Changeset 18319 for gsdl/trunk/perllib

Timestamp:

2009-01-06T18:40:53+13:00 (15 years ago)

Author:

ak19

Message:

Now plugins provide the option of base64 encoding or url encoding filenames that are to be renamed (when copied into the archives dir). Previously renamed files would always be url-encoded. URL-encoding is the default now for most plugins except MP3Plugin and OggVorbisPlugin, where the default is base64 encoding. Base64 encoding filenames upon renaming them was introduced so that more files that browsers try to open in external applications can open them, since url encoding does not seem to be implemented the same everywhere (for instance, windows media player is unable to handle url-encoded wmv filenames when such files are launched in it through the browser).

Location:

gsdl/trunk/perllib

Files:

• doc.pm (modified) (3 diffs)
• strings.properties (modified) (1 diff)
• unicode.pm (modified) (3 diffs)
• util.pm (modified) (2 diffs)

gsdl/trunk/perllib/doc.pm

@@ -49 +49 @@
 sub new {
     my $class = shift (@_);
-    my ($source_filename, $doc_type) = @_;
+    my ($source_filename, $doc_type, $rename_method) = @_;
 
 
@@ -67 +67 @@
     if (defined $source_filename) {
     $source_filename = &util::filename_within_collection($source_filename);
-    $self->set_source_filename ($source_filename);
+    print STDERR "******doc.pm::new(): no file rename method provided\n" unless $rename_method;
+    $self->set_source_filename ($source_filename, $rename_method);
     }
 
@@ -144 +145 @@
 sub set_source_filename {
     my $self = shift (@_);
-    my ($source_filename) = @_;
+    my ($source_filename, $rename_method) = @_;
 
     # Since the gsdlsourcefilename element goes into the doc.xml it has 
     # to be utf8. However, it should also *represent* the source filename 
     # (in the import directory) which may not be utf8 at all. 
-    # For instance, if this meta element (gsdlsourcefilename) will be 
-    # used by other applications that parse doc.xml in order to locate 
-    # gsdlsourcefilename. Therefore, the solution is to URLencode the real
-    # filename as this is a binary-to-text encoding meaning that the 
-    # resulting string is ASCII (utf8).
-    
-#    print STDERR "******URLencoding the gsdl_source_filename $source_filename ";
-    
+    # For instance, if this meta element (gsdlsourcefilename) will be used 
+    # by other applications that parse doc.xml in order to locate 
+    # gsdlsourcefilename. Therefore, the solution is to URLencode or base64 
+    # encode the real filename as this is a binary-to-text encoding meaning
+    # that the resulting string is ASCII (utf8). Decoding will give the original.
+    
+#    print STDERR "******URL/base64 encoding the gsdl_source_filename $source_filename ";
+
     # URLencode just the gsdl_source_filename, not the directory. Then prepend dir 
     my ($srcfilename,$dirname,$suffix) 
     = &File::Basename::fileparse($source_filename, "\\.[^\\.]+\$");
 #    print STDERR "-> $srcfilename -> ";
-    $srcfilename = &unicode::url_encode($srcfilename.$suffix);
+    $srcfilename = &util::rename_file($srcfilename.$suffix, $rename_method);
     $source_filename = &util::filename_cat($dirname, $srcfilename);
 #    print STDERR "$source_filename\n";

gsdl/trunk/perllib/strings.properties

@@ -716 +716 @@
 BasePlugin.stems:stems
 
+BasePlugin.file_rename_method:The method to be used in renaming the copy of the imported file and associated files.
+
+BasePlugin.rename_method.url:Use url encoding in renaming files and associated files.
+
+BasePlugin.rename_method.base64:Use base64 encoding in renaming imported files and associated files.
+
 BibTexPlugin.desc:BibTexPlugin reads bibliography files in BibTex format. BibTexPlugin creates a document object for every reference in the file. It is a subclass of SplitTextFile, so if there are multiple records, all are read.
 

gsdl/trunk/perllib/unicode.pm

@@ -600 +600 @@
 
 # When a filename on the filesystem is already URL-encoded, the
-# URL to it will have %25s in in place of every % sign, so that 
+# URL to it will have %25s in place of every % sign, so that 
 # URLs in html pages can refer to the URL-encoded filename.
 # This method changes the URL reference back into the actual 
@@ -606 +606 @@
 sub url_to_filename {
     my ($text) =@_;
-    $text =~ s/%25/%/g;
+    $text =~ s/%25/%/g if is_url_encoded($text);
     return $text;
 }
@@ -619 +619 @@
 # a url more than once this way.
 sub filename_to_url {
-    my ($text) =@_;
-    
-    if($text !~ m/%25/) {
-    $text =~ s/%/%25/g;
+    my ($text, $rename_method) = @_;
+    
+    if(!defined $rename_method || $rename_method eq "url") {
+    if($text !~ m/%25/) {
+        $text =~ s/%/%25/g;
+    }
     }
     return $text;

gsdl/trunk/perllib/util.pm

@@ -28 +28 @@
 use File::Copy;
 use File::Basename;
+use MIME::Base64; # for base64 encoding
 
 use strict;
@@ -994 +995 @@
 }
 
+# returns the given filename converted using either URL encoding or 
+# base64 encoding, as specified by $rename_method
+sub rename_file {
+    my ($filename, $rename_method)  = @_;
+    
+    my ($tailname,$dirname,$suffix) = File::Basename::fileparse($filename, "\\.[^\\.]+\$");
+
+    if (!$rename_method) {
+    print STDERR "WARNING: no file renaming method specified. Defaulting to using URL encoding...\n";
+    # Debugging information
+    my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(1);
+    print STDERR "Calling method; $cfilename:$cline $cpackage->$csubr\n";
+    }
+
+    if (!$rename_method || $rename_method eq "url") {
+    $tailname = &unicode::url_encode($tailname);
+    }
+    elsif ($rename_method eq "base64") {
+    $tailname = &MIME::Base64::encode_base64($tailname);
+    $tailname =~ s/\s*//sg;      # for some reason it adds spaces not just at end but also in middle
+    }
+        
+    $filename = "$tailname$suffix";
+    $filename = "$dirname$filename" if ($dirname ne "./");
+
+    return $filename;
+}
+
 1;