Safe Haskell | None |
---|
NOTE: This module should be considered internal, and will be hidden in future releases.
- data ShakespeareSettings = ShakespeareSettings {
- varChar :: Char
- urlChar :: Char
- intChar :: Char
- toBuilder :: Exp
- wrap :: Exp
- unwrap :: Exp
- justVarInterpolation :: Bool
- preConversion :: Maybe PreConvert
- modifyFinalValue :: Maybe Exp
- data PreConvert = PreConvert {}
- data WrapInsertion = WrapInsertion {}
- data PreConversion
- = ReadProcess String [String]
- | Id
- defaultShakespeareSettings :: ShakespeareSettings
- shakespeare :: ShakespeareSettings -> QuasiQuoter
- shakespeareFile :: ShakespeareSettings -> FilePath -> Q Exp
- shakespeareFileReload :: ShakespeareSettings -> FilePath -> Q Exp
- shakespeareFromString :: ShakespeareSettings -> String -> Q Exp
- shakespeareUsedIdentifiers :: ShakespeareSettings -> String -> [(Deref, VarType)]
- type RenderUrl url = url -> QueryParameters -> Text
- data VarType
- = VTPlain
- | VTUrl
- | VTUrlParam
- | VTMixin
- data Deref
- type Parser = Parsec String [String]
- preFilter :: Maybe FilePath -> ShakespeareSettings -> String -> IO String
- shakespeareRuntime :: ShakespeareSettings -> FilePath -> [(Deref, VarExp url)] -> Shakespeare url
- pack' :: String -> Text
Documentation
data ShakespeareSettings Source
ShakespeareSettings | |
|
data PreConvert Source
Coffeescript, TypeScript, and other languages compiles down to Javascript. Previously we waited until the very end, at the rendering stage to perform this compilation. Lets call is a post-conversion This had the advantage that all Haskell values were inserted first: for example a value could be inserted that Coffeescript would compile into Javascript. While that is perhaps a safer approach, the advantage is not used in practice: it was that way mainly for ease of implementation. The down-side is the template must be compiled down to Javascript during every request. If instead we do a pre-conversion to compile down to Javascript, we only need to perform the compilation once.
The problem then is the insertion of Haskell values: we need a hole for them. This can be done with variables known to the language. During the pre-conversion we first modify all Haskell insertions So #{a} is change to shakespeare_var_a Then we can place the Haskell values in a function wrapper that exposes those variables: (function(shakespeare_var_a){ ... shakespeare_var_a ...}) TypeScript can compile that, and then we tack an application of the Haskell values onto the result: (#{a})
preEscapeIgnoreBalanced is used to not insert backtacks for variable already inside strings or backticks. coffeescript will happily ignore the interpolations, and backticks would not be treated as escaping in that context. preEscapeIgnoreLine was added to ignore comments (which in Coffeescript begin with a '#')
shakespeareFile :: ShakespeareSettings -> FilePath -> Q ExpSource
low-level
shakespeareUsedIdentifiers :: ShakespeareSettings -> String -> [(Deref, VarType)]Source
Determine which identifiers are used by the given template, useful for creating systems like yesod devel.
Internal
shakespeareRuntime :: ShakespeareSettings -> FilePath -> [(Deref, VarExp url)] -> Shakespeare urlSource