2806aee · typeable/pandoc

git.iko.soy/typeable/pandoc

Added --email-obfuscation option.

+ Added writer option for email obfuscation. + Implemented email obfuscation options in HTML writer. + Added option to option parser. + Documented in README and pandoc man page. + Resolves Issue #97. git-svn-id: https://pandoc.googlecode.com/svn/trunk@1523 788f1e2b-df1e-0410-8736-df70ead52e1b

commit2806aee9b210c338037176bff467b76a394b120f authorfiddlosopher <fiddlosopher@788f1e2b-df1e-0410-8736-df70ead52e1b> dateSat 24 Jan 2009, 19:58 (18 years ago) parent46a3b22

5 files changed+110−60

compare with master

unifiedsplit

Main.hs+42−23 README+8 Text/Pandoc/Shared.hs+41−32 Text/Pandoc/Writers/HTML.hs+11−5 man/man1/pandoc.1.md+8

Main.hs+42−23

@@ -32,7 +32,7 @@ writers.

3232module Main where

3333import Text.Pandoc

3434import Text.Pandoc.ODT

35−import Text.Pandoc.Shared ( HTMLMathMethod (..), splitBy )

35+import Text.Pandoc.Shared ( HTMLMathMethod (..), splitBy, ObfuscationMethod (..) )

3636import Text.Pandoc.Highlighting ( languages )

3737import System.Environment ( getArgs, getProgName, getEnvironment )

3838import System.Exit ( exitWith, ExitCode (..) )

@@ -156,6 +156,7 @@ data Opt = Opt

156156 , optWrapText :: Bool -- ^ Wrap text

157157 , optSanitizeHTML :: Bool -- ^ Sanitize HTML

158158 , optPlugins :: [Pandoc -> IO Pandoc] -- ^ Plugins to apply

159+ , optEmailObfuscation :: ObfuscationMethod

159160#ifdef _CITEPROC

160161 , optBiblioFile :: String

161162 , optBiblioFormat :: String

@@ -191,6 +192,7 @@ defaultOpts = Opt

191192 , optWrapText = True

192193 , optSanitizeHTML = False

193194 , optPlugins = []

195+ , optEmailObfuscation = JavascriptObfuscation

194196#ifdef _CITEPROC

195197 , optBiblioFile = []

196198 , optBiblioFormat = []

@@ -301,6 +303,19 @@ options =

301303 (\opt -> return opt { optSanitizeHTML = True }))

302304 "" -- "Sanitize HTML"

303305

306+ , Option "" ["email-obfuscation"]

307+ (ReqArg

308+ (\arg opt -> do

309+ method <- case arg of

310+ "references" -> return ReferenceObfuscation

311+ "javascript" -> return JavascriptObfuscation

312+ "none" -> return NoObfuscation

313+ _ -> hPutStrLn stderr ("Error: Unknown obfuscation method: " ++ arg) >>

314+ exitWith (ExitFailure 6)

315+ return opt { optEmailObfuscation = method })

316+ "none|javascript|references")

317+ "" -- "Method for obfuscating email in HTML"

318+

304319 , Option "" ["toc", "table-of-contents"]

305320 (NoArg

306321 (\opt -> return opt { optTableOfContents = True }))

@@ -533,10 +548,11 @@ main = do

533548 , optWrapText = wrap

534549 , optSanitizeHTML = sanitize

535550 , optPlugins = plugins

551+ , optEmailObfuscation = obfuscationMethod

536552#ifdef _CITEPROC

537− , optBiblioFile = biblioFile

538− , optBiblioFormat = biblioFormat

539− , optCslFile = cslFile

553+ , optBiblioFile = biblioFile

554+ , optBiblioFormat = biblioFormat

555+ , optCslFile = cslFile

540556#endif

541557 } = opts

542558

@@ -614,25 +630,28 @@ main = do

614630 let header = (if (customHeader == "DEFAULT")

