{-# LANGUAGE BangPatterns #-}
{-# LANGUAGE GeneralizedNewtypeDeriving #-}
{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE PackageImports #-}
{-# LANGUAGE ScopedTypeVariables #-}
module Text.Templating.Heist.Internal where
------------------------------------------------------------------------------
import Blaze.ByteString.Builder
import Control.Applicative
import Control.Arrow hiding (loop)
import Control.Exception (SomeException)
import Control.Monad
import Control.Monad.CatchIO
import Control.Monad.Trans
import qualified Data.Attoparsec.Text as AP
import Data.ByteString (ByteString)
import qualified Data.ByteString as B
import qualified Data.ByteString.Char8 as BC
import Data.Either
import qualified Data.Foldable as F
import Data.List
import qualified Data.HashMap.Strict as Map
import Data.Maybe
import Data.Monoid
import qualified Data.Text as T
import Data.Text (Text)
import Prelude hiding (catch)
import System.Directory.Tree hiding (name)
import System.FilePath
import qualified Text.XmlHtml as X
------------------------------------------------------------------------------
import Text.Templating.Heist.Types
------------------------------------------------------------------------------
-- | Mappends a doctype to the state.
addDoctype :: Monad m => [X.DocType] -> HeistT m ()
addDoctype dt = do
modifyTS (\s -> s { _doctypes = _doctypes s `mappend` dt })
------------------------------------------------------------------------------
-- HeistState functions
------------------------------------------------------------------------------
------------------------------------------------------------------------------
-- | Adds an on-load hook to a `HeistState`.
addOnLoadHook :: (Monad m) =>
(Template -> IO Template)
-> HeistState m
-> HeistState m
addOnLoadHook hook ts = ts { _onLoadHook = _onLoadHook ts >=> hook }
------------------------------------------------------------------------------
-- | Adds a pre-run hook to a `HeistState`.
addPreRunHook :: (Monad m) =>
(Template -> m Template)
-> HeistState m
-> HeistState m
addPreRunHook hook ts = ts { _preRunHook = _preRunHook ts >=> hook }
------------------------------------------------------------------------------
-- | Adds a post-run hook to a `HeistState`.
addPostRunHook :: (Monad m) =>
(Template -> m Template)
-> HeistState m
-> HeistState m
addPostRunHook hook ts = ts { _postRunHook = _postRunHook ts >=> hook }
------------------------------------------------------------------------------
-- | Binds a new splice declaration to a tag name within a 'HeistState'.
bindSplice :: Monad m =>
Text -- ^ tag name
-> Splice m -- ^ splice action
-> HeistState m -- ^ source state
-> HeistState m
bindSplice n v ts = ts {_spliceMap = Map.insert n v (_spliceMap ts)}
------------------------------------------------------------------------------
-- | Binds a set of new splice declarations within a 'HeistState'.
bindSplices :: Monad m =>
[(Text, Splice m)] -- ^ splices to bind
-> HeistState m -- ^ start state
-> HeistState m
bindSplices ss ts = foldl' (flip id) ts acts
where
acts = map (uncurry bindSplice) ss
------------------------------------------------------------------------------
-- | Sets the current template file.
setCurTemplateFile :: Monad m
=> Maybe FilePath -> HeistState m -> HeistState m
setCurTemplateFile fp ts = ts { _curTemplateFile = fp }
------------------------------------------------------------------------------
-- | Converts 'Text' to a splice returning a single 'TextNode'.
textSplice :: (Monad m) => Text -> Splice m
textSplice t = return [X.TextNode t]
------------------------------------------------------------------------------
-- | Runs the parameter node's children and returns the resulting node list.
-- By itself this function is a simple passthrough splice that makes the
-- spliced node disappear. In combination with locally bound splices, this
-- function makes it easier to pass the desired view into your splices.
runChildren :: Monad m => Splice m
runChildren = runNodeList . X.childNodes =<< getParamNode
------------------------------------------------------------------------------
-- | Binds a list of splices before using the children of the spliced node as
-- a view.
runChildrenWith :: (Monad m)
=> [(Text, Splice m)]
-- ^ List of splices to bind before running the param nodes.
