2019-05-13 08:05:28 +02:00
|
|
|
{-# LANGUAGE NamedFieldPuns #-}
|
|
|
|
module PDF (
|
2019-05-16 22:41:14 +02:00
|
|
|
Document(..)
|
|
|
|
, parseDocument
|
|
|
|
, render
|
2019-05-13 08:05:28 +02:00
|
|
|
) where
|
|
|
|
|
|
|
|
import Data.ByteString.Lazy.Char8 (ByteString)
|
2019-05-16 11:01:50 +02:00
|
|
|
import qualified Data.ByteString.Lazy.Char8 as BS (
|
|
|
|
drop, findIndex, head, isPrefixOf, last, length, span, unpack
|
|
|
|
)
|
2019-05-13 08:05:28 +02:00
|
|
|
import Data.ByteString.Lazy.Char8.Util (previous, subBS)
|
|
|
|
import Data.Int (Int64)
|
2019-05-13 18:22:05 +02:00
|
|
|
import qualified Data.Map as Map (lookup)
|
2019-05-14 18:42:11 +02:00
|
|
|
import PDF.Body (populate)
|
2019-05-16 22:41:14 +02:00
|
|
|
import qualified PDF.EOL as EOL (Style(..), charset, parser)
|
2019-05-17 16:14:06 +02:00
|
|
|
import PDF.Object (
|
2019-05-18 09:01:13 +02:00
|
|
|
Content(..), DirectObject(..), InputStructure(..), Name(..), Number(..)
|
|
|
|
, Structure(..)
|
2019-05-17 16:14:06 +02:00
|
|
|
, eofMarker, magicNumber, structure
|
|
|
|
)
|
2019-05-16 22:41:14 +02:00
|
|
|
import qualified PDF.Output as Output (render, string)
|
|
|
|
import PDF.Output (Output(..), nextLine)
|
2019-05-13 08:05:28 +02:00
|
|
|
import Text.Parsec
|
|
|
|
import Text.Parsec.ByteString.Lazy (Parser)
|
|
|
|
import Text.Parsec.Pos (newPos)
|
|
|
|
import Text.ParserCombinators.Parsec.Error (Message(..), newErrorMessage)
|
2019-05-16 22:41:14 +02:00
|
|
|
import Text.Printf (printf)
|
2019-05-13 08:05:28 +02:00
|
|
|
|
|
|
|
data Document = Document {
|
|
|
|
pdfVersion :: String
|
2019-05-16 22:41:14 +02:00
|
|
|
, eolStyle :: EOL.Style
|
2019-05-13 18:22:05 +02:00
|
|
|
, contents :: [Content]
|
2019-05-13 08:05:28 +02:00
|
|
|
} deriving Show
|
|
|
|
|
2019-05-16 22:41:14 +02:00
|
|
|
instance Output Document where
|
|
|
|
output (Document {pdfVersion, contents}) =
|
|
|
|
Output.string (printf "%%PDF-%s" pdfVersion)
|
|
|
|
`nextLine` output contents
|
|
|
|
|
|
|
|
render :: Document -> ByteString
|
|
|
|
render document@(Document {eolStyle}) = Output.render eolStyle document
|
|
|
|
|
2019-05-13 08:05:28 +02:00
|
|
|
version :: Parser String
|
2019-05-16 22:41:14 +02:00
|
|
|
version = string magicNumber *> many (noneOf EOL.charset)
|
2019-05-13 08:05:28 +02:00
|
|
|
|
2019-05-13 18:22:05 +02:00
|
|
|
parseError :: String -> Either ParseError a
|
|
|
|
parseError errorMessage =
|
|
|
|
Left $ newErrorMessage (Message errorMessage) (newPos "" 0 0)
|
|
|
|
|
2019-05-13 08:05:28 +02:00
|
|
|
check :: Bool -> String -> Either ParseError ()
|
2019-05-13 18:22:05 +02:00
|
|
|
check test errorMessage = if test then return () else parseError errorMessage
|
2019-05-13 08:05:28 +02:00
|
|
|
|
2019-05-16 22:41:14 +02:00
|
|
|
readStartXref :: EOL.Style -> ByteString -> Either ParseError Int64
|
2019-05-13 08:05:28 +02:00
|
|
|
readStartXref eolStyle input =
|
|
|
|
check (eofMarker `BS.isPrefixOf` (BS.drop eofMarkerPosition input))
|
|
|
|
"Badly formed document : missing EOF marker at the end"
|
|
|
|
>> return (read . BS.unpack $ subBS startXrefPosition startXrefLength input)
