Support all readers and writers that pandoc does

This uses pandoc's own getReader and getWriter functions, so it doesn't
need to be constantly updated with newer formats.

Furthermore, libpandoc also now supports readers and writers that are
not pure.

Signed-off-by: Shahbaz Youssefi <ShabbyX@gmail.com>

.gitignore

@@ -1,3 +1,3 @@
 /dist
 bin
-TestResult.xml
+*.swp

README.md

@@ -93,48 +93,13 @@ the buffer and `user_data` is the last argument of the `pandoc` function,
 similar to `user_data` of the reader.  The writer function must write the
 contents of the buffer as the output of the conversion by Pandoc.
 
+
 #### Input and Output Formats
 
 Input and output formats depend on Pandoc version the library is built
-against.  They are passed as strings.  Possible values include (TODO: needs
-to be verified and possibly extended):
-
-- For reader:
-
-  * docbook
-  * html
-  * latex
-  * markdown
-  * mediawiki
-  * native
-  * rst
-  * ~~texmath~~
-  * textile
-
-- For writer:
-
-  * asciidoc
-  * context
-  * docbook
-  * ~~docx~~
-  * ~~epub~~
-  * ~~fb2~~
-  * html
-  * latex
-  * man
-  * markdown
-  * mediawiki
-  * native
-  * ~~odt~~
-  * opendocument
-  * org
-  * rst
-  * rtf
-  * texinfo
-  * textile
-
-Note: Some read and write types supported by Pandoc striked above are not yet
-supported by libpandoc.
+against.  They are passed as strings.  Reader and writer names are the same
+as understood by Pandoc, for example `"html"` or `"markdown"`.
+
 
 #### JSON Settings
 
@@ -158,7 +123,8 @@ fields that have non-default values have to be provided.
 ## Changelog
 
   * 0.8
-    - Switched to JSON internal representation instead of XML
+    - Switched to JSON format for settings instead of XML
+    - Completed support for all Pandoc readers and writers
     - Updated to Pandoc version 1.16 and higher
   * 0.7 - Updated to Pandoc version 1.13 and higher
   * 0.6 - Updated to Pandoc version 1.10 and higher

libpandoc.cabal

@@ -7,7 +7,7 @@ Maintainer:             Shahbaz Youssefi <shabbyx@gmail.com>
 Synopsis:               Pandoc as a shared object or DLL.
 Build-Type:             Simple
 Executable libpandoc.so
-  If os(windows)
+  if os(windows)
     CPP-Options:        -DWIN32
   Extensions:           ForeignFunctionInterface
   Build-Depends:        base              >= 4.6,
@@ -32,6 +32,6 @@ Executable libpandoc.so
   Ghc-Options:          -no-hs-main -shared
   if !os(windows)
     Ghc-Options:        -dynamic
-    extra-libraries:    HSrts-ghc$compiler
+    extra-libraries:    HSrts-ghc7.10.2
   else
     extra-libraries:    HSrts

src/LibPandoc.hs

@@ -26,22 +26,21 @@ module LibPandoc (pandoc, LibPandocSettings(..), defaultLibPandocSettings) where
 
 import           Control.Arrow              ((>>>))
 import           Control.Exception          (catch, Exception(..), SomeException(..))
-import           Control.Monad.Except       (MonadError(..))
+import           Control.Monad              ((>=>), liftM)
 import qualified Data.ByteString.Lazy.Char8 as BLC
 import qualified Data.Char                  as Char
 import qualified Data.List                  as List
 import qualified Data.Map                   as Map
 import           Data.Maybe
-import           Data.String (IsString)
 import           Data.Typeable              (typeOf)
 import           Foreign
 import           Foreign.C.String
 import           Foreign.C.Types
 import           LibPandoc.IO
 import           LibPandoc.Settings
-import           System.IO.Unsafe
 import           Text.Pandoc
 import           Text.Pandoc.Error
+import           Text.Pandoc.MediaBag
 import           Text.JSON
 import           Text.JSON.Generic          (toJSON,fromJSON)
 
@@ -51,55 +50,9 @@ type CPandoc = CInt -> CString -> CString -> CString
              -> IO CString
 
 foreign export ccall "pandoc" pandoc     :: CPandoc
-foreign export ccall "increase" increase :: CInt -> IO CInt
 foreign import ccall "dynamic" peekReader :: FunPtr CReader -> CReader
 foreign import ccall "dynamic" peekWriter :: FunPtr CWriter -> CWriter
 
-increase :: CInt -> IO CInt
-increase x = return (x + 1)
-
-readNativeWrapper :: ReaderOptions -> String -> Either PandocError Pandoc
-readNativeWrapper options = readNative
-
-getInputFormat :: String -> Maybe (ReaderOptions -> String -> Either PandocError Pandoc)
-getInputFormat x =
-    case map Char.toLower x of
-      "docbook"    -> Just readDocBook
-      "html"       -> Just readHtml
-      "latex"      -> Just readLaTeX
-      "markdown"   -> Just readMarkdown
-      "mediawiki"  -> Just readMediaWiki
-      "native"     -> Just readNativeWrapper
-      "rst"        -> Just readRST
---      "texmath"    -> Just readTeXMath  TODO: disabled until I figure out how to convert it to ReaderOptions -> String -> Pandoc
-      "textile"    -> Just readTextile
-      _            -> Nothing
-
-getOutputFormat :: String -> Maybe (WriterOptions -> Pandoc -> String)
-getOutputFormat x =
-    case map Char.toLower x of
-      "asciidoc"     -> Just writeAsciiDoc
-      "context"      -> Just writeConTeXt
-      "docbook"      -> Just writeDocbook
---      "docx"         -> Just writeDocx  TODO: The following are disabled because they return IO types
---      "epub"         -> Just writeEPUB  TODO: Which I do not know yet how to mix with the non IO type
---      "fb2"          -> Just writeFB2
-      "html"         -> Just writeHtmlString
-      "latex"        -> Just writeLaTeX
-      "man"          -> Just writeMan
-      "markdown"     -> Just writeMarkdown
-      "mediawiki"    -> Just writeMediaWiki
-      "native"       -> Just writeNative
---      "odt"          -> Just writeODT
-      "opendocument" -> Just writeOpenDocument
-      "org"          -> Just writeOrg
-      "rst"          -> Just writeRST
-      "rtf"          -> Just writeRTF
-      "texinfo"      -> Just writeTexinfo
-      "textile"      -> Just writeTextile
-      _              -> Nothing
-
-
 -- | Gives preferential treatment to first argument (should be user options)
 joinJSON :: JSValue -> JSValue -> JSValue
 joinJSON (JSArray a) (JSArray b) = JSArray (List.zipWith joinJSON a b)
@@ -124,17 +77,34 @@ getSettings settings = do
 
