Changeset 28488 for gs2-extensions

Timestamp:

2013-10-17T17:03:01+13:00 (11 years ago)

Author:

davidb

Message:

Further work was necessary to better process the data passing through to be suitable escaped for TTL

File:

• gs2-extensions/apache-jena/trunk/src/perllib/jenaTDBBuildproc.pm (modified) (4 diffs)

gs2-extensions/apache-jena/trunk/src/perllib/jenaTDBBuildproc.pm

@@ -85 +85 @@
     my ($hashmap,$value) = @_;
     
-    print STDERR "*** checking value = '$value'\n";
-
-    print STDERR "*** lookup = ", $hashmap->{$value}, "\n";
-
     my $lookup = (defined $hashmap->{$value}) ? $hashmap->{$value} : "\@$value\@";
 
@@ -179 +175 @@
 }
 
+sub make_ttl_safe
+{
+    my ($front,$str,$back) = @_;
+    
+    $str =~ s/\\/\\\\/g;
+    
+    $str =~ s/\&#x([0-9A-F]+);/chr(hex($1))/eig;
+    $str =~ s/\&#([0-9]+);/chr($1)/eig;
+
+    $str =~ s/[\r\n]+/\\n/g;
+
+    return "$front$str$back";
+}
 
 sub textedit {
@@ -198 +207 @@
 
     my $section_text = &docprint::get_section_xml($doc_obj,$doc_obj->get_top_section());
+
+#    $section_text =~ s/[\r\n]+$//s; # remove very last newline char
+
+##    $section_text =~ s/&#x([0-9A-F]+);/chr(hex($1))/eig;
+##    $section_text =~ s/\&#([0-9]+);/chr($1)/eig;
+
+#    $section_text =~ s/\\/\\\\/g;
+
+#    $section_text =~ s/\&#x([0-9A-F]+);/chr(hex($1))/eig;
+#    $section_text =~ s/\&#([0-9]+);/chr($1)/eig;
+
+#    $section_text =~ s/(\r|\n)+/<br \/>/gs;
+#    $section_text =~ s/[\r\n]+/ /gs;
+
+##    $section_text =~ s/\n/ AND /gs;
+
+#    open(DOUT,">/tmp/debug.xml") || die "Failed to open";
+#    print DOUT $section_text;
+#    print DOUT "\n";
+#    close DOUT;
+#    exit -1;
+
+    $section_text =~ s/(<Metadata[^>]*>)(.*?)(<\/Metadata>)/&make_ttl_safe($1,$2,$3)/gse;
+
+## $1&make_ttl_safe($2)$3
+
+##    print STDERR "*** st = $section_text\n\n";
+
+
     print $outhandler $section_text;
 
@@ -212 +250 @@
 
     my $cmd = "gs-triplestore-add $collection \"$tmp_doc_filename\"";
-    
+        
     my $status = system($cmd);
     if ($status != 0) {