633a9ec · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

LaTeX writer: avoid `{}` after control sequences when escaping.

633a9ecJohn MacFarlane8 years ago

src/Text/Pandoc/Writers/LaTeX.hs+36−26

compare with master

unifiedsplit

@@ -39,8 +39,8 @@ import Prelude

39import Control.Applicative ((<|>))

40import Control.Monad.State.Strict

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

42−import Data.Char (isAlphaNum, isAscii, isDigit, isLetter, isPunctuation, ord,

42+import Data.Char (isAlphaNum, isAscii, isDigit, isLetter, isSpace,

43− toLower)

43+ isPunctuation, ord, toLower)

44import Data.List (foldl', intercalate, intersperse, isInfixOf, nubBy,

45 stripPrefix, (\\), uncons)

46import Data.Maybe (catMaybes, fromMaybe, isJust, mapMaybe, isNothing)

@@ -345,10 +345,20 @@ stringToLaTeX context zs = do

345 Just cmd -> ((cmd ++ "{" ++ [c] ++ "}") ++)

346 <$> go opts ctx (drop 1 xs) -- drop combining accent

347 Nothing -> (c:) <$> go opts ctx xs

348+ let emitcseq cs = do

349+ rest <- go opts ctx xs

350+ case rest of

351+ c:_ | isLetter c

352+ , ctx == TextString

353+ -> return (cs <> " " <> rest)

354+ | isSpace c -> return (cs <> "{}" <> rest)

355+ | ctx == TextString

356+ -> return (cs <> rest)

357+ _ -> return (cs <> "{}" <> rest)

348 case x of

358 case x of

349 '{' -> emits "\\{"

359 '{' -> emits "\\{"

350 '}' -> emits "\\}"

360 '}' -> emits "\\}"

351− '`' | ctx == CodeString -> emits "\\textasciigrave{}"

361+ '`' | ctx == CodeString -> emitcseq "\\textasciigrave"

352 '$' | not isUrl -> emits "\\$"

362 '$' | not isUrl -> emits "\\$"

353 '%' -> emits "\\%"

363 '%' -> emits "\\%"

354 '&' -> emits "\\&"

364 '&' -> emits "\\&"

@@ -358,19 +368,19 @@ stringToLaTeX context zs = do

358 -- prevent adjacent hyphens from forming ligatures

368 -- prevent adjacent hyphens from forming ligatures

359 ('-':_) -> emits "-\\/"

369 ('-':_) -> emits "-\\/"

360 _ -> emitc '-'

370 _ -> emitc '-'

361− '~' | not isUrl -> emits "\\textasciitilde{}"

371+ '~' | not isUrl -> emitcseq "\\textasciitilde"

362 '^' -> emits "\\^{}"

372 '^' -> emits "\\^{}"

363 '\\'| isUrl -> emitc '/' -- NB. / works as path sep even on Windows

373 '\\'| isUrl -> emitc '/' -- NB. / works as path sep even on Windows

364− | otherwise -> emits "\\textbackslash{}"

374+ | otherwise -> emitcseq "\\textbackslash"

365− '|' | not isUrl -> emits "\\textbar{}"

375+ '|' | not isUrl -> emitcseq "\\textbar"

366− '<' -> emits "\\textless{}"

376+ '<' -> emitcseq "\\textless"

367− '>' -> emits "\\textgreater{}"

377+ '>' -> emitcseq "\\textgreater"

368 '[' -> emits "{[}" -- to avoid interpretation as

378 '[' -> emits "{[}" -- to avoid interpretation as

369 ']' -> emits "{]}" -- optional arguments

379 ']' -> emits "{]}" -- optional arguments

370− '\'' | ctx == CodeString -> emits "\\textquotesingle{}"

380+ '\'' | ctx == CodeString -> emitcseq "\\textquotesingle"

371 '\160' -> emits "~"

381 '\160' -> emits "~"

372 '\x202F' -> emits "\\,"

382 '\x202F' -> emits "\\,"

373− '\x2026' -> emits "\\ldots{}"

