Given a decimal number N
as a string of digits, how do I check if it\'s divisible by M
using regular expressions only, without con
This is an old question, but none of the existing answers include code. I wrote code to compute these regular expressions back in 2010, which is online here and has a link to the commented source code, so I thought it might be useful to add a link to it here.
The technique is basically the one described in eh9’s answer, except that I compute a regular expression directly from the DFA using state elimination and then apply some simplifications to the result. They’re plain regular expressions, without recursion. (Using recursion would make the results much shorter, but I think it’s interesting that it’s possible without.)
The results aren’t quite as verbose as eh9’s estimate would suggest. For example, here is a regular expression that matches multiples of 7 in base 10:
^(0|7|[18]5*4|(2|9|[18]5*6)(3|[29]5*6)*(1|8|[29]5*4)|(3|[18]5*[07]|(2|9|[18]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(5|65*4|(0|7|65*6)(3|[29]5*6)*(1|8|[29]5*4))|(4|[18]5*[18]|(2|9|[18]5*6)(3|[29]5*6)*(5|[29]5*[18])|(3|[18]5*[07]|(2|9|[18]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(2|9|35*4|(4|35*6)(3|[29]5*6)*(1|8|[29]5*4)|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(5|65*4|(0|7|65*6)(3|[29]5*6)*(1|8|[29]5*4)))|(5|[18]5*[29]|(2|9|[18]5*6)(3|[29]5*6)*(6|[29]5*[29])|(3|[18]5*[07]|(2|9|[18]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))|(4|[18]5*[18]|(2|9|[18]5*6)(3|[29]5*6)*(5|[29]5*[18])|(3|[18]5*[07]|(2|9|[18]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(0|7|35*[29]|(4|35*6)(3|[29]5*6)*(6|[29]5*[29])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))))(4|[07]5*[29]|(1|8|[07]5*6)(3|[29]5*6)*(6|[29]5*[29])|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))|(3|[07]5*[18]|(1|8|[07]5*6)(3|[29]5*6)*(5|[29]5*[18])|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(0|7|35*[29]|(4|35*6)(3|[29]5*6)*(6|[29]5*[29])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))))*(6|[07]5*4|(1|8|[07]5*6)(3|[29]5*6)*(1|8|[29]5*4)|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(5|65*4|(0|7|65*6)(3|[29]5*6)*(1|8|[29]5*4))|(3|[07]5*[18]|(1|8|[07]5*6)(3|[29]5*6)*(5|[29]5*[18])|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(2|9|35*4|(4|35*6)(3|[29]5*6)*(1|8|[29]5*4)|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(5|65*4|(0|7|65*6)(3|[29]5*6)*(1|8|[29]5*4))))|(6|[18]5*3|(2|9|[18]5*6)(3|[29]5*6)*(0|7|[29]5*3)|(3|[18]5*[07]|(2|9|[18]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(4|65*3|(0|7|65*6)(3|[29]5*6)*(0|7|[29]5*3))|(4|[18]5*[18]|(2|9|[18]5*6)(3|[29]5*6)*(5|[29]5*[18])|(3|[18]5*[07]|(2|9|[18]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(1|8|35*3|(4|35*6)(3|[29]5*6)*(0|7|[29]5*3)|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(4|65*3|(0|7|65*6)(3|[29]5*6)*(0|7|[29]5*3)))|(5|[18]5*[29]|(2|9|[18]5*6)(3|[29]5*6)*(6|[29]5*[29])|(3|[18]5*[07]|(2|9|[18]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))|(4|[18]5*[18]|(2|9|[18]5*6)(3|[29]5*6)*(5|[29]5*[18])|(3|[18]5*[07]|(2|9|[18]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(0|7|35*[29]|(4|35*6)(3|[29]5*6)*(6|[29]5*[29])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))))(4|[07]5*[29]|(1|8|[07]5*6)(3|[29]5*6)*(6|[29]5*[29])|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))|(3|[07]5*[18]|(1|8|[07]5*6)(3|[29]5*6)*(5|[29]5*[18])|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(0|7|35*[29]|(4|35*6)(3|[29]5*6)*(6|[29]5*[29])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))))*(5|[07]5*3|(1|8|[07]5*6)(3|[29]5*6)*(0|7|[29]5*3)|