Add identifier/pattern modules

2 files changed, 138 insertions, 0 deletions

diff --git a/src/Hakyll/Core/Identifier.hs b/src/Hakyll/Core/Identifier.hs
new file mode 100644
index 0000000..609e722
--- /dev/null
+++ b/src/Hakyll/Core/Identifier.hs
@@ -0,0 +1,47 @@
+-- | An identifier is a type used to uniquely identify a resource, target...
+--
+-- One can think of an identifier as something similar to a file path. An
+-- identifier is a path as well, with the different elements in the path
+-- separated by @/@ characters. Examples of identifiers are:
+--
+-- * @posts/foo.markdown@
+--
+-- * @index@
+--
+-- * @error/404@
+--
+module Hakyll.Core.Identifier
+    ( Identifier (..)
+    , parseIdentifier
+    , toFilePath
+    ) where
+
+import Control.Arrow (second)
+
+import GHC.Exts (IsString, fromString)
+import System.FilePath (joinPath)
+
+-- | An identifier used to uniquely identify a value
+--
+newtype Identifier = Identifier {unIdentifier :: [String]}
+                   deriving (Eq, Ord)
+
+instance Show Identifier where
+    show = toFilePath
+
+instance IsString Identifier where
+    fromString = parseIdentifier
+
+-- | Parse an identifier from a string
+--
+parseIdentifier :: String -> Identifier
+parseIdentifier = Identifier . filter (not . null) . split'
+  where
+    split' [] = [[]]
+    split' str = let (pre, post) = second (drop 1) $ break (== '/') str
+                 in pre : split' post
+
+-- | Convert an identifier to a relative 'FilePath'
+--
+toFilePath :: Identifier -> FilePath
+toFilePath = joinPath . unIdentifier
diff --git a/src/Hakyll/Core/Identifier/Pattern.hs b/src/Hakyll/Core/Identifier/Pattern.hs
new file mode 100644
index 0000000..02b023f
--- /dev/null
+++ b/src/Hakyll/Core/Identifier/Pattern.hs
@@ -0,0 +1,91 @@
+-- | Module providing pattern matching and capturing on 'Identifier's.
+--
+-- TODO: Documentation
+--
+module Hakyll.Core.Identifier.Pattern
+    ( Pattern
+    , parsePattern
+    , match
+    , doesMatch
+    , matches
+    ) where
+
+import Data.List (intercalate)
+import Control.Monad (msum)
+import Data.Maybe (isJust)
+
+import GHC.Exts (IsString, fromString)
+
+import Hakyll.Core.Identifier
+
+-- | One base element of a pattern
+--
+data PatternComponent = CaptureOne
+                      | CaptureMany
+                      | Literal String
+                      deriving (Eq)
+
+instance Show PatternComponent where
+    show CaptureOne = "*"
+    show CaptureMany = "**"
+    show (Literal s) = s
+
+-- | Type that allows matching on identifiers
+--
+newtype Pattern = Pattern {unPattern :: [PatternComponent]}
+                deriving (Eq)
+
+instance Show Pattern where
+    show = intercalate "/" . map show . unPattern
+
+instance IsString Pattern where
+    fromString = parsePattern
+
+-- | Parse a pattern from a string
+--
+parsePattern :: String -> Pattern
+parsePattern = Pattern . map toPattern . unIdentifier . parseIdentifier
+  where
+    toPattern x | x == "*"  = CaptureOne
+                | x == "**" = CaptureMany
+                | otherwise = Literal x
+
+-- | Match an identifier against a pattern, generating a list of captures
+--
+match :: Pattern -> Identifier -> Maybe [[String]]
+match (Pattern p) (Identifier i) = match' p i
+
+-- | Check if an identifier matches a pattern
+--
+doesMatch :: Pattern -> Identifier -> Bool
+doesMatch p = isJust . match p
+
+-- | Given a list of identifiers, retain only those who match the given pattern
+--
+matches :: Pattern -> [Identifier] -> [Identifier]
+matches p = filter (doesMatch p)
+
+-- | Split a list at every possible point, generate a list of (init, tail) cases
+--
+splits :: [a] -> [([a], [a])]
+splits ls = reverse $ splits' [] ls
+  where
+    splits' lx ly = (lx, ly) : case ly of
+        []       -> []
+        (y : ys) -> splits' (lx ++ [y]) ys
+
+-- | Internal verion of 'match'
+--
+match' :: [PatternComponent] -> [String] -> Maybe [[String]]
+match' [] [] = Just []  -- An empty match
+match' [] _ = Nothing   -- No match
+match' _ [] = Nothing   -- No match
+match' (m : ms) (s : ss) = case m of
+    -- Take one string and one literal, fail on mismatch
+    Literal l -> if s == l then match' ms ss else Nothing
+    -- Take one string and one capture
+    CaptureOne -> fmap ([s] :) $ match' ms ss
+    -- Take one string, and one or many captures
+    CaptureMany ->
+        let take' (i, t) = fmap (i :) $ match' ms t
+        in msum $ map take' $ splits (s : ss)