diff --git a/Text/Subtitles/Datatypes.hs b/Text/Subtitles/Datatypes.hs deleted file mode 100644 --- a/Text/Subtitles/Datatypes.hs +++ /dev/null @@ -1,70 +0,0 @@ --- | --- Module : Text.Subtitles.Datatypes --- Copyright : Ruben Astudillo 2012 --- License : BSD3 --- --- Maintainer : ruben.astud@gmail.com --- Portability : unknown --- --- Common ADT for the project. Also serves as a place to provide instance --- declarations for the ADTs. - -module Text.Subtitles.Datatypes - ( - -- * Re-exported datatypes. - module Data.Attoparsec.Text, - module Data.Text, - - -- * Datatypes - Subtitles, - Line(..), - Range(..), - Time(..) - ) where - -import Data.List (intercalate) -import Data.Text (Text, unpack) -import Data.Attoparsec.Text (Parser) - -data Time = Time - { hour :: Int - , minutes :: Int - , seconds :: Int - , frame :: Int - } deriving (Eq, Ord) - -data Range = Range - { from :: Time - , to :: Time - } deriving (Eq, Ord) - --- | The core of the parser. each one of the constructor representing one part --- of the Line -data Line = Line - { index :: Int - , range :: Range - , subs :: Text - } deriving (Eq, Ord) - -type Subtitles = [Line] - -instance Show Time where - show (Time h m s f) = concat [showT h, ":", showT m, ":", showT s, ",", showF f] - -instance Show Range where - show (Range f t) = concat [show f, " --> ", show t] - -instance Show Line where - show (Line i t s) = intercalate "\n" (show i : show t : [unpack s]) ++ "\n" - -{- showT stands for showTime. In the parsing process "00" is read as "0", and - - that tailing 0 is lost unless we add it manually when showing -} -showT :: Int -> String -showT a | a < 10 = '0' : show a - | otherwise = show a - -{- showF stands for showFrame -} -showF :: Int -> String -showF a | a < 10 = '0':'0': show a - | a < 100 = '0': show a - | otherwise = show a diff --git a/Text/Subtitles/SRT.hs b/Text/Subtitles/SRT.hs --- a/Text/Subtitles/SRT.hs +++ b/Text/Subtitles/SRT.hs @@ -14,10 +14,10 @@ -- $example -- * Re-exported Datatypes - module Text.Subtitles.Datatypes, + module Text.Subtitles.SRT.Datatypes, -- * Main parsers - parseSubtitles, + parseSRT, parseSingleLine ) where @@ -26,12 +26,12 @@ import Data.Attoparsec.Text import qualified Data.Text as T -- in project modules -import Text.Subtitles.Datatypes +import Text.Subtitles.SRT.Datatypes -- $example -- -- All the sections of a Line have their corresponding ADT in --- "Text.Subtitles.Datatypes" +-- "Text.Subtitles.SRT.Datatypes" -- -- >2 -- >00:00:50,050 --> 00:00:52,217 @@ -50,14 +50,14 @@ -- |Main Parser, gives you a list of all the Lines of the subtitle. It fails if -- the subtitle doesn't have any Lines. -parseSubtitles :: Parser Subtitles -parseSubtitles = many1 parseSingleLine +parseSRT :: Parser Subtitles +parseSRT = many1 parseSingleLine -- |The individual Line parser. given the upper example return the -- corresponding Line representation parseSingleLine :: Parser Line parseSingleLine = - Line <$> parseIndex <*> parseRange <* eol <*> parseSubs T.empty + Line <$> parseIndex <*> parseRange <* eol <*> parseDialog T.empty parseIndex :: Parser Int parseIndex = decimal <* eol @@ -80,13 +80,13 @@ {- return the dialog checking for newlines that could be in there. that why is - written in a monad instead of applicative. more efficient version welcome -} -parseSubs :: Text -> Parser Text -parseSubs t = do +parseDialog :: Text -> Parser Text +parseDialog t = do line <- takeWhile1 (not . isEndOfLine) endOfLine let lineState = T.append t (T.snoc line '\n') next <- anyChar case next of '\n' -> return lineState - _ -> parseSubs (T.snoc lineState next) + _ -> parseDialog (T.snoc lineState next) diff --git a/Text/Subtitles/SRT/Datatypes.hs b/Text/Subtitles/SRT/Datatypes.hs new file mode 100644 --- /dev/null +++ b/Text/Subtitles/SRT/Datatypes.hs @@ -0,0 +1,70 @@ +-- | +-- Module : Text.Subtitles.SRT.Datatypes +-- Copyright : Ruben Astudillo 2012 +-- License : BSD3 +-- +-- Maintainer : ruben.astud@gmail.com +-- Portability : unknown +-- +-- ADT for .srt files. Also serves as a place to provide instance +-- declarations for the ADTs. + +module