Changeset 16638 for gsdl/trunk/perllib/plugins/HTMLPlugin.pm

Timestamp:

2008-08-04T12:32:24+12:00 (16 years ago)

Author:

kjdon

Message:

modified store_block_files: includes script (js) files, don't add a block entry if have empty filename, check for %XX in url and resolve to get true filename if necessary

File:

• gsdl/trunk/perllib/plugins/HTMLPlugin.pm (modified) (3 diffs)

gsdl/trunk/perllib/plugins/HTMLPlugin.pm

@@ -585 +585 @@
     my @embed_matches = ($$textref =~ m/<embed[^>]*?src\s*=\s*($attval)[^>]*>/igs);
     my @tabbg_matches = ($$textref =~ m/<(?:table|tr|td)[^>]*?background\s*=\s*($attval)[^>]*>/igs);
-
-    foreach my $link (@img_matches, @usemap_matches, @link_matches, @embed_matches, @tabbg_matches) {
+    my @script_matches = ($$textref =~ m/<script[^>]*?src\s*=\s*($attval)[^>]*>/igs);
+
+    foreach my $link (@img_matches, @usemap_matches, @link_matches, @embed_matches, @tabbg_matches, @script_matches) {
 
     # remove quotes from link at start and end if necessary
@@ -595 +596 @@
 
     $link =~ s/\#.*$//s; # remove any anchor names, e.g. foo.html#name becomes foo.html
+    # some links may just be anchor names
+    next unless ($link =~ /\S+/);
 
     if ($link !~ m@^/@ && $link !~ m/^([A-Z]:?)\\/) {
@@ -602 +605 @@
     }
     $link = $self->eval_dir_dots($link);
+
+    # Replace %XX's in URL with decoded value if required. 
+    # Note that the filename may include the %XX in some situations
+    if ($link =~ m/\%[A-F0-9]{2}/i) {
+        if (!-e $link) {
+        $link =~ s/\%([A-F0-9]{2})/pack('C', hex($1))/ige;
+        }
+    }
     
     $block_hash->{'file_blocks'}->{$link} = 1;