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(4|65*3|(0|7|65*6)(3|[29]5*6)*(0|7|[29]5*3))|(3|[07]5*[18]|(1|8|[07]5*6)(3|[29]5*6)*(5|[29]5*[18])|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(1|8|35*3|(4|35*6)(3|[29]5*6)*(0|7|[29]5*3)|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(4|65*3|(0|7|65*6)(3|[29]5*6)*(0|7|[29]5*3)))))(2|9|45*3|(5|45*6)(3|[29]5*6)*(0|7|[29]5*3)|(6|45*[07]|(5|45*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(4|65*3|(0|7|65*6)(3|[29]5*6)*(0|7|[29]5*3))|(0|7|45*[18]|(5|45*6)(3|[29]5*6)*(5|[29]5*[18])|(6|45*[07]|(5|45*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(1|8|35*3|(4|35*6)(3|[29]5*6)*(0|7|[29]5*3)|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(4|65*3|(0|7|65*6)(3|[29]5*6)*(0|7|[29]5*3)))|(1|8|45*[29]|(5|45*6)(3|[29]5*6)*(6|[29]5*[29])|(6|45*[07]|(5|45*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))|(0|7|45*[18]|(5|45*6)(3|[29]5*6)*(5|[29]5*[18])|(6|45*[07]|(5|45*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(0|7|35*[29]|(4|35*6)(3|[29]5*6)*(6|[29]5*[29])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))))(4|[07]5*[29]|(1|8|[07]5*6)(3|[29]5*6)*(6|[29]5*[29])|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))|(3|[07]5*[18]|(1|8|[07]5*6)(3|[29]5*6)*(5|[29]5*[18])|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(0|7|35*[29]|(4|35*6)(3|[29]5*6)*(6|[29]5*[29])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))))*(5|[07]5*3|(1|8|[07]5*6)(3|[29]5*6)*(0|7|[29]5*3)|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(4|65*3|(0|7|65*6)(3|[29]5*6)*(0|7|[29]5*3))|(3|[07]5*[18]|(1|8|[07]5*6)(3|[29]5*6)*(5|[29]5*[18])|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(1|8|35*3|(4|35*6)(3|[29]5*6)*(0|7|[29]5*3)|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(4|65*3|(0|7|65*6)(3|[29]5*6)*(0|7|[29]5*3)))))*(3|45*4|(5|45*6)(3|[29]5*6)*(1|8|[29]5*4)|(6|45*[07]|(5|45*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(5|65*4|(0|7|65*6)(3|[29]5*6)*(1|8|[29]5*4))|(0|7|45*[18]|(5|45*6)(3|[29]5*6)*(5|[29]5*[18])|(6|45*[07]|(5|45*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(2|9|35*4|(4|35*6)(3|[29]5*6)*(1|8|[29]5*4)|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(5|65*4|(0|7|65*6)(3|[29]5*6)*(1|8|[29]5*4)))|(1|8|45*[29]|(5|45*6)(3|[29]5*6)*(6|[29]5*[29])|(6|45*[07]|(5|45*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))|(0|7|45*[18]|(5|45*6)(3|[29]5*6)*(5|[29]5*[18])|(6|45*[07]|(5|45*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(0|7|35*[29]|(4|35*6)(3|[29]5*6)*(6|[29]5*[29])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))))(4|[07]5*[29]|(1|8|[07]5*6)(3|[29]5*6)*(6|[29]5*[29])|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))|(3|[07]5*[18]|(1|8|[07]5*6)(3|[29]5*6)*(5|[29]5*[18])|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(0|7|35*[29]|(4|35*6)(3|[29]5*6)*(6|[29]5*[29])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(3|65*[29]|(0|7|65*6)(3|[29]5*6)*(6|[29]5*[29]))))*(6|[07]5*4|(1|8|[07]5*6)(3|[29]5*6)*(1|8|[29]5*4)|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(5|65*4|(0|7|65*6)(3|[29]5*6)*(1|8|[29]5*4))|(3|[07]5*[18]|(1|8|[07]5*6)(3|[29]5*6)*(5|[29]5*[18])|(2|9|[07]5*[07]|(1|8|[07]5*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))(6|35*[18]|(4|35*6)(3|[29]5*6)*(5|[29]5*[18])|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(2|9|65*[18]|(0|7|65*6)(3|[29]5*6)*(5|[29]5*[18])))*(2|9|35*4|(4|35*6)(3|[29]5*6)*(1|8|[29]5*4)|(5|35*[07]|(4|35*6)(3|[29]5*6)*(4|[29]5*[07]))(1|8|65*[07]|(0|7|65*6)(3|[29]5*6)*(4|[29]5*[07]))*(5|65*4|(0|7|65*6)(3|[29]5*6)*(1|8|[29]5*4))))))*$
which is “only” 12,733 characters, and works fine in the implementations I’ve tried.
