diff --git a/app/Main.hs b/app/Main.hs
index 23a86b9..afdee16 100644
--- a/app/Main.hs
+++ b/app/Main.hs
@@ -2,44 +2,25 @@ module Main
     ( main
     ) where
 
-import Control.Monad (foldM_)
-import Data.Int (Int64)
-import qualified Data.ByteString as ByteString
-import qualified Data.ByteString.Char8 as Char8
-import Data.ByteString (ByteString)
+import Data.Bifunctor (Bifunctor(..))
 import qualified Text.Megaparsec as Megaparsec
-import Data.Foldable (find)
 import qualified Data.Text.Lazy.Builder as Text.Builder
-import qualified Data.Text.Lazy.Builder.Int as Text.Builder
-import qualified Data.Text.Lazy as Text.Lazy
 import qualified Data.Text.Lazy.IO as Text.Lazy
-import qualified Data.Text.Encoding as Text
-import GHC.Records (HasField(..))
-import Graphics.Fountainhead.Dumper (DumpError(..), dumpTrueType)
+import Graphics.Fountainhead (parseFontDirectoryFromFile)
+import Graphics.Fountainhead.Dumper (DumpError(..), dumpTables)
 -- TODO: kern table since format 1.
 -- For details on subtable format see examples in TrueType reference.
-import Graphics.Fountainhead.Parser
-    ( fontDirectoryP
-    , os2TableP
-    , parseTable
-    , shortLocaTableP
-    )
-import Graphics.Fountainhead.TrueType
-    ( FontDirectory(..)
-    , OffsetSubtable(..)
-    , TableDirectory(..)
-    )
 import System.Environment (getArgs)
 import System.Exit (exitWith)
 import GHC.IO.Exception (ExitCode(..))
 
 fontMain :: FilePath -> IO ()
 fontMain fontFile = do
-    putStrLn $ "Dumping File:" <> fontFile <> "\n\n"
+    putStrLn ("Dumping File:" <> fontFile <> "\n\n")
 
-    ttfContents <- ByteString.readFile fontFile
-    
-    case dumpTrueType ttfContents fontFile of
+    (processedState, initialResult) <- parseFontDirectoryFromFile fontFile
+
+    case first DumpParseError initialResult >>= dumpTables processedState of
         Right fontDump -> Text.Lazy.putStrLn $ Text.Builder.toLazyText fontDump
         Left e
             | DumpParseError bundle <- e -> putStr
diff --git a/fountainhead.cabal b/fountainhead.cabal
index e02f11b..1271e18 100644
--- a/fountainhead.cabal
+++ b/fountainhead.cabal
@@ -21,21 +21,22 @@ extra-source-files:
 
 common dependencies
   build-depends:
-    text ^>= 2.0
+    bytestring ^>= 0.11.0,
+    text ^>= 2.0,
+    zlib ^>= 0.6.3
+  default-language: Haskell2010
 
 library
   import: dependencies
   exposed-modules:
+    Graphics.Fountainhead
     Graphics.Fountainhead.Dumper
     Graphics.Fountainhead.Parser
-    Graphics.Fountainhead.PDF
     Graphics.Fountainhead.Type
     Graphics.Fountainhead.TrueType
-  hs-source-dirs:
-    src
+  hs-source-dirs: src
   build-depends:
     base >= 4.16 && < 5,
-    bytestring ^>= 0.11.0,
     containers ^>= 0.6.5,
     megaparsec ^>= 9.3,
     time ^>= 1.12,
@@ -53,7 +54,6 @@ executable fountainhead
     ExplicitForAll
   build-depends:
     base,
-    bytestring,
     containers,
     parser-combinators,
     vector,
@@ -62,4 +62,4 @@ executable fountainhead
     megaparsec,
     fountainhead
   hs-source-dirs: app
-  default-language: Haskell2010
+  ghc-options: -Wall
diff --git a/src/Graphics/Fountainhead.hs b/src/Graphics/Fountainhead.hs
new file mode 100644
index 0000000..f965680
--- /dev/null
+++ b/src/Graphics/Fountainhead.hs
@@ -0,0 +1,50 @@
+{- This Source Code Form is subject to the terms of the Mozilla Public License,
+   v. 2.0. If a copy of the MPL was not distributed with this file, You can
+   obtain one at https://mozilla.org/MPL/2.0/. -}
+
+module Graphics.Fountainhead
+    ( parseFontDirectoryFromFile
+    ) where
+
+import qualified Codec.Compression.Zlib as Zlib
+import Data.ByteString (ByteString)
+import qualified Data.ByteString as ByteString
+import qualified Data.ByteString.Lazy as ByteString.Lazy
+import Data.Void (Void)
+import Graphics.Fountainhead.Parser (ParseErrorBundle, fontDirectoryP)
+import Graphics.Fountainhead.TrueType (FontDirectory(..))
