module Data.Csv.Encoding
(
HasHeader(..)
, decode
, decodeByName
, Quoting(..)
, encode
, encodeByName
, encodeDefaultOrderedByName
, DecodeOptions(..)
, defaultDecodeOptions
, decodeWith
, decodeByNameWith
, EncodeOptions(..)
, defaultEncodeOptions
, encodeWith
, encodeByNameWith
, encodeDefaultOrderedByNameWith
, encodeRecord
, encodeNamedRecord
, recordSep
) where
import Blaze.ByteString.Builder (Builder, fromByteString, fromWord8,
toLazyByteString, toByteString)
import Blaze.ByteString.Builder.Char8 (fromString)
import Control.Applicative ((<|>), optional)
import Data.Attoparsec.ByteString.Char8 (endOfInput)
import qualified Data.Attoparsec.ByteString.Lazy as AL
import qualified Data.ByteString as B
import qualified Data.ByteString.Char8 as B8
import qualified Data.ByteString.Lazy as L
import qualified Data.ByteString.Lazy.Char8 as BL8
import qualified Data.HashMap.Strict as HM
import Data.Vector (Vector)
import qualified Data.Vector as V
import Data.Word (Word8)
import Prelude hiding (unlines)
import Data.Csv.Compat.Monoid ((<>))
import qualified Data.Csv.Conversion as Conversion
import Data.Csv.Conversion (FromNamedRecord, FromRecord, ToNamedRecord,
ToRecord, parseNamedRecord, parseRecord, runParser,
toNamedRecord, toRecord)
import Data.Csv.Parser hiding (csv, csvWithHeader)
import qualified Data.Csv.Parser as Parser
import Data.Csv.Types hiding (toNamedRecord)
import qualified Data.Csv.Types as Types
import Data.Csv.Util (blankLine, endOfLine)
#if !MIN_VERSION_base(4,8,0)
import Control.Applicative ((*>), pure)
import Data.Monoid (mconcat, mempty)
#endif
decode :: FromRecord a
=> HasHeader
-> L.ByteString
-> Either String (Vector a)
decode = decodeWith defaultDecodeOptions
decodeByName :: FromNamedRecord a
=> L.ByteString
-> Either String (Header, Vector a)
decodeByName = decodeByNameWith defaultDecodeOptions
encode :: ToRecord a => [a] -> L.ByteString
encode = encodeWith defaultEncodeOptions
encodeByName :: ToNamedRecord a => Header -> [a] -> L.ByteString
encodeByName = encodeByNameWith defaultEncodeOptions
encodeDefaultOrderedByName :: (Conversion.DefaultOrdered a, ToNamedRecord a) =>
[a] -> L.ByteString
encodeDefaultOrderedByName = encodeDefaultOrderedByNameWith defaultEncodeOptions
decodeWith :: FromRecord a
=> DecodeOptions
-> HasHeader
-> L.ByteString
-> Either String (Vector a)
decodeWith = decodeWithC csv
idDecodeWith :: DecodeOptions -> HasHeader -> L.ByteString
-> Either String (Vector (Vector B.ByteString))
idDecodeWith = decodeWithC Parser.csv
decodeWithC :: (DecodeOptions -> AL.Parser a) -> DecodeOptions -> HasHeader
-> BL8.ByteString -> Either String a
decodeWithC p !opts hasHeader = decodeWithP parser
where parser = case hasHeader of
HasHeader -> header (decDelimiter opts) *> p opts
NoHeader -> p opts
decodeByNameWith :: FromNamedRecord a
=> DecodeOptions
-> L.ByteString
-> Either String (Header, Vector a)
decodeByNameWith !opts = decodeWithP (csvWithHeader opts)
data Quoting
= QuoteNone
| QuoteMinimal
| QuoteAll
deriving (Eq, Show)
data EncodeOptions = EncodeOptions
{
encDelimiter :: !Word8
, encUseCrLf :: !Bool
, encIncludeHeader :: !Bool
, encQuoting :: !Quoting
} deriving (Eq, Show)
defaultEncodeOptions :: EncodeOptions
defaultEncodeOptions = EncodeOptions
{ encDelimiter = 44
, encUseCrLf = True
, encIncludeHeader = True
, encQuoting = QuoteMinimal
}
encodeWith :: ToRecord a => EncodeOptions -> [a] -> L.ByteString
encodeWith opts
| validDelim (encDelimiter opts) =
toLazyByteString
. unlines (recordSep (encUseCrLf opts))
. map (encodeRecord (encQuoting opts) (encDelimiter opts)
. toRecord)
| otherwise = encodeOptionsError
validDelim :: Word8 -> Bool
validDelim delim = delim `notElem` [cr, nl, dquote]
where
nl = 10
cr = 13
dquote = 34
encodeOptionsError :: a
encodeOptionsError = error $ "Data.Csv: " ++
"The 'encDelimiter' must /not/ be the quote character (i.e. " ++
"\") or one of the record separator characters (i.e. \\n or " ++
"\\r)"
encodeRecord :: Quoting -> Word8 -> Record -> Builder