PS. To see how much less interesting it is using recursion, here’s a Perl program that tests for divisibility by 7 using a recursive regular expression:
my $re = qr{
^($|[07](?1)|[18](?2)|[29](?3)|3(?4)|4(?5)|5(?6)|6(?7))
|(?!)(?:
([07](?4)|[18](?5)|[29](?6)|3(?7)|4(?1)|5(?2)|6(?3))
| ([07](?7)|[18](?1)|[29](?2)|3(?3)|4(?4)|5(?5)|6(?6))
| ([07](?3)|[18](?4)|[29](?5)|3(?6)|4(?7)|5(?1)|6(?2))
| ([07](?6)|[18](?7)|[29](?1)|3(?2)|4(?3)|5(?4)|6(?5))
| ([07](?2)|[18](?3)|[29](?4)|3(?5)|4(?6)|5(?7)|6(?1))
| ([07](?5)|[18](?6)|[29](?7)|3(?1)|4(?2)|5(?3)|6(?4))
)
}x;
while (<>) {
print(/$re/ ? "matches\n" : "does not match\n");
}
This is just a direct expression of the DFA as a recursive regular expression.
If you are allowed to modify the string and repeat, you can do one step of long division at a time. sed syntax for 7: reapply until you get the remainder. Stop when you have a single 0, 1, 2, 3, 4, 5, or 6.
s/^0//
s/^7/0/
s/^8/1/
s/^9/2/
s/^([18]4|[29][18]|35|4[29]|56|63)/0/
s/^([18]5|[29][29]|36|43|5[07]|64)/1/
s/^([18]6|[29]3|3[07]|44|5[18]|65)/2/
s/^([18][07]|[29]4|3[18]|45|5[29]|66)/3/
s/^([18][18]|[29]5|3[29]|46|53|6[07])/4/
s/^([18][29]|[29]6|33|4[07]|54|6[18])/5/
s/^([18]3|[29][07]|34|4[18]|55|6[29])/6/
But it's very weak sauce. Easier to do away with the "regex" altogether and just read in a character at a time and switch to the appropriate state. If that's not an option, then I suspect you are out of luck for 7 and 13, although 11 might still be possible.
If your numbers are unary based, you can use this regex: s/1{divisor}//g
then test if the number is empty.
Here is a Perl way to do it
my @divs = (2,3,5,7,11,13);
for my $num(2..26) {
my $unary = '1'x$num; # convert num to unary
print "\n$num can be divided by : ";
for(@divs) {
my $test = $unary;
$test =~ s/1{$_}//g;
print "$_, " unless $test;
}
}
output:
2 can be divided by : 2,
3 can be divided by : 3,
4 can be divided by : 2,
5 can be divided by : 5,
6 can be divided by : 2, 3,
7 can be divided by : 7,
8 can be divided by : 2,
9 can be divided by : 3,
10 can be divided by : 2, 5,
11 can be divided by : 11,
12 can be divided by : 2, 3,
13 can be divided by : 13,
14 can be divided by : 2, 7,
15 can be divided by : 3, 5,
16 can be divided by : 2,
17 can be divided by :
18 can be divided by : 2, 3,
19 can be divided by :
20 can be divided by : 2, 5,
21 can be divided by : 3, 7,
22 can be divided by : 2, 11,
23 can be divided by :
24 can be divided by : 2, 3,
25 can be divided by : 5,
26 can be divided by : 2, 13,
As interesting as this question is, I don't believe it's possible for anything other than the "obvious" ones you list.
Most of the divisibility rules require mathematical manipulation.
You could use a lookahead to test the string for more than one requirement, so that you can combine pairs of the "obvious" ones together (2 x 3, 3 x 5, etc.):
Matching a 6-letter word is easy with
\b\w{6}\b
. Matching a word containing "cat" is equally easy:\b\w*cat\w*\b
.Combining the two, we get:
(?=\b\w{6}\b)\b\w*cat\w*\b
Analyze this regular expression with RegexBuddy. Easy! Here's how this works. At each character position in the string where the regex is attempted, the engine will first attempt the regex inside the positive lookahead. This sub-regex, and therefore the lookahead, matches only when the current character position in the string is at the start of a 6-letter word in the string. If not, the lookahead will fail, and the engine will continue trying the regex from the start at the next character position in the string.
