31c030e · typeable/pandoc

git.iko.soy/typeable/pandoc

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

as inline links, rather than reference links. (Addresses Issue #4.) git-svn-id: https://pandoc.googlecode.com/svn/trunk@554 788f1e2b-df1e-0410-8736-df70ead52e1b

commit31c030e3a55c3bc9d4a606e9658dd30a9aa1c938 authorfiddlosopher <fiddlosopher@788f1e2b-df1e-0410-8736-df70ead52e1b> dateSat 3 Mar 2007, 18:19 (20 years ago) parentf99cedd

6 files changed+31−9

compare with master

unifiedsplit

README+3 man/man1/html2markdown.1+2 man/man1/pandoc.1+3 src/Main.hs+15−6 src/Text/Pandoc/Readers/HTML.hs+6−3 src/Text/Pandoc/Shared.hs+2

README+3

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

241241`--strict` specifies that strict markdown syntax is to be used, without

242242pandoc's usual extensions and variants (described below).

243243

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

245+than reference links.

246+

244247`-R` or `--parse-raw` causes the HTML and LaTeX readers to parse HTML

245248codes and LaTeX environments that it can't translate as raw HTML or

246249LaTeX. 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.

3232.B \-\-strict

3333Use strict markdown syntax, with no extensions or variants.

3434.TP

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

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

3537.TP

3638.B \-R, \-\-parse-raw

3739Parse untranslatable HTML codes as raw HTML.

man/man1/pandoc.1+3

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

105105.B \-\-strict

106106Use strict markdown syntax, with no extensions or variants.

107107.TP

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

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

110+.TP

108111.B \-R, \-\-parse-raw

109112Parse untranslatable HTML codes and LaTeX environments as raw HTML

110113or LaTeX, instead of ignoring them.

src/Main.hs+15−6

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

118118 , optDumpArgs :: Bool -- ^ Output command-line arguments

119119 , optIgnoreArgs :: Bool -- ^ Ignore command-line arguments

120120 , optStrict :: Bool -- ^ Use strict markdown syntax

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

121122 }

122123

123124-- | Defaults for command-line options.

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

143144 , optDumpArgs = False

144145 , optIgnoreArgs = False

145146 , optStrict = False

147+ , optInlineLinks = False

146148 }

147149

148150-- | A list of functions, each transforming the options data structure

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

188190 (\opt -> return opt { optStrict = True } ))

189191 "" -- "Use strict markdown syntax with no extensions"

190192

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

194+ (NoArg

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

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

197+

191198 , Option "R" ["parse-raw"]

192199 (NoArg

193200 (\opt -> return opt { optParseRaw = True }))

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

398405 , optDumpArgs = dumpArgs

399406 , optIgnoreArgs = ignoreArgs

400407 , optStrict = strict

408+ , optInlineLinks = inlineLinks

401409 } = opts

402410

403411 if dumpArgs

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

440448 let removeCRs str = filter (/= '\r') str -- remove DOS-style line endings

441449 let filter = tabFilter . addBlank . removeCRs

442450 let startParserState =

443− defaultParserState { stateParseRaw = parseRaw,

444− stateTabStop = tabStop,

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

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

447− stateColumns = columns,

448− stateStrict = strict }

451+ defaultParserState { stateParseRaw = parseRaw,

452+ stateTabStop = tabStop,

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

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

455+ stateColumns = columns,

456+ stateStrict = strict,

457+ stateInlineLinks = inlineLinks }

449458 let csslink = if (css == "")

450459 then ""

451460 else "<link rel=\"stylesheet\" href=\"" ++ css ++

src/Text/Pandoc/Readers/HTML.hs+6−3

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

449449 then Just (decodeEntities contents)

450450 else extractAttribute name rest

451451

452−link = try (do

452+link = try $ do

453453 (tag, attributes) <- htmlTag "a"

454454 url <- case (extractAttribute "href" attributes) of

455455 Just url -> do {return url}

456456 Nothing -> fail "no href"

457457 let title = fromMaybe "" (extractAttribute "title" attributes)

458458 label <- inlinesTilEnd "a"

459− ref <- generateReference url title

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

459+ state <- getState

460+ ref <- if stateInlineLinks state

461+ then return (Src url title)

462+ else generateReference url title

463+ return $ Link (normalizeSpaces label) ref

461464

462465image = try (do

463466 (tag, attributes) <- htmlTag "img"

src/Text/Pandoc/Shared.hs+2

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

133133 stateSmart :: Bool, -- ^ Use smart typography

134134 stateColumns :: Int, -- ^ Number of columns in

135135 -- terminal (used for tables)

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

136137 stateHeaderTable :: [HeaderType] -- ^ List of header types used,

137138 -- in what order (rst only)

138139 }

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

155156 stateStrict = False,

156157 stateSmart = False,

157158 stateColumns = 80,

159+ stateInlineLinks = False,

158160 stateHeaderTable = [] }

159161

160162-- | Indent string as a block.