Recently I have been trying to create in Haskell a regex interpretor. What I did was create a new data type with all possible constructors (for sequence, *, ^, intervals, etc) and then define a matcher function. It works wonders but my problem is that I have to convert the input (the String, for example "a(b*)(c|d)ef") to my data type ("Seq (Sym a) (Seq (Rep Sym b) (Seq (Or Sym c Sym d) Sym ef))"). I am having trouble with this part of the problem (I tried creating a new data type, a parsing tree, but I failed completely). Any ideas on how I could solve it?
Asked
Active
Viewed 1,157 times
4
rid
- 61,078
- 31
- 152
- 193
Iulia Muntianu
- 145
- 1
- 14
-
In case you're not building this just for fun, there is also Text.Regex – Jani Hartikainen Apr 18 '12 at 12:52
-
1http://www.haskell.org/haskellwiki/Parsec: I don't know its details, but it's a really good library for parsing... playing with it also teaches you many things about monads. – Riccardo T. Apr 18 '12 at 12:53
2 Answers
8
The canonical approach is to use a parser combinator library, such as Parsec. Parser combinator libraries (like parser generators) let you write descriptions of your grammar, yielding a parser from strings to tokens in that language.
You simply have to encode your grammar as a Parsec function.
As an example, see this previous SO question: Using Parsec to parse regular expressions
Community
- 1
- 1
Don Stewart
- 137,316
- 36
- 365
- 468
-
@Jani Hartikainen: Unfortunately I am not allowed to use Text.Regex – Iulia Muntianu Apr 18 '12 at 12:56
-
Thank you, I will try to use Parsec and see what I manage to do – Iulia Muntianu Apr 18 '12 at 12:58
4
That's an interesting article (a play) on the implementation of regular expressions:
JJJ
- 2,731
- 1
- 12
- 23