Here is a generic straight forward recursive brute force long division. It's not optimized and definitely not elegant lol. It's in JavaScript (below is a test html page with the code):
<!doctype html>
<html>
<head>
<script type="text/javascript">
function isNDivisibleByM(N, M) {
var copyN = N;
var MLength = (""+M).length;
var multiples = [];
for(var x = 0; x < M; x++)
multiples[x] = [];
for(var i = M, x=0; i < M*10; x=0) {
for(var j = i; j < i+M; j++, x++)
multiples[x].push(j);
i+=M;
}
var REs = [];
for(var x = 0; x < M; x++)
REs[x] = new RegExp("^("+multiples[x].join("|")+")");
while(N.length >= MLength) {
var sameLen = (N.length == MLength);
for(var x = 0; x < M; x++)
N = N.replace(REs[x], (x==0)?"":(""+x));
N = N.replace(/^0/g, "");
if(sameLen) break;
}
N = N.replace(/^0/g, "");
var numericN = parseInt(copyN);
if(N.length == 0) {
if(numericN%M!=0) {
console.error("Wrong claim: " + copyN + " NOT divisible by " + M);
}
return true;
}
if(numericN%M==0 && N.length != 0) {
console.error("Missed claim: " + copyN + " IS divisible by " + M + " - " + N + " is: " + copyN);
}
return false;
}
function run() {
alert(isNDivisibleByM((""+document.getElementById("N").value), parseInt(document.getElementById("M").value)));
}
</script>
</head>
<body>
<label>N:</label><input type="text" id="N" />
<label>M:</label><input type="text" id="M" />
<button onclick="run()">Is N divisible by M?</button>
</body>
</html>
The idea is for M = 7 (example):
while(numStr.length > 1) {
numStr = numStr.replace(/^(14|21|35|42|56|63)/, "");
numStr = numStr.replace(/^(15|22|36|43|50|64)/, "1");
numStr = numStr.replace(/^(16|23|30|44|51|65)/, "2");
numStr = numStr.replace(/^(10|24|31|45|52|66)/, "3");
numStr = numStr.replace(/^(11|25|32|46|53|60)/, "4");
numStr = numStr.replace(/^(12|26|33|40|54|61)/, "5");
numStr = numStr.replace(/^(13|20|34|41|55|62)/, "6");
}
The perhaps-surprising result is that such a regular expression always exists. The much-less-surprising one is that it's usually not useful.
The existence result comes from the correspondence between deterministic finite automata (DFA) and regular expressions. So let's make a DFA. Denote the modulus by N (it doesn't need to be prime) and denote the numerical base by B, which is 10 for ordinary decimal numbers. The DFA with N states labelled 0 through N-1. The initial state is 0. The symbols of the DFA are the digits 0 through B-1. The states represent the remainder of the left-prefix of the input string, interpreted as an integer, when divided by N. The edges represent the change of state when you add a digit to the right. Arithmetically, this is the state map S(state,digit) = B * state + digit (modulo N). The accepting state is 0, since a zero remainder indicates divisibility. So we have a DFA. The languages recognized by the DFA are the same as that recognized by regular expressions, so one exists. So while this is interesting, it's not helpful, since it doesn't tell you much about how to determine the expression.
If you want a generic algorithm, it's easy to build such a DFA at run-time and populate its state table by direct computation. Initialization is just a pair of nested loops with run time O(M * N). Recognition with the machine is a constant time per input character. This is perfectly fast, but doesn't use a regexp library, if that's what you really need.
In getting toward an actual regular expression, we need to look at Fermat's Little Theorem. From the theorem, we know that B^(N-1) == 1 (modulo N). For example, when N=7 and B=10, what this means is that every block of 6 digits is equivalent to some single digit in the range 0 .. 6 for the purpose of divisibility. The exponent can be smaller than N-1; in general it's a factor of the Euler totient function of N. Call the size of the block D. There are N regular expressions for blocks of D digits, each one representing a particular equivalence class of remainders modulo N. At most, these expressions have length O(B^D), which is large. For N=7 that's a set of regular expressions a million characters long; I would imagine that would break most regexp libraries.
This relates to how the expression in the example code works; the expression (?1)
is matching strings that are equal to 0 (mod 3). This works with N=3 since 10^1 == 1 (mod 3) which means that A0B == AB (mod 3). This is more complicated when the exponent is greater than 1, but the principle is the same. (Note that the example code uses a recognizer which is more than just regular expressions, strictly speaking.) The expressions [0369]
, [147]
, and [258]
are the regular expressions for the digits 0, 1, and 2 in a modulo 3 expression. Generalizing, you would use the regexp-digits as above in an analogous manner.
I'm not providing code because (1) it would take longer to write than this answer has been, and (2) I really doubt it would execute in any known implementation.