Based on the comic strip. A "tool" for exploring and solving passwords based on hints and inter-password shared character sequences.
I got bored with it though, since guessing passwords isn't that fun when you can't really know if you are right or wrong.
The code itself is pretty solid, but it is not in a really enduser-friendly state.
For fun. That's how I roll.
Handling 10 GiB of text was kind of tricky. I thought I could just dump it all in a relational database and play around with it from there, but it turns out 250 million rows with multiple joins was more than my laptop could handle with any reasonable speed.
I ended up distilling out only the data I needed with some bash scripting and minimal scripts to massage text streams. After throwing away the more uncommon passwords, that gave me a few hundred megabytes of password hints, and some 6 MiB of password frequency data. This preprocessing step takes about 2 hours, compared to over 24 hours to just import a CSV file into Postgres/MySQL.
The frequency data and solutions can easily be handled by a relational database, while the hints can easily be read from text files named by the password hash.
Isn't this a huge security risk?
Not really. It was a long time ago, and the passwords aren't really difficult to guess manually, when there are hints like "drowssap backwards + 1".