Changeset 12424

Timestamp:

2006-08-09T15:16:46+12:00 (18 years ago)

Author:

mdewsnip

Message:

Turned "use strict" on and tried to tidy up some confusing code.

Location:

trunk/gsdl/perllib

Files:

• lucenebuildproc.pm (modified) (4 diffs)
• mgppbuildproc.pm (modified) (3 diffs)

trunk/gsdl/perllib/lucenebuildproc.pm

@@ -33 +33 @@
 use mgppbuildproc; 
 use ghtml;
+use strict;
+no strict 'refs'; # allow filehandles to be variables and viceversa
+
 
 sub BEGIN {
@@ -161 +164 @@
 
         my $new_text = ""; 
-        my $tmp_text = "";
 
         # we get allfields by default - do nothing
@@ -171 +173 @@
         my $shortname = "";
         my $metadata = $doc_obj->get_all_metadata ($section);
-        foreach $pair (@$metadata) {
+        foreach my $pair (@$metadata) {
             my ($mfield, $mvalue) = (@$pair);
             # check fields here, maybe others dont want - change to use dontindex!!
@@ -209 +211 @@
         
         my @metadata_list = ();
-        foreach $submeta (split /,/, $real_field) {
+        foreach my $submeta (split /,/, $real_field) {
             if ($submeta eq "text") {
-            if ($self->{'indexing_text'}) { #tag the text with <Text>...</Text>, add the <Paragraph> tags and strip out html if needed
+            my $section_text = $doc_obj->get_text($section);
+            if ($self->{'indexing_text'}) {
+                            # tag the text with <Text>...</Text>, add the <Paragraph> tags and strip out html if needed
                 $new_text .= "$parastarttag<$shortname index=\"1\">\n";
-                $tmp_text .= $doc_obj->get_text ($section);
                 if ($parastarttag ne "") {
-                $tmp_text = $self->preprocess_text($tmp_text, $self->{'strip_html'}, "</$shortname>$paraendtag$parastarttag<$shortname index=\"1\">");
-                } else {
+                $section_text = $self->preprocess_text($section_text, $self->{'strip_html'}, "</$shortname>$paraendtag$parastarttag<$shortname index=\"1\">");
+                }
+                else {
                 # we don't want to individually tag each paragraph if not doing para indexing
-                $tmp_text = $self->preprocess_text($tmp_text, $self->{'strip_html'}, "");
-                } 
-                $new_text .= "$tmp_text</$shortname>$paraendtag\n";
-            }
-            else { # leave html stuff in, but escape the tags, and dont add Paragraph tags - never retrieve paras at the moment
-                $tmp_text .= $doc_obj->get_text ($section);
-                &ghtml::htmlsafe($tmp_text);
-                $new_text .= $tmp_text;
-                
+                $section_text = $self->preprocess_text($section_text, $self->{'strip_html'}, "");
+                }
+                $new_text .= "$section_text</$shortname>$paraendtag\n";
+            }
+            else {
+                            # leave html stuff in, but escape the tags, and don't add Paragraph tags - never retrieve paras at the moment
+                &ghtml::htmlsafe($section_text);
+                $new_text .= $section_text;
             }
             }

trunk/gsdl/perllib/mgppbuildproc.pm

@@ -31 +31 @@
 
 use basebuildproc;
+use strict;
+no strict 'refs'; # allow filehandles to be variables and viceversa
+
 
 BEGIN {
@@ -257 +260 @@
         
         my $new_text = ""; 
-        my $tmp_text = "";
 
         # we get allfields by default - do nothing
@@ -305 +307 @@
         }
         my @metadata_list = ();
-        foreach $submeta (split /,/, $real_field) {
+        foreach my $submeta (split /,/, $real_field) {
             if ($submeta eq "text") {
-            if ($self->{'indexing_text'}) { #tag the text with <Text>...</Text>, add the <Paragraph> tags and strip out html if needed
+            my $section_text = $doc_obj->get_text($section);
+            if ($self->{'indexing_text'}) {
+                # tag the text with <Text>...</Text>, add the <Paragraph> tags and strip out html if needed
                 $new_text .= "$paratag<$shortname>\n";
-                $tmp_text .= $doc_obj->get_text ($section);
                 if ($paratag ne "") {
-                $tmp_text = $self->preprocess_text($tmp_text, $self->{'strip_html'}, "</$shortname>$paratag<$shortname>");
-                } else {
-                $tmp_text = $self->preprocess_text($tmp_text, $self->{'strip_html'}, "");
+                $section_text = $self->preprocess_text($section_text, $self->{'strip_html'}, "</$shortname>$paratag<$shortname>");
                 }
-                $new_text .= "$tmp_text</$shortname>\n";
+                else {
+                $section_text = $self->preprocess_text($section_text, $self->{'strip_html'}, "");
+                }
+                $new_text .= "$section_text</$shortname>\n";
             }
-            else { # leave html stuff in, and dont add Paragraph tags - never retrieve paras at the moment
-                $new_text .= $doc_obj->get_text ($section);
+            else {
+                            # leave html stuff in, and don't add Paragraph tags - never retrieve paras at the moment
+                $new_text .= $section_text;
             }
-            } else {
+            }
+            else {
             my @section_metadata = @{$doc_obj->get_metadata ($section, $submeta)};
             if ($section ne $doc_obj->get_top_section() && $self->{'indexing_text'} && defined ($self->{'sections_index_document_metadata'})) {