Regex Utils

A set of utils to work with Regular Expressions in Python

WARNING: most functions are still under development, please report any issues you find, or feel free to contribute!

Main Features

• Generate strings that match a given regex
• Intersect two regular expressions
• Negate a regular expression (experimental)
• Convert a NFA back into a regex

Use Python native sre_parse to parse a regex, then convert it into a Nondeterministic Finite Automaton

Once you have this NFA, you can use all the features in this package. The regex module offers an abstraction for NFA-based operations.

Getting Started

First of all, install the package:

pip install regex-utils

Then, you can use all functions in the regex module by importing it:

from regex_utils import regex


print(regex.from_string("abc").generate_sample())
# Output: "abc"

String Generation

Generate strings that match the given regex, by performing a random walk over the generated NFA.

regex.from_string("[a-z]{5}").generate_sample()
# Output: "xfkdy"

Intersection

Intersect two regular expressions.

This is the equivalent of having to match both regex, one after the other. The advantage is that you can now use the to_string function to generate the resulting regular expression.

This is also useful if you want to “compile” lookarounds into the regex itself.

Finally, you can use intersection to generate strings of arbitrary length. For example, if you had the ab+c* regex, and you wanted a minimum length of 3 characters and a maximum length of 10 character for your string, you could intersect the regex like this:

r = regex.intersect("ab+c*", ".{3,10}")

You could then generate strings using the resulting regex:

r.generate_sample()
# Returns: abc   -- Warning: this is not deterministic

To get the final regex, you can also use the to_string function

regex.intersect("a*b*", "\w{5}").to_string()
# Returns: '(?:(?:(?:(?:[ab]b|aa)b|aaa)b|aaaa)b|aaaaa)'

Please notice that Intersections could result in empty regex. For example, if you tried to intersect two regex with nothing in common, you would receive an empty regex as a result.

regex.intersect("[a-z]", "[^a-z]").to_string()
# Returns: ''

Negation (experimental)

Complement the original NFA by converting accepting states to non-accepting states, and add all missing transitions (this is used to generate random strings). The logic is similar to a Negative Lookahead in PCRE.

WARNING: this feature is currently experimental, and it contains some known bugs for specific scenarios.

Please open a issue if you find any specific bugs related to this feature 🙏

To String

Once you have the resulting NFA, you can get the regex back in plain text, so that you can use it in other tools.

This could be useful for example if you wanted to get the intersection of two regex, or the negation of one, and put it in your application.

regex.intersect("a*b*", "\w{5}").to_string()
# Returns: '(?:(?:(?:(?:[ab]b|aa)b|aaa)b|aaaa)b|aaaaa)'

An interesting use case is converting Lookaheads into a regex that can be used in non-PCRE compliant engines, such as the native ones in Go or Rust. This specific feature is also a work in progress for this package. For example, if you wanted strings that match the (?!abc).* regex, you could write it like this:

regex.negate("abc").to_string()
# Returns: '(?:(?:[^a]|a(?:[^b]|b[^c])))(?:\\.)*|(?:ab?)?'