How to use it

The syntax is a little bit different than python’s re, and similar to that of pyparsing, you have to more-or-less explicitly build the syntax tree of your regular expression. For instance:

“ab” is Literal(“a”) + Literal(“b”)

“a*” is Star(Literal(“a”))

“(ab)+|(bb)*?” is:

a = Literal(“a”) b = Literal(“b”) regex = Plus(a + b) | Star(b + b, greedy=False)

You can also assign a group to any sub-match and later on retrieve the matched content, for instance:

regex = Group(Plus(a + b), “foobar”) | (b + b) m = match(regex, “abab”) print m.span(“foobar”) # prints (0, 4)

For more, check out the examples in the examples folder.

How we use it

At Machinalis we use REfO for applications similar to that in examples/words.py, check it out!

About the implementation

I use a Thompson-like virtual machine aproach, which ensures polynomial time worst-case complexity. See examples/poly_time.py for an example of this.

The implementation is heavily based on Russ Cox notes, see http://swtch.com/~rsc/regexp/regexp2.html for the source.

If you go to read the code, some glossary:

• RE – regular expression

• VM – virtual machine

• Epsilon transitions – All VM instructions that do not consume a symbol

  or stop the thread (for example an Accept).

Acknowledgements

Thanks Russ Cox for sharing the awesome info and insights on your web site.

Thanks Javier Mansilla for reviewing the code and being enthusiastic about it.

Thanks Machinalis for everything :)