
+        foreach my $subfield (@subfields){
+        if ($i == 0){
+            ##print STDERR $subfield."\n";
+            push(@{$marc_mapping->{$code}},"info");
+            push(@{$marc_mapping->{$code}},$subfield);
+                 
+            $i++;
+        } 
+         else{
+             $subfield =~ /(\w)\s/;
+             ##print STDERR "$1=>$'\n";
+             push(@{$marc_mapping->{$code}},$1);
+                     push(@{$marc_mapping->{$code}},$'); 
+         }
+        }
+    }
+    }
+
+
+     foreach my $mm ( @$metadata_mapping )
+    {
+    my $marc_field = $mm->{'marc'};
+        
+    my $matched_field = $marc_mapping->{$marc_field};
+    my $subfield = undef;
+
+    if (defined $matched_field){
+        ## test whether this field has subfield
+        if ($marc_field =~ /\d\d\d(\w)/){
+        $subfield = $1;
+        }
+        my $metaname = $mm->{'gsdl'};
+ 
+        my $metavalue;
+        if (defined $subfield){
+        my %mapped_subfield = {@$matched_field};
+        $metavalue = $mapped_subfield{$subfield};
+        }
+        else{ ## get all values except info
+        my $i =0;
+        foreach my $value (@$matched_field){
+            if ($i%2 != 0 and $i != 1){
+            $metavalue .= $value." ";
+            }
+            $i++;
+        }
+        }
+        
+        ## escape [ and ]
+        $metavalue =~ s/\[/\\\[/g;
+        $metavalue =~ s/\]/\\\]/g;
+        ##print STDERR  "$metaname=$metavalue\n";
+        $doc_obj->add_metadata ($section, $metaname, $metavalue) ;          
+    }
+       
+    }
+
+}
+
+
 1;