Package

purescript-eulalie

Repository
bodil/purescript-eulalie
License
LGPL-3.0+
Uploaded by
bodil

Fast string parser combinators for PureScript.

Documentation

Read the API documentation.

Usage

Eulalie works on the principle of constructing parsers from smaller parsers using various combinator functions.

A parser is a function which takes an input Stream, and returns a ParseResult value which can be either a success or an error.

The type of parsers is defined like this:

newtype Parser i o = Parser (Stream i -> ParseResult i o)

Data Types

newtype Stream i = Stream i { buffer :: Array i, cursor :: Int }

A Stream just contains an array of input data, and an index into this array. We use this structure instead of passing arrays around as input because array operations are expensive, while any operation on the Stream can be performed in linear time, and while many Streams will be created during a parse operation, we only ever keep a single copy of the array they wrap.

data ParseResult i o = Success (ParseSuccess i o)
                     | Error (ParseError i)

A ParseResult is what's returned from a parser, and signals whether it succeeded or failed. It wraps one of two result values, ParseSuccess and ParseError.

type ParseSuccess i o =
  { value :: o
  , next :: Stream i
  , start :: Stream i
  , matched :: Array i
  }

A ParseSuccess contains four properties: the value we parsed (an arbitrary value), the next input to be parsed (a Stream), the point in the stream where we started parsing (also a Stream), and the sub-array that was matched by this parser.

type ParseError i =
  { input :: Stream i
  , expected :: Set String
  , fatal :: Boolean
  }

Finally, a ParseError simply contains an input property (a Stream) which points to the exact position where the parsing failed, and a set of string descriptions of expected inputs. It also contains a fatal flag, which signifies to the either combinator that we should stop parsing immediately instead of trying further parsers.

Parser Combinators

The most basic parsers form the building blocks from which you can assemble more complex parsers:

  • succeed :: forall i o. o -> Parser i o makes a parser which doesn't consume input, just returns the provided value wrapped in a ParseSuccess.
  • fail :: forall i o. Parser i o is a parser which consumes no input and returns a ParseError.
  • item :: forall i. Parser i i is a parser which consumes one arbitrary input value and returns it as a ParseSuccess.

The two fundamental parser combinators are:

  • seq :: forall i a b. Parser i a -> (a -> Parser i b) -> Parser i b is used to combine multiple parsers in a sequence. It takes a parser, and a function which will be called with the result of the parser if it succeeded, and must return another parser, which will be run on the remaining input. The result of the combined parser will be the result of this last parser, or the first error encountered.

    (This corresponds to the bind/>>= method on the Monad type class.)

  • either :: forall i o. Parser i o -> Parser i o -> Parser i o makes a parser which will first try the first provided parser, and returns its result if it succeeds. If it fails, it will run the second parser on the same input, and return its result directly, whether or not it succeeded.

    If you've heard the term "backtracking" in relation to parsers, this is handled automatically by the either function, and you don't need to worry about it.

    (This corresponds to the alt/<|> method on the Alt type class.)

Using these, you can construct more advanced parser combinators. Some particularly useful combinators are predefined:

  • sat :: forall i. (i -> Boolean) -> Parser i i makes a parser which will match one input value only if the provided predicate function returns true for it.
  • many :: forall i o. Parser i o -> Parser i (List o) makes a parser which will match the provided parser zero or more times.
  • many1 :: forall i o. Parser i o -> Parser i (List o) works just like many, but requires at minimum one match.
  • char :: Char -> Parser Char Char makes a parser which matches a specific single character.
  • string :: String -> Parser Char String makes a parser which matches the provided string exactly (which is done by using the char parser for each Char in the string).

Other predefined parsers are digit, space, alphanum, letter, upper and lower, which match one character of their respective types, and their inverse counterparts, notDigit, notSpace, notAlphanum, notLetter, notUpper and notLower. There are also whitespace matchers spaces and spaces1, and their opposites, notSpaces and notSpaces1.

Do Notation

Because parsers implement the Monad type class, and the seq combinator is actually the monadic bind operation, you can combine parsers using do notation, like this:

import Data.Eulalie.Parser as P
import Data.Eulalie.String as S

-- a parser for the string "hi <your name>!",
-- returning the "<your name>" part.

-- without do notation:
myParser = P.seq (S.string "hi") \_ ->
  P.seq S.spaces1 \_ ->
    P.seq S.notSpaces1 \name ->
      P.seq S.string "!" \_ ->
        P.succeed name

-- with do notation:
myParser = do
  S.string "hi"
  S.spaces1
  name <- S.notSpaces1
  S.string "!"
  return name

Alternatives

Because Parser implements the MonadPlus type class (and either is the implementation for alt), you can also use this alternative syntax for trying several parsers until one succeeds:

import Control.Alt ((<|>))
import Data.Eulalie.Parser as P
import Data.Eulalie.String as S

-- a parser matching one of the strings "lol", "rofl" and "lmao"

-- using `either`:
myParser = P.either (S.string "lol")
  (P.either (S.string "rofl" S.String "lmao"))

-- using <|>:
myParser = S.string "lol" <|> S.string "rofl" <|> S.string "lmao"

Monoids

For a Parser i o where o is a monoid, there's a type class implementation for (Monoid o) => Monoid (Parser i o), so that you can treat parsers for monoids like they're monoids too.

What this means, practically, is that because strings are monoids, you can do things like this for parsers of type Parser Char String, concatenating the results of each parser into a final result:

import Data.Eulalie.Parser as P
import Data.Eulalie.String as S
import Data.Foldable (fold)

-- a parser matching the whole string "hi <your name>!"

-- using the semigroup append operator:
myParser = S.string "hi" <> S.spaces1 <> S.notSpaces1 <> S.string "!"

-- using a fold over a list of parsers:
myParser = fold [S.string "hi", S.spaces1, S.notSpaces1, S.string "!"]

A Working Example

This is how you might write a parser for the first line of an HTTP request:

import Data.Eulalie.Parser as P
import Data.Eulalie.String as S
import Data.Eulalie.Char as C
import Data.Eulalie.Stream (stream)
import Data.String (toCharArray)

type HTTPRequest = { method :: String,
                     path :: String,
                     version :: String }

parser :: Parser Char HTTPRequest
parser = do
  -- Parse a sequence of 1 or more upper case letters
  method <- C.many1 C.upper
  -- Consume 1 or more spaces
  S.spaces1
  -- Parse a sequence of 1 or more non-whitespace characters
  path <- S.notSpaces1
  -- Consume 1 or more spaces
  S.spaces1
  -- Match the string "HTTP/"
  S.string "HTTP/"
  -- Parse the version string
  version <- C.many1 C.digit <> S.string "." <> C.many1 C.digit
  -- Return the final parsed value
  return { method, path, version }

result = P.parse parser (stream $ toCharArray "GET /lol.gif HTTP/1.0")
-- { method:  "GET",
--   path:    "/lol.gif",
--   version: "1.0" }

Licence

Copyright 2015 Bodil Stokke

This program is free software: you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details.

You should have received a copy of the GNU Lesser General Public License along with this program. If not, see http://www.gnu.org/licenses/.