Portability | non-portable (local universal quantification) |
---|---|

Stability | provisional |

Maintainer | libraries@haskell.org |

This is a library of parser combinators, originally written by Koen Claessen. It parses all alternatives in parallel, so it never keeps hold of the beginning of the input string, a common source of space leaks with other parsers. The '(+++)' choice combinator is genuinely commutative; it makes no difference which branch is "shorter".

- data ReadP a
- get :: ReadP Char
- look :: ReadP String
- (+++) :: ReadP a -> ReadP a -> ReadP a
- (<++) :: ReadP a -> ReadP a -> ReadP a
- gather :: ReadP a -> ReadP (String, a)
- pfail :: ReadP a
- eof :: ReadP ()
- satisfy :: (Char -> Bool) -> ReadP Char
- char :: Char -> ReadP Char
- string :: String -> ReadP String
- munch :: (Char -> Bool) -> ReadP String
- munch1 :: (Char -> Bool) -> ReadP String
- skipSpaces :: ReadP ()
- choice :: [ReadP a] -> ReadP a
- count :: Int -> ReadP a -> ReadP [a]
- between :: ReadP open -> ReadP close -> ReadP a -> ReadP a
- option :: a -> ReadP a -> ReadP a
- optional :: ReadP a -> ReadP ()
- many :: ReadP a -> ReadP [a]
- many1 :: ReadP a -> ReadP [a]
- skipMany :: ReadP a -> ReadP ()
- skipMany1 :: ReadP a -> ReadP ()
- sepBy :: ReadP a -> ReadP sep -> ReadP [a]
- sepBy1 :: ReadP a -> ReadP sep -> ReadP [a]
- endBy :: ReadP a -> ReadP sep -> ReadP [a]
- endBy1 :: ReadP a -> ReadP sep -> ReadP [a]
- chainr :: ReadP a -> ReadP (a -> a -> a) -> a -> ReadP a
- chainl :: ReadP a -> ReadP (a -> a -> a) -> a -> ReadP a
- chainl1 :: ReadP a -> ReadP (a -> a -> a) -> ReadP a
- chainr1 :: ReadP a -> ReadP (a -> a -> a) -> ReadP a
- manyTill :: ReadP a -> ReadP end -> ReadP [a]
- type ReadS a = String -> [(a, String)]
- readP_to_S :: ReadP a -> ReadS a
- readS_to_P :: ReadS a -> ReadP a

# The `ReadP`

type

# Primitive operations

Look-ahead: returns the part of the input that is left, without consuming it.

(<++) :: ReadP a -> ReadP a -> ReadP aSource

Local, exclusive, left-biased choice: If left parser locally produces any result at all, then right parser is not used.

gather :: ReadP a -> ReadP (String, a)Source

Transforms a parser into one that does the same, but
in addition returns the exact characters read.
IMPORTANT NOTE: `gather`

gives a runtime error if its first argument
is built using any occurrences of readS_to_P.

# Other operations

satisfy :: (Char -> Bool) -> ReadP CharSource

Consumes and returns the next character, if it satisfies the specified predicate.

munch :: (Char -> Bool) -> ReadP StringSource

Parses the first zero or more characters satisfying the predicate. Always succeds, exactly once having consumed all the characters Hence NOT the same as (many (satisfy p))

munch1 :: (Char -> Bool) -> ReadP StringSource

Parses the first one or more characters satisfying the predicate. Fails if none, else succeeds exactly once having consumed all the characters Hence NOT the same as (many1 (satisfy p))

Skips all whitespace.

count :: Int -> ReadP a -> ReadP [a]Source

`count n p`

parses `n`

occurrences of `p`

in sequence. A list of
results is returned.

between :: ReadP open -> ReadP close -> ReadP a -> ReadP aSource

`between open close p`

parses `open`

, followed by `p`

and finally
`close`

. Only the value of `p`

is returned.

option :: a -> ReadP a -> ReadP aSource

`option x p`

will either parse `p`

or return `x`

without consuming
any input.

sepBy :: ReadP a -> ReadP sep -> ReadP [a]Source

`sepBy p sep`

parses zero or more occurrences of `p`

, separated by `sep`

.
Returns a list of values returned by `p`

.

sepBy1 :: ReadP a -> ReadP sep -> ReadP [a]Source

`sepBy1 p sep`

parses one or more occurrences of `p`

, separated by `sep`

.
Returns a list of values returned by `p`

.

