8f408e4 · typeable/pandoc

git.iko.soy/typeable/pandoc

Added --ascii option.

Currently supported only in HTML writer.

commit8f408e4b7b24bfcf0f6a48522a2d77b0b878d34d authorJohn MacFarlane <jgm@berkeley.edu> dateSun 6 Feb 2011, 09:27 (16 years ago) parent9b38b69

4 files changed+38−19

compare with master

unifiedsplit

README+5 src/Text/Pandoc/Shared.hs+2 src/Text/Pandoc/Writers/HTML.hs+20−17 src/pandoc.hs+11−2

README+5

@@ -292,6 +292,11 @@ Options

292292`--columns`=*NUMBER*

293293: Specify length of lines in characters (for text wrapping).

294294

295+`--ascii`

296+: Use only ascii characters in output. Currently supported only

297+ for HTML output (which uses numerical entities instead of

298+ UTF-8 when this option is selected).

299+

295300`--email-obfuscation=`*none|javascript|references*

296301: Specify a method for obfuscating `mailto:` links in HTML documents.

297302 *none* leaves `mailto:` links as they are. *javascript* obfuscates

src/Text/Pandoc/Shared.hs+2

@@ -490,6 +490,7 @@ data WriterOptions = WriterOptions

490490 , writerHtml5 :: Bool -- ^ Produce HTML5

491491 , writerChapters :: Bool -- ^ Use "chapter" for top-level sects

492492 , writerListings :: Bool -- ^ Use listings package for code

493+ , writerAscii :: Bool -- ^ Avoid non-ascii characters

493494 } deriving Show

494495

495496-- | Default writer options.

@@ -522,6 +523,7 @@ defaultWriterOptions =

522523 , writerHtml5 = False

523524 , writerChapters = False

524525 , writerListings = False

526+ , writerAscii = False

525527 }

526528

527529--

src/Text/Pandoc/Writers/HTML.hs+20−17

@@ -43,6 +43,7 @@ import Data.List ( isPrefixOf, intersperse )

4343import Data.Maybe ( catMaybes )

4444import Control.Monad.State

4545import Text.XHtml.Transitional hiding ( stringToHtml, unordList, ordList )

46+import qualified Text.XHtml.Transitional as XHtml

4647import Text.TeXMath

4748import Text.XML.Light.Output

4849

@@ -60,8 +61,10 @@ defaultWriterState = WriterState {stNotes= [], stMath = False, stHighlighting =

6061

6162-- | Modified version of Text.XHtml's stringToHtml.

6263-- Use unicode characters wherever possible.

63−stringToHtml :: String -> Html

64−stringToHtml = primHtml . escapeStringForXML

64+stringToHtml :: WriterOptions -> String -> Html

65+stringToHtml opts = if writerAscii opts

66+ then XHtml.stringToHtml

67+ else primHtml . escapeStringForXML

6568

6669-- | Hard linebreak.

6770nl :: WriterOptions -> Html

@@ -219,7 +222,7 @@ elementToListItem _ (Blk _) = return Nothing

219222elementToListItem opts (Sec _ num id' headerText subsecs) = do

220223 let sectnum = if writerNumberSections opts

221224 then (thespan ! [theclass "toc-section-number"] << showSecNum num) +++

222− stringToHtml " "

225+ stringToHtml opts" "

223226 else noHtml

224227 txt <- liftM (sectnum +++) $ inlineListToHtml opts headerText

225228 subHeads <- mapM (elementToListItem opts) subsecs >>= return . catMaybes

@@ -301,7 +304,7 @@ obfuscateLink opts txt s =

301304 linkText ++ "+'<\\/'+'a'+'>');\n// -->\n")) +++

302305 noscript (primHtml $ obfuscateString altText)

303306 _ -> error $ "Unknown obfuscation method: " ++ show meth

304− _ -> anchor ! [href s] $ stringToHtml txt -- malformed email

307+ _ -> anchor ! [href s] $ stringToHtml opts txt -- malformed email

305308

306309-- | Obfuscate character as entity.