+import qualified Text.Megaparsec as Megaparsec
+import Text.Megaparsec (PosState(..), State(..))
+import System.IO (IOMode(..), SeekMode(..), hFileSize, hSeek, withBinaryFile)
+
+parseFontDirectoryFromFile :: String
+    -> IO (State ByteString Void, Either ParseErrorBundle FontDirectory)
+parseFontDirectoryFromFile fontFile =
+    withBinaryFile fontFile ReadMode withFontHandle
+  where
+    withFontHandle fontHandle = doParsing
+        <$> readFontContents fontHandle
+    doParsing ttfContents =
+        let initialState = Megaparsec.State
+                { stateInput = ttfContents
+                , stateOffset = 0
+                , statePosState = Megaparsec.PosState
+                    { pstateInput = ttfContents
+                    , pstateOffset = 0
+                    , pstateSourcePos = Megaparsec.initialPos fontFile
+                    , pstateTabWidth = Megaparsec.defaultTabWidth
+                    , pstateLinePrefix = ""
+                    }
+                , stateParseErrors = []
+                }
+         in Megaparsec.runParser' fontDirectoryP initialState 
+    readFontContents fontHandle = do
+        firstBytes <- ByteString.unpack <$> ByteString.hGet fontHandle 2
+        hSeek fontHandle AbsoluteSeek 0
+        fileSize <- fromIntegral <$> hFileSize fontHandle
+        case firstBytes of
+            0x78 : [secondByte]
+                | secondByte `elem` [0x01, 0x9c, 0x5e, 0xda] ->
+                    ByteString.Lazy.toStrict . Zlib.decompress
+                        <$> ByteString.Lazy.hGet fontHandle fileSize
+            _ -> ByteString.hGetContents fontHandle
diff --git a/src/Graphics/Fountainhead/Dumper.hs b/src/Graphics/Fountainhead/Dumper.hs
index 46655cb..b060de7 100644
--- a/src/Graphics/Fountainhead/Dumper.hs
+++ b/src/Graphics/Fountainhead/Dumper.hs
@@ -22,6 +22,7 @@ module Graphics.Fountainhead.Dumper
     , dumpMaxp
     , dumpOs2
     , dumpPost
+    , dumpTables
     , dumpTrueType
     , dumpOffsetTable
     ) where
@@ -91,7 +92,8 @@ import Graphics.Fountainhead.Parser
     , maxpTableP
     , nameTableP
     , os2TableP
-    , postTableP, cvTableP
+    , postTableP
+    , cvTableP
     )
 import Graphics.Fountainhead.Type (Fixed32(..), ttfEpoch)
 import Data.Foldable (Foldable(..), find)
diff --git a/src/Graphics/Fountainhead/PDF.hs b/src/Graphics/Fountainhead/PDF.hs
deleted file mode 100644
index 1ed42fc..0000000
--- a/src/Graphics/Fountainhead/PDF.hs
+++ /dev/null
@@ -1,316 +0,0 @@
-{- This Source Code Form is subject to the terms of the Mozilla Public License,
-   v. 2.0. If a copy of the MPL was not distributed with this file, You can
-   obtain one at https://mozilla.org/MPL/2.0/. -}
-
-{-# LANGUAGE EmptyDataDecls #-}
-
-module Graphics.Fountainhead.PDF
-    ( Dictionary(..)
-    , E5
-    , Header(..)
-    , Link(..)
-    , Name(..)
-    , Object(..)
-    , Sink
-    , TextString(..)
-    , Trailer(..)
-    , Type(..)
-    , UncoatedString(..)
-    , XRefSection(..)
-    , XRefEntry(..)
