Changeset 24586

Timestamp:

2011-09-13T20:54:35+12:00 (13 years ago)

Author:

ak19

Message:

Dr Bainbridge fixed the unicode bug that Diego had found when using an hfile to specify the Hierarchy of a classifier. The hfile was so far being read as a regular cfg file read, but the read operation needed to read in UTF8 mode to preserve the encodings of characters in the file.

Location:

main/trunk/greenstone2/perllib

Files:

• cfgread.pm (modified) (1 diff)
• classify/HFileHierarchy.pm (modified) (1 diff)

main/trunk/greenstone2/perllib/cfgread.pm

@@ -137 +137 @@
     } else {
     print STDERR "cfgread::read_cfg_file couldn't read the cfg file $filename\n";
+    }
+
+    return $data;
+}
+
+# If the cfg file contains unicode characters, use this method to read from it
+# Used by HFileHierarchy classifier, since an HFile is read as a cfg file, but 
+# can contain unicode characters.
+sub read_cfg_file_unicode {
+    my ($filename, $stringexp, $arrayexp, $hashexp, $arrayarrayexp,
+    $hashhashexp) = @_;
+    my ($line);
+    my $data = {};
+
+    if (open (COLCFG, $filename)) {
+    binmode(COLCFG,":utf8");
+    while (defined ($line = &read_cfg_line('COLCFG'))) {
+        if (scalar(@$line) >= 2) {
+
+        #map { decode("utf8",$_) } @$line; #use Encode;
+
+        my $key = shift (@$line);
+        if (defined $stringexp && $key =~ /$stringexp/) {
+            $data->{$key} = shift (@$line);
+
+        } elsif (defined $arrayexp && $key =~ /$arrayexp/) {
+            push (@{$data->{$key}}, @$line);
+
+        } elsif (defined $hashexp && $key =~ /$hashexp/) {
+            my $k = shift @$line;
+            my $v = shift @$line;
+            $data->{$key}->{$k} = $v;
+        } elsif (defined $arrayarrayexp && $key =~ /$arrayarrayexp/) {
+            if (!defined $data->{$key}) {
+            $data->{$key} = [];
+            }
+            push (@{$data->{$key}}, $line);
+        }
+        elsif (defined $hashhashexp && $key =~ /$hashhashexp/) {
+            my $k = shift @$line;
+            my $p = shift @$line;
+            my $v = shift @$line;
+            if (!defined $v) {
+            $v = $p;
+            $p = 'default';
+            }
+            $data->{$key}->{$k}->{$p} = $v;
+        }
+        }
+    }
+    close (COLCFG);
+
+    } else {
+    print STDERR "cfgread::read_cfg_file_unicode couldn't read the cfg file $filename\n";
     }
 

main/trunk/greenstone2/perllib/classify/HFileHierarchy.pm

@@ -163 +163 @@
     my $subjectfile = $self->{'subjectfile'};
     if (defined $subjectfile) {
-    # read in the subject file
-    my $list = &cfgread::read_cfg_file ($self->{'subjectfile'}, undef, '^[^#]?\S');
+    # read in the subject file, but read in unicode mode to preserve special characters
+    my $list = &cfgread::read_cfg_file_unicode ($self->{'subjectfile'}, undef, '^[^#]?\S');
     # $list is a hash that is indexed by the descriptor. The contents of this
     # hash is a list of two items. The first item is the OID and the second item