endBy :: ReadP a -> ReadP sep -> ReadP [a]Source

`endBy p sep`

parses zero or more occurrences of `p`

, separated and ended
by `sep`

.

endBy1 :: ReadP a -> ReadP sep -> ReadP [a]Source

`endBy p sep`

parses one or more occurrences of `p`

, separated and ended
by `sep`

.

chainr :: ReadP a -> ReadP (a -> a -> a) -> a -> ReadP aSource

`chainr p op x`

parses zero or more occurrences of `p`

, separated by `op`

.
Returns a value produced by a *right* associative application of all
functions returned by `op`

. If there are no occurrences of `p`

, `x`

is
returned.

chainl :: ReadP a -> ReadP (a -> a -> a) -> a -> ReadP aSource

`chainl p op x`

parses zero or more occurrences of `p`

, separated by `op`

.
Returns a value produced by a *left* associative application of all
functions returned by `op`

. If there are no occurrences of `p`

, `x`

is
returned.

chainl1 :: ReadP a -> ReadP (a -> a -> a) -> ReadP aSource

Like `chainl`

, but parses one or more occurrences of `p`

.

chainr1 :: ReadP a -> ReadP (a -> a -> a) -> ReadP aSource

Like `chainr`

, but parses one or more occurrences of `p`

.

manyTill :: ReadP a -> ReadP end -> ReadP [a]Source

`manyTill p end`

parses zero or more occurrences of `p`

, until `end`

succeeds. Returns a list of values returned by `p`

.

# Running a parser

readP_to_S :: ReadP a -> ReadS aSource

Converts a parser into a Haskell ReadS-style function.
This is the main way in which you can "run" a `ReadP`

parser:
the expanded type is
` readP_to_S :: ReadP a -> String -> [(a,String)] `

readS_to_P :: ReadS a -> ReadP aSource

Converts a Haskell ReadS-style function into a parser. Warning: This introduces local backtracking in the resulting parser, and therefore a possible inefficiency.

# Properties

The following are QuickCheck specifications of what the combinators do. These can be seen as formal specifications of the behavior of the combinators.

We use bags to give semantics to the combinators.

type Bag a = [a]

Equality on bags does not care about the order of elements.

(=~) :: Ord a => Bag a -> Bag a -> Bool xs =~ ys = sort xs == sort ys

A special equality operator to avoid unresolved overloading when testing the properties.

(=~.) :: Bag (Int,String) -> Bag (Int,String) -> Bool (=~.) = (=~)

Here follow the properties:

prop_Get_Nil = readP_to_S get [] =~ [] prop_Get_Cons c s = readP_to_S get (c:s) =~ [(c,s)] prop_Look s = readP_to_S look s =~ [(s,s)] prop_Fail s = readP_to_S pfail s =~. [] prop_Return x s = readP_to_S (return x) s =~. [(x,s)] prop_Bind p k s = readP_to_S (p >>= k) s =~. [ ys'' | (x,s') <- readP_to_S p s , ys'' <- readP_to_S (k (x::Int)) s' ] prop_Plus p q s = readP_to_S (p +++ q) s =~. (readP_to_S p s ++ readP_to_S q s) prop_LeftPlus p q s = readP_to_S (p <++ q) s =~. (readP_to_S p s +<+ readP_to_S q s) where [] +<+ ys = ys xs +<+ _ = xs prop_Gather s = forAll readPWithoutReadS $ \p -> readP_to_S (gather p) s =~ [ ((pre,x::Int),s') | (x,s') <- readP_to_S p s , let pre = take (length s - length s') s ] prop_String_Yes this s = readP_to_S (string this) (this ++ s) =~ [(this,s)] prop_String_Maybe this s = readP_to_S (string this) s =~ [(this, drop (length this) s) | this `isPrefixOf` s] prop_Munch p s = readP_to_S (munch p) s =~ [(takeWhile p s, dropWhile p s)] prop_Munch1 p s = readP_to_S (munch1 p) s =~ [(res,s') | let (res,s') = (takeWhile p s, dropWhile p s), not (null res)] prop_Choice ps s = readP_to_S (choice ps) s =~. readP_to_S (foldr (+++) pfail ps) s prop_ReadS r s = readP_to_S (readS_to_P r) s =~. r s