9328f4c · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

LaTeX and ConTeXt writers: support lang attribute on divs and spans

9328f4cmb2111 years ago

README+11−6

compare with master

unifiedsplit

@@ -1047,12 +1047,21 @@ Language variables

1047 format stored in the additional variables `babel-lang`,

1048 `polyglossia-lang` (LaTeX) and `context-lang` (ConTeXt).

1049

1050+ Native pandoc `span`s and `div`s with the lang attribute

1051+ (value in BCP 47) can be used to switch the language in

1052+ that range.

1053+

1050`otherlangs`

1054`otherlangs`

1051: a list of other languages used in the document

1055: a list of other languages used in the document

1052 in the YAML metadata, according to [BCP 47]. For example:

1056 in the YAML metadata, according to [BCP 47]. For example:

1053 `otherlangs: [en-GB, fr]`.

1057 `otherlangs: [en-GB, fr]`.

1058+ This is automatically generated from the `lang` attributes

1059+ in all `span`s and `div`s but can be overriden.

1054− Currently only used by `xelatex` through the generated

1060+ Currently only used by LaTeX through the generated

1055− `polyglossia-otherlangs` variable.

1061+ `babel-otherlangs` and `polyglossia-otherlangs` variables.

1062+ The LaTeX writer outputs polyglossia commands in the text but

1063+ the `babel-newcommands` variable contains mappings for them

1064+ to the corresponding babel.

1056

1065

1057`dir`

1066`dir`

1058: the base direction of the document, either `rtl` (right-to-left)

1067: the base direction of the document, either `rtl` (right-to-left)

@@ -1065,10 +1074,6 @@ Language variables

1065 (e.g. the browser, when generating HTML) supports the

1074 (e.g. the browser, when generating HTML) supports the

1066 [Unicode Bidirectional Algorithm].

1075 [Unicode Bidirectional Algorithm].

1067

1076

1068− LaTeX and ConTeXt assume by default that all text is left-to-right.

1069− Setting `dir: ltr` enables bidirectional text handling in a document

1070− whose base direction is left-to-right but contains some right-to-left script.

1071−

1072 When using LaTeX for bidirectional documents, only the `xelatex` engine

1077 When using LaTeX for bidirectional documents, only the `xelatex` engine

1073 is fully supported (use `--latex-engine=xelatex`).

1078 is fully supported (use `--latex-engine=xelatex`).

1074

1079

src/Text/Pandoc/Writers/ConTeXt.hs+27−16

@@ -157,17 +157,21 @@ blockToConTeXt (CodeBlock _ str) =

157blockToConTeXt (RawBlock "context" str) = return $ text str <> blankline

158blockToConTeXt (RawBlock _ _ ) = return empty

159blockToConTeXt (Div (ident,_,kvs) bs) = do

160− contents <- blockListToConTeXt bs

160+ let align dir txt = "\\startalignment[" <> dir <> "]" $$ txt $$ "\\stopalignment"

161− let contents' = if null ident

161+ let wrapRef txt = if null ident

162− then contents

162+ then txt

163− else ("\\reference" <> brackets (text $ toLabel ident) <>

163+ else ("\\reference" <> brackets (text $ toLabel ident) <>

164− braces empty <> "%") $$ contents

164+ braces empty <> "%") $$ txt

165− let align dir = blankline <> "\\startalignment[" <> dir <> "]"

166− $$ contents' $$ "\\stopalignment" <> blankline

167− return $ case lookup "dir" kvs of

165+ wrapDir = case lookup "dir" kvs of

168− Just "rtl" -> align "righttoleft"

166+ Just "rtl" -> align "righttoleft"

169− Just "ltr" -> align "lefttoright"

167+ Just "ltr" -> align "lefttoright"

170− _ -> contents'

168+ _ -> id

169+ wrapLang txt = case lookup "lang" kvs of

170+ Just lng -> "\\start\\language["

171+ <> text (fromBcp47' lng) <> "]" $$ txt $$ "\\stop"