615631 then defaultHeader

616632 else customHeader) ++ csslink ++ includeHeader

617− let writerOptions = WriterOptions { writerStandalone = standalone',

618− writerHeader = header,

619− writerTitlePrefix = titlePrefix,

620− writerTabStop = tabStop,

621− writerTableOfContents = toc &&

622− (not strict) &&

623− writerName' /= "s5",

624− writerHTMLMathMethod = mathMethod,

625− writerS5 = (writerName' == "s5"),

626− writerIgnoreNotes = False,

627− writerIncremental = incremental,

628− writerNumberSections = numberSections,

629− writerIncludeBefore = includeBefore,

630− writerIncludeAfter = includeAfter,

631− writerStrictMarkdown = strict,

632− writerReferenceLinks = referenceLinks,

633− writerWrapText = wrap,

634− writerLiterateHaskell = "+lhs" `isSuffixOf` writerName' ||

635− lhsExtension [outputFile] }

633+ let writerOptions = WriterOptions { writerStandalone = standalone',

634+ writerHeader = header,

635+ writerTitlePrefix = titlePrefix,

636+ writerTabStop = tabStop,

637+ writerTableOfContents = toc &&

638+ (not strict) &&

639+ writerName' /= "s5",

640+ writerHTMLMathMethod = mathMethod,

641+ writerS5 = (writerName' == "s5"),

642+ writerIgnoreNotes = False,

643+ writerIncremental = incremental,

644+ writerNumberSections = numberSections,

645+ writerIncludeBefore = includeBefore,

646+ writerIncludeAfter = includeAfter,

647+ writerStrictMarkdown = strict,

648+ writerReferenceLinks = referenceLinks,

649+ writerWrapText = wrap,

650+ writerLiterateHaskell = "+lhs" `isSuffixOf` writerName' ||

651+ lhsExtension [outputFile],

652+ writerEmailObfuscation = if strict

653+ then ReferenceObfuscation

654+ else obfuscationMethod }

636655

637656 if isNonTextOutput writerName' && outputFile == "-"

638657 then do hPutStrLn stderr ("Error: Cannot write " ++ writerName ++ " output to stdout.\n" ++

README+8

@@ -383,6 +383,14 @@ For further documentation, see the `pandoc(1)` man page.

383383 are omitted. URIs in links and images are also checked against a

384384 whitelist of URI schemes.

385385

386+`--email-obfuscation`*=none|javascript|references*

387+: specifies a method for obfuscating `mailto:` links in HTML documents.

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

389+ them using javascript. *references* obfuscates them by printing their

390+ letters as decimal or hexadecimal character references. If `--strict`

391+ is specified, *references* is used regardless of the presence

392+ of this option.

393+

386394`--dump-args`

387395: is intended to make it easier to create wrapper scripts that use

388396 Pandoc. It causes Pandoc to dump information about the arguments

Text/Pandoc/Shared.hs+41−32

@@ -99,6 +99,7 @@ module Text.Pandoc.Shared (

9999 isHeaderBlock,

100100 -- * Writer options

101101 HTMLMathMethod (..),

102+ ObfuscationMethod (..),

102103 WriterOptions (..),

103104 defaultWriterOptions,

104105 -- * File handling

@@ -889,45 +890,53 @@ data HTMLMathMethod = PlainMath

889890 | MimeTeX String -- url of mimetex.cgi

890891 deriving (Show, Read, Eq)

891892

893+-- | Methods for obfuscating email addresses in HTML.

894+data ObfuscationMethod = NoObfuscation

895+ | ReferenceObfuscation

896+ | JavascriptObfuscation

897+ deriving (Show, Read, Eq)

898+

892899-- | Options for writers

893900data WriterOptions = WriterOptions

894− { writerStandalone :: Bool -- ^ Include header and footer

895− , writerHeader :: String -- ^ Header for the document

896− , writerTitlePrefix :: String -- ^ Prefix for HTML titles

897− , writerTabStop :: Int -- ^ Tabstop for conversion btw spaces and tabs

898− , writerTableOfContents :: Bool -- ^ Include table of contents

899− , writerS5 :: Bool -- ^ We're writing S5

900− , writerHTMLMathMethod :: HTMLMathMethod -- ^ How to print math in HTML

901− , writerIgnoreNotes :: Bool -- ^ Ignore footnotes (used in making toc)

902− , writerIncremental :: Bool -- ^ Incremental S5 lists

903− , writerNumberSections :: Bool -- ^ Number sections in LaTeX

904− , writerIncludeBefore :: String -- ^ String to include before the body

905− , writerIncludeAfter :: String -- ^ String to include after the body

906− , writerStrictMarkdown :: Bool -- ^ Use strict markdown syntax

907− , writerReferenceLinks :: Bool -- ^ Use reference links in writing markdown, rst

908− , writerWrapText :: Bool -- ^ Wrap text to line length

909− , writerLiterateHaskell :: Bool -- ^ Write as literate haskell

901+ { writerStandalone :: Bool -- ^ Include header and footer

902+ , writerHeader :: String -- ^ Header for the document

903+ , writerTitlePrefix :: String -- ^ Prefix for HTML titles

904+ , writerTabStop :: Int -- ^ Tabstop for conversion btw spaces and tabs

905+ , writerTableOfContents :: Bool -- ^ Include table of contents

906+ , writerS5 :: Bool -- ^ We're writing S5

907+ , writerHTMLMathMethod :: HTMLMathMethod -- ^ How to print math in HTML

908+ , writerIgnoreNotes :: Bool -- ^ Ignore footnotes (used in making toc)

909+ , writerIncremental :: Bool -- ^ Incremental S5 lists

910+ , writerNumberSections :: Bool -- ^ Number sections in LaTeX

911+ , writerIncludeBefore :: String -- ^ String to include before the body

912+ , writerIncludeAfter :: String -- ^ String to include after the body

913+ , writerStrictMarkdown :: Bool -- ^ Use strict markdown syntax

914+ , writerReferenceLinks :: Bool -- ^ Use reference links in writing markdown, rst

915+ , writerWrapText :: Bool -- ^ Wrap text to line length

916+ , writerLiterateHaskell :: Bool -- ^ Write as literate haskell

917+ , writerEmailObfuscation :: ObfuscationMethod -- ^ How to obfuscate emails

910918 } deriving Show

911919

912920-- | Default writer options.

