@@ -1,4 +1,12 @@
1-- | Module providing pattern matching and capturing on 'Identifier's.
1-- | Module providing pattern matching and capturing on 'Identifier's.
2+-- 'Pattern's come in two kinds:
4+-- * Simple glob patterns, like @foo\/*@;
6+-- * Custom, arbitrary predicates of the type @Identifier -> Bool@.
8+-- They both have advantages and disadvantages. By default, globs are used,
9+-- unless you construct your 'Pattern' using the 'predicate' function.
3-- A very simple pattern could be, for example, @foo\/bar@. This pattern will
11-- A very simple pattern could be, for example, @foo\/bar@. This pattern will
4-- only match the exact @foo\/bar@ identifier.
12-- only match the exact @foo\/bar@ identifier.
@@ -20,15 +28,17 @@
21-- * @foo\/*.html@ will match all HTML files in the @foo\/@ directory.
29-- * @foo\/*.html@ will match all HTML files in the @foo\/@ directory.
23−-- The 'match' function allows the user to get access to the elements captured
31+-- The 'capture' function allows the user to get access to the elements captured
24-- by the capture elements in the pattern.
32-- by the capture elements in the pattern.
26module Hakyll.Core.Identifier.Pattern
34module Hakyll.Core.Identifier.Pattern
28− , parsePattern
36+ , parseGlob
29− , match
37+ , predicate
30− , doesMatch
38+ , regex
32 , fromCapture
42 , fromCapture
33 , fromCaptureString
43 , fromCaptureString
34 , fromCaptures
44 , fromCaptures
@@ -37,32 +47,39 @@ module Hakyll.Core.Identifier.Pattern
37import Data.List (isPrefixOf, inits, tails)
47import Data.List (isPrefixOf, inits, tails)
38import Control.Arrow ((&&&), (>>>))
48import Control.Arrow ((&&&), (>>>))
39import Control.Monad (msum)
49import Control.Monad (msum)
40−import Data.Maybe (isJust)
50+import Data.Maybe (isJust, fromMaybe)
41−import Data.Monoid (mempty, mappend)
51+import Data.Monoid (Monoid, mempty, mappend)
43import GHC.Exts (IsString, fromString)
53import GHC.Exts (IsString, fromString)
54+import Text.Regex.PCRE ((=~~))
45import Hakyll.Core.Identifier
56import Hakyll.Core.Identifier
47-- | One base element of a pattern
58-- | One base element of a pattern
49−data PatternComponent = Capture
60+data GlobComponent = Capture
50− | CaptureMany
61+ | CaptureMany
51− | Literal String
62+ | Literal String
52− deriving (Eq, Show)
63+ deriving (Eq, Show)
54-- | Type that allows matching on identifiers
65-- | Type that allows matching on identifiers
56−newtype Pattern = Pattern {unPattern :: [PatternComponent]}
67+data Pattern = Glob [GlobComponent]
57− deriving (Eq, Show)
68+ | Predicate (Identifier -> Bool)
59instance IsString Pattern where
70instance IsString Pattern where
60− fromString = parsePattern
71+ fromString = parseGlob
73+instance Monoid Pattern where
74+ mempty = Predicate (const True)
75+ g@(Glob _) `mappend` x = Predicate (matches g) `mappend` x
76+ x `mappend` g@(Glob _) = x `mappend` Predicate (matches g)
77+ Predicate f `mappend` Predicate g = Predicate $ \i -> f i && g i
62-- | Parse a pattern from a string
79-- | Parse a pattern from a string
64−parsePattern :: String -> Pattern
81+parseGlob :: String -> Pattern
65−parsePattern = Pattern . parse' -- undefined -- Pattern . map toPattern . unIdentifier . parseIdentifier
82+parseGlob = Glob . parse'
67 parse' str =
84 parse' str =
68 let (chunk, rest) = break (`elem` "\\*") str
85 let (chunk, rest) = break (`elem` "\\*") str
@@ -72,20 +89,34 @@ parsePattern = Pattern . parse' -- undefined -- Pattern . map toPattern . unIden
72 ('*' : xs) -> Literal chunk : Capture : parse' xs
89 ('*' : xs) -> Literal chunk : Capture : parse' xs
73 xs -> Literal chunk : Literal xs : []
90 xs -> Literal chunk : Literal xs : []
75−-- | Match an identifier against a pattern, generating a list of captures
92+-- | Create a 'Pattern' from an arbitrary predicate
96+-- > predicate (\i -> matches "foo/*" i && not (matches "foo/bar" i))
98+predicate :: (Identifier -> Bool) -> Pattern
101+-- | Create a 'Pattern' from a regex
105+-- > regex "^foo/[^x]*$
77−match :: Pattern -> Identifier -> Maybe [Identifier]
107+regex :: String -> Pattern
78−match p (Identifier i) = fmap (map Identifier) $ match' (unPattern p) i
108+regex str = predicate $ fromMaybe False . (=~~ str) . toFilePath
80-- | Check if an identifier matches a pattern
110-- | Check if an identifier matches a pattern
82−doesMatch :: Pattern -> Identifier -> Bool
112+matches :: Pattern -> Identifier -> Bool
83−doesMatch p = isJust . match p
113+matches (Glob p) = isJust . capture (Glob p)
114+matches (Predicate p) = (p $)
85-- | Given a list of identifiers, retain only those who match the given pattern
116-- | Given a list of identifiers, retain only those who match the given pattern
87−matches :: Pattern -> [Identifier] -> [Identifier]
118+filterMatches :: Pattern -> [Identifier] -> [Identifier]
88−matches p = filter (doesMatch p)
119+filterMatches = filter . matches
90-- | Split a list at every possible point, generate a list of (init, tail)
121-- | Split a list at every possible point, generate a list of (init, tail)
91-- cases. The result is sorted with inits decreasing in length.
