Get rid of sanitize_filename for archive extraction as it can't possi…

…bly save us anymore

+ add an extra sanity check to fully fix corrupt files on file serving

lib/LANraragi/Model/Archive.pm

@@ -166,21 +166,28 @@ sub serve_page {
 
     my $logger = get_logger( "File Serving", "lanraragi" );
 
+    $logger->debug("Page /$id/$path was requested");
+
     my $tempfldr = get_temp();
-    my $file     = LANraragi::Utils::Archive::sanitize_filename( $tempfldr . "/$id/$path" );
+    my $file     = $tempfldr . "/$id/$path";
 
     if ( -e $file ) {
 
         # Freshly created files might not be complete yet.
         # We have to wait before trying to serve them out...
         my $last_size = 0;
         my $size      = -s $file;
+        my $timeout   = 0;
         while (1) {
             $logger->debug("Waiting for file to be fully written ($size, previously $last_size)");
             usleep(10000);    # 10ms
+            $timeout += 10;   # Sanity check in case the file remains at 0 bytes forever
             $last_size = $size;
             $size      = -s $file;
-            last if ( $last_size eq $size );    # If the size hasn't changed since the last loop, it's likely the file is ready.
+
+            # If the size hasn't changed since the last loop, it's likely the file is ready.
+            last
+              if ( $last_size eq $size && ( $size ne 0 || $timeout > 1000 ) );
         }
 
     } else {

lib/LANraragi/Model/Reader.pm

@@ -73,7 +73,7 @@ sub build_reader_JSON {
     # To investigate further, perhaps with custom sorting algorithms?
     @images = sort { &expand($a) cmp &expand($b) } @images;
 
-    $self->LRR_LOGGER->debug( "Files found in archive: \n " . Dumper @images );
+    $self->LRR_LOGGER->debug( "Files found in archive (encoding might be incorrect): \n " . Dumper @images );
 
     # Build a browser-compliant filepath array from @images
     my @images_browser;
@@ -92,8 +92,6 @@ sub build_reader_JSON {
         # Then we bring the slashes back.
         $imgpath =~ s!%2F!/!g;
 
-        $self->LRR_LOGGER->debug("Will be extracted to disk as: $imgpath");
-
         # Bundle this path into an API call which will be used by the browser
         push @images_browser, "./api/archives/$id/page?path=$imgpath";
     }

lib/LANraragi/Utils/Archive.pm

@@ -47,37 +47,14 @@ sub generate_thumbnail {
     undef $img;
 }
 
-# sanitize_filename(filename)
-# Converts extracted filenames to an ascii variant to avoid extra filesystem headaches.
-sub sanitize_filename {
-
-    my $filename = $_[0];
-    eval {
-        # Try a guess to regular japanese encodings first
-        $filename = decode( "Guess", $filename );
-    };
-
-    # Fallback to utf8
-    $filename = decode_utf8($filename) if $@;
-
-    # Re-encode the result to ASCII and move the file to said result name.
-    # Use Encode's coderef feature to map non-ascii characters to their Unicode codepoint equivalent.
-    $filename = encode( "ascii", $filename, sub { sprintf "%04X", shift } );
-
-    if ( length $filename > 254 ) {
-        $filename = substr( $filename, 0, 254 );
-    }
-
-    return $filename;
-}
-
 # extract_archive(path, archive_to_extract, force)
 # Extract the given archive to the given path.
 # This sub won't re-extract files already present in the destination unless force = 1.
 sub extract_archive {
 
     my ( $destination, $to_extract, $force_extract ) = @_;
     my $logger = get_logger( "Archive", "lanraragi" );
+    $logger->debug("Fully extracting archive $to_extract");
 
     # PDFs are handled by Ghostscript (alas)
     if ( is_pdf($to_extract) ) {
@@ -92,14 +69,16 @@ sub extract_archive {
             if ($force_extract) { return 1; }
 
             my $filename = $e->pathname;
-            $filename = sanitize_filename($filename);
             if ( -e "$destination/$filename" ) {
                 $logger->debug("$filename already exists in $destination");
                 return 0;
             }
+            $logger->debug("Extracting $filename");
 
             # Pre-emptively create the file to signal we're working on it
-            open( my $fh, ">", "$destination/$filename" ) or return 0;
+            open( my $fh, ">", "$destination/$filename" )
+              or
+              $logger->error("Couldn't create placeholder file $destination/$filename (might be a folder?), moving on nonetheless");
             close $fh;
             return 1;
         }
@@ -112,16 +91,6 @@ sub extract_archive {
     my $result_dir = $ae->to;
     my $cwd        = getcwd();
 
-    # Rename extracted files and folders to an encoded version for easier handling
-    finddepth(
-        sub {
-            unless ( $_ eq '.' ) {
-                move( $_, sanitize_filename($_) );
-            }
-        },
-        $result_dir
-    );
-
     # chdir back to the base cwd in case finddepth died midway
     chdir $cwd;
 
@@ -291,18 +260,27 @@ sub extract_single_file {
     } else {
 
         my $contents = "";
-        my $peek = Archive::Libarchive::Peek->new( filename => $archive );
-        $contents = $peek->file($filepath);
+        my $peek     = Archive::Libarchive::Peek->new( filename => $archive );
+        my @files    = $peek->files;
+
+        for my $name (@files) {
+            my $decoded_name = LANraragi::Utils::Database::redis_decode($name);
+
+            # This sub can receive either encoded or raw filenames, so we have to test for both.
+            if ( $decoded_name eq $filepath || $name eq $filepath ) {
+                $logger->debug("Found file $filepath in archive $archive");
+                $contents = $peek->file($name);
+                last;
+            }
+        }
 
         open( my $fh, '>', $outfile )
           or die "Could not open file '$outfile' $!";
         print $fh $contents;
         close $fh;
     }
 
-    my $fixed_name = sanitize_filename($outfile);
-    move( $outfile, $fixed_name );
-    return $fixed_name;
+    return $outfile;
 }
 
 # extract_file_from_archive($archive, $file)

tools/Documentation/.gitignore

@@ -0,0 +1 @@
+.gitbook

tools/Documentation/basic-operations/archives.md

@@ -11,6 +11,7 @@ You can customize display options to show the index in compact mode, as well as
 ![Customized compact mode to sort by ratings](../.screenshots/ratings.png)  
 
 The topmost **carousel** view will show random archives from your current search in both thumbnail and compact modes -- It can also be configured to show New or Untagged archives instead.  
+
 ![Index page of a regular LRR install in compact mode](<https://raw.githubusercontent.com/Difegue/LANraragi/dev/tools/\_screenshots/archive\_list.png>)
 
 {% hint style="info" %}