Beefy Boxes and Bandwidth Generously Provided by pair Networks
Just another Perl shrine
 
PerlMonks  

Re: Bloom::Filter Usage [patch]

by thpfft (Chaplain)
on Apr 20, 2004 at 19:11 UTC ( [id://346758]=note: print w/replies, xml ) Need Help??


in reply to Re: Re: Bloom::Filter Usage
in thread Bloom::Filter Usage

On closer inspection, Bloom::Filter is a bit broken, unless I'm badly misreading it:

* The add() method doesn't actually add anything, because the new() method doesn't initialise the $self->{contents} hashref. it needs to be changed so that either the add method assigns directly to $self->{contents}, or the new() method changes 'keys' to 'contents'. that's probably just a typo.

* because $self->{contents} is not populated, _calculate_filter_length() always returns undef, so the default filter length is always used.

* because $self->{contents} is not populated, build_filter doesn't build a filter anyway.

* the use of == to compare bitstrings in check() is generating warnings, as you've seen. Its purpose is to test that ($a & $b) is the same as $a, ie that all the on bits in $a are also on in $b, and never mind the off bits. Someone better than me at pack(), ie anyone at all, will know how to test equivalence using the bitstrings themselves: meanwhile, you can make it work by testing with the unpacked versions.

* And anyway, it's not incremental. Every time you add a new key, the whole filter is rebuilt by way of a loop on $self->{contents}. This makes it more or less useless for your purposes: you presumably want an iterative check($_) || add($_) mechanism that can be placed in the equivalent of a while(<HUGE_FILE>) loop. As it stands, Bloom::Filter will perform another loop across your whole dataset-so-far for each input line, which might slow things down a bit.

You will need to roll your own, I think, unless the author can be persuaded to accommodate both approaches as well as fixing the bugs, but if you patch Bloom::Filter with this, at least your test script should work:

--- Filter.orig.pm Tue Apr 20 20:01:37 2004 +++ Filter.pm Tue Apr 20 20:04:34 2004 @@ -25,7 +25,7 @@ min_length => 20, filter_length => 20, %params, - keys => {} + contents => {} }, $class; } @@ -52,7 +52,7 @@ my $list = $self->{contents}; foreach my $add ( @addresses ) { # convert email to SHA1 hash if necessary - $add = sha1_base64( $add ) if $add =~ /@/o; + #$add = sha1_base64( $add ) if $add =~ /@/o; $list->{$add}++; } $self->{reindex_flag} = 1; @@ -83,7 +83,11 @@ # A match occurs if every bit we check is on foreach my $key ( @keys ) { my $mask = $self->_make_bitmask( $key ); - push @result, ($mask == ( $mask & $self->{filter} )); + #push @result, ($mask == ( $mask & $self->{filter} )); + + my $m = unpack("b*", $mask); + push @result, ($m == unpack("b*", ($mask & $self->{filter}))) +; + } return ( wantarray() ? @result : $result[0] ); }

Replies are listed 'Best First'.
Re: Re: Bloom::Filter Usage [patch]
by FamousLongAgo (Friar) on Apr 21, 2004 at 16:05 UTC
    I am the module author...
    Sorry - looks like an update I thought I had made to CPAN before publishing the article never got through. There should be a v0.02 now on CPAN (same as the code in Listing 1 linked from the article) that has the bugs fixed.

Log In?
Username:
Password:

What's my password?
Create A New User
Domain Nodelet?
Node Status?
node history
Node Type: note [id://346758]
help
Chatterbox?
and the web crawler heard nothing...

How do I use this?Last hourOther CB clients
Other Users?
Others wandering the Monastery: (3)
As of 2024-04-25 14:46 GMT
Sections?
Information?
Find Nodes?
Leftovers?
    Voting Booth?

    No recent polls found