src/Main.hs · typeable/pandoc

git.iko.soy/typeable/pandoc

on cfaf0c1 help sign in

typeable/pandoc/src/Main.hs

335 lines 14 kB history raw

1{-

4This program is free software; you can redistribute it and/or modify

5it under the terms of the GNU General Public License as published by

6the Free Software Foundation; either version 2 of the License, or

7(at your option) any later version.

9This program is distributed in the hope that it will be useful,

10but WITHOUT ANY WARRANTY; without even the implied warranty of

11MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the

12GNU General Public License for more details.

14You should have received a copy of the GNU General Public License

15along with this program; if not, write to the Free Software

16Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA

17-}

19{- |

20 Module : Main

22 License : GNU GPL, version 2 or above

24 Maintainer : John MacFarlane <jgm at berkeley dot edu>

25 Stability : alpha

26 Portability : portable

28Parses command-line options and calls the appropriate readers and

29writers.

30-}

31module Main where

32import Text.Pandoc.UTF8 ( decodeUTF8, encodeUTF8 )

33import Text.Pandoc.Readers.Markdown ( readMarkdown )

34import Text.Pandoc.Readers.HTML ( readHtml )

35import Text.Pandoc.Writers.S5 ( s5CSS, s5Javascript, writeS5 )

36import Text.Pandoc.Writers.RST ( writeRST )

37import Text.Pandoc.Readers.RST ( readRST )

38import Text.Pandoc.ASCIIMathML ( asciiMathMLScript )

39import Text.Pandoc.Writers.HTML ( writeHtml )

40import Text.Pandoc.Writers.LaTeX ( writeLaTeX )

41import Text.Pandoc.Readers.LaTeX ( readLaTeX )

42import Text.Pandoc.Writers.RTF ( writeRTF )

43import Text.Pandoc.Writers.Markdown ( writeMarkdown )

44import Text.Pandoc.Writers.DefaultHeaders ( defaultHtmlHeader,

45 defaultRTFHeader, defaultS5Header, defaultLaTeXHeader )

46import Text.Pandoc.Definition

47import Text.Pandoc.Shared

48import System ( exitWith, getArgs, getProgName )

49import System.Exit

50import System.Console.GetOpt

51import System.IO

52import Data.Maybe ( fromMaybe )

53import Data.List ( isPrefixOf )

54import Char ( toLower )

55import Control.Monad ( (>>=) )

57version :: String

58version = "0.3"

60-- | Association list of formats and readers.

61readers :: [(String, ParserState -> String -> Pandoc)]

62readers = [("native" , readPandoc)

63 ,("markdown" , readMarkdown)

64 ,("rst" , readRST)

65 ,("html" , readHtml)

66 ,("latex" , readLaTeX)

67 ]

69-- | Reader for native Pandoc format.

70readPandoc :: ParserState -> String -> Pandoc

71readPandoc state input = read input

73-- | Association list of formats and pairs of writers and default headers.

74writers :: [ ( String, ( WriterOptions -> Pandoc -> String, String ) ) ]

75writers = [("native" , (writeDoc, ""))

76 ,("html" , (writeHtml, defaultHtmlHeader))

77 ,("s5" , (writeS5, defaultS5Header))

78 ,("latex" , (writeLaTeX, defaultLaTeXHeader))

79 ,("markdown" , (writeMarkdown, ""))

80 ,("rst" , (writeRST, ""))

81 ,("rtf" , (writeRTF, defaultRTFHeader))

82 ]

84-- | Writer for Pandoc native format.

85writeDoc :: WriterOptions -> Pandoc -> String

86writeDoc options = prettyPandoc

88-- | Data structure for command line options.

89data Opt = Opt

90 { optPreserveTabs :: Bool -- ^ If @False@, convert tabs to spaces

91 , optTabStop :: Int -- ^ Number of spaces per tab

92 , optStandalone :: Bool -- ^ If @True@, include header, footer

93 , optReader :: ParserState -> String -> Pandoc -- ^ Read format

94 , optWriter :: WriterOptions -> Pandoc -> String -- ^ Write fmt

95 , optParseRaw :: Bool -- ^ If @True@, parse unconvertable

96 -- HTML and TeX

97 , optCSS :: String -- ^ CSS file to link to

98 , optIncludeInHeader :: String -- ^ File to include in header

99 , optIncludeBeforeBody :: String -- ^ File to include at top of body

100 , optIncludeAfterBody :: String -- ^ File to include at end of body

101 , optCustomHeader :: String -- ^ Custom header to use, or "DEFAULT"

102 , optDefaultHeader :: String -- ^ Default header

103 , optTitlePrefix :: String -- ^ Optional prefix for HTML title

104 , optNumberSections :: Bool -- ^ If @True@, number sections in LaTeX

105 , optIncremental :: Bool -- ^ If @True@, incremental lists in S5

106 , optSmart :: Bool -- ^ If @True@, use smart typography

107 , optASCIIMathML :: Bool -- ^ If @True@, use ASCIIMathML in HTML

108 }

109

110-- | Defaults for command-line options.

