aboutsummaryrefslogtreecommitdiff
path: root/haddock-library/src/Documentation/Haddock/Parser
diff options
context:
space:
mode:
authorBen Gamari <ben@smart-cactus.org>2021-01-05 15:54:42 -0500
committerBen Gamari <ben@smart-cactus.org>2021-01-05 16:14:31 -0500
commit1e56f63c3197e7ca1c1e506e083c2bad25d08793 (patch)
tree7152e5a53fe1c18e6fd5044d5aa3168ab99c3cc6 /haddock-library/src/Documentation/Haddock/Parser
parent1d657cf377b5f147b08aafb3ab3a5d11be538331 (diff)
parent665226f384ee9b0a66a98638ede9eff845f6c45b (diff)
Merge remote-tracking branch 'origin/ghc-8.10' into ghc-9.0
Diffstat (limited to 'haddock-library/src/Documentation/Haddock/Parser')
-rw-r--r--haddock-library/src/Documentation/Haddock/Parser/Identifier.hs160
-rw-r--r--haddock-library/src/Documentation/Haddock/Parser/Monad.hs16
-rw-r--r--haddock-library/src/Documentation/Haddock/Parser/Util.hs14
3 files changed, 181 insertions, 9 deletions
diff --git a/haddock-library/src/Documentation/Haddock/Parser/Identifier.hs b/haddock-library/src/Documentation/Haddock/Parser/Identifier.hs
new file mode 100644
index 00000000..b8afb951
--- /dev/null
+++ b/haddock-library/src/Documentation/Haddock/Parser/Identifier.hs
@@ -0,0 +1,160 @@
+{-# LANGUAGE BangPatterns #-}
+{-# LANGUAGE ViewPatterns #-}
+-- |
+-- Module : Documentation.Haddock.Parser.Identifier
+-- Copyright : (c) Alec Theriault 2019,
+-- License : BSD-like
+--
+-- Maintainer : haddock@projects.haskell.org
+-- Stability : experimental
+-- Portability : portable
+--
+-- Functionality for parsing identifiers and operators
+
+module Documentation.Haddock.Parser.Identifier (
+ Identifier(..),
+ parseValid,
+) where
+
+import Documentation.Haddock.Types ( Namespace(..) )
+import Documentation.Haddock.Parser.Monad
+import qualified Text.Parsec as Parsec
+import Text.Parsec.Pos ( updatePosChar )
+import Text.Parsec ( State(..)
+ , getParserState, setParserState )
+
+import Data.Text (Text)
+import qualified Data.Text as T
+
+import Data.Char (isAlpha, isAlphaNum)
+import Control.Monad (guard)
+import Data.Maybe
+import CompatPrelude
+
+-- | Identifier string surrounded with namespace, opening, and closing quotes/backticks.
+data Identifier = Identifier !Namespace !Char String !Char
+ deriving (Show, Eq)
+
+parseValid :: Parser Identifier
+parseValid = do
+ s@State{ stateInput = inp, statePos = pos } <- getParserState
+
+ case takeIdentifier inp of
+ Nothing -> Parsec.parserFail "parseValid: Failed to match a valid identifier"
+ Just (ns, op, ident, cl, inp') ->
+ let posOp = updatePosChar pos op
+ posIdent = T.foldl updatePosChar posOp ident
+ posCl = updatePosChar posIdent cl
+ s' = s{ stateInput = inp', statePos = posCl }
+ in setParserState s' $> Identifier ns op (T.unpack ident) cl
+
+
+-- | Try to parse a delimited identifier off the front of the given input.