|
|
|
|
where
|
|
|
|
(eolOffset, eolLastByte) = case eolStyle of
|
2019-05-16 22:41:14 +02:00
|
|
|
EOL.CRLF -> (2, '\n')
|
|
|
|
EOL.CR -> (1, '\r')
|
2019-05-13 08:05:28 +02:00
|
|
|
_ -> (1, '\n')
|
|
|
|
eofMarkerPosition =
|
|
|
|
BS.length input - BS.length eofMarker
|
|
|
|
- if BS.last input == BS.last eofMarker then 0 else eolOffset
|
|
|
|
startXrefPosition =
|
2019-05-13 11:34:15 +02:00
|
|
|
previous eolLastByte (eofMarkerPosition - eolOffset - 1) input + 1
|
2019-05-13 08:05:28 +02:00
|
|
|
startXrefLength = eofMarkerPosition - eolOffset - startXrefPosition
|
|
|
|
|
2019-05-16 22:41:14 +02:00
|
|
|
findNextLine :: ByteString -> Int64
|
|
|
|
findNextLine input =
|
2019-05-16 11:01:50 +02:00
|
|
|
let (line, eolPrefixed) = BS.span notInEol input in
|
|
|
|
let nextNotInEol = BS.findIndex notInEol eolPrefixed in
|
|
|
|
BS.length line + (maybe (BS.length eolPrefixed) id nextNotInEol)
|
|
|
|
where
|
2019-05-16 22:41:14 +02:00
|
|
|
notInEol = not . (`elem` EOL.charset)
|
2019-05-13 18:22:05 +02:00
|
|
|
|
2019-05-16 22:41:14 +02:00
|
|
|
findNextSection :: Int64 -> ByteString -> Int64
|
|
|
|
findNextSection offset input =
|
2019-05-16 11:01:50 +02:00
|
|
|
case BS.findIndex (== BS.head eofMarker) input of
|
|
|
|
Nothing -> 0
|
|
|
|
Just delta ->
|
|
|
|
let newInput = BS.drop delta input in
|
|
|
|
let newOffset = offset + delta in
|
|
|
|
if BS.isPrefixOf eofMarker newInput
|
2019-05-16 22:41:14 +02:00
|
|
|
then newOffset + findNextLine newInput
|
|
|
|
else findNextSection (newOffset + 1) (BS.drop 1 newInput)
|
2019-05-15 19:12:38 +02:00
|
|
|
|
2019-05-18 09:01:13 +02:00
|
|
|
readStructures :: Int64 -> ByteString -> Either ParseError [InputStructure]
|
2019-05-17 16:14:06 +02:00
|
|
|
readStructures startXref input =
|
|
|
|
parse structure ('@':show startXref) (BS.drop startXref input) >>= stopOrFollow
|
2019-05-13 18:22:05 +02:00
|
|
|
where
|
2019-05-18 09:01:13 +02:00
|
|
|
stopOrFollow s@(Structure {trailer}) =
|
|
|
|
case Map.lookup (Name "Prev") trailer of
|
|
|
|
Nothing -> Right [InputStructure (findNextLine input) s]
|
2019-05-17 16:14:06 +02:00
|
|
|
Just (NumberObject (Number newStartXref)) ->
|
2019-05-16 11:01:50 +02:00
|
|
|
let offset = truncate newStartXref in
|
2019-05-16 22:41:14 +02:00
|
|
|
let startOffset = findNextSection offset (BS.drop offset input) in
|
2019-05-18 09:01:13 +02:00
|
|
|
(InputStructure startOffset s:) <$> (readStructures offset input)
|
2019-05-13 18:22:05 +02:00
|
|
|
Just v -> parseError $ "Bad value for Prev entry in trailer: " ++ show v
|
2019-05-13 08:05:28 +02:00
|
|
|
|
2019-05-16 11:01:50 +02:00
|
|
|
parseDocument :: ByteString -> Either ParseError Document
|
|
|
|
parseDocument input = do
|
2019-05-16 22:41:14 +02:00
|
|
|
(pdfVersion, eolStyle) <- parse ((,) <$> version <*> EOL.parser) "" input
|
2019-05-16 11:01:50 +02:00
|
|
|
startXref <- readStartXref eolStyle input
|
2019-05-17 16:14:06 +02:00
|
|
|
structuresRead <- readStructures startXref input
|
|
|
|
let contents = populate input <$> structuresRead
|
2019-05-16 22:41:14 +02:00
|
|
|
return $ Document {pdfVersion, eolStyle, contents}
|