8f408e4 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

Added --ascii option.

8f408e4John MacFarlane16 years ago

README+5

compare with master

unifiedsplit

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

292`--columns`=*NUMBER*

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

294

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+

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

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

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

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

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

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

src/Text/Pandoc/Shared.hs+2

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

490 , writerHtml5 :: Bool -- ^ Produce HTML5

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

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

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

493 } deriving Show

494 } deriving Show

494

495

495-- | Default writer options.

496-- | Default writer options.

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

522 , writerHtml5 = False

523 , writerHtml5 = False

523 , writerChapters = False

524 , writerChapters = False

524 , writerListings = False

525 , writerListings = False

526+ , writerAscii = False

525 }

527 }

526

528

527--

529--

src/Text/Pandoc/Writers/HTML.hs+20−17

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

43import Data.Maybe ( catMaybes )

44import Control.Monad.State

45import Text.XHtml.Transitional hiding ( stringToHtml, unordList, ordList )

46+import qualified Text.XHtml.Transitional as XHtml

46import Text.TeXMath

47import Text.TeXMath

47import Text.XML.Light.Output

48import Text.XML.Light.Output

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

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

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

62-- Use unicode characters wherever possible.

63-- Use unicode characters wherever possible.

63−stringToHtml :: String -> Html

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

65+stringToHtml opts = if writerAscii opts

66+ then XHtml.stringToHtml

64−stringToHtml = primHtml . escapeStringForXML

67+ else primHtml . escapeStringForXML

66-- | Hard linebreak.

69-- | Hard linebreak.

67nl :: WriterOptions -> Html

70nl :: WriterOptions -> Html

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

219elementToListItem opts (Sec _ num id' headerText subsecs) = do

222elementToListItem opts (Sec _ num id' headerText subsecs) = do

220 let sectnum = if writerNumberSections opts

223 let sectnum = if writerNumberSections opts

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

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

222− stringToHtml " "

225+ stringToHtml opts" "

223 else noHtml

226 else noHtml

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

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

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

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

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

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

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

302 noscript (primHtml $ obfuscateString altText)

305 noscript (primHtml $ obfuscateString altText)

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

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

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

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

305

308

306-- | Obfuscate character as entity.

309-- | Obfuscate character as entity.

307obfuscateChar :: Char -> String

310obfuscateChar :: Char -> String

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

353 else unlines . lines

356 else unlines . lines

354 in return $ pre ! attrs $ thecode <<

357 in return $ pre ! attrs $ thecode <<

355 (replicate (length leadingBreaks) br +++

358 (replicate (length leadingBreaks) br +++

356− [stringToHtml $ addBird rawCode'])

359+ [stringToHtml opts $ addBird rawCode'])

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

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

358 return h

361 return h

359blockToHtml opts (BlockQuote blocks) =

362blockToHtml opts (BlockQuote blocks) =

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

379 secnum <- liftM stSecNum get

382 secnum <- liftM stSecNum get

380 let contents' = if writerNumberSections opts

383 let contents' = if writerNumberSections opts

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

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

382− stringToHtml " " +++ contents

385+ stringToHtml opts " " +++ contents

383 else contents

386 else contents

384 let contents'' = if writerTableOfContents opts

387 let contents'' = if writerTableOfContents opts

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

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

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

512inlineToHtml :: WriterOptions -> Inline -> State WriterState Html

515inlineToHtml :: WriterOptions -> Inline -> State WriterState Html

513inlineToHtml opts inline =

516inlineToHtml opts inline =

514 case inline of

517 case inline of

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

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

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

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

517 (LineBreak) -> return br

520 (LineBreak) -> return br

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

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

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

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

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

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

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

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

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

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

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

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

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

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

525 Left _ -> return

528 Left _ -> return

526 $ thecode ! (attrsToHtml opts attr)

529 $ thecode ! (attrsToHtml opts attr)

527− $ stringToHtml str

530+ $ stringToHtml opts str

528 Right h -> return h

531 Right h -> return h

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

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

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

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

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

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

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

535 (Quoted quoteType lst) ->

538 (Quoted quoteType lst) ->

536 let (leftQuote, rightQuote) = case quoteType of

539 let (leftQuote, rightQuote) = case quoteType of

537− SingleQuote -> (stringToHtml "‘",

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

538− stringToHtml "’")

541+ stringToHtml opts "’")

539− DoubleQuote -> (stringToHtml "“",

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

540− stringToHtml "”")

543+ stringToHtml opts "”")

541 in do contents <- inlineListToHtml opts lst

544 in do contents <- inlineListToHtml opts lst

542 return $ leftQuote +++ contents +++ rightQuote

545 return $ leftQuote +++ contents +++ rightQuote

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

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

src/pandoc.hs+11−2

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

123 , optBibliography :: [String]

124 , optCslFile :: FilePath

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

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

126 }

127 }

127

128

128-- | Defaults for command-line options.

129-- | Defaults for command-line options.

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

166 , optBibliography = []

167 , optBibliography = []

167 , optCslFile = ""

168 , optCslFile = ""

168 , optListings = False

169 , optListings = False

170+ , optAscii = False

169 }

171 }

170

172

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

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

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

347 "NUMBER")

349 "NUMBER")

348 "" -- "Length of line in characters"

350 "" -- "Length of line in characters"

349

351

352+ , Option "" ["ascii"]

353+ (NoArg

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

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

356+

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

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

351 (ReqArg

358 (ReqArg

352 (\arg opt -> do

359 (\arg opt -> do

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

681 , optBibliography = reffiles

688 , optBibliography = reffiles

682 , optCslFile = cslfile

689 , optCslFile = cslfile

683 , optCiteMethod = citeMethod

690 , optCiteMethod = citeMethod

684− , optListings = listings

691+ , optListings = listings

692+ , optAscii = ascii

685 } = opts

693 } = opts

686

694

687 when dumpArgs $

695 when dumpArgs $

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

803 writerHtml5 = html5 &&

811 writerHtml5 = html5 &&

804 "html" `isPrefixOf` writerName',

812 "html" `isPrefixOf` writerName',

805 writerChapters = chapters,

813 writerChapters = chapters,

806− writerListings = listings }

814+ writerListings = listings,

815+ writerAscii = ascii }

807

816

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

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

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

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