-    , arrayType
-    , headerToPdf
-    , dictionaryToPdf
-    , dictionaryType
-    , linkToPdf
-    , linkType
-    , nameToPdf
-    , nameType
-    , nextName
-    , objectToPdf
-    , sinkWithLength
-    , stringType
-    , textStringToPdf
-    , textType
-    , trailerToPdf
-    , typeToPdf
-    , uncoatedStringToPdf
-    , xrefEntryToPdf
-    , xrefSectionToPdf
-    , writeObject
-    ) where
-
-import Control.Monad.IO.Class (MonadIO(..))
-import Control.Monad.Trans.State (StateT, get, gets, put)
-import Data.ByteString (ByteString)
-import qualified Data.ByteString as ByteString
-import qualified Data.ByteString.Builder as ByteString.Builder
-import qualified Data.ByteString.Builder as ByteString (Builder)
-import qualified Data.ByteString.Lazy as ByteString.Lazy
-import Data.Char (ord)
-import Data.Fixed (Fixed(..), HasResolution(..), showFixed)
-import Data.Vector (Vector)
-import qualified Data.Vector as Vector
-
--- | The header in the first line of a PDF file contains a PDF version number
--- consisting of a major and a minor version.
-data Header = Header Int Int
-    deriving (Eq, Show)
-
--- | See t'Header'.
-headerToPdf :: Header -> ByteString.Builder
-headerToPdf (Header major minor)
-    = ByteString.Builder.string7 "%PDF-"
-    <> ByteString.Builder.intDec major
-    <> ByteString.Builder.char7 '.'
-    <> ByteString.Builder.intDec minor
-    <> ByteString.Builder.char7 '\n'
-
--- | A name object is an atomic symbol uniquely defined by a sequence of
--- characters.
-newtype Name = Name String
-    deriving (Eq, Show)
-
--- | See t'Name'.
-nameToPdf :: Name -> ByteString.Builder
-nameToPdf (Name name) = ByteString.Builder.char7 '/'
-    <> ByteString.Builder.string7 name
-
--- | A dictionary object is an associative table containing pairs of objects.
-newtype Dictionary = Dictionary (Vector (Name, Type))
-
--- | See t'Dictionary'.
-dictionaryToPdf :: Dictionary -> IO ByteString.Builder
-dictionaryToPdf (Dictionary valuePairs) = do
-    pairs <- traverse pairToPdf valuePairs
-    pure
-        $ ByteString.Builder.string7 "<<"
-        <> unwordBuilder pairs
-        <> ByteString.Builder.string7 ">>"
-  where
-    pairToPdf (name, value) = do
-        value' <- typeToPdf value
-        pure $ nameToPdf name <> ByteString.Builder.char7 ' ' <> value'
-
--- | Hexadecimal data.
-newtype UncoatedString = UncoatedString String
-    deriving (Eq, Show)
-
--- | See t'UncoatedString'.
-uncoatedStringToPdf :: UncoatedString -> ByteString.Builder
-uncoatedStringToPdf (UncoatedString uncoatedString)
-    = ByteString.Builder.char7 '<'
-    <> ByteString.Builder.string8 uncoatedString
-    <> ByteString.Builder.char7 '>'
-
--- | A sequence of literal characters.
-newtype TextString = TextString String
-    deriving (Eq, Show)
-
--- | See t'TextString'.
-textStringToPdf :: TextString -> ByteString.Builder
-textStringToPdf (TextString textString)
-    = ByteString.Builder.char7 '('
-    <> ByteString.Builder.stringUtf8 textString 
-    <> ByteString.Builder.char7 ')'
-
--- Resolution of 10^-5 = .001.
-data E5
-
-instance HasResolution E5
-  where
-    resolution _ = 100000
-
--- | Reference to an inderect object, consisting of the object name and
--- revision.
-data Link = Link Int Int
-    deriving (Eq, Show)
-
--- | See t'Link'.
-linkToPdf :: Link -> ByteString.Builder
-linkToPdf (Link name revision)
-    = ByteString.Builder.intDec name
-    <> ByteString.Builder.char7 ' '
-    <> ByteString.Builder.intDec revision
-    <> ByteString.Builder.string8 " R"
-
--- | Basic types of object.