383+ '\x2026' -> emitcseq "\\ldots"

374 '\x2018' | ligatures -> emits "`"

384 '\x2018' | ligatures -> emits "`"

375 '\x2019' | ligatures -> emits "'"

385 '\x2019' | ligatures -> emits "'"

376 '\x201C' | ligatures -> emits "``"

386 '\x201C' | ligatures -> emits "``"

@@ -379,22 +389,22 @@ stringToLaTeX context zs = do

379 '\x2013' | ligatures -> emits "--"

389 '\x2013' | ligatures -> emits "--"

380 _ | writerPreferAscii opts

390 _ | writerPreferAscii opts

381 -> case x of

391 -> case x of

382− 'ı' -> emits "\\i "

392+ 'ı' -> emitcseq "\\i"

383− 'ȷ' -> emits "\\j "

393+ 'ȷ' -> emitcseq "\\j"

384− 'å' -> emits "\\aa "

394+ 'å' -> emitcseq "\\aa"

385− 'Å' -> emits "\\AA "

395+ 'Å' -> emitcseq "\\AA"

386− 'ß' -> emits "\\ss "

396+ 'ß' -> emitcseq "\\ss"

387− 'ø' -> emits "\\o "

397+ 'ø' -> emitcseq "\\o"

388− 'Ø' -> emits "\\O "

398+ 'Ø' -> emitcseq "\\O"

389− 'Ł' -> emits "\\L "

399+ 'Ł' -> emitcseq "\\L"

390− 'ł' -> emits "\\l "

400+ 'ł' -> emitcseq "\\l"

391− 'æ' -> emits "\\ae "

401+ 'æ' -> emitcseq "\\ae"

392− 'Æ' -> emits "\\AE "

402+ 'Æ' -> emitcseq "\\AE"

393− 'œ' -> emits "\\oe "

403+ 'œ' -> emitcseq "\\oe"

394− 'Œ' -> emits "\\OE "

404+ 'Œ' -> emitcseq "\\OE"

395− '£' -> emits "\\pounds "

405+ '£' -> emitcseq "\\pounds"

396− '€' -> emits "\\euro "

406+ '€' -> emitcseq "\\euro"

398 _ -> emitc x

408 _ -> emitc x

399 | otherwise -> emitc x

409 | otherwise -> emitc x

400

410

test/command/1126.md+2−3

@@ -5,9 +5,8 @@ A&=&B,\\

5C&=&D

6\end{eqnarray}

7^D

8−\textbackslash{}begin\{eqnarray\}

9−A\&=\&B,\textbackslash{}\textbackslash{} C\&=\&D

8+\textbackslash begin\{eqnarray\} A\&=\&B,\textbackslash\textbackslash{}

10−\textbackslash{}end\{eqnarray\}

9+C\&=\&D \textbackslash end\{eqnarray\}

11```

10```

13```

12```

test/command/ascii.md+1−1

@@ -9,7 +9,7 @@ pandoc -t html --ascii

9pandoc -t latex --ascii

10äéıå

11^D

12−\"{a}\'{e}\i \r{a}

12+\"{a}\'{e}\i\r{a}

13```

15```

test/writer.latex+2−2

@@ -719,7 +719,7 @@ Subscripts: H\textsubscript{2}O, H\textsubscript{23}O,

719H\textsubscript{many~of~them}O.

720

721These should not be superscripts or subscripts, because of the unescaped

722−spaces: a\^{}b c\^{}d, a\textasciitilde{}b c\textasciitilde{}d.

722+spaces: a\^{}b c\^{}d, a\textasciitilde b c\textasciitilde d.

723

724\begin{center}\rule{0.5\linewidth}{\linethickness}\end{center}

725

@@ -741,7 +741,7 @@ Some dashes: one---two --- three---four --- five.

741

742Dashes between numbers: 5--7, 255--66, 1987--1999.

743

744−Ellipses\ldots{}and\ldots{}and\ldots{}.

744+Ellipses\ldots and\ldots and\ldots.

745

746\begin{center}\rule{0.5\linewidth}{\linethickness}\end{center}

747