Text.Subtitles.SRT.Datatypes + ( + -- * Re-exported datatypes. + module Data.Attoparsec.Text, + module Data.Text, + + -- * Datatypes + Subtitles, + Line(..), + Range(..), + Time(..) + ) where + +import Data.List (intercalate) +import Data.Text (Text, unpack) +import Data.Attoparsec.Text (Parser) + +data Time = Time + { hour :: Int + , minutes :: Int + , seconds :: Int + , frame :: Int + } deriving (Eq, Ord) + +data Range = Range + { from :: Time + , to :: Time + } deriving (Eq, Ord) + +-- | The core of the parser. each one of the constructor representing one part +-- of the Line +data Line = Line + { index :: Int + , range :: Range + , dialog :: Text + } deriving (Eq, Ord) + +type Subtitles = [Line] + +instance Show Time where + show (Time h m s f) = concat [showT h, ":", showT m, ":", showT s, ",", showF f] + +instance Show Range where + show (Range f t) = concat [show f, " --> ", show t] + +instance Show Line where + show (Line i t s) = intercalate "\n" (show i : show t : [unpack s]) ++ "\n" + +{- showT stands for showTime. In the parsing process "00" is read as "0", and + - that tailing 0 is lost unless we add it manually when showing -} +showT :: Int -> String +showT a | a < 10 = '0' : show a + | otherwise = show a + +{- showF stands for showFrame -} +showF :: Int -> String +showF a | a < 10 = '0':'0': show a + | a < 100 = '0': show a + | otherwise = show a diff --git a/Text/Subtitles/SUB.hs b/Text/Subtitles/SUB.hs new file mode 100644 --- /dev/null +++ b/Text/Subtitles/SUB.hs @@ -0,0 +1,49 @@ +-- | +-- Module : Text.Subtitles.SUB +-- Copyright : Ruben Astudillo 2012 +-- License : BSD3 +-- +-- Maintainer : ruben.astud@gmail.com +-- Portability : unknown +-- +-- A basic parser for .sub files (subtitles) based on 'Attoparsec' and 'Text' + +module Text.Subtitles.SUB + ( + -- * Terminology of the module + -- $example + + -- *Re-exported Datatypes + module Text.Subtitles.SUB.Datatypes, + + -- * Main Parsers + parseSingleLine, + parseSUB + ) where + +import Control.Applicative +import Data.Attoparsec.Text + +import Text.Subtitles.SUB.Datatypes + +-- $example +-- +-- Refering to the parts of a single line of a subtitle file. +-- +-- >{50}{100}Drama here! +-- +-- the first to numbers correspond to the frame in which the text Drama here! +-- is displayed. + +frame :: Parser Frame +frame = char '{' *> decimal <* char '}' + +-- | Given the example return the corresponding Line representation. At the +-- moment this not handles modifiers as underlines or bold text +parseSingleLine :: Parser Line +parseSingleLine = Line <$> frame <*> frame <*> takeTill isEndOfLine + +-- | Main parser of .sub files, given a .sub file it return a list of the dialog +-- lines +parseSUB :: Parser Subtitles +parseSUB = sepBy1 parseSingleLine endOfLine diff --git a/Text/Subtitles/SUB/Datatypes.hs b/Text/Subtitles/SUB/Datatypes.hs new file mode 100644 --- /dev/null +++ b/Text/Subtitles/SUB/Datatypes.hs @@ -0,0 +1,36 @@ +-- | +-- Module : Text.Subtitles.SUB.Datatypes +-- Copyright : Ruben Astudillo 2012 +-- License : BSD3 +-- +-- Maintainer : ruben.astud@gmail.com +-- Portability : unknown +-- +-- ADT for .sub files. Also serves as a place to provide instance +-- declarations for the ADTs. + +module Text.Subtitles.SUB.Datatypes + ( + -- * Re-exported datatypes + module Data.Attoparsec.Text, + module Data.Text, + + -- * Datatypes + Frame, + Subtitles, + Line(..) + ) where + +import Data.Text (Text) +import Data.Attoparsec.Text (Parser) + +type Subtitles = [Line] + +{- Should this be a newtype? or just a type? -} +type Frame = Int + +data Line = Line { startFrame :: Frame, + finalFrame :: Frame, + dialog :: Text } + deriving (Show) + diff --git a/subtitleParser.cabal b/subtitleParser.cabal --- a/subtitleParser.cabal +++ b/subtitleParser.cabal @@ -1,5 +1,5 @@ name: subtitleParser -version: 0.1.1 +version: 0.2 license: BSD3 license-file: LICENSE category: Text, Parsing @@ -29,7 +29,9 @@ attoparsec exposed-modules: Text.Subtitles.SRT - Text.Subtitles.Datatypes + Text.Subtitles.SRT.Datatypes + Text.Subtitles.SUB + Text.Subtitles.SUB.Datatypes ghc-options: -Wall