#Basic Regex, 656813 bytes [safe!]
Basic Regex, 656813 bytes [safe!]
The regex to end all regexes. One final hurrah into the night.
Testable under PCRE, Perl, Python and many others.
bzip2'd and base64-encoded version on Pastebin: http://pastebin.com/9kprSWBn (Pastebin didn't want the raw version because it was too big).
To make sure you get the right regex, you can verify that its MD5 hash is
c121a7604c6f819d3805231c6241c4ef
or check that it begins with
^(?:.*[^!0-9@-Za-z].*|.{,255}|.{257,}|.[U-Za-z].{34}[12569@CDGHKLOPSTWXabefijmnqruvyz].{8}[02468@BDFHJLNPRTVXZbdfhjlnprtvxz].{210}
and ends with
.{56}[7-9@-DM-Tc-js-z].{121}[3-6A-DI-LQ-TYZabg-jo-rw-z].{28}[!0-9@-T].{48})$
The key is still a nice comfortable 256 bytes.
I tested this regex with Python, but note that this regex doesn't use any special features of Python. Indeed, with the exception of (?:) (as a grouping mechanism), it actually uses no special features of any regex engine at all: just basic character classes, repetitions, and anchoring. Thus, it should be testable in a great number of regular expression engines.
Well, actually, I can still crank the difficulty up, assuming someone doesn't just instantly solve the smaller problems...but I wager people will have trouble with a 1GB regex...
After 72 hours, this submission remains uncracked! Thus, I am now revealing the key to make the submission safe. This is the first safe submission, after over 30 submissions were cracked in a row by persistent robbers.
Match: Massive Regex Problem Survives The Night!
Non-match: rae4q9N4gMXG3QkjV1lvbfN!wI4unaqJtMXG9sqt2Tb!0eonbKx9yUt3xcZlUo5ZDilQO6Wfh25vixRzgWUDdiYgw7@J8LgYINiUzEsIjc1GPV1jpXqGcbS7JETMBAqGSlFC3ZOuCJroqcBeYQtOiEHRpmCM1ZPyRQg26F5Cf!5xthgWNiK!8q0mS7093XlRo7YJTgZUXHEN!tXXhER!Kenf8jRFGaWu6AoQpj!juLyMuUO5i0V5cz7knpDX0nsL
Regex explanation:
The regex was generated from a "hard" 3SAT problem with a deliberately-introduced random solution. This problem was generated using the algorithm from [Jia, Moore & Strain, 2007]: "Generating Hard Satisfiable Formulas by Hiding Solutions Deceptively". Six boolean variables are packed into each byte of the key, for a total of 1536 variables.
The regex itself is quite simple: it expresses each of 7680 3SAT clauses as a an inverted condition (by de Morgan's laws), and matches any string that does not meet one of the 3SAT clauses. Therefore, the key is a string which does not match the regex, i.e. one that satisfies every one of the clauses.
