Allow specifying a metadata provider

In addition to parsing metadata header & metadata files it will
allow any custom routine to extract metadata from the resource files,
for example with Pandoc

5 files changed, 38 insertions, 17 deletions

diff --git a/lib/Hakyll/Core/Configuration.hs b/lib/Hakyll/Core/Configuration.hs
index 52b23ec..4e32caa 100644
--- a/lib/Hakyll/Core/Configuration.hs
+++ b/lib/Hakyll/Core/Configuration.hs
@@ -8,14 +8,15 @@ module Hakyll.Core.Configuration
 
 
 --------------------------------------------------------------------------------
-import           Data.Default     (Default (..))
-import           Data.List        (isPrefixOf, isSuffixOf)
-import           System.Directory (canonicalizePath)
-import           System.Exit      (ExitCode)
-import           System.FilePath  (isAbsolute, normalise, takeFileName)
-import           System.IO.Error  (catchIOError)
-import           System.Process   (system)
+import           Data.Default         (Default (..))
+import           Data.List            (isPrefixOf, isSuffixOf)
+import           System.Directory     (canonicalizePath)
+import           System.Exit          (ExitCode)
+import           System.FilePath      (isAbsolute, normalise, takeFileName)
+import           System.IO.Error      (catchIOError)
+import           System.Process       (system)
 
+import           Hakyll.Core.Metadata
 
 --------------------------------------------------------------------------------
 data Configuration = Configuration
@@ -78,6 +79,10 @@ data Configuration = Configuration
       -- One can also override the port as a command line argument:
       -- ./site preview -p 1234
       previewPort          :: Int
+    , -- | Function to extract metadata from the files
+      --
+      -- By default it returns empty metadata
+      provideMetadata      :: FilePath -> IO Metadata
     }
 
 --------------------------------------------------------------------------------
@@ -98,6 +103,7 @@ defaultConfiguration = Configuration
     , inMemoryCache        = True
     , previewHost          = "127.0.0.1"
     , previewPort          = 8000
+    , provideMetadata      = const (return mempty)
     }
   where
     ignoreFile' path
diff --git a/lib/Hakyll/Core/Provider.hs b/lib/Hakyll/Core/Provider.hs
index 384f5b1..65e1507 100644
--- a/lib/Hakyll/Core/Provider.hs
+++ b/lib/Hakyll/Core/Provider.hs
@@ -5,6 +5,7 @@ module Hakyll.Core.Provider
     ( -- * Constructing resource providers
       Internal.Provider
     , newProvider
+    , newProvider'
 
       -- * Querying resource properties
     , Internal.resourceList
@@ -24,20 +25,29 @@ module Hakyll.Core.Provider
 
 
 --------------------------------------------------------------------------------
+import           Hakyll.Core.Metadata
 import qualified Hakyll.Core.Provider.Internal      as Internal
 import qualified Hakyll.Core.Provider.MetadataCache as Internal
 import           Hakyll.Core.Store                  (Store)
 
 
 --------------------------------------------------------------------------------
--- | Create a resource provider
+-- | Create a resource provider with the void metadata provider
 newProvider :: Store                   -- ^ Store to use
             -> (FilePath -> IO Bool)   -- ^ Should we ignore this file?
             -> FilePath                -- ^ Search directory
             -> IO Internal.Provider    -- ^ Resulting provider
-newProvider store ignore directory = do
+newProvider store ignore = newProvider' store ignore (const $ return mempty)
+
+-- | Create a resource provider
+newProvider' :: Store                   -- ^ Store to use
+            -> (FilePath -> IO Bool)   -- ^ Should we ignore this file?
+            -> (FilePath -> IO Metadata)   -- ^ Metadata provider
+            -> FilePath                -- ^ Search directory
+            -> IO Internal.Provider    -- ^ Resulting provider
+newProvider' store ignore metadata directory = do
     -- Delete metadata cache where necessary
-    p <- Internal.newProvider store ignore directory
+    p <- Internal.newProvider store ignore metadata directory
     mapM_ (Internal.resourceInvalidateMetadataCache p) $
         filter (Internal.resourceModified p) $ Internal.resourceList p
     return p
diff --git a/lib/Hakyll/Core/Provider/Internal.hs b/lib/Hakyll/Core/Provider/Internal.hs
index c298653..777decb 100644
--- a/lib/Hakyll/Core/Provider/Internal.hs
+++ b/lib/Hakyll/Core/Provider/Internal.hs
@@ -45,6 +45,7 @@ import           System.Time            (formatCalendarTime, toCalendarTime)
 