+--
+-- This tries to match as many valid Haskell identifiers/operators as possible,
+-- to the point of sometimes accepting invalid things (ex: keywords). Some
+-- considerations:
+--
+-- - operators and identifiers can have module qualifications
+-- - operators can be wrapped in parens (for prefix)
+-- - identifiers can be wrapped in backticks (for infix)
+-- - delimiters are backticks or regular ticks
+-- - since regular ticks are also valid in identifiers, we opt for the
+-- longest successful parse
+--
+-- This function should make /O(1)/ allocations
+takeIdentifier :: Text -> Maybe (Namespace, Char, Text, Char, Text)
+takeIdentifier input = listToMaybe $ do
+
+ -- Optional namespace
+ let (ns, input') = case T.uncons input of
+ Just ('v', i) -> (Value, i)
+ Just ('t', i) -> (Type, i)
+ _ -> (None, input)
+
+ -- Opening tick
+ (op, input'') <- maybeToList (T.uncons input')
+ guard (op == '\'' || op == '`')
+
+ -- Identifier/operator
+ (ident, input''') <- wrapped input''
+
+ -- Closing tick
+ (cl, input'''') <- maybeToList (T.uncons input''')
+ guard (cl == '\'' || cl == '`')
+
+ return (ns, op, ident, cl, input'''')
+
+ where
+
+ -- | Parse out a wrapped, possibly qualified, operator or identifier
+ wrapped t = do
+ (c, t' ) <- maybeToList (T.uncons t)
+ -- Tuples
+ case c of
+ '(' | Just (c', _) <- T.uncons t'
+ , c' == ',' || c' == ')'
+ -> do let (commas, t'') = T.span (== ',') t'
+ (')', t''') <- maybeToList (T.uncons t'')
+ return (T.take (T.length commas + 2) t, t''')
+
+ -- Parenthesized
+ '(' -> do (n, t'' ) <- general False 0 [] t'
+ (')', t''') <- maybeToList (T.uncons t'')
+ return (T.take (n + 2) t, t''')
+
+ -- Backticked
+ '`' -> do (n, t'' ) <- general False 0 [] t'
+ ('`', t''') <- maybeToList (T.uncons t'')
+ return (T.take (n + 2) t, t''')
+
+ -- Unadorned
+ _ -> do (n, t'' ) <- general False 0 [] t
+ return (T.take n t, t'')
+
+ -- | Parse out a possibly qualified operator or identifier
+ general :: Bool -- ^ refuse inputs starting with operators
+ -> Int -- ^ total characters \"consumed\" so far
+ -> [(Int, Text)] -- ^ accumulated results
+ -> Text -- ^ current input
+ -> [(Int, Text)] -- ^ total characters parsed & what remains
+ general !identOnly !i acc t
+ -- Starts with an identifier (either just an identifier, or a module qual)
+ | Just (n, rest) <- identLike t
+ = if T.null rest
+ then acc
+ else case T.head rest of
+ '`' -> (n + i, rest) : acc
+ ')' -> (n + i, rest) : acc
+ '.' -> general False (n + i + 1) acc (T.tail rest)
+ '\'' -> let (m, rest') = quotes rest
+ in general True (n + m + 1 + i) ((n + m + i, rest') : acc) (T.tail rest')
+ _ -> acc
+
+ -- An operator
+ | Just (n, rest) <- optr t
+ , not identOnly
+ = (n + i, rest) : acc
+
+ -- Anything else
+ | otherwise
+ = acc
+
+ -- | Parse an identifier off the front of the input
+ identLike t
+ | T.null t = Nothing
+ | isAlpha (T.head t) || '_' == T.head t
+ = let !(idt, rest) = T.span (\c -> isAlphaNum c || c == '_') t
+ !(octos, rest') = T.span (== '#') rest
+ in Just (T.length idt + T.length octos, rest')
+ | otherwise = Nothing
+
+ -- | Parse all but the last quote off the front of the input
+ -- PRECONDITION: T.head t == '\''
+ quotes :: Text -> (Int, Text)
+ quotes t = let !n = T.length (T.takeWhile (== '\'') t) - 1
+ in (n, T.drop n t)
+
+ -- | Parse an operator off the front of the input
+ optr t = let !(op, rest) = T.span isSymbolChar t
+ in if T.null op then Nothing else Just (T.length op, rest)
diff --git a/haddock-library/src/Documentation/Haddock/Parser/Monad.hs b/haddock-library/src/Documentation/Haddock/Parser/Monad.hs
index 8f5bd217..7c73a168 100644
--- a/haddock-library/src/Documentation/Haddock/Parser/Monad.hs
+++ b/haddock-library/src/Documentation/Haddock/Parser/Monad.hs
@@ -4,6 +4,18 @@
{-# LANGUAGE TypeFamilies #-}
{-# LANGUAGE BangPatterns #-}
{-# LANGUAGE TypeSynonymInstances #-}
+-- |
+-- Module : Documentation.Haddock.Parser.Monad
+-- Copyright : (c) Alec Theriault 2018-2019,
+-- License : BSD-like
+--
+-- Maintainer : haddock@projects.haskell.org
+-- Stability : experimental
+-- Portability : portable
+--
+-- Defines the Parsec monad over which all parsing is done and also provides
+-- more efficient versions of the usual parsec combinator functions (but
+-- specialized to 'Text').