-> Splice m
-- ^ Returns the passed in view.
runChildrenWith splices = localTS (bindSplices splices) runChildren
------------------------------------------------------------------------------
-- | Wrapper around runChildrenWith that applies a transformation function to
-- the second item in each of the tuples before calling runChildrenWith.
runChildrenWithTrans :: (Monad m)
=> (b -> Splice m)
-- ^ Splice generating function
-> [(Text, b)]
-- ^ List of tuples to be bound
-> Splice m
runChildrenWithTrans f = runChildrenWith . map (second f)
------------------------------------------------------------------------------
-- | Like runChildrenWith but using constant templates rather than dynamic
-- splices.
runChildrenWithTemplates :: (Monad m) => [(Text, Template)] -> Splice m
runChildrenWithTemplates = runChildrenWithTrans return
------------------------------------------------------------------------------
-- | Like runChildrenWith but using literal text rather than dynamic splices.
runChildrenWithText :: (Monad m) => [(Text, Text)] -> Splice m
runChildrenWithText = runChildrenWithTrans textSplice
------------------------------------------------------------------------------
-- | Maps a splice generating function over a list and concatenates the
-- results.
mapSplices :: (Monad m)
=> (a -> Splice m)
-- ^ Splice generating function
-> [a]
-- ^ List of items to generate splices for
-> Splice m
-- ^ The result of all splices concatenated together.
mapSplices f vs = liftM concat $ mapM f vs
{-# INLINE mapSplices #-}
------------------------------------------------------------------------------
-- | Convenience function for looking up a splice.
lookupSplice :: Monad m =>
Text
-> HeistState m
-> Maybe (Splice m)
lookupSplice nm ts = Map.lookup nm $ _spliceMap ts
{-# INLINE lookupSplice #-}
------------------------------------------------------------------------------
-- | Converts a path into an array of the elements in reverse order. If the
-- path is absolute, we need to remove the leading slash so the split doesn't
-- leave @\"\"@ as the last element of the TPath.
--
-- FIXME @\"..\"@ currently doesn't work in paths, the solution is non-trivial
splitPathWith :: Char -> ByteString -> TPath
splitPathWith s p = if BC.null p then [] else (reverse $ BC.split s path)
where
path = if BC.head p == s then BC.tail p else p
-- | Converts a path into an array of the elements in reverse order using the
-- path separator of the local operating system. See 'splitPathWith' for more
-- details.
splitLocalPath :: ByteString -> TPath
splitLocalPath = splitPathWith pathSeparator
-- | Converts a path into an array of the elements in reverse order using a
-- forward slash (/) as the path separator. See 'splitPathWith' for more
-- details.
splitTemplatePath :: ByteString -> TPath
splitTemplatePath = splitPathWith '/'
------------------------------------------------------------------------------
-- | Does a single template lookup without cascading up.
singleLookup :: TemplateMap
-> TPath
-> ByteString
-> Maybe (DocumentFile, TPath)
singleLookup tm path name = fmap (\a -> (a,path)) $ Map.lookup (name:path) tm
------------------------------------------------------------------------------
-- | Searches for a template by looking in the full path then backing up into
-- each of the parent directories until the template is found.
traversePath :: TemplateMap
-> TPath
-> ByteString
-> Maybe (DocumentFile, TPath)
traversePath tm [] name = fmap (\a -> (a,[])) (Map.lookup [name] tm)
traversePath tm path name =
singleLookup tm path name `mplus`
traversePath tm (tail path) name
------------------------------------------------------------------------------
-- | Returns 'True' if the given template can be found in the heist state.
hasTemplate :: Monad m =>
ByteString
-> HeistState m
-> Bool
hasTemplate nameStr ts = isJust $ lookupTemplate nameStr ts
------------------------------------------------------------------------------
-- | Convenience function for looking up a template.
lookupTemplate :: Monad m =>
ByteString
-> HeistState m
-> Maybe (DocumentFile, TPath)
lookupTemplate nameStr ts =
f (_templateMap ts) path name
where (name:p) = case splitTemplatePath nameStr of
[] -> [""]
ps -> ps
ctx = if B.isPrefixOf "/" nameStr then [] else _curContext ts
path = p ++ ctx
f = if '/' `BC.elem` nameStr
then singleLookup
else traversePath
------------------------------------------------------------------------------
-- | Sets the templateMap in a HeistState.
setTemplates :: Monad m => TemplateMap -> HeistState m -> HeistState m
setTemplates m ts = ts { _templateMap = m }
------------------------------------------------------------------------------
-- | Adds a template to the heist state.