172+ Nothing -> txt

173+ wrapBlank txt = blankline <> txt <> blankline

174+ fmap (wrapBlank . wrapLang . wrapDir . wrapRef) $ blockListToConTeXt bs

171blockToConTeXt (BulletList lst) = do

175blockToConTeXt (BulletList lst) = do

172 contents <- mapM listItemToConTeXt lst

176 contents <- mapM listItemToConTeXt lst

173 return $ ("\\startitemize" <> if isTightList lst

177 return $ ("\\startitemize" <> if isTightList lst

@@ -346,11 +350,15 @@ inlineToConTeXt (Note contents) = do

346 else text "\\startbuffer " <> nest 2 contents' <>

350 else text "\\startbuffer " <> nest 2 contents' <>

347 text "\\stopbuffer\\footnote{\\getbuffer}"

351 text "\\stopbuffer\\footnote{\\getbuffer}"

348inlineToConTeXt (Span (_,_,kvs) ils) = do

352inlineToConTeXt (Span (_,_,kvs) ils) = do

349− contents <- inlineListToConTeXt ils

350− return $ case lookup "dir" kvs of

353+ let wrapDir txt = case lookup "dir" kvs of

351− Just "rtl" -> braces $ "\\righttoleft " <> contents

354+ Just "rtl" -> braces $ "\\righttoleft " <> txt

352− Just "ltr" -> braces $ "\\lefttoright " <> contents

355+ Just "ltr" -> braces $ "\\lefttoright " <> txt

353− _ -> contents

356+ _ -> txt

357+ wrapLang txt = case lookup "lang" kvs of

358+ Just lng -> "\\start\\language[" <> text (fromBcp47' lng)

359+ <> "]" <> txt <> "\\stop "

360+ Nothing -> txt

361+ fmap (wrapLang . wrapDir) $ inlineListToConTeXt ils

354

362

355-- | Craft the section header, inserting the secton reference, if supplied.

363-- | Craft the section header, inserting the secton reference, if supplied.

356sectionHeader :: Attr

364sectionHeader :: Attr

@@ -377,6 +385,9 @@ sectionHeader (ident,classes,_) hdrLevel lst = do

377 then char '\\' <> chapter <> braces contents

385 then char '\\' <> chapter <> braces contents

378 else contents <> blankline

386 else contents <> blankline

379

387

388+fromBcp47' :: String -> String

389+fromBcp47' = fromBcp47 . splitBy (=='-')

390+

380-- Takes a list of the constituents of a BCP 47 language code

391-- Takes a list of the constituents of a BCP 47 language code

381-- and irons out ConTeXt's exceptions

392-- and irons out ConTeXt's exceptions

382-- https://tools.ietf.org/html/bcp47#section-2.1

393-- https://tools.ietf.org/html/bcp47#section-2.1

src/Text/Pandoc/Writers/LaTeX.hs+83−16

@@ -39,7 +39,7 @@ import Text.Pandoc.Templates

39import Text.Printf ( printf )

40import Network.URI ( isURI, unEscapeString )

41import Data.Aeson (object, (.=))

42−import Data.List ( (\\), isInfixOf, stripPrefix, intercalate, intersperse )

42+import Data.List ( (\\), isInfixOf, stripPrefix, intercalate, intersperse, nub, nubBy )

43import Data.Char ( toLower, isPunctuation, isAscii, isLetter, isDigit, ord )

44import Data.Maybe ( fromMaybe )

45import qualified Data.Text as T

@@ -145,6 +145,7 @@ pandocToLaTeX options (Pandoc meta blocks) = do

145 st <- get

146 titleMeta <- stringToLaTeX TextString $ stringify $ docTitle meta

147 authorsMeta <- mapM (stringToLaTeX TextString . stringify) $ docAuthors meta

148+ let docLangs = nub $ query (extract "lang") blocks

148 let context = defField "toc" (writerTableOfContents options) $

149 let context = defField "toc" (writerTableOfContents options) $

149 defField "toc-depth" (show (writerTOCDepth options -

150 defField "toc-depth" (show (writerTOCDepth options -

150 if stBook st

151 if stBook st

@@ -179,18 +180,48 @@ pandocToLaTeX options (Pandoc meta blocks) = do

179 Biblatex -> defField "biblio-title" biblioTitle .

180 Biblatex -> defField "biblio-title" biblioTitle .

180 defField "biblatex" True

181 defField "biblatex" True

181 _ -> id) $

182 _ -> id) $

183+ -- set lang to something so polyglossia/babel is included

184+ defField "lang" (if null docLangs then ""::String else "en") $

185+ defField "otherlangs" docLangs $

186+ defField "dir" (if (null $ query (extract "dir") blocks)

