[erlang-questions] regexp sux!

Robert Virding robert.virding@REDACTED
Wed May 30 02:57:37 CEST 2007

Not really, the tokeniser does that. It means you would have to write 
regexp engine in such a way that it's state is explicit and not implicit 
in a call stack. Doing that now. The only real problem would be how you 
specify that there was no input left to be scanned.


Christian S wrote:
> Would it be easy to make regexps incremental, in such that i can
> feed a regexp a string, and then find out if the string contained enough
> for the regexp to match/halt, or if i need to continue feeding it with
> more data.
> Imagine I am receiving data blocks with an active gen_tcp socket, and
> my potential regexp match is across the boundary of two received data
> blocks.
> I'm imagining an interface somewhat similar to md5_init/md5_update but
> with an exit to let me know there was a match, at which point i would
> get the matched part and the remaining part returned to me so i can
> continue matching on the latter when i have processed the matched
> block.
> On 5/21/07, Darius Bacon <darius@REDACTED> wrote:
>> Luke Gorrie <luke.gorrie@REDACTED> wrote:
>>> Darius Bacon <darius@REDACTED> writes:
>>>> Here's a lightly-tested hack -- I'm not familiar with the existing
>>>> libraries so this is from scratch.
>>> You bloody showoff. :-) Works great!
>> Guilty. :-)
>>> I had to add . (dot) but nothing else really obvious is missing.
>>> Feel free to draw my attention to anything I'm overlooking though :-)
>> There's no attempt to conform to any standard, and the regex parser
>> ought to be replaced with a real one that understands char ranges,
>> inverted char classes, escaping, etc.
>>> Can I steal this code and use it at work?
>> Sure, I guess an MIT license would be appropriate. I'd appreciate
>> getting back any improvements/tests -- don't know when I'll get around
>> to the quickcheck tests.
>>> NB: I would still feel more comfortable if this were quickcheck'd a
>>> bit for divergence etc. I can't help but think that a free quickcheck
>>> clone is likely to surface soon. I hope that John & Thomas are happy
>>> with this idea or if not that they make their one accessible to us
>>> random hackers on the list somehow soon :-)
>> Their automatic error-case shrinking sounds really cool and useful.
>> Darius
>> _______________________________________________
>> erlang-questions mailing list
>> erlang-questions@REDACTED
>> http://www.erlang.org/mailman/listinfo/erlang-questions
> _______________________________________________
> erlang-questions mailing list
> erlang-questions@REDACTED
> http://www.erlang.org/mailman/listinfo/erlang-questions

More information about the erlang-questions mailing list