Regular expressions versus lexical analyzers in Haskell

Question

I'm getting started with Haskell and I'm trying to use the Alex tool to create regular expressions and I'm a little bit lost; my first inconvenience was the compile part. How I have to do to compile a file with Alex?. Then, I think that I have to import into my code the modules that alex generates, but not sure. If someone can help me, I would be very greatful!

Answer 1

Why do you want to use alex to create regular expressions? If all you want is to do some regex matching etc, you should look at the regex-base package.

Answer 2

If it is plain Regex you want, the API is specified in text.regex.base. Then there are the implementations text.regex.Posix , text.regex.pcre and several others. The Haddoc documentation is a bit slim, however the basics are described in Real World Haskell, chapter 8. Some more indepth stuff is descriped in this SO question.

Answer 3

You can specify regular expression functions in Alex.

Here for example, a regex in Alex to match floating point numbers:

$space       = [\ \t\xa0]
$digit       = 0-9
$octit       = 0-7
$hexit       = [$digit A-F a-f]

@sign        = [\-\+]
@decimal     = $digit+
@octal       = $octit+
@hexadecimal = $hexit+
@exponent    = [eE] [\-\+]? @decimal

@number      = @decimal
             | @decimal \. @decimal @exponent?
             | @decimal @exponent
             | 0[oO] @octal
             | 0[xX] @hexadecimal

lex :-

   @sign? @number { strtod }

When we match the floating point number, we dispatch to a parsing function to operate on that captured string, which we can then wrap and expose to the user as a parsing function:

readDouble :: ByteString -> Maybe (Double, ByteString)
readDouble str = case alexScan (AlexInput '\n' str) 0 of
    AlexEOF            -> Nothing
    AlexError _        -> Nothing
    AlexToken (AlexInput _ rest) n _ ->
       case strtod (B.unsafeTake n str) of d -> d `seq` Just $! (d , rest)

A nice consequence of using Alex for this regex matching is that the performance is good, as the regex engine is compiled statically. It can also be exposed as a regular Haskell library built with cabal. For the full implementation, see bytestring-lexing.

The general advice on when to use a lexer instead of a regex matcher would be that, if you have a grammar for the lexemes you're trying to match, as I did for floating point, use Alex. If you don't, and the structure is more ad hoc, use a regex engine.