insertTemplate :: Monad m =>
TPath
-> DocumentFile
-> HeistState m
-> HeistState m
insertTemplate p t st =
setTemplates (Map.insert p t (_templateMap st)) st
------------------------------------------------------------------------------
-- | Adds an HTML format template to the heist state.
addTemplate :: Monad m
=> ByteString
-- ^ Path that the template will be referenced by
-> Template
-- ^ The template's DOM nodes
-> Maybe FilePath
-- ^ An optional path to the actual file on disk where the
-- template is stored
-> HeistState m
-> HeistState m
addTemplate n t mfp st =
insertTemplate (splitTemplatePath n) doc st
where
doc = DocumentFile (X.HtmlDocument X.UTF8 Nothing t) mfp
------------------------------------------------------------------------------
-- | Adds an XML format template to the heist state.
addXMLTemplate :: Monad m
=> ByteString
-- ^ Path that the template will be referenced by
-> Template
-- ^ The template's DOM nodes
-> Maybe FilePath
-- ^ An optional path to the actual file on disk where the
-- template is stored
-> HeistState m
-> HeistState m
addXMLTemplate n t mfp st =
insertTemplate (splitTemplatePath n) doc st
where
doc = DocumentFile (X.XmlDocument X.UTF8 Nothing t) mfp
------------------------------------------------------------------------------
-- | Stops the recursive processing of splices. Consider the following
-- example:
--
-- >
-- >
-- > ...
-- >
-- >
--
-- Assume that @\"foo\"@ is bound to a splice procedure. Running the @foo@
-- splice will result in a list of nodes @L@. Normally @foo@ will recursively
-- scan @L@ for splices and run them. If @foo@ calls @stopRecursion@, @L@
-- will be included in the output verbatim without running any splices.
stopRecursion :: Monad m => HeistT m ()
stopRecursion = modifyTS (\st -> st { _recurse = False })
------------------------------------------------------------------------------
-- | Sets the current context
setContext :: Monad m => TPath -> HeistT m ()
setContext c = modifyTS (\st -> st { _curContext = c })
------------------------------------------------------------------------------
-- | Gets the current context
getContext :: Monad m => HeistT m TPath
getContext = getsTS _curContext
------------------------------------------------------------------------------
-- | Gets the full path to the file holding the template currently being
-- processed. Returns Nothing if the template is not associated with a file
-- on disk or if there is no template being processed.
getTemplateFilePath :: Monad m => HeistT m (Maybe FilePath)
getTemplateFilePath = getsTS _curTemplateFile
------------------------------------------------------------------------------
-- | Performs splice processing on a single node.
runNode :: Monad m => X.Node -> Splice m
runNode (X.Element nm at ch) = do
newAtts <- mapM attSubst at
let n = X.Element nm newAtts ch
s <- liftM (lookupSplice nm) getTS
maybe (runKids newAtts) (recurseSplice n) s
where
runKids newAtts = do
newKids <- runNodeList ch
return [X.Element nm newAtts newKids]
runNode n = return [n]
------------------------------------------------------------------------------
-- | Helper function for substituting a parsed attribute into an attribute
-- tuple.
attSubst :: (Monad m) => (t, Text) -> HeistT m (t, Text)
attSubst (n,v) = do
v' <- parseAtt v
return (n,v')
------------------------------------------------------------------------------
-- | Parses an attribute for any identifier expressions and performs
-- appropriate substitution.
parseAtt :: (Monad m) => Text -> HeistT m Text
parseAtt bs = do
let ast = case AP.feed (AP.parse attParser bs) "" of
(AP.Done _ res) -> res
(AP.Fail _ _ _) -> []
(AP.Partial _) -> []
chunks <- mapM cvt ast
return $ T.concat chunks
where
cvt (Literal x) = return x
cvt (Ident x) =
localParamNode (const $ X.Element x [] []) $ getAttributeSplice x
------------------------------------------------------------------------------
-- | AST to hold attribute parsing structure. This is necessary because
-- attoparsec doesn't support parsers running in another monad.
data AttAST = Literal Text
| Ident Text
deriving (Show)
------------------------------------------------------------------------------
-- | Parser for attribute variable substitution.