307310obfuscateChar :: Char -> String

@@ -353,7 +356,7 @@ blockToHtml opts (CodeBlock (id',classes,keyvals) rawCode) = do

353356 else unlines . lines

354357 in return $ pre ! attrs $ thecode <<

355358 (replicate (length leadingBreaks) br +++

356− [stringToHtml $ addBird rawCode'])

359+ [stringToHtml opts $ addBird rawCode'])

357360 Right h -> modify (\st -> st{ stHighlighting = True }) >>

358361 return h

359362blockToHtml opts (BlockQuote blocks) =

@@ -379,7 +382,7 @@ blockToHtml opts (Header level lst) = do

379382 secnum <- liftM stSecNum get

380383 let contents' = if writerNumberSections opts

381384 then (thespan ! [theclass "header-section-number"] << showSecNum secnum) +++

382− stringToHtml " " +++ contents

385+ stringToHtml opts " " +++ contents

383386 else contents

384387 let contents'' = if writerTableOfContents opts

385388 then anchor ! [href $ "#" ++ writerIdentifierPrefix opts ++ "TOC"] $ contents'

@@ -512,19 +515,19 @@ inlineListToHtml opts lst =

512515inlineToHtml :: WriterOptions -> Inline -> State WriterState Html

513516inlineToHtml opts inline =

514517 case inline of

515− (Str str) -> return $ stringToHtml str

516− (Space) -> return $ stringToHtml " "

518+ (Str str) -> return $ stringToHtml opts str

519+ (Space) -> return $ stringToHtml opts " "

517520 (LineBreak) -> return br

518− (EmDash) -> return $ stringToHtml "—"

519− (EnDash) -> return $ stringToHtml "–"

520− (Ellipses) -> return $ stringToHtml "…"

521− (Apostrophe) -> return $ stringToHtml "’"

521+ (EmDash) -> return $ stringToHtml opts "—"

522+ (EnDash) -> return $ stringToHtml opts "–"

523+ (Ellipses) -> return $ stringToHtml opts "…"

524+ (Apostrophe) -> return $ stringToHtml opts "’"

522525 (Emph lst) -> inlineListToHtml opts lst >>= return . emphasize

523526 (Strong lst) -> inlineListToHtml opts lst >>= return . strong

524527 (Code attr str) -> case highlightHtml True attr str of

525528 Left _ -> return

526529 $ thecode ! (attrsToHtml opts attr)

527− $ stringToHtml str

530+ $ stringToHtml opts str

528531 Right h -> return h

529532 (Strikeout lst) -> inlineListToHtml opts lst >>=

530533 return . (thespan ! [thestyle "text-decoration: line-through;"])

@@ -534,10 +537,10 @@ inlineToHtml opts inline =

534537 (Subscript lst) -> inlineListToHtml opts lst >>= return . sub

535538 (Quoted quoteType lst) ->

536539 let (leftQuote, rightQuote) = case quoteType of

537− SingleQuote -> (stringToHtml "‘",

538− stringToHtml "’")

539− DoubleQuote -> (stringToHtml "“",

540− stringToHtml "”")

540+ SingleQuote -> (stringToHtml opts "‘",

541+ stringToHtml opts "’")

542+ DoubleQuote -> (stringToHtml opts "“",

543+ stringToHtml opts "”")

541544 in do contents <- inlineListToHtml opts lst

542545 return $ leftQuote +++ contents +++ rightQuote

543546 (Math t str) -> modify (\st -> st {stMath = True}) >>

src/pandoc.hs+11−2

@@ -123,6 +123,7 @@ data Opt = Opt

123123 , optBibliography :: [String]

124124 , optCslFile :: FilePath

125125 , optListings :: Bool -- ^ Use listings package for code blocks

126+ , optAscii :: Bool -- ^ Avoid using nonascii characters

126127 }

127128

128129-- | Defaults for command-line options.

@@ -166,6 +167,7 @@ defaultOpts = Opt

166167 , optBibliography = []

167168 , optCslFile = ""

168169 , optListings = False

170+ , optAscii = False

169171 }

170172

171173-- | A list of functions, each transforming the options data structure

@@ -347,6 +349,11 @@ options =

347349 "NUMBER")

348350 "" -- "Length of line in characters"

349351

352+ , Option "" ["ascii"]

353+ (NoArg

354+ (\opt -> return opt { optAscii = True }))

355+ "" -- "Avoid using non-ascii characters in output"

356+

350357 , Option "" ["email-obfuscation"]

351358 (ReqArg

352359 (\arg opt -> do

@@ -681,7 +688,8 @@ main = do

681688 , optBibliography = reffiles

682689 , optCslFile = cslfile

683690 , optCiteMethod = citeMethod

684− , optListings = listings

691+ , optListings = listings

692+ , optAscii = ascii

685693 } = opts

686694

687695 when dumpArgs $

@@ -803,7 +811,8 @@ main = do

803811 writerHtml5 = html5 &&

804812 "html" `isPrefixOf` writerName',

805813 writerChapters = chapters,

806− writerListings = listings }

814+ writerListings = listings,

815+ writerAscii = ascii }

807816

808817 when (isNonTextOutput writerName' && outputFile == "-") $

809818 do UTF8.hPutStrLn stderr ("Error: Cannot write " ++ writerName ++ " output to stdout.\n" ++