 --------------------------------------------------------------------------------
 import           Hakyll.Core.Identifier
+import           Hakyll.Core.Metadata
 import           Hakyll.Core.Store      (Store)
 import qualified Hakyll.Core.Store      as Store
 import           Hakyll.Core.Util.File
@@ -95,16 +96,19 @@ data Provider = Provider
       providerOldFiles  :: Map Identifier ResourceInfo
     , -- | Underlying persistent store for caching
       providerStore     :: Store
-    } deriving (Show)
+      -- | A custom function to extract metadata from files
+    , providerMetadata  :: FilePath -> IO Metadata
+    }
 
 
 --------------------------------------------------------------------------------
 -- | Create a resource provider
 newProvider :: Store                  -- ^ Store to use
             -> (FilePath -> IO Bool)  -- ^ Should we ignore this file?
+            -> (FilePath -> IO Metadata)  -- ^ Metadata provider
             -> FilePath               -- ^ Search directory
             -> IO Provider            -- ^ Resulting provider
-newProvider store ignore directory = do
+newProvider store ignore metadata directory = do
     list <- map fromFilePath <$> getRecursiveContents ignore directory
     let universe = S.fromList list
     files <- fmap (maxmtime . M.fromList) $ forM list $ \identifier -> do
@@ -116,7 +120,7 @@ newProvider store ignore directory = do
     oldFiles <- fromMaybe mempty . Store.toMaybe <$> Store.get store oldKey
     oldFiles `deepseq` Store.set store oldKey files
 
-    return $ Provider directory files oldFiles store
+    return $ Provider directory files oldFiles store metadata
   where
     oldKey = ["Hakyll.Core.Provider.Internal.newProvider", "oldFiles"]
 
diff --git a/lib/Hakyll/Core/Provider/Metadata.hs b/lib/Hakyll/Core/Provider/Metadata.hs
index c74627b..1630b0c 100644
--- a/lib/Hakyll/Core/Provider/Metadata.hs
+++ b/lib/Hakyll/Core/Provider/Metadata.hs
@@ -27,7 +27,7 @@ import           Hakyll.Core.Identifier
 import           Hakyll.Core.Metadata
 import           Hakyll.Core.Provider.Internal
 import           System.IO                     as IO
-import           System.IO.Error               (modifyIOError, ioeSetLocation)
+import           System.IO.Error               (ioeSetLocation, modifyIOError)
 
 
 --------------------------------------------------------------------------------
@@ -42,7 +42,8 @@ loadMetadata p identifier = do
         Nothing  -> return mempty
         Just mi' -> loadMetadataFile $ resourceFilePath p mi'
 
-    return (md <> emd, body)
+    mdp <- providerMetadata p fp
+    return (md <> emd <> mdp, body)
   where
     normal = setVersion Nothing identifier
     fp     = resourceFilePath p identifier
diff --git a/lib/Hakyll/Core/Runtime.hs b/lib/Hakyll/Core/Runtime.hs
index 922b676..68970cd 100644
--- a/lib/Hakyll/Core/Runtime.hs
+++ b/lib/Hakyll/Core/Runtime.hs
@@ -47,8 +47,8 @@ run config logger rules = do
     Logger.message logger "Creating store..."
     store <- Store.new (inMemoryCache config) $ storeDirectory config
     Logger.message logger "Creating provider..."
-    provider <- newProvider store (shouldIgnoreFile config) $
-        providerDirectory config
+    provider <- newProvider' store (shouldIgnoreFile config) (provideMetadata config)
+        (providerDirectory config)
     Logger.message logger "Running rules..."
     ruleSet  <- runRules rules provider