 pandoc :: CPandoc
 pandoc bufferSize input output settings reader writer userData = do
-  let r = peekReader reader
-      w = peekWriter writer
+  let cr = peekReader reader
+      cw = peekWriter writer
   i <- peekCString input
   o <- peekCString output
   s <- getSettings settings
-  case (getInputFormat i, getOutputFormat o) of
-   (Nothing, _)            -> newCString "Invalid input format."
-   (_, Nothing)            -> newCString "Invalid output format."
-   (Just read, Just write) ->
-     do let run = read (readerOptions s) >>> handleError >>> write (writerOptions s)
-        result <- tryMaybe (transform (decodeInt bufferSize) run r w userData)
+
+  read <- return $ getReader i >>= \rd -> case rd of
+    -- if the reader returns just a string, add an empty media bag to it
+    StringReader r -> Right $ \o s -> r o s >>= \p -> return $ p >>= \x -> Right (x, mempty::MediaBag)
+    -- otherwise, the output (Pandoc, MediaBag) is already fine
+    ByteStringReader r -> Right $ \o s -> r o (BLC.pack s)
+  -- Note: currently, the media bag is actually later thrown away.  Perhaps in the future there could be support for it
+
+  write <- return $ getWriter o >>= \wr -> case wr of
+    -- if the writer returns just a string, add an IO
+    PureStringWriter w -> Right $ \o s -> return $ w o s
+    -- if it returns an IO string, it's fine
+    IOStringWriter w -> Right w
+    -- if it returns an IO byte string, convert it to IO string
+    IOByteStringWriter w -> Right $ \o s -> w o s >>= \s -> return $ BLC.unpack s
+
+  case (read, write) of
+   (Left e, _)            -> newCString e
+   (_, Left e)            -> newCString e
+   (Right r, Right w) ->
+     do let run = (r (readerOptions s) >=> return . liftM fst) >>> liftM handleError >>> \x -> x >>= w (writerOptions s)
+        -- run takes the media bag out of reader, as it is currently unused.  Since the reader returns IO, everything is lifted
+        result <- tryMaybe (transform (decodeInt bufferSize) run cr cw userData)
         case result of
          Just (SomeException res) -> newCString (show (typeOf res) ++ ": " ++ show res)
          Nothing -> return nullPtr

src/LibPandoc/IO.hs

@@ -41,7 +41,7 @@ type CWriter = CString -> CInt -> Ptr () -> IO ()
 -- | parameter is the size of the buffer in bytes.  The number of
 -- | bytes returned by the reader and passed to the writer should not
 -- | exceed the buffer size.
-transform :: Int -> (String -> String) -> CReader -> CWriter -> Ptr () -> IO ()
+transform :: Int -> (String -> IO String) -> CReader -> CWriter -> Ptr () -> IO ()
 transform bufferSize transformer reader writer userData =
     withBuffer bufferSize $ \rbuf ->
     withBuffer bufferSize $ \wbuf ->
@@ -65,8 +65,15 @@ readStream (Buffer (buf, size)) reader userData = unsafeInterleaveIO result wher
       k <- reader buf userData
       fmap (Utf8.decodeString s ++) (loop (decodeInt k))
 
-writeStream :: Buffer -> CWriter -> String -> Ptr() -> IO ()
-writeStream (Buffer (buf, size)) writer text userData = loop text where
+{--
+writeStream' :: Buffer -> CWriter -> IO String -> Ptr() -> IO ()
+writeStream' buf writer text userData = do
+    text' <- text
+    writeStream buf writer text' userData
+--}
+
+writeStream :: Buffer -> CWriter -> IO String -> Ptr() -> IO ()
+writeStream (Buffer (buf, size)) writer text userData = text >>= loop where
     buffer = castPtr buf
     loop text = do
       let (head, tail) = splitAt (div size 4) text
@@ -78,7 +85,7 @@ writeStream (Buffer (buf, size)) writer text userData = loop text where
         _  -> loop tail
 
 decodeInt :: CInt -> Int
-decodeInt x = fromInteger (toInteger x)
+decodeInt = fromInteger . toInteger
 
 encodeInt :: Int -> CInt
-encodeInt x = fromInteger (toInteger x)
+encodeInt = fromInteger . toInteger

src/pandoc.h

@@ -39,7 +39,7 @@ void pandoc_exit();
 /*
  * Calls `pandoc` with given input and output formats and streams.
  * Returns a `NULL` on success, or a `NULL`-terminated error message
- * on failure.  Settings is an XML string conforming to a schema
+ * on failure.  Settings is a JSON string conforming to a schema
  * distributed with `libpandoc`.  Settings can be `NULL`.  All strings
  * should be encoded as UTF-8.  User data is any pointer.
  */