module Documentation.Haddock.Parser.Monad where
@@ -17,7 +29,6 @@ import qualified Data.Text as T
import Data.Text ( Text )
import Control.Monad ( mfilter )
-import Data.Functor ( ($>) )
import Data.String ( IsString(..) )
import Data.Bits ( Bits(..) )
import Data.Char ( ord )
@@ -25,7 +36,9 @@ import Data.List ( foldl' )
import Control.Applicative as App
import Documentation.Haddock.Types ( Version )
+
import Prelude hiding (takeWhile)
+import CompatPrelude
-- | The only bit of information we really care about truding along with us
-- through parsing is the version attached to a @\@since@ annotation - if
@@ -96,7 +109,6 @@ takeWhile f = do
s' = s{ stateInput = inp', statePos = pos' }
setParserState s' $> t
-
-- | Like 'takeWhile', but fails if no characters matched.
--
-- Equivalent to @fmap T.pack . Parsec.many1@, but more efficient.
diff --git a/haddock-library/src/Documentation/Haddock/Parser/Util.hs b/haddock-library/src/Documentation/Haddock/Parser/Util.hs
index 98570c22..eef744d8 100644
--- a/haddock-library/src/Documentation/Haddock/Parser/Util.hs
+++ b/haddock-library/src/Documentation/Haddock/Parser/Util.hs
@@ -31,16 +31,16 @@ import Prelude hiding (takeWhile)
import Data.Char (isSpace)
-- | Characters that count as horizontal space
-horizontalSpace :: [Char]
-horizontalSpace = " \t\f\v\r"
+horizontalSpace :: Char -> Bool
+horizontalSpace c = isSpace c && c /= '\n'
-- | Skip and ignore leading horizontal space
skipHorizontalSpace :: Parser ()
-skipHorizontalSpace = Parsec.skipMany (Parsec.oneOf horizontalSpace)
+skipHorizontalSpace = Parsec.skipMany (Parsec.satisfy horizontalSpace)
-- | Take leading horizontal space
-takeHorizontalSpace :: Parser Text
-takeHorizontalSpace = takeWhile (`elem` horizontalSpace)
+takeHorizontalSpace :: Parser Text
+takeHorizontalSpace = takeWhile horizontalSpace
makeLabeled :: (String -> Maybe String -> a) -> Text -> a
makeLabeled f input = case T.break isSpace $ removeEscapes $ T.strip input of
@@ -60,10 +60,10 @@ removeEscapes = T.unfoldr go
-- | Consume characters from the input up to and including the given pattern.
-- Return everything consumed except for the end pattern itself.
-takeUntil :: Text -> Parser Text
+takeUntil :: Text -> Parser Text
takeUntil end_ = T.dropEnd (T.length end_) <$> requireEnd (scan p (False, end)) >>= gotSome
where
- end = T.unpack end_
+ end = T.unpack end_
p :: (Bool, String) -> Char -> Maybe (Bool, String)
p acc c = case acc of