111startOpt :: Opt

112startOpt = Opt

113 { optPreserveTabs = False

114 , optTabStop = 4

115 , optStandalone = False

116 , optReader = readMarkdown

117 , optWriter = writeHtml

118 , optParseRaw = False

119 , optCSS = ""

120 , optIncludeInHeader = ""

121 , optIncludeBeforeBody = ""

122 , optIncludeAfterBody = ""

123 , optCustomHeader = "DEFAULT"

124 , optDefaultHeader = defaultHtmlHeader

125 , optTitlePrefix = ""

126 , optNumberSections = False

127 , optIncremental = False

128 , optSmart = False

129 , optASCIIMathML = False

130 }

131

132-- | A list of functions, each transforming the options data structure in response

133-- to a command-line option.

134options :: [OptDescr (Opt -> IO Opt)]

135options =

136 [ Option "v" ["version"]

137 (NoArg

138 (\_ -> do

139 hPutStrLn stderr ("Version " ++ version)

140 exitWith ExitSuccess))

141 "Print version"

142

143 , Option "h" ["help"]

144 (NoArg

145 (\_ -> do

146 prg <- getProgName

147 hPutStrLn stderr (usageInfo (prg ++ " [OPTIONS] [FILES] - convert FILES from one markup format to another\nIf no OPTIONS specified, converts from markdown to html.\nIf no FILES specified, input is read from STDIN.\nOptions:") options)

148 exitWith ExitSuccess))

149 "Show help"

150

151 , Option "fr" ["from","read"]

152 (ReqArg

153 (\arg opt -> case (lookup (map toLower arg) readers) of

154 Just reader -> return opt { optReader = reader }

155 Nothing -> error ("Unknown reader: " ++ arg) )

156 "FORMAT")

157 ("Source format (" ++

158 (concatMap (\(name, fn) -> " " ++ name) readers) ++ " )")

159

160 , Option "tw" ["to","write"]

161 (ReqArg

162 (\arg opt -> case (lookup (map toLower arg) writers) of

163 Just (writer, defaultHeader) ->

164 return opt { optWriter = writer,

165 optDefaultHeader = defaultHeader }

166 Nothing -> error ("Unknown writer: " ++ arg) )

167 "FORMAT")

168 ("Output format (" ++ (concatMap (\(name, fn) -> " " ++ name) writers) ++ " )")

169

170 , Option "s" ["standalone"]

171 (NoArg

172 (\opt -> return opt { optStandalone = True }))

173 "Include needed header and footer on output"

174

175 , Option "p" ["preserve-tabs"]

176 (NoArg

177 (\opt -> return opt { optPreserveTabs = True }))

178 "Preserve tabs instead of converting to spaces"

179

180 , Option "" ["tab-stop"]

181 (ReqArg

182 (\arg opt -> return opt { optTabStop = (read arg) } )

183 "TABSTOP")

184 "Tab stop (default 4)"

185

186 , Option "R" ["parse-raw"]

187 (NoArg

188 (\opt -> return opt { optParseRaw = True }))

189 "Parse untranslatable HTML codes and LaTeX environments as raw"

190

191 , Option "S" ["smart"]

192 (NoArg

193 (\opt -> return opt { optSmart = True }))

194 "Use smart quotes, dashes, and ellipses in HTML output"

195

196 , Option "m" ["asciimathml"]

197 (NoArg

198 (\opt -> return opt { optASCIIMathML = True,

199 optStandalone = True }))

200 "Use ASCIIMathML script in html output"

201

202 , Option "i" ["incremental"]

203 (NoArg

204 (\opt -> return opt { optIncremental = True }))

205 "Make list items display incrementally in S5"

206

207 , Option "N" ["number-sections"]

208 (NoArg

209 (\opt -> return opt { optNumberSections = True }))

210 "Number sections in LaTeX"

211

212 , Option "c" ["css"]