913921defaultWriterOptions :: WriterOptions

914922defaultWriterOptions =

915− WriterOptions { writerStandalone = False

916− , writerHeader = ""

917− , writerTitlePrefix = ""

918− , writerTabStop = 4

919− , writerTableOfContents = False

920− , writerS5 = False

921− , writerHTMLMathMethod = PlainMath

922− , writerIgnoreNotes = False

923− , writerIncremental = False

924− , writerNumberSections = False

925− , writerIncludeBefore = ""

926− , writerIncludeAfter = ""

927− , writerStrictMarkdown = False

928− , writerReferenceLinks = False

929− , writerWrapText = True

930− , writerLiterateHaskell = False

923+ WriterOptions { writerStandalone = False

924+ , writerHeader = ""

925+ , writerTitlePrefix = ""

926+ , writerTabStop = 4

927+ , writerTableOfContents = False

928+ , writerS5 = False

929+ , writerHTMLMathMethod = PlainMath

930+ , writerIgnoreNotes = False

931+ , writerIncremental = False

932+ , writerNumberSections = False

933+ , writerIncludeBefore = ""

934+ , writerIncludeAfter = ""

935+ , writerStrictMarkdown = False

936+ , writerReferenceLinks = False

937+ , writerWrapText = True

938+ , writerLiterateHaskell = False

939+ , writerEmailObfuscation = JavascriptObfuscation

931940 }

932941

933942--

Text/Pandoc/Writers/HTML.hs+11−5

@@ -193,10 +193,13 @@ parseMailto ('m':'a':'i':'l':'t':'o':':':addr) =

193193 in Just (name', domain)

194194parseMailto _ = Nothing

195195

196−-- | Obfuscate a "mailto:" link using Javascript.

196+-- | Obfuscate a "mailto:" link.

197197obfuscateLink :: WriterOptions -> String -> String -> Html

198+obfuscateLink opts txt s | writerEmailObfuscation opts == NoObfuscation =

199+ anchor ! [href s] << txt

198200obfuscateLink opts txt s =

199− let s' = map toLower s

201+ let meth = writerEmailObfuscation opts

202+ s' = map toLower s

200203 in case parseMailto s' of

201204 (Just (name', domain)) ->

202205 let domain' = substitute "." " dot " domain

@@ -206,17 +209,20 @@ obfuscateLink opts txt s =

206209 then ("'<code>'+e+'</code>'", name' ++ " at " ++ domain')

207210 else ("'" ++ txt ++ "'", txt ++ " (" ++ name' ++ " at " ++

208211 domain' ++ ")")

209− in if writerStrictMarkdown opts

210− then -- need to use primHtml or &'s are escaped to & in URL

212+ in case meth of

213+ ReferenceObfuscation ->

214+ -- need to use primHtml or &'s are escaped to & in URL

211215 primHtml $ "<a href=\"" ++ (obfuscateString s')

212216 ++ "\">" ++ (obfuscateString txt) ++ "</a>"

213− else (script ! [thetype "text/javascript"] $

217+ JavascriptObfuscation ->

218+ (script ! [thetype "text/javascript"] $

214219 primHtml ("\n<!--\nh='" ++

215220 obfuscateString domain ++ "';a='" ++ at' ++ "';n='" ++

216221 obfuscateString name' ++ "';e=n+a+h;\n" ++

217222 "document.write('<a h'+'ref'+'=\"ma'+'ilto'+':'+e+'\">'+" ++

218223 linkText ++ "+'<\\/'+'a'+'>');\n// -->\n")) +++

219224 noscript (primHtml $ obfuscateString altText)

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

220226 _ -> anchor ! [href s] $ primHtml txt -- malformed email

221227

222228-- | Obfuscate character as entity.

man/man1/pandoc.1.md+8

@@ -146,6 +146,14 @@ to Pandoc. Or use `html2markdown`(1), a wrapper around `pandoc`.

146146 are omitted. URIs in links and images are also checked against a

147147 whitelist of URI schemes.

148148

149+\--email-obfuscation=*none|javascript|references*

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

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

152+ them using javascript. *references* obfuscates them by printing their

153+ letters as decimal or hexadecimal character references.

154+ If `--strict` is specified, *references* is used regardless of the

155+ presence of this option.

156+

149157\--toc, \--table-of-contents

150158: Include an automatically generated table of contents (HTML, markdown,

151159 RTF) or an instruction to create one (LaTeX, reStructuredText).