X-Git-Url: http://git.ieval.ro/?a=blobdiff_plain;f=lib%2FSlob.pm;h=818ff785ed7fe6fd298fe3a9ff0c7142aa4713f8;hb=d50f39582b61e51d104a3fc4799163855d5a289b;hp=8298becd0e964bb8e1e3508ff9f811793cc67cdd;hpb=0e15b49616308e77f3f6ede65711ea70a876af47;p=slob.git

diff --git a/lib/Slob.pm b/lib/Slob.pm
index 8298bec..818ff78 100644
--- a/lib/Slob.pm
+++ b/lib/Slob.pm
@@ -10,9 +10,30 @@ use constant MAGIC => "!-1SLOB\x1F";
 use Carp qw/croak verbose/;
 use Encode;
 
+use Compress::Raw::Lzma;
+
+our %UNCOMPRESS = (
+	'' => sub { $_[0] },
+	'lzma2' => sub {
+		my ($input) = @_;
+		my ($lzma2, $code, $output);
+		($lzma2, $code) = Compress::Raw::Lzma::RawDecoder->new(Filter => Lzma::Filter::Lzma2());
+		die "Error creating LZMA2 decoder: $code\n" unless $code == LZMA_OK;
+		$code = $lzma2->code($input, $output);
+		die "Did not reach end of stream" if $code == LZMA_OK;
+		die "Error decoding LZMA2: $code" if $code != LZMA_STREAM_END;
+		$output
+	}
+);
+
 sub new {
 	my ($class, $path) = @_;
-	open my $fh, '<', $path or croak "Cannot open \"$path\": $!";
+	my $fh;
+	if (ref $path eq 'IO') {
+		$fh = $path
+	} else {
+		open $fh, '<', $path or croak "Cannot open \"$path\": $!"
+	}
 	my $self = bless {path => $path, fh => $fh}, $class;
 	$self->{header} = $self->read_header;
 	$self
@@ -106,7 +127,7 @@ sub ftell {
 
 sub uncompress {
 	my ($self, $data) = @_;
-	$data
+	$UNCOMPRESS{$self->{header}{compression}}->($data)
 }
 
 sub read_header {
@@ -119,7 +140,7 @@ sub read_header {
 	$self->{encoding} = $encoding;
 
 	my $compression = $self->read_tiny_text;
-	die "Compression not yet supported" if $compression;
+	die "Compression '$compression' not yet supported" unless exists $UNCOMPRESS{$compression};
 	my %tags = $self->read_tags;
 	my @content_types = $self->read_content_types;
 	my $blob_count = $self->read_int;
@@ -241,7 +262,70 @@ Slob - Read .slob dictionaries (as used by Aard 2)
 
 =head1 DESCRIPTION
 
-No documentation yet, see SYNOPSIS.
+Slob is a compressed read-only format for storing natural language
+dictionaries. It is used in Aard 2. C<Slob.pm> is a module that reads
+dictionaries in slob format.
+
+The following methods are available:
+
+=over
+
+=item Slob->B<new>(I<$path>)
+=item Slob->B<new>(I<$fh>)
+
+Create a new slob reader reading from the given path or filehandle.
+
+=item $slob->B<ref_count>
+
+The number of refs (keys) in the dictionary.
+
+=item $slob->B<seek_and_read_ref>(I<$index>)
+
+Read the ref (key) at the given index. Returns a hashref with the
+following keys:
+
+=over
+
+=item key
+
+The key
+
+=item bin_index
+
+The storage bin that contains the value for this key
+
+=item item_index
+
+The index in the bin_index storage bin of the value for this key
+
+=item fragment
+
+HTML fragment that, when applied to the HTML value, points to the
+definition of the key.
+
+=back
+
+=item $slob->B<seek_and_read_storage_bin>(I<$index>)
+
+Read the storage bin with the given index. Returns the storage bin,
+which can later be given to B<get_entry_of_storage_bin>.
+
+=item $slob->B<get_entry_of_storage_bin>(I<$bin>, I<$index>)
+
+Given a storage bin (as returned by C<seek_and_read_storage_bin>) and
+item index, returns the value at the index i nthe storage bin.
+
+=item $slob->B<seek_and_read_ref_and_data>($index)
+
+Convenience method that returns the key and value at a given index.
+Returns a hashref like C<seek_and_read_ref> with an extra key,
+I<data>, which is the value of the key.
+
+=back
+
+=head1 SEE ALSO
+
+L<https://github.com/itkach/slob>
 
 =head1 AUTHOR