213 (ReqArg

214 (\arg opt -> return opt { optCSS = arg,

215 optStandalone = True })

216 "CSS")

217 "Link to CSS style sheet"

218

219 , Option "H" ["include-in-header"]

220 (ReqArg

221 (\arg opt -> do

222 text <- readFile arg

223 return opt { optIncludeInHeader = text,

224 optStandalone = True })

225 "FILENAME")

226 "File to include at end of header (implies -s)"

227

228 , Option "B" ["include-before-body"]

229 (ReqArg

230 (\arg opt -> do

231 text <- readFile arg

232 return opt { optIncludeBeforeBody = text })

233 "FILENAME")

234 "File to include before document body"

235

236 , Option "A" ["include-after-body"]

237 (ReqArg

238 (\arg opt -> do

239 text <- readFile arg

240 return opt { optIncludeAfterBody = text })

241 "FILENAME")

242 "File to include after document body"

243

244 , Option "" ["custom-header"]

245 (ReqArg

246 (\arg opt -> do

247 text <- readFile arg

248 return opt { optCustomHeader = text,

249 optStandalone = True })

250 "FILENAME")

251 "File to use for custom header (implies -s)"

252

253 , Option "T" ["title-prefix"]

254 (ReqArg

255 (\arg opt -> return opt { optTitlePrefix = arg,

256 optStandalone = True })

257 "STRING")

258 "String to prefix to HTML window title"

259

260 , Option "D" ["print-default-header"]

261 (ReqArg

262 (\arg opt -> do

263 let header = case (lookup arg writers) of

264 Just (writer, head) -> head

265 Nothing -> error ("Unknown reader: " ++ arg)

266 hPutStrLn stdout header

267 exitWith ExitSuccess)

268 "FORMAT")

269 "Print default header for FORMAT"

270 ]

271main = do

272

273 args <- getArgs

274 let (actions, sources, errors) = getOpt RequireOrder options args

275

276 -- thread option data structure through all supplied option actions

277 opts <- foldl (>>=) (return startOpt) actions

278

279 let Opt { optPreserveTabs = preserveTabs

280 , optTabStop = tabStop

281 , optStandalone = standalone

282 , optReader = reader

283 , optWriter = writer

284 , optParseRaw = parseRaw

285 , optCSS = css

286 , optIncludeInHeader = includeHeader

287 , optIncludeBeforeBody = includeBefore

288 , optIncludeAfterBody = includeAfter

289 , optCustomHeader = customHeader

290 , optDefaultHeader = defaultHeader

291 , optTitlePrefix = titlePrefix

292 , optNumberSections = numberSections

293 , optIncremental = incremental

294 , optSmart = smart

295 , optASCIIMathML = asciiMathML

296 } = opts

297

298 let writingS5 = (defaultHeader == defaultS5Header)

299 let tabFilter = if preserveTabs then id else (tabsToSpaces tabStop)

300 let addBlank str = str ++ "\n\n"

301 let removeCRs str = filter (/= '\r') str -- remove DOS-style line endings

302 let filter = tabFilter . addBlank . removeCRs

303 let startParserState = defaultParserState { stateParseRaw = parseRaw,

304 stateTabStop = tabStop,

305 stateStandalone = standalone }

306 let csslink = if (css == "")

307 then ""

308 else "<link rel=\"stylesheet\" href=\"" ++ css ++

309 "\" type=\"text/css\" media=\"all\" />\n"

310 let asciiMathMLLink = if asciiMathML then asciiMathMLScript else ""

311 let header = (if (customHeader == "DEFAULT")

312 then defaultHeader

313 else customHeader) ++

314 csslink ++ asciiMathMLLink ++ includeHeader

315 let writerOptions = WriterOptions { writerStandalone = standalone,

316 writerHeader = header,

317 writerTitlePrefix = titlePrefix,

318 writerSmart = smart,

319 writerTabStop = tabStop,

320 writerS5 = writingS5,

321 writerIncremental = incremental,

322 writerNumberSections = numberSections,

323 writerIncludeBefore = includeBefore,

324 writerIncludeAfter = includeAfter }

325

326 (readSources sources) >>= (putStr . encodeUTF8 . (writer writerOptions) .

327 (reader startParserState) . filter .

328 decodeUTF8 . (joinWithSep "\n"))

329

330 where

331 readSources [] = mapM readSource ["-"]

332 readSources sources = mapM readSource sources

333 readSource "-" = getContents

334 readSource source = readFile source

335