187+ then ""::String

188+ else "ltr") $

182 metadata

189 metadata

183 let toPolyObj lang = object [ "name" .= T.pack name

190 let toPolyObj lang = object [ "name" .= T.pack name

184 , "options" .= T.pack opts ]

191 , "options" .= T.pack opts ]

185 where

192 where

186 (name, opts) = toPolyglossia lang

193 (name, opts) = toPolyglossia lang

187 let lang = maybe [] (splitBy (=='-')) $ getField "lang" context

194 let lang = maybe [] (splitBy (=='-')) $ getField "lang" context

195+ otherlangs = maybe [] (map $ splitBy (=='-')) $ getField "otherlangs" context

188 let context' =

196 let context' =

189 defField "babel-lang" (toBabel lang)

197 defField "babel-lang" (toBabel lang)

198+ $ defField "babel-otherlangs" (map toBabel otherlangs)

199+ $ defField "babel-newcommands" (concatMap (\(poly, babel) ->

200+ -- \textspanish and \textgalician are already used by babel

201+ -- save them as \oritext... and let babel use that

202+ if poly `elem` ["spanish", "galician"]

203+ then "\\let\\oritext" ++ poly ++ "\\text" ++ poly ++ "\n" ++

204+ "\\AddBabelHook{" ++ poly ++ "}{beforeextras}" ++

205+ "{\\renewcommand{\\text" ++ poly ++ "}{\\oritext"

206+ ++ poly ++ "}}\n" ++

207+ "\\AddBabelHook{" ++ poly ++ "}{afterextras}" ++

208+ "{\\renewcommand{\\text" ++ poly ++ "}[2][]{\\foreignlanguage{"

209+ ++ poly ++ "}{##2}}}\n"

210+ else "\\newcommand{\\text" ++ poly ++ "}[2][]{\\foreignlanguage{"

211+ ++ babel ++ "}{#2}}\n" ++

212+ "\\newenvironment{" ++ poly ++ "}[1]{\\begin{otherlanguage}{"

213+ ++ babel ++ "}}{\\end{otherlanguage}}\n"

214+ )

215+ -- eliminate duplicates that have same polyglossia name

216+ $ nubBy (\a b -> fst a == fst b)

217+ -- find polyglossia and babel names of languages used in the document

218+ $ map (\l ->

219+ let lng = splitBy (=='-') l

220+ in (fst $ toPolyglossia lng, toBabel lng)

221+ )

222+ docLangs )

190 $ defField "polyglossia-lang" (toPolyObj lang)

223 $ defField "polyglossia-lang" (toPolyObj lang)

191− $ defField "polyglossia-otherlangs"

224+ $ defField "polyglossia-otherlangs" (map toPolyObj otherlangs)

192− (maybe [] (map $ toPolyObj . splitBy (=='-')) $

193− getField "otherlangs" context)

194 $ defField "latex-dir-rtl" (case (getField "dir" context)::Maybe String of

225 $ defField "latex-dir-rtl" (case (getField "dir" context)::Maybe String of

195 Just "rtl" -> True

226 Just "rtl" -> True

196 _ -> False)

227 _ -> False)

@@ -340,15 +371,24 @@ blockToLaTeX (Div (identifier,classes,kvs) bs) = do

340 then empty

371 then empty

341 else "\\hyperdef{}" <> braces (text ref) <>

372 else "\\hyperdef{}" <> braces (text ref) <>

342 braces ("\\label" <> braces (text ref))

373 braces ("\\label" <> braces (text ref))

343− contents' <- blockListToLaTeX bs

344− let align dir = inCmd "begin" dir $$ contents' $$ inCmd "end" dir

374+ let align dir txt = inCmd "begin" dir $$ txt $$ inCmd "end" dir

345− let contents = case lookup "dir" kvs of

375+ let wrapDir = case lookup "dir" kvs of

346− Just "rtl" -> align "RTL"

376+ Just "rtl" -> align "RTL"

347− Just "ltr" -> align "LTR"

377+ Just "ltr" -> align "LTR"

348− _ -> contents'

378+ _ -> id

379+ wrapLang txt = case lookup "lang" kvs of

380+ Just lng -> let (l, o) = toPolyglossiaEnv lng

381+ ops = if null o

382+ then ""

