Note: If you're looking for the NodeJS module go to https://github.com/krisives/bloomer

 ____ ____ ____ ____ ____ ____ ____ 
||B |||l |||o |||o |||m |||e |||r ||
||__|||__|||__|||__|||__|||__|||__||
|/__\|/__\|/__\|/__\|/__\|/__\|/__\|

Bloomer is a utility library to help website authors guide their users into picking better passwords. It generates a Bloom Filter of compromised passwords, which are obtained from public security lists on the Internet.

The password lists used by Bloomer do not contain identifiable user information, such as an e-mail address - just the passwords. The final resulting bloom filter data does not contain any plaintext passwords and is the result of SHA1, MD5, and SHA256 crytpographic hashes.

The goal of the code written is to be simple and easy to understand. I tried to provide comments that explain things like key distribution, uniformity, etc. I also adapt the constants used in the Wikipedia article, making it easy to see how the parameters k, m, and n affect the data structure.

• Download the source code from this GitHub - You will need PHP to generate a bloom filter, otherwise skip the next step and just use the "bloom.hex"

• Generate the bloom filter data by running:

    php generate.php
  

• Grab "bloom.hex" and "bloomer.js"

• After including "bloomer.js" create an instance:

    var bloomer = new Bloomer();
  

• Feed the data into the Javascript object

    bloomer.setHex($('#bloomhex').val()); // Data from a DOM node
  

• Check if a string is in the bloom filter

    if (bloomer.check("test")) {
  

Drop a .txt file into the lists/ directory and run php generate.php again. More work is needed to adjust the parameters and how large the bloom filter is, but this should be fixed soon.

PHP internally is really bad at memory management. Giving it just 153MB of string data can cause it to allocate gigabytes of memory. Bloomer keeps a unique index of all the passwords added while generating the filter in an effort to maximize the accuracy of the error rate estimate.

You can avoid this problem by disabling the unique index of Bloomer:

$bloomer = new Bloomer(array(
    'unique' => 0
));

Now Bloomer will not track duplicates, and the error rate display is inaccurate with respect to how many duplicate passwords there are (since adding duplicates to the filter doesn't actually affect it's error)

Press the fork button!