Changeset 9998

Timestamp:

2005-06-01T15:47:06+12:00 (19 years ago)

Author:

mdewsnip

Message:

Changed this to work on Windows now that multiread sets binmode, and fixed the bug where the Keywords metadata was not being added correctly.

File:

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

trunk/gsdl/perllib/plugins/ISISPlug.pm

@@ -90 +90 @@
 # This plugin splits the input text at the "----------" lines
 sub get_default_split_exp {
-    return q^\n----------\n^;
+    return q^\r?\n----------\r?\n^;
 }
 
@@ -149 +149 @@
 
     # Map the tag numbers to tag names, using the FDT mapping
-    $$textref =~ s/\ntag=(\d+) /\ntag=$fdtmapping{$1}{'title'} /g;
+    $$textref =~ s/\r?\ntag=(\d+) /\ntag=$fdtmapping{$1}{'title'} /g;
 
     # Remove the line at the start so it is split and processed properly
@@ -206 +206 @@
         my $metadatafieldname = $tagname . $subfieldname;
         my $metadatafieldvalue = $1;
-
-        # Escape any '<' and '>' characters so they appear correctly in the final collection
-        $metadatafieldvalue =~ s/\</&lt;/g;
-        $metadatafieldvalue =~ s/\>/&gt;/g;
         
         # Handle Keywords specially
@@ -215 +211 @@
             my $keywordmetadatavalue = $metadatafieldvalue;
             my $keywordlist = "";
-            while ($keywordmetadatavalue =~ s/\<([^\>]+)\>//) {
+            while ($keywordmetadatavalue =~ s/\<(.+?)\>//) {
             my $keyword = $1;
             $doc_obj->add_utf8_metadata($cursection, $metadatafieldname, $keyword); 
@@ -225 +221 @@
         }
 
-        else {
+        # Escape any '<' and '>' characters so they appear correctly in the final collection
+        $metadatafieldvalue =~ s/\</&lt;/g;
+        $metadatafieldvalue =~ s/\>/&gt;/g;
+
+        # We have already added Keywords metadata above
+        unless ($metadatafieldname eq "Keywords") {
             $doc_obj->add_utf8_metadata($cursection, $metadatafieldname, $metadatafieldvalue); 
         }