attParser :: AP.Parser [AttAST]
attParser = liftM ($! []) (loop id)
where
append !dl !x = dl . (x:)
loop !dl = go id
where
finish subDL = let !txt = T.concat $! subDL []
lit = Literal $! T.concat $! subDL []
in return $! if T.null txt
then dl
else append dl lit
go !subDL = (gobbleText >>= go . append subDL)
<|> (AP.endOfInput *> finish subDL)
<|> (escChar >>= go . append subDL)
<|> (do
idp <- identParser
dl' <- finish subDL
loop $! append dl' idp)
gobbleText = AP.takeWhile1 (AP.notInClass "\\$")
escChar = AP.char '\\' *> (T.singleton <$> AP.anyChar)
identParser = AP.char '$' *> (ident <|> return (Literal "$"))
ident = (AP.char '{' *> (Ident <$> AP.takeWhile (/='}')) <* AP.string "}")
------------------------------------------------------------------------------
-- | Gets the attribute value. If the splice's result list contains non-text
-- nodes, this will translate them into text nodes with nodeText and
-- concatenate them together.
--
-- Originally, this only took the first node from the splices's result list,
-- and only if it was a text node. This caused problems when the splice's
-- result contained HTML entities, as they would split a text node. This was
-- then fixed to take the first consecutive bunch of text nodes, and return
-- their concatenation. This was seen as more useful than throwing an error,
-- and more intuitive than trying to render all the nodes as text.
--
-- However, it was decided in the end to render all the nodes as text, and
-- then concatenate them. If a splice returned
-- \"some \text\<\/b\> foobar\", the user would almost certainly want
-- \"some text foobar\" to be rendered, and Heist would probably seem
-- annoyingly limited for not being able to do this. If the user really did
-- want it to render \"some \", it would probably be easier for them to
-- accept that they were silly to pass more than that to be substituted than
-- it would be for the former user to accept that
-- \"some \text\<\/b\> foobar\" is being rendered as \"some \" because
-- it's \"more intuitive\".
getAttributeSplice :: Monad m => Text -> HeistT m Text
getAttributeSplice name = do
s <- liftM (lookupSplice name) getTS
nodes <- maybe (return []) id s
return $ T.concat $ map X.nodeText nodes
------------------------------------------------------------------------------
-- | Performs splice processing on a list of nodes.
runNodeList :: Monad m => [X.Node] -> Splice m
runNodeList = mapSplices runNode
{-# INLINE runNodeList #-}
------------------------------------------------------------------------------
-- | The maximum recursion depth. (Used to prevent infinite loops.)
mAX_RECURSION_DEPTH :: Int
mAX_RECURSION_DEPTH = 50
------------------------------------------------------------------------------
-- | Checks the recursion flag and recurses accordingly. Does not recurse
-- deeper than mAX_RECURSION_DEPTH to avoid infinite loops.
recurseSplice :: Monad m => X.Node -> Splice m -> Splice m
recurseSplice node splice = do
result <- localParamNode (const node) splice
ts' <- getTS
if _recurse ts' && _recursionDepth ts' < mAX_RECURSION_DEPTH
then do modRecursionDepth (+1)
res <- runNodeList result
restoreTS ts'
return res
else return result
where
modRecursionDepth :: Monad m => (Int -> Int) -> HeistT m ()
modRecursionDepth f =
modifyTS (\st -> st { _recursionDepth = f (_recursionDepth st) })
------------------------------------------------------------------------------
-- | Looks up a template name runs a 'HeistT' computation on it.
lookupAndRun :: Monad m
=> ByteString
-> ((DocumentFile, TPath) -> HeistT m (Maybe a))
-> HeistT m (Maybe a)
lookupAndRun name k = do
ts <- getTS
let mt = lookupTemplate name ts
let curPath = join $ fmap (dfFile . fst) mt
modifyTS (setCurTemplateFile curPath)
maybe (return Nothing) k mt
------------------------------------------------------------------------------
-- | Looks up a template name evaluates it by calling runNodeList.
evalTemplate :: Monad m
=> ByteString
-> HeistT m (Maybe Template)
evalTemplate name = lookupAndRun name
(\(t,ctx) -> localTS (\ts -> ts {_curContext = ctx})
(liftM Just $ runNodeList $ X.docContent $ dfDoc t))
------------------------------------------------------------------------------
-- | Sets the document type of a 'X.Document' based on the 'HeistT'
-- value.
