Changeset 16692 for gsdl/trunk/perllib/plugins/MARCXMLPlugin.pm

Timestamp:

2008-08-11T10:44:04+12:00 (16 years ago)

Author:

kjdon

Message:

code to read in marc mapping files moved from MARCXMLPlugin to marcmapping.pm, and its now also used by MARCPlugin.pm so that MARCplugin can use qualified dublin core

File:

• gsdl/trunk/perllib/plugins/MARCXMLPlugin.pm (modified) (5 diffs)

gsdl/trunk/perllib/plugins/MARCXMLPlugin.pm

@@ -31 +31 @@
 
 use ReadXMLFile;
+use marcmapping;
 
 use strict;
@@ -85 +86 @@
 
 
-sub _parse_marc_metadata_mapping
-{
-    my $self = shift(@_);
-    my ($mm_file,$metadata_mapping) = @_;
-
-    my $outhandle = $self->{'outhandle'};
-
-    if (open(MMIN, "<$mm_file"))
-    {
-    my $l=0;
-    my $line;
-    while (defined($line=<MMIN>))
-    {
-        $l++;
-        chomp $line;
-        $line =~ s/#.*$//; # strip out any comments, including end of line
-        next if ($line =~ m/^\s*$/); 
-        $line =~ s/\s+$//; # remove any white space at end of line
-
-        my $parse_error_count = 0;
-        if ($line =~ m/^-(\d+)\s*$/) {
-        # special "remove" rule syntax
-        my $marc_info = $1;
-        if (defined $metadata_mapping->{$marc_info}) {
-            delete $metadata_mapping->{$marc_info};
-        }
-        else {
-            print $outhandle "Parse Warning: Did not file pre-existing rule $marc_info to remove";
-            print $outhandle " on line $l of $mm_file:\n";
-            print $outhandle "  $line\n";
-        }
-        }
-        elsif ($line =~ m/^(.*?)->\s*([\w\^]+)$/)
-        {
-        my $lhs = $1;
-        my $gsdl_info = $2;
-
-        my @fields = split(/,\s*/,$lhs);
-        my $f;
-        while ($f  = shift (@fields)) {
-            $f =~ s/\s+$//; # remove any white space at end of line
-
-            if ($f =~ m/^(\d+)\-(\d+)$/) {
-            # number range => genrate number in range and
-            # push on to array
-            push(@fields,$1..$2);
-            next;
-            }
-
-            if ($f =~ m/^(\d+)((?:(?:\$|\^)\w)*)\s*$/) {
-
-            my $marc_info = $1;
-            my $opt_sub_fields = $2;
-
-            if ($opt_sub_fields ne "") {            
-                my @sub_fields = split(/\$|\^/,$opt_sub_fields);
-                shift @sub_fields; # skip first entry, which is blank
-
-                foreach my $sub_field (@sub_fields) {
-                $metadata_mapping->{$marc_info."\$".$sub_field} = $gsdl_info;
-                }
-            }
-            else {
-                # no subfields to worry about
-                $marc_info =~ s/\^/\$/;
-                $metadata_mapping->{$marc_info} = $gsdl_info;
-            }
-            }
-            else {
-            $parse_error_count++;
-            }
-        }
-        }
-        else
-        {
-        $parse_error_count++;
-        }
-
-        if ($parse_error_count>0) {
-        
-        print $outhandle "Parse Error: $parse_error_count syntax error(s) on line $l of $mm_file:\n";
-        print $outhandle "  $line\n";
-        }
-    }
-    close(MMIN);
-    }
-    else
-    {
-    print STDERR "Unable to open $mm_file: $!\n";
-    }
-}
-
-
-sub parse_marc_metadata_mapping
-{
-    my $self = shift(@_);
-    my ($mm_file_or_files) = @_;
-
-    my $metadata_mapping = {};
-
-    if (ref ($mm_file_or_files) eq 'SCALAR') {
-    my $mm_file = $mm_file_or_files;
-    $self->_parse_marc_metadata_mapping($mm_file,$metadata_mapping);
-    }
-    else {
-    my $mm_files = $mm_file_or_files;
-    
-    # Need to process files in reverse order.  This is so in the
-    # case where we have both a "collect" and "main" version,
-    # the "collect" one tops up the main one
-
-    my $mm_file;
-    while ($mm_file = pop(@$mm_files)) {
-        $self->_parse_marc_metadata_mapping($mm_file,$metadata_mapping);
-    }
-    }
-
-    return $metadata_mapping;
-}
-
-
 sub init {
     my $self = shift (@_);
@@ -216 +96 @@
     }
 
-    # read in the metadata mapping file
+    # read in the metadata mapping files
     my $mm_files = &util::locate_config_files($self->{'metadata_mapping_file'}); 
 
@@ -233 +113 @@
     }
     else {
-    $self->{'metadata_mapping'} = $self->parse_marc_metadata_mapping($mm_files);
+    $self->{'metadata_mapping'} = &marcmapping::parse_marc_metadata_mapping($mm_files, $outhandle);
     }
 
@@ -316 +196 @@
     }
     $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Plugin", "$self->{'plugin_type'}");
-    $doc_obj->add_metadata($doc_obj->get_top_section(), "FileFormat", "XML");
+    $doc_obj->add_metadata($doc_obj->get_top_section(), "FileFormat", "MARCXML");
 
     my $outhandle = $self->{'outhandle'};