aboutsummaryrefslogtreecommitdiff
path: root/haddock-library/src/Documentation/Haddock/Parser.hs
diff options
context:
space:
mode:
authorAlec Theriault <alec.theriault@gmail.com>2019-02-06 01:01:41 -0800
committerAlec Theriault <alec.theriault@gmail.com>2019-02-25 00:42:46 -0800
commita5199600c39d25d7b71dcb2328000c1c49ad95a2 (patch)
tree787057c0315d1adf98cab3769ad47b63cb3c0a94 /haddock-library/src/Documentation/Haddock/Parser.hs
parentdd47029cb29c80b1ab4db520c9c2ce4dca37f833 (diff)
Better identifier parsing
* '(<|>)' and '`elem`' now get parsed and rendered properly as links * 'DbModule'/'DbUnitId' now properly get split apart into two links * tuple names now get parsed properly * some more small niceties... The identifier parsing code is more precise and more efficient (although to be fair: it is also longer and in its own module). On the rendering side, we need to pipe through information about backticks/parens/neither all the way through from renaming to the backends. In terms of impact: a total of 35 modules in the entirety of the bootlib + ghc lib docs change. The only "regression" is things like '\0'. These should be changed to @\\0@ (the path by which this previously worked seems accidental).
Diffstat (limited to 'haddock-library/src/Documentation/Haddock/Parser.hs')
-rw-r--r--haddock-library/src/Documentation/Haddock/Parser.hs63
1 files changed, 4 insertions, 59 deletions
diff --git a/haddock-library/src/Documentation/Haddock/Parser.hs b/haddock-library/src/Documentation/Haddock/Parser.hs
index e9b1c496..36c8bb5b 100644
--- a/haddock-library/src/Documentation/Haddock/Parser.hs
+++ b/haddock-library/src/Documentation/Haddock/Parser.hs
@@ -27,8 +27,7 @@ module Documentation.Haddock.Parser (
import Control.Applicative
import Control.Arrow (first)
import Control.Monad
-import Data.Char (chr, isUpper, isAlpha, isAlphaNum, isSpace)
-import Data.Foldable (asum)
+import Data.Char (chr, isUpper, isAlpha, isSpace)
import Data.List (intercalate, unfoldr, elemIndex)
import Data.Maybe (fromMaybe, mapMaybe)
import Data.Monoid
@@ -37,6 +36,7 @@ import Documentation.Haddock.Doc
import Documentation.Haddock.Markup ( markup, plainMarkup )
import Documentation.Haddock.Parser.Monad
import Documentation.Haddock.Parser.Util
+import Documentation.Haddock.Parser.Identifier
import Documentation.Haddock.Types
import Prelude hiding (takeWhile)
import qualified Prelude as P
@@ -47,37 +47,10 @@ import Text.Parsec (try)
import qualified Data.Text as T
import Data.Text (Text)
-#if MIN_VERSION_base(4,9,0)
-import Text.Read.Lex (isSymbolChar)
-#else
-import Data.Char (GeneralCategory (..),
- generalCategory)
-#endif
-- $setup
-- >>> :set -XOverloadedStrings
-#if !MIN_VERSION_base(4,9,0)
--- inlined from base-4.10.0.0
-isSymbolChar :: Char -> Bool
-isSymbolChar c = not (isPuncChar c) && case generalCategory c of
- MathSymbol -> True
- CurrencySymbol -> True
- ModifierSymbol -> True
- OtherSymbol -> True
- DashPunctuation -> True
- OtherPunctuation -> c `notElem` ("'\"" :: String)
- ConnectorPunctuation -> c /= '_'
- _ -> False
- where
- -- | The @special@ character class as defined in the Haskell Report.
- isPuncChar :: Char -> Bool
- isPuncChar = (`elem` (",;()[]{}`" :: String))
-#endif
-
--- | Identifier string surrounded with opening and closing quotes/backticks.
-data Identifier = Identifier !Namespace !Char String !Char
-
-- | Drops the quotes/backticks around all identifiers, as if they
-- were valid but still 'String's.
toRegular :: DocH mod Identifier -> DocH mod String
@@ -838,34 +811,6 @@ autoUrl = mkLink <$> url
mkHyperlink lnk = Hyperlink (T.unpack lnk) Nothing
-
--- | Parses strings between identifier delimiters. Consumes all input that it
--- deems to be valid in an identifier. Note that it simply blindly consumes
--- characters and does no actual validation itself.
-parseValid :: Parser String
-parseValid = p some
- where
- idChar = Parsec.satisfy (\c -> isAlphaNum c || isSymbolChar c || c == '_')
-
- p p' = do
- vs <- p' idChar
- c <- peekChar'
- case c of
- '`' -> return vs
- '\'' -> choice' [ (\x -> vs ++ "'" ++ x) <$> ("'" *> p many), return vs ]
- _ -> fail "outofvalid"
-
--- | Parses identifiers with help of 'parseValid'. Asks GHC for
--- 'String' from the string it deems valid.
+-- | Parses identifiers with help of 'parseValid'.
identifier :: Parser (DocH mod Identifier)
-identifier = do
- ns <- asum [ Value <$ Parsec.char 'v'
- , Type <$ Parsec.char 't'
- , pure None
- ]
- o <- idDelim
- vid <- parseValid
- e <- idDelim
- return $ DocIdentifier (Identifier ns o vid e)
- where
- idDelim = Parsec.oneOf "'`"
+identifier = DocIdentifier <$> parseValid