31c030e · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

Added --inline-links option to force links in HTML to be parsed

31c030efiddlosopher20 years ago

README+3

compare with master

unifiedsplit

@@ -241,6 +241,9 @@ preserved, rather than converted to spaces (the default).

241`--strict` specifies that strict markdown syntax is to be used, without

242pandoc's usual extensions and variants (described below).

243

244+`--inline-links` causes links in HTML to be parsed as inline links, rather

245+than reference links.

246+

244`-R` or `--parse-raw` causes the HTML and LaTeX readers to parse HTML

247`-R` or `--parse-raw` causes the HTML and LaTeX readers to parse HTML

245codes and LaTeX environments that it can't translate as raw HTML or

248codes and LaTeX environments that it can't translate as raw HTML or

246LaTeX. Raw HTML can be printed in markdown, reStructuredText, HTML,

249LaTeX. Raw HTML can be printed in markdown, reStructuredText, HTML,

man/man1/html2markdown.1+2

@@ -32,6 +32,8 @@ Write output to \fIFILE\fR instead of STDOUT.

32.B \-\-strict

33Use strict markdown syntax, with no extensions or variants.

34.TP

35+.B \-\-inline\-links

36+Parse links in HTML as inline links, rather than reference links.

35.TP

37.TP

36.B \-R, \-\-parse-raw

38.B \-R, \-\-parse-raw

37Parse untranslatable HTML codes as raw HTML.

39Parse untranslatable HTML codes as raw HTML.

man/man1/pandoc.1+3

@@ -105,6 +105,9 @@ Specify tab stop (default is 4).

105.B \-\-strict

106Use strict markdown syntax, with no extensions or variants.

107.TP

108+.B \-\-inline\-links

109+Parse links in HTML as inline links, rather than reference links.

110+.TP

108.B \-R, \-\-parse-raw

111.B \-R, \-\-parse-raw

109Parse untranslatable HTML codes and LaTeX environments as raw HTML

112Parse untranslatable HTML codes and LaTeX environments as raw HTML

110or LaTeX, instead of ignoring them.

113or LaTeX, instead of ignoring them.

src/Main.hs+15−6

@@ -118,6 +118,7 @@ data Opt = Opt

118 , optDumpArgs :: Bool -- ^ Output command-line arguments

119 , optIgnoreArgs :: Bool -- ^ Ignore command-line arguments

120 , optStrict :: Bool -- ^ Use strict markdown syntax

121+ , optInlineLinks :: Bool -- ^ Use inline links in parsing HTML

121 }

122 }

122

123

123-- | Defaults for command-line options.

124-- | Defaults for command-line options.

@@ -143,6 +144,7 @@ defaultOpts = Opt

143 , optDumpArgs = False

144 , optDumpArgs = False

144 , optIgnoreArgs = False

145 , optIgnoreArgs = False

145 , optStrict = False

146 , optStrict = False

147+ , optInlineLinks = False

146 }

148 }

147

149

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

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

@@ -188,6 +190,11 @@ options =

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

190 (\opt -> return opt { optStrict = True } ))

189 "" -- "Use strict markdown syntax with no extensions"

191 "" -- "Use strict markdown syntax with no extensions"

190

192

193+ , Option "" ["inline-links"]

194+ (NoArg

195+ (\opt -> return opt { optInlineLinks = True } ))

196+ "" -- "Use inline links in parsing HTML"

197+

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

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

192 (NoArg

199 (NoArg

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

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

@@ -398,6 +405,7 @@ main = do

398 , optDumpArgs = dumpArgs

405 , optDumpArgs = dumpArgs

399 , optIgnoreArgs = ignoreArgs

406 , optIgnoreArgs = ignoreArgs

400 , optStrict = strict

407 , optStrict = strict

408+ , optInlineLinks = inlineLinks

401 } = opts

409 } = opts

402

410

403 if dumpArgs

411 if dumpArgs

@@ -440,12 +448,13 @@ main = do

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

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

441 let filter = tabFilter . addBlank . removeCRs

449 let filter = tabFilter . addBlank . removeCRs

442 let startParserState =

450 let startParserState =

443− defaultParserState { stateParseRaw = parseRaw,

451+ defaultParserState { stateParseRaw = parseRaw,

444− stateTabStop = tabStop,

452+ stateTabStop = tabStop,

445− stateStandalone = standalone && (not strict),

453+ stateStandalone = standalone && (not strict),

446− stateSmart = smart || writerName' == "latex",

454+ stateSmart = smart || writerName' == "latex",

447− stateColumns = columns,

455+ stateColumns = columns,

448− stateStrict = strict }

456+ stateStrict = strict,

457+ stateInlineLinks = inlineLinks }

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

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

450 then ""

459 then ""

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

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

src/Text/Pandoc/Readers/HTML.hs+6−3

@@ -449,15 +449,18 @@ extractAttribute name ((attrName, contents):rest) =

449 then Just (decodeEntities contents)

450 else extractAttribute name rest

451

452−link = try (do

452+link = try $ do

453 (tag, attributes) <- htmlTag "a"

454 url <- case (extractAttribute "href" attributes) of

455 Just url -> do {return url}

456 Nothing -> fail "no href"

457 let title = fromMaybe "" (extractAttribute "title" attributes)

458 label <- inlinesTilEnd "a"

459+ state <- getState

460+ ref <- if stateInlineLinks state

461+ then return (Src url title)

459− ref <- generateReference url title

462+ else generateReference url title

460− return (Link (normalizeSpaces label) ref))

463+ return $ Link (normalizeSpaces label) ref

461

464

462image = try (do

465image = try (do

463 (tag, attributes) <- htmlTag "img"

466 (tag, attributes) <- htmlTag "img"

src/Text/Pandoc/Shared.hs+2

@@ -133,6 +133,7 @@ data ParserState = ParserState

133 stateSmart :: Bool, -- ^ Use smart typography

134 stateColumns :: Int, -- ^ Number of columns in

135 -- terminal (used for tables)

136+ stateInlineLinks :: Bool, -- ^ Parse html links as inline

136 stateHeaderTable :: [HeaderType] -- ^ List of header types used,

137 stateHeaderTable :: [HeaderType] -- ^ List of header types used,

137 -- in what order (rst only)

138 -- in what order (rst only)

138 }

139 }

@@ -155,6 +156,7 @@ defaultParserState =

155 stateStrict = False,

156 stateStrict = False,

156 stateSmart = False,

157 stateSmart = False,

157 stateColumns = 80,

158 stateColumns = 80,

159+ stateInlineLinks = False,

158 stateHeaderTable = [] }

160 stateHeaderTable = [] }

159

161

160-- | Indent string as a block.

162-- | Indent string as a block.