fixDocType :: Monad m => X.Document -> HeistT m X.Document
fixDocType d = do
dts <- getsTS _doctypes
return $ d { X.docType = listToMaybe dts }
------------------------------------------------------------------------------
-- | Same as evalWithHooks, but returns the entire 'X.Document' rather than
-- just the nodes. This is the right thing to do if we are starting at the
-- top level.
evalWithHooksInternal :: Monad m
=> ByteString
-> HeistT m (Maybe X.Document)
evalWithHooksInternal name = lookupAndRun name $ \(t,ctx) -> do
addDoctype $ maybeToList $ X.docType $ dfDoc t
ts <- getTS
nodes <- lift $ _preRunHook ts $ X.docContent $ dfDoc t
putTS (ts {_curContext = ctx})
res <- runNodeList nodes
restoreTS ts
newNodes <- lift (_postRunHook ts res)
newDoc <- fixDocType $ (dfDoc t) { X.docContent = newNodes }
return (Just newDoc)
------------------------------------------------------------------------------
-- | Looks up a template name evaluates it by calling runNodeList. This also
-- executes pre- and post-run hooks and adds the doctype.
evalWithHooks :: Monad m
=> ByteString
-> HeistT m (Maybe Template)
evalWithHooks name = liftM (liftM X.docContent) (evalWithHooksInternal name)
------------------------------------------------------------------------------
-- | Binds a list of constant string splices.
bindStrings :: Monad m
=> [(Text, Text)]
-> HeistState m
-> HeistState m
bindStrings pairs ts = foldr (uncurry bindString) ts pairs
------------------------------------------------------------------------------
-- | Binds a single constant string splice.
bindString :: Monad m
=> Text
-> Text
-> HeistState m
-> HeistState m
bindString n = bindSplice n . textSplice
------------------------------------------------------------------------------
-- | Renders a template with the specified parameters. This is the function
-- to use when you want to "call" a template and pass in parameters from
-- inside a splice. If the template does not exist, this version simply
-- returns an empty list.
callTemplate :: Monad m
=> ByteString -- ^ The name of the template
-> [(Text, Splice m)] -- ^ Association list of
-- (name,value) parameter pairs
-> HeistT m Template
callTemplate name params = do
modifyTS $ bindSplices params
liftM (maybe [] id) $ evalTemplate name
------------------------------------------------------------------------------
-- | Like callTemplate except the splices being bound are constant text
-- splices.
callTemplateWithText :: Monad m
=> ByteString -- ^ The name of the template
-> [(Text, Text)] -- ^ Association list of
-- (name,value) parameter pairs
-> HeistT m Template
callTemplateWithText name params = do
modifyTS $ bindStrings params
liftM (maybe [] id) $ evalTemplate name
------------------------------------------------------------------------------
-- Gives the MIME type for a 'X.Document'
mimeType :: X.Document -> ByteString
mimeType d = case d of
(X.HtmlDocument e _ _) -> "text/html;charset=" `BC.append` enc e
(X.XmlDocument e _ _) -> "text/xml;charset=" `BC.append` enc e
where
enc X.UTF8 = "utf-8"
-- Should not include byte order designation for UTF-16 since
-- rendering will include a byte order mark. (RFC 2781, Sec. 3.3)
enc X.UTF16BE = "utf-16"
enc X.UTF16LE = "utf-16"
------------------------------------------------------------------------------
-- | Renders a template from the specified HeistState to a 'Builder'. The
-- MIME type returned is based on the detected character encoding, and whether
-- the root template was an HTML or XML format template. It will always be
-- @text/html@ or @text/xml@. If a more specific MIME type is needed for a
-- particular XML application, it must be provided by the application.
renderTemplate :: Monad m
=> HeistState m
-> ByteString
-> m (Maybe (Builder, MIMEType))
renderTemplate ts name = evalHeistT tpl (X.TextNode "") ts
where tpl = do mt <- evalWithHooksInternal name
case mt of
Nothing -> return Nothing
Just doc -> return $ Just $ (X.render doc, mimeType doc)
------------------------------------------------------------------------------
-- | Renders a template with the specified arguments passed to it. This is a
-- convenience function for the common pattern of calling renderTemplate after
-- using bindString, bindStrings, or bindSplice to set up the arguments to the
-- template.