383+ else brackets $ text o

384+ in inCmd "begin" (text l) <> ops

385+ $$ blankline <> txt <> blankline

386+ $$ inCmd "end" (text l)

387+ Nothing -> txt

349− if beamer && "notes" `elem` classes -- speaker notes

388+ wrapNotes txt = if beamer && "notes" `elem` classes

350− then return $ "\\note" <> braces contents

389+ then "\\note" <> braces txt -- speaker notes

351− else return (linkAnchor $$ contents)

390+ else linkAnchor $$ txt

391+ fmap (wrapDir . wrapLang . wrapNotes) $ blockListToLaTeX bs

352blockToLaTeX (Plain lst) =

392blockToLaTeX (Plain lst) =

353 inlineListToLaTeX $ dropWhile isLineBreakOrSpace lst

393 inlineListToLaTeX $ dropWhile isLineBreakOrSpace lst

354-- title beginning with fig: indicates that the image is a figure

394-- title beginning with fig: indicates that the image is a figure

@@ -759,9 +799,12 @@ inlineToLaTeX (Span (id',classes,kvs) ils) = do

759 (if noSmallCaps then inCmd "textnormal" else id) .

799 (if noSmallCaps then inCmd "textnormal" else id) .

760 (if rtl then inCmd "RL" else id) .

800 (if rtl then inCmd "RL" else id) .

761 (if ltr then inCmd "LR" else id) .

801 (if ltr then inCmd "LR" else id) .

762− (if not (noEmph || noStrong || noSmallCaps || rtl || ltr)

802+ (case lookup "lang" kvs of

763− then braces

803+ Just lng -> let (l, o) = toPolyglossiaEnv lng

804+ ops = if null o then "" else brackets (text o)

805+ in \c -> char '\\' <> "text" <> text l <> ops <> braces c

806+ Nothing -> id)

764− else id)) `fmap` inlineListToLaTeX ils

807+ ) `fmap` inlineListToLaTeX ils

765inlineToLaTeX (Emph lst) =

808inlineToLaTeX (Emph lst) =

766 inlineListToLaTeX lst >>= return . inCmd "emph"

809 inlineListToLaTeX lst >>= return . inCmd "emph"

767inlineToLaTeX (Strong lst) =

810inlineToLaTeX (Strong lst) =

@@ -1002,6 +1045,30 @@ getListingsLanguage :: [String] -> Maybe String

1002getListingsLanguage [] = Nothing

1045getListingsLanguage [] = Nothing

1003getListingsLanguage (x:xs) = toListingsLanguage x <|> getListingsLanguage xs

1046getListingsLanguage (x:xs) = toListingsLanguage x <|> getListingsLanguage xs

1004

1047

1048+-- Extract a key from divs and spans

1049+extract :: String -> Block -> [String]

1050+extract key (Div attr _) = lookKey key attr

1051+extract key (Plain ils) = concatMap (extractInline key) ils

1052+extract key (Para ils) = concatMap (extractInline key) ils

1053+extract key (Header _ _ ils) = concatMap (extractInline key) ils

1054+extract _ _ = []

1055+

1056+-- Extract a key from spans

1057+extractInline :: String -> Inline -> [String]

1058+extractInline key (Span attr _) = lookKey key attr

1059+extractInline _ _ = []

1060+

1061+-- Look up a key in an attribute and give a list of its values

1062+lookKey :: String -> Attr -> [String]

1063+lookKey key (_,_,kvs) = maybe [] words $ lookup key kvs

1064+

1065+-- In environments \Arabic instead of \arabic is used

1066+toPolyglossiaEnv :: String -> (String, String)

1067+toPolyglossiaEnv l =

1068+ case toPolyglossia $ (splitBy (=='-')) l of

1069+ ("arabic", o) -> ("Arabic", o)

1070+ x -> x

1071+

1005-- Takes a list of the constituents of a BCP 47 language code and

1072-- Takes a list of the constituents of a BCP 47 language code and

1006-- converts it to a Polyglossia (language, options) tuple

1073-- converts it to a Polyglossia (language, options) tuple

1007-- http://mirrors.concertpass.com/tex-archive/macros/latex/contrib/polyglossia/polyglossia.pdf

1074-- http://mirrors.concertpass.com/tex-archive/macros/latex/contrib/polyglossia/polyglossia.pdf