README.md

regex

Experimental implementation of regex; construct NFA/DFA from regex.

Example

Construct NFA/DFA for the regex (a|b)*

import regex

a = regex.NFA.char('a') # construct NFA of regex 'a'
b = regex.NFA.char('b') # construct NFA of regex 'b'
union = a.union(b) # construct NFA of regex 'a|b'
closure = union.closure() # construct NFA of regex '(a|b)*' (kleene closure)

print(closure.to_graphviz()) 

Output:

Digraph G {
        node [shape=circle]
        rankdir = LR
        0 [shape=plaintext, label="start"]
        3 [shape=doublecircle]

        0 -> 6
        1 ->  2 [label = " ε"]
        2 ->  3 [label = " ε"]
        4 ->  5 [label = " ε"]
        6 ->  3 [label = " ε"]
        2 ->  4 [label = " ε"]
        7 ->  1 [label = " b"]
        6 ->  4 [label = " ε"]
        8 ->  2 [label = " ε"]
        4 ->  7 [label = " ε"]
        5 ->  8 [label = " a"]
}

NFA:

dfa = regex.subset_construction(nfa)
print(dfa.to_graphviz())

Output:

Digraph G {
        node [shape=circle]
        rankdir = LR
        0 [shape=plaintext, label="start"]
        3 [shape=doublecircle]
        2 [shape=doublecircle]
        1 [shape=doublecircle]

        0 -> 1
        1 ->  2 [label = " a"]
        3 ->  3 [label = " b"]
        3 ->  2 [label = " a"]
        2 ->  2 [label = " a"]
        2 ->  3 [label = " b"]
        1 ->  3 [label = " b"]
}

DFA:

min_dfa = regex.dfa_minimize(dfa)
print(min_dfa.to_graphviz())

Output:

Digraph G {
        node [shape=circle]
        rankdir = LR
        0 [shape=plaintext, label="start"]
        1 [shape=doublecircle]

        0 -> 1
        1 ->  1 [label = " b"]
        1 ->  1 [label = " a"]
}

Minimal DFA:

Utility function is provided to simplify the construction of NFA, escape ('\') and extensions ('[]^?') not supported, use carefully:

# Build NFA from regex
nfa = regex.parse('(a|b)*')

Known issue: abc* will faultily parsed as (abc)*.

Reference

Main reference: Engineering: A Compiler 2nd Edition