Changeset 24932 for main

Timestamp:

2012-01-10T16:18:19+13:00 (12 years ago)

Author:

ak19

Message:

Diego noticed how the metadata in a toplevel metadata.xml, which specifies metadata for files in import's subfolders, does not get attached to the files on Windows, while this works on Linux. It had to do with the difference between the file slashes used on the OS versus the URL-type fileslashes used in the metadata.xml Diego had constructed. This has now been fixed and Dr Bainbridge came up with a tidier solution of a new method in util.pm that would handle the details.

Location:

main/trunk/greenstone2/perllib

Files:

• plugins/DirectoryPlugin.pm (modified) (3 diffs)
• util.pm (modified) (4 diffs)

main/trunk/greenstone2/perllib/plugins/DirectoryPlugin.pm

@@ -382 +382 @@
     $dirname = &util::filename_cat ($base_dir, $file) if $base_dir =~ /\w/;
     }
-
+    
     my $directory_ok = $self->check_directory_path($dirname);
     return $directory_ok unless (defined $directory_ok && $directory_ok == 1);
@@ -424 +424 @@
 
 
-    my $os_dirsep = &util::get_os_dirsep();
+    my $os_dirsep = &util::get_os_dirsep();
     my $dirsep    = &util::get_dirsep();
     my $base_dir_regexp = $base_dir;
     $base_dir_regexp =~ s/\//$os_dirsep/g;
-    my $local_dirname = $dirname;
-    
-    $local_dirname =~ s/^$base_dir_regexp($os_dirsep)*//;
+       
+    # Want to get relative path of local_dirname within the base_directory
+    # but with URL style slashes. 
+    my $local_dirname = &util::filename_within_directory_url_format($dirname, $base_dir);
+
     # if we are in import folder, then local_dirname will be empty
     if ($local_dirname ne "") {
-    # look for extra metadata passed down from higher folders 
-    $local_dirname .= $dirsep;
+    # look for extra metadata passed down from higher folders   
+    $local_dirname .= "/"; # closing slash must be URL type slash also and not $dirsep;
     if (defined $self->{'subdir_extrametakeys'}->{$local_dirname}) {
         my $extrakeys = $self->{'subdir_extrametakeys'}->{$local_dirname};
@@ -491 +493 @@
 
         my $subdir_extrametakeys = $self->{'subdir_extrametakeys'};
-
+        
         my $subdir_rec = { 're' => $subdir_re, 'md' => $md, 'mf' => $mf };
 

main/trunk/greenstone2/perllib/util.pm

@@ -845 +845 @@
     $filename =~ s@\(@\\(@g; # escape brackets
     $filename =~ s@\)@\\)@g; # escape brackets
+    $filename =~ s@\[@\\[@g; # escape brackets
+    $filename =~ s@\]@\\]@g; # escape brackets
     
     return $filename;
@@ -857 +859 @@
     $filename =~ s@\\\(@(@g; # remove RE syntax for ( => "\(" turns into "("
     $filename =~ s@\\\)@)@g; # remove RE syntax for ) => "\)" turns into ")"
+    $filename =~ s@\\\[@[@g; # remove RE syntax for [ => "\[" turns into "["
+    $filename =~ s@\\\]@]@g; # remove RE syntax for ] => "\]" turns into "]"
     return $filename;
 }
@@ -1032 +1036 @@
 }
 
-
+# If filename is relative to within_dir, returns the relative path of filename to that directory
+# with slashes in the filename returned as they were in the original (absolute) filename.
 sub filename_within_directory
 {
@@ -1049 +1054 @@
     return $filename;
 }
+
+# If filename is relative to within_dir, returns the relative path of filename to that directory in URL format.
+# Filename and within_dir can be any type of slashes, but will be compared as URLs (i.e. unix-style slashes).
+# The subpath returned will also be a URL type filename.
+sub filename_within_directory_url_format
+{
+    my ($filename,$within_dir) = @_;
+    
+    # convert parameters only to / slashes if Windows
+    
+    my $filename_urlformat = $filename;
+    my $within_dir_urlformat = $within_dir;
+    
+    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
+        # Only need to worry about Windows, as Unix style directories already in url-format
+        # Convert Windows style \ => /
+        $filename_urlformat =~ s@\\@/@g;
+        $within_dir_urlformat =~ s@\\@/@g;
+    }
+    
+    #if ($within_dir_urlformat !~ m/\/$/) {
+        # make sure directory ends with a slash
+        #$within_dir_urlformat .= "/";
+    #}
+    
+    my $within_dir_urlformat_re = &filename_to_regex($within_dir_urlformat); # escape any special RE characters, such as brackets
+    
+    #print STDERR "@@@@@ $filename_urlformat =~ $within_dir_urlformat_re\n";
+    
+    # dir prefix may or may not end with a slash (this is discarded when extracting the sub-filepath)
+    if ($filename_urlformat =~ m/^$within_dir_urlformat_re(?:\/)*(.*)$/) {
+        $filename_urlformat = $1;
+    }
+    
+    return $filename_urlformat;
+}
+
 
 sub filename_within_collection