renderWithArgs :: Monad m
=> [(Text, Text)]
-> HeistState m
-> ByteString
-> m (Maybe (Builder, MIMEType))
renderWithArgs args ts = renderTemplate (bindStrings args ts)
------------------------------------------------------------------------------
-- Template loading
------------------------------------------------------------------------------
------------------------------------------------------------------------------
-- | Type synonym for parsers.
type ParserFun = String -> ByteString -> Either String X.Document
------------------------------------------------------------------------------
-- | Reads an HTML or XML template from disk.
getDocWith :: ParserFun -> String -> IO (Either String DocumentFile)
getDocWith parser f = do
bs <- catch (liftM Right $ B.readFile f)
(\(e::SomeException) -> return $ Left $ show e)
let eitherDoc = either Left (parser f) bs
return $ either (\s -> Left $ f ++ " " ++ s)
(\d -> Right $ DocumentFile d (Just f)) eitherDoc
------------------------------------------------------------------------------
-- | Reads an HTML template from disk.
getDoc :: String -> IO (Either String DocumentFile)
getDoc = getDocWith X.parseHTML
------------------------------------------------------------------------------
-- | Reads an XML template from disk.
getXMLDoc :: String -> IO (Either String DocumentFile)
getXMLDoc = getDocWith X.parseXML
------------------------------------------------------------------------------
-- | Loads a template with the specified path and filename. The
-- template is only loaded if it has a ".tpl" or ".xtpl" extension.
loadTemplate :: String -- ^ path of the template root
-> String -- ^ full file path (includes the template root)
-> IO [Either String (TPath, DocumentFile)] --TemplateMap
loadTemplate templateRoot fname
| isHTMLTemplate = do
c <- getDoc fname
return [fmap (\t -> (splitLocalPath $ BC.pack tName, t)) c]
| isXMLTemplate = do
c <- getXMLDoc fname
return [fmap (\t -> (splitLocalPath $ BC.pack tName, t)) c]
| otherwise = return []
where -- tName is path relative to the template root directory
isHTMLTemplate = ".tpl" `isSuffixOf` fname
isXMLTemplate = ".xtpl" `isSuffixOf` fname
correction = if last templateRoot == '/' then 0 else 1
extLen = if isHTMLTemplate then 4 else 5
tName = drop ((length templateRoot)+correction) $
-- We're only dropping the template root, not the whole path
take ((length fname) - extLen) fname
------------------------------------------------------------------------------
-- | Traverses the specified directory structure and builds a HeistState by
-- loading all the files with a ".tpl" or ".xtpl" extension.
loadTemplates :: Monad m => FilePath -> HeistState m
-> IO (Either String (HeistState m))
loadTemplates dir ts = do
d <- readDirectoryWith (loadTemplate dir) dir
let tlist = F.fold (free d)
errs = lefts tlist
case errs of
[] -> liftM Right $ foldM loadHook ts $ rights tlist
_ -> return $ Left $ unlines errs
------------------------------------------------------------------------------
-- | Runs a template modifying function on a DocumentFile.
runHook :: Monad m => (Template -> m Template)
-> DocumentFile
-> m DocumentFile
runHook f t = do
n <- f $ X.docContent $ dfDoc t
return $ t { dfDoc = (dfDoc t) { X.docContent = n } }
------------------------------------------------------------------------------
-- | Runs the onLoad hook on the template and returns the 'HeistState'
-- with the result inserted.
loadHook :: Monad m => HeistState m -> (TPath, DocumentFile)
-> IO (HeistState m)
loadHook ts (tp, t) = do
t' <- runHook (_onLoadHook ts) t
return $ insertTemplate tp t' ts
------------------------------------------------------------------------------
-- | Adds a path prefix to all the templates in the 'HeistState'. If you
-- want to add multiple levels of directories, separate them with slashes as
-- in "foo/bar". Using an empty string as a path prefix will leave the
-- 'HeistState' unchanged.
addTemplatePathPrefix :: ByteString -> HeistState m -> HeistState m
addTemplatePathPrefix dir ts
| B.null dir = ts
| otherwise = ts { _templateMap = Map.fromList $
map (\(x,y) -> (f x, y)) $
Map.toList $
_templateMap ts
}
where
f ps = ps++splitTemplatePath dir