122-- cases. The result is sorted with inits decreasing in length.
@@ -93,30 +124,35 @@ matches p = filter (doesMatch p)
93splits :: [a] -> [([a], [a])]
124splits :: [a] -> [([a], [a])]
94splits = inits &&& tails >>> uncurry zip >>> reverse
125splits = inits &&& tails >>> uncurry zip >>> reverse
96−-- | Internal verion of 'match'
127+-- | Match a glob against a pattern, generating a list of captures
129+capture :: Pattern -> Identifier -> Maybe [Identifier]
130+capture (Glob p) (Identifier i) = fmap (map Identifier) $ capture' p i
131+capture (Predicate _) _ = Nothing
133+-- | Internal verion of 'capture'
98−match' :: [PatternComponent] -> String -> Maybe [String]
135+capture' :: [GlobComponent] -> String -> Maybe [String]
99−match' [] [] = Just [] -- An empty match
136+capture' [] [] = Just [] -- An empty match
100−match' [] _ = Nothing -- No match
137+capture' [] _ = Nothing -- No match
101−-- match' _ [] = Nothing -- No match
102−match' (Literal l : ms) str
138+capture' (Literal l : ms) str
103 -- Match the literal against the string
139 -- Match the literal against the string
104− | l `isPrefixOf` str = match' ms $ drop (length l) str
140+ | l `isPrefixOf` str = capture' ms $ drop (length l) str
105 | otherwise = Nothing
141 | otherwise = Nothing
106−match' (Capture : ms) str =
142+capture' (Capture : ms) str =
107 -- Match until the next /
143 -- Match until the next /
108 let (chunk, rest) = break (== '/') str
144 let (chunk, rest) = break (== '/') str
109− in msum $ [ fmap (i :) (match' ms (t ++ rest)) | (i, t) <- splits chunk ]
145+ in msum $ [ fmap (i :) (capture' ms (t ++ rest)) | (i, t) <- splits chunk ]
110−match' (CaptureMany : ms) str =
146+capture' (CaptureMany : ms) str =
111 -- Match everything
147 -- Match everything
112− msum $ [ fmap (i :) (match' ms t) | (i, t) <- splits str ]
148+ msum $ [ fmap (i :) (capture' ms t) | (i, t) <- splits str ]
114-- | Create an identifier from a pattern by filling in the captures with a given
150-- | Create an identifier from a pattern by filling in the captures with a given
117-- Example:
153-- Example:
119−-- > fromCapture (parsePattern "tags/*") (parseIdentifier "foo")
155+-- > fromCapture (parseGlob "tags/*") (parseIdentifier "foo")
121-- Result:
157-- Result:
@@ -128,7 +164,7 @@ fromCapture pattern = fromCaptures pattern . repeat
128-- | Simplified version of 'fromCapture' which takes a 'String' instead of an
164-- | Simplified version of 'fromCapture' which takes a 'String' instead of an
129-- 'Identifier'
165-- 'Identifier'
131−-- > fromCaptureString (parsePattern "tags/*") "foo"
167+-- > fromCaptureString (parseGlob "tags/*") "foo"
133-- Result:
169-- Result:
@@ -141,11 +177,19 @@ fromCaptureString pattern = fromCapture pattern . parseIdentifier
141-- given list of strings
177-- given list of strings
143fromCaptures :: Pattern -> [Identifier] -> Identifier
179fromCaptures :: Pattern -> [Identifier] -> Identifier
144−fromCaptures (Pattern []) _ = mempty
180+fromCaptures (Glob p) = fromCaptures' p
181+fromCaptures (Predicate _) = error $
182+ "Hakyll.Core.Identifier.Pattern.fromCaptures: fromCaptures called on a " ++
183+ "predicate instead of a glob"
185+-- | Internally used version of 'fromCaptures'
187+fromCaptures' :: [GlobComponent] -> [Identifier] -> Identifier
188+fromCaptures' [] _ = mempty
145−fromCaptures (Pattern (m : ms)) [] = case m of
189+fromCaptures' (m : ms) [] = case m of
146− Literal l -> Identifier l `mappend` fromCaptures (Pattern ms) []
190+ Literal l -> Identifier l `mappend` fromCaptures' ms []
147− _ -> error $ "Hakyll.Core.Identifier.Pattern.fromCaptures: "
191+ _ -> error $ "Hakyll.Core.Identifier.Pattern.fromCaptures': "
148 ++ "identifier list exhausted"
192 ++ "identifier list exhausted"
149−fromCaptures (Pattern (m : ms)) ids@(i : is) = case m of
193+fromCaptures' (m : ms) ids@(i : is) = case m of
150− Literal l -> Identifier l `mappend` fromCaptures (Pattern ms) ids
194+ Literal l -> Identifier l `mappend` fromCaptures' ms ids
151− _ -> i `mappend` fromCaptures (Pattern ms) is
195+ _ -> i `mappend` fromCaptures' ms is