-data Type
-    = DictionaryType Dictionary
-    | ArrayType (Vector Type)
-    | LinkType Link
-    | NameType Name
-    | IntegerType Int
-    | RealType (Fixed E5)
-    | StreamType Dictionary (IO ByteString)
-    | StringType UncoatedString
-    | TextType TextString
-    | NullType
-
--- | See t'Type'.
-typeToPdf :: Type -> IO ByteString.Builder
-typeToPdf (DictionaryType dictionary) = dictionaryToPdf dictionary
-typeToPdf (ArrayType values) = do
-    converted <- traverse typeToPdf values
-    pure
-        $ ByteString.Builder.char7 '['
-        <> unwordBuilder converted
-        <> ByteString.Builder.char7 ']'
-typeToPdf (LinkType link) = pure $ linkToPdf link
-typeToPdf (NameType name) = pure $ nameToPdf name
-typeToPdf (IntegerType pdfInteger) = pure $ ByteString.Builder.intDec pdfInteger
-typeToPdf (StreamType dictionary producer) = do
-    streamContents <- producer
-    producedDictionary <- dictionaryToPdf dictionary
-    pure
-        $ producedDictionary
-        <> ByteString.Builder.string8 "\nstream\n"
-        <> ByteString.Builder.byteString streamContents
-        <> ByteString.Builder.string8 "\nendstream"
-typeToPdf (StringType string) = pure $ uncoatedStringToPdf string
-typeToPdf (TextType text) = pure $ textStringToPdf text
-typeToPdf (RealType realType) =
-    pure $ ByteString.Builder.string7 $ showFixed True realType
-typeToPdf NullType = pure $ ByteString.Builder.string7 "null"
-
--- | Object number, generation number and object contents.
-data Object = Object Int Int Type
-
--- | See t'Object'.
-objectToPdf :: Object -> IO ByteString.Builder
-objectToPdf (Object name revision type') = do
-    producedType <- typeToPdf type'
-    pure $ ByteString.Builder.intDec name
-        <> ByteString.Builder.char7 ' '
-        <> ByteString.Builder.intDec revision
-        <> ByteString.Builder.string7 " obj\n"
-        <> producedType
-        <> ByteString.Builder.string7 "\nendobj\n"
-
--- | Shortcut to create a t'Dictionary' type.
-dictionaryType :: [(Name, Type)] -> Type
-dictionaryType = DictionaryType . Dictionary . Vector.fromList
-
--- | Shortcut to create an t'Array' type.
-arrayType :: [Type] -> Type
-arrayType = ArrayType . Vector.fromList
-
--- | Shortcut to create a t'Name' type.
-nameType :: String -> Type
-nameType = NameType . Name
-
--- | Shortcut to create a t'UncoatedString' type.
-stringType :: String -> Type
-stringType = StringType . UncoatedString
-
--- | Shortcut to create a t'TextString' type.
-textType :: String -> Type
-textType = TextType . TextString
-
--- | Shortcut to create a t'Link' type.
-linkType :: Int -> Int -> Type
-linkType name revision = LinkType $ Link name revision
-
--- | Byte offset of an object in the file, generation number and whether this is
--- an in-use entry.
-data XRefEntry = XRefEntry Int Int Bool
-    deriving (Eq, Show)
-
--- | See t'XRefEntry'.
-xrefEntryToPdf :: XRefEntry -> ByteString.Builder
-xrefEntryToPdf (XRefEntry offset generation True)
-    = pad 10 offset
-    <> ByteString.Builder.char7 ' '
-    <> pad 5 generation
-    <> ByteString.Builder.string7 " n"
-xrefEntryToPdf (XRefEntry offset generation False)
-    = pad 10 offset
-    <> ByteString.Builder.char7 ' '
-    <> pad 5 generation
-    <> ByteString.Builder.string7 " f"
-
--- | Cross-reference table containing information about the indirect objects in
--- the file.
-newtype XRefSection = XRefSection
-    { unXRefSection :: Vector XRefEntry
-    } deriving (Eq, Show)
-
--- | See t'XRefSection'.
-xrefSectionToPdf :: XRefSection -> ByteString.Builder
-xrefSectionToPdf (XRefSection entries)
-    = ByteString.Builder.string7 "xref\n0 "
-    <> ByteString.Builder.intDec (length entries)
-    <> newline
-    <> Vector.foldMap (newline <>) (xrefEntryToPdf <$> entries)
-    <> newline
-
-instance Semigroup XRefSection
-  where
-    (XRefSection lhs) <> (XRefSection rhs) = XRefSection $ lhs <> rhs
-
-instance Monoid XRefSection
-  where
-    mempty = XRefSection mempty
-
--- | A trailer giving the location of the cross-reference table and of certain
--- special objects within the body of the file.