encodeRecord qtng delim = mconcat . intersperse (fromWord8 delim)
. map fromByteString . map (escape qtng delim) . V.toList
encodeNamedRecord :: Header -> Quoting -> Word8 -> NamedRecord -> Builder
encodeNamedRecord hdr qtng delim =
encodeRecord qtng delim . namedRecordToRecord hdr
escape :: Quoting -> Word8 -> B.ByteString -> B.ByteString
escape !qtng !delim !s
| (qtng == QuoteMinimal &&
B.any (\ b -> b == dquote || b == delim || b == nl || b == cr || b == sp) s
) || qtng == QuoteAll
= toByteString $
fromWord8 dquote
<> B.foldl
(\ acc b -> acc <> if b == dquote
then fromByteString "\"\""
else fromWord8 b)
mempty
s
<> fromWord8 dquote
| otherwise = s
where
dquote = 34
nl = 10
cr = 13
sp = 32
encodeByNameWith :: ToNamedRecord a => EncodeOptions -> Header -> [a]
-> L.ByteString
encodeByNameWith opts hdr v
| validDelim (encDelimiter opts) =
toLazyByteString (rows (encIncludeHeader opts))
| otherwise = encodeOptionsError
where
rows False = records
rows True = encodeRecord (encQuoting opts) (encDelimiter opts) hdr <>
recordSep (encUseCrLf opts) <> records
records = unlines (recordSep (encUseCrLf opts))
. map (encodeNamedRecord hdr (encQuoting opts) (encDelimiter opts)
. toNamedRecord)
$ v
encodeDefaultOrderedByNameWith ::
forall a. (Conversion.DefaultOrdered a, ToNamedRecord a) =>
EncodeOptions -> [a] -> L.ByteString
encodeDefaultOrderedByNameWith opts v
| validDelim (encDelimiter opts) =
toLazyByteString (rows (encIncludeHeader opts))
| otherwise = encodeOptionsError
where
hdr = (Conversion.headerOrder (undefined :: a))
rows False = records
rows True = encodeRecord (encQuoting opts) (encDelimiter opts) hdr <>
recordSep (encUseCrLf opts) <> records
records = unlines (recordSep (encUseCrLf opts))
. map (encodeNamedRecord hdr (encQuoting opts) (encDelimiter opts)
. toNamedRecord)
$ v
namedRecordToRecord :: Header -> NamedRecord -> Record
namedRecordToRecord hdr nr = V.map find hdr
where
find n = case HM.lookup n nr of
Nothing -> moduleError "namedRecordToRecord" $
"header contains name " ++ show (B8.unpack n) ++
" which is not present in the named record"
Just v -> v
moduleError :: String -> String -> a
moduleError func msg = error $ "Data.Csv.Encoding." ++ func ++ ": " ++ msg
recordSep :: Bool -> Builder
recordSep False = fromWord8 10
recordSep True = fromString "\r\n"
unlines :: Builder -> [Builder] -> Builder
unlines _ [] = mempty
unlines sep (b:bs) = b <> sep <> unlines sep bs
intersperse :: Builder -> [Builder] -> [Builder]
intersperse _ [] = []
intersperse sep (x:xs) = x : prependToAll sep xs
prependToAll :: Builder -> [Builder] -> [Builder]
prependToAll _ [] = []
prependToAll sep (x:xs) = sep <> x : prependToAll sep xs
decodeWithP :: AL.Parser a -> L.ByteString -> Either String a
decodeWithP p s =
case AL.parse p s of
AL.Done _ v -> Right v
AL.Fail left _ msg -> Left errMsg
where
errMsg = "parse error (" ++ msg ++ ") at " ++
(if BL8.length left > 100
then (take 100 $ BL8.unpack left) ++ " (truncated)"
else show (BL8.unpack left))
csv :: FromRecord a => DecodeOptions -> AL.Parser (V.Vector a)
csv !opts = do
vals <- records
_ <- optional endOfLine
endOfInput
return $! V.fromList vals
where
records = do
!r <- record (decDelimiter opts)
if blankLine r
then (endOfLine *> records) <|> pure []
else case runParser (parseRecord r) of
Left msg -> fail $ "conversion error: " ++ msg
Right val -> do
!vals <- (endOfLine *> records) <|> pure []
return (val : vals)
csvWithHeader :: FromNamedRecord a => DecodeOptions
-> AL.Parser (Header, V.Vector a)
csvWithHeader !opts = do
!hdr <- header (decDelimiter opts)
vals <- records hdr
_ <- optional endOfLine
endOfInput
let !v = V.fromList vals
return (hdr, v)
where
records hdr = do
!r <- record (decDelimiter opts)
if blankLine r
then (endOfLine *> records hdr) <|> pure []
else case runParser (convert hdr r) of
Left msg -> fail $ "conversion error: " ++ msg
Right val -> do
!vals <- (endOfLine *> records hdr) <|> pure []
return (val : vals)
convert hdr = parseNamedRecord . Types.toNamedRecord hdr