Changeset 862

Timestamp:

2000-01-21T16:59:04+13:00 (24 years ago)

Author:

sjboddie

Message:

fixed a couple of bugs that were preventing muliple document gml files
from being processed correctly

File:

• trunk/gsdl/perllib/plugins/GMLPlug.pm (modified) (6 diffs)

trunk/gsdl/perllib/plugins/GMLPlug.pm

@@ -86 +86 @@
 
     # read in the document
-    my $gml = "";
-    my $line = "";
-
     if ($gz) {
     if (!open (INFILE, "zcat $fullname |")) {
@@ -102 +99 @@
 
     undef $/;
-    while (defined ($line = <INFILE>)) {
-    $gml .= $line;
-    }
+    my $gml = <INFILE>;
     $/ = "\n";
-
     close (INFILE);
-
+    
     my @gml_sections = split("</gsdlsection>",$gml);
     $gml = shift(@gml_sections);
 
     my $no_docs = 0;
-    my $src_filename = "";
-
-    while (1)
-    {
+#    my $src_filename = ""; #### don't appear to use this anymore - not sure if that's right
+
+    while (1) {
     # create a new document
     my $doc_obj = new doc ();
@@ -123 +116 @@
     # process the document
     my $firstsection = 1;
-    while (1) 
-    {
-        my $tags = "";
-        my $text = "";
-
-        my @indenting_sections = split("<gsdlsection",$gml);
+    while (1) {
+        my ($tags, $text) = ("", "");
+
+        my @indenting_sections = split("<gsdlsection", $gml);
         shift(@indenting_sections); # first entry is trivially empty
-####        print STDERR "**** no indenting sections = ", scalar(@indenting_sections), "\n";
-
-        foreach $gml (@indenting_sections)
-        {
-####        print STDERR "\n\n\n\n!!!!!!!!!!! gml = $gml\n\n\n";
-
-        if ($gml =~ m/^\s*([^>]*)>(.*)$/so)
-        {
-            $tags = $1 if (defined $1);
+
+        foreach $gml (@indenting_sections) {
+
+        if ($gml =~ /^\s*([^>]*)>(.*)$/so) {
+            $tags = $1 if defined $1;
             $text = &GMLPlug::_unescape_text($2);
-        }
-        else
-        {
+
+        } else {
             print STDERR "GMLPlug::read - error in file $fullname\n";
             print STDERR "text: \"$gml\"\n";
@@ -149 +135 @@
 
         # create the section (unless this is the first section)
-        if ($firstsection) 
-        {
+        if ($firstsection) {
             $firstsection = 0;
-            
-            print STDERR "  0 of $src_filename\n" if ($no_docs==1);
-            $tags =~ m/gsdlsourcefilename\s*=\s*(\"([^\"]*)\")|(\w+)/o;
-            $src_filename = $2 || $3;
-            print STDERR "  $no_docs of $src_filename\n" if ($no_docs>=1);
-        }
-        else
-        {
-####            print STDERR "*** tags = $tags\n";
-
-            if ($tags =~ s/gsdlnum\s*=\s*\"?(\d+)\"?//o)
-            {
+#           $tags =~ /gsdlsourcefilename\s*=\s*(?:\"([^\"]*)\")/o;
+#           $src_filename = $2 || $3;
+
+        } else {
+
+            $tags =~ s/gsdlnum\s*=\s*\"?(\d+)\"?//o;
+            if (defined $1) {
             $section .= ".$1";
             $doc_obj->create_named_section($section);
-            } 
-            else 
-            {
+            } else {
             $section = $doc_obj->insert_section($doc_obj->get_end_child($section));
             }
         }
         
-
         # add the tags
-        while ((defined $tags) 
-               && ($tags =~ s/^\s*(\w+)\s*=\s*\"([^\"]*)\"//o)) 
-        {
-            $doc_obj->add_utf8_metadata($section, $1, 
-                        &GMLPlug::_unescape_text($2)) 
+        while ((defined $tags) && ($tags =~ s/^\s*(\w+)=\"([^\"]*)\"//o)) {
+            $doc_obj->add_utf8_metadata($section, $1, &GMLPlug::_unescape_text($2)) 
             if (defined $1 and defined $2);
         }
@@ -189 +163 @@
 
         $gml = shift(@gml_sections); # get next bit of data
-        last if (!defined $gml);
-###     print "####### before section = $section\n";
+        last unless defined $gml;
         $section = $doc_obj->get_parent_section ($section);
-###     print "####### after section = $section\n";
-        #### last if ($section eq "");
-    }
-
-    # add the associated files
-    $assoc_files = $doc_obj->get_metadata($doc_obj->get_top_section(), "gsdlassocfile");
-    my ($assoc_file_info);
-    foreach $assoc_file_info (@$assoc_files) {
-    my ($assoc_file, $mime_type, $dir) = split (":", $assoc_file_info);
-    $dir = "" unless defined $dir;
-    $doc_obj->associate_file(&util::filename_cat($parent_dir, $assoc_file),
-                 &util::filename_cat($dir, $assoc_file), $mime_type);
+        last if $section eq ""; # back to top level again (more than one document in gml file)
+    }
+
+    # add the associated files
+    $assoc_files = $doc_obj->get_metadata($doc_obj->get_top_section(), "gsdlassocfile");
+    my ($assoc_file_info);
+    foreach $assoc_file_info (@$assoc_files) {
+        my ($assoc_file, $mime_type, $dir) = split (":", $assoc_file_info);
+        $dir = "" unless defined $dir;
+        $doc_obj->associate_file(&util::filename_cat($parent_dir, $assoc_file),
+                     &util::filename_cat($dir, $assoc_file), $mime_type);
     }
     $doc_obj->delete_metadata($doc_obj->get_top_section(), "gsdlassocfile");
-
+    
     # add metadata
     foreach $field (keys(%$metadata)) {
@@ -218 +190 @@
         }
     }
-
-
+    
     # assume the document has an OID
-
+    
     # process the document
     $processor->process($doc_obj, $file);
-
+    
     $no_docs++;
-
-    last if (defined $maxdocs && $maxdocs =~ /\d/ && $no_docs >= $maxdocs);
-
-    last if (!defined $gml);
+    last if ($maxdocs > -1 && $no_docs >= $maxdocs);
+    last unless defined $gml;
     }