-data Trailer = Trailer Dictionary Int
-
--- | See t'Trailer'.
-trailerToPdf :: Trailer -> IO ByteString.Builder
-trailerToPdf (Trailer dictionary startxref) = do
-    producedDictionary <- dictionaryToPdf dictionary
-    pure $ ByteString.Builder.string7 "trailer "
-        <> producedDictionary
-        <> ByteString.Builder.string7 "\nstartxref\n"
-        <> ByteString.Builder.intDec startxref
-        <> ByteString.Builder.string7 "\n%%EOF\n"
-
-pad :: Int -> Int -> ByteString.Builder
-pad length' number =
-    let asString = ByteString.Builder.intDec number
-        numberLength = builderLength asString
-        padding = ByteString.Builder.byteString
-            $ ByteString.replicate (length' - numberLength) zero
-     in padding <> asString
-  where
-    zero = fromIntegral $ ord '0'
-    builderLength = fromIntegral
-        . ByteString.Lazy.length
-        . ByteString.Builder.toLazyByteString
-
-unwordBuilder :: Vector ByteString.Builder -> ByteString.Builder
-unwordBuilder = Vector.foldMap (ByteString.Builder.char7 ' ' <>)
-
-newline :: ByteString.Builder
-newline = ByteString.Builder.char7 '\n'
-
-type Sink = ByteString.Lazy.ByteString -> IO ()
-
--- | Creates a new object using the provided value, writes the object to the
--- sink, and returns a reference to that object.
---
--- For example if the passed value is a dictionary, the created object could be
--- @
--- 2 0 obj <<…>> endobj
--- @
--- where "2 0" identifies the object. The name (2) is generated using the state,
--- the revision is always 0.
-writeObject :: Sink -> Type -> StateT (Int, XRefSection) IO Link
-writeObject sink object = do
-    (previousLength, XRefSection refs) <- get
-    let objectName = length refs
-    objectContents <- liftIO $ objectToPdf $ Object objectName 0 object
-    writtenLength <- liftIO $ sinkWithLength sink objectContents
-    put (previousLength + writtenLength, XRefSection $ Vector.snoc refs (XRefEntry previousLength 0 True))
-    pure $ Link objectName 0
-
--- | Gets the name of the object which will be generated next.
-nextName :: StateT (Int, XRefSection) IO Int
-nextName = gets (length . unXRefSection . snd)
-
--- | Writes the data into the sink and returns the number of the written bytes.
-sinkWithLength :: Sink -> ByteString.Builder -> IO Int
-sinkWithLength sink data' =
-    let lazyData = ByteString.Builder.toLazyByteString data'
-     in sink lazyData >> pure (fromIntegral $ ByteString.Lazy.length lazyData)
diff --git a/src/Graphics/Fountainhead/Parser.hs b/src/Graphics/Fountainhead/Parser.hs
index 590f95f..bb74ea7 100644
--- a/src/Graphics/Fountainhead/Parser.hs
+++ b/src/Graphics/Fountainhead/Parser.hs
@@ -12,6 +12,7 @@
 -- | Font parser.
 module Graphics.Fountainhead.Parser
     ( Parser
+    , ParseErrorBundle
     , cmapTableP
     , cvTableP
     , f2Dot14P
@@ -147,6 +148,7 @@ import qualified Text.Megaparsec as Megaparsec
 import qualified Text.Megaparsec.Byte.Binary as Megaparsec.Binary
 
 type Parser = Megaparsec.Parsec Void ByteString
+type ParseErrorBundle = Megaparsec.ParseErrorBundle ByteString Void
 
 -- * Font directory
 
@@ -942,7 +944,7 @@ parseTable
     :: TableDirectory
     -> Parser a
     -> Megaparsec.State ByteString Void
-    -> Either (Megaparsec.ParseErrorBundle ByteString Void) a
+    -> Either ParseErrorBundle a
 parseTable TableDirectory{ offset, length = length' } parser state = snd
     $ Megaparsec.runParser' parser 
     $ state