a1a57bc · typeable/pandoc

git.iko.soy/typeable/pandoc

T.P.Shared: remove `backslashEscapes`, `escapeStringUsing`.

[API change] These are inefficient association list lookups. Replace with more efficient functions in the writers that used them (with 10-25% performance improvements in haddock, org, rtf, texinfo writers).

commita1a57bce4e32cc26b968bcc2847a8e8da30f725b authorJohn MacFarlane <jgm@berkeley.edu> dateSat 20 Mar 2021, 00:02 (6 years ago) parentceadf33

8 files changed+77−47

compare with master

unifiedsplit

src/Text/Pandoc/Shared.hs−13 src/Text/Pandoc/Writers/AsciiDoc.hs+8−3 src/Text/Pandoc/Writers/ConTeXt.hs+5−1 src/Text/Pandoc/Writers/Haddock.hs+13−2 src/Text/Pandoc/Writers/LaTeX.hs+13−1 src/Text/Pandoc/Writers/Org.hs+9−6 src/Text/Pandoc/Writers/RTF.hs+16−10 src/Text/Pandoc/Writers/Texinfo.hs+13−11

src/Text/Pandoc/Shared.hs−13

@@ -26,8 +26,6 @@ module Text.Pandoc.Shared (

2626 findM,

2727 -- * Text processing

2828 tshow,

29− backslashEscapes,

30− escapeStringUsing,

3129 elemText,

3230 notElemText,

3331 stripTrailingNewlines,

@@ -184,17 +182,6 @@ findM p = foldr go (pure Nothing)

184182tshow :: Show a => a -> T.Text

185183tshow = T.pack . show

186184

187−-- | Returns an association list of backslash escapes for the

188−-- designated characters.

189−backslashEscapes :: [Char] -- ^ list of special characters to escape

190− -> [(Char, T.Text)]

191−backslashEscapes = map (\ch -> (ch, T.pack ['\\',ch]))

192−

193−-- | Escape a string of characters, using an association list of

194−-- characters and strings.

195−escapeStringUsing :: [(Char, T.Text)] -> T.Text -> T.Text

196−escapeStringUsing tbl = T.concatMap $ \c -> fromMaybe (T.singleton c) $ lookup c tbl

197−

198185-- | @True@ exactly when the @Char@ appears in the @Text@.

199186elemText :: Char -> T.Text -> Bool

200187elemText c = T.any (== c)

src/Text/Pandoc/Writers/AsciiDoc.hs+8−3

@@ -105,8 +105,11 @@ pandocToAsciiDoc opts (Pandoc meta blocks) = do

105105

106106-- | Escape special characters for AsciiDoc.

107107escapeString :: Text -> Text

108−escapeString = escapeStringUsing escs

109− where escs = backslashEscapes "{"

108+escapeString t

109+ | T.any (== '{') t = T.concatMap escChar t

110+ | otherwise = t

111+ where escChar '{' = "\\{"

112+ escChar c = T.singleton c

110113

111114-- | Ordered list start parser for use in Para below.

112115olMarker :: Parser Text ParserState Char

@@ -496,7 +499,9 @@ inlineToAsciiDoc opts (Quoted qt lst) = do

496499 | otherwise -> [Str "``"] ++ lst ++ [Str "''"]

497500inlineToAsciiDoc _ (Code _ str) = do

498501 isAsciidoctor <- gets asciidoctorVariant

499− let contents = literal (escapeStringUsing (backslashEscapes "`") str)

502+ let escChar '`' = "\\'"

503+ escChar c = T.singleton c

504+ let contents = literal (T.concatMap escChar str)

500505 return $

501506 if isAsciidoctor

502507 then text "`+" <> contents <> "+`"

src/Text/Pandoc/Writers/ConTeXt.hs+5−1

@@ -434,9 +434,13 @@ inlineToConTeXt (Link _ txt (src, _)) = do

434434 put $ st {stNextRef = next + 1}

435435 let ref = "url" <> tshow next

436436 contents <- inlineListToConTeXt txt

437+ let escChar '#' = "\\#"

438+ escChar '%' = "\\%"

439+ escChar c = T.singleton c

440+ let escContextURL = T.concatMap escChar

437441 return $ "\\useURL"

438442 <> brackets (literal ref)

439− <> brackets (literal $ escapeStringUsing [('#',"\\#"),('%',"\\%")] src)

443+ <> brackets (literal $ escContextURL src)

440444 <> (if isAutolink

441445 then empty

442446 else brackets empty <> brackets contents)

src/Text/Pandoc/Writers/Haddock.hs+13−2

@@ -15,6 +15,7 @@ Haddock: <http://www.haskell.org/haddock/doc/html/>

1515-}

1616module Text.Pandoc.Writers.Haddock (writeHaddock) where

1717import Control.Monad.State.Strict

18+import Data.Char (isAlphaNum)

1819import Data.Default

1920import Data.Text (Text)

2021import qualified Data.Text as T

@@ -71,8 +72,18 @@ notesToHaddock opts notes =

7172

7273-- | Escape special characters for Haddock.

7374escapeString :: Text -> Text

74−escapeString = escapeStringUsing haddockEscapes

75− where haddockEscapes = backslashEscapes "\\/'`\"@<"

75+escapeString t

76+ | T.all isAlphaNum t = t

77+ | otherwise = T.concatMap escChar t

78+ where

79+ escChar '\\' = "\\\\"

80+ escChar '/' = "\\/"

81+ escChar '\'' = "\\'"

82+ escChar '`' = "\\`"

83+ escChar '"' = "\\\""

84+ escChar '@' = "\\@"

85+ escChar '<' = "\\<"

86+ escChar c = T.singleton c

7687

7788-- | Convert Pandoc block element to haddock.

7889blockToHaddock :: PandocMonad m

src/Text/Pandoc/Writers/LaTeX.hs+13−1

@@ -825,7 +825,19 @@ inlineToLaTeX (Code (_,classes,kvs) str) = do

825825 let chr = case "!\"'()*,-./:;?@" \\ T.unpack str of

826826 (c:_) -> c

827827 [] -> '!'

828− let str' = escapeStringUsing (backslashEscapes "\\{}%~_&#^") str

828+ let isEscapable '\\' = True

829+ isEscapable '{' = True

830+ isEscapable '}' = True

831+ isEscapable '%' = True

832+ isEscapable '~' = True

833+ isEscapable '_' = True

834+ isEscapable '&' = True

835+ isEscapable '#' = True

836+ isEscapable '^' = True

837+ isEscapable _ = False

838+ let escChar c | isEscapable c = T.pack ['\\',c]

839+ | otherwise = T.singleton c

840+ let str' = T.concatMap escChar str

829841 -- we always put lstinline in a dummy 'passthrough' command

830842 -- (defined in the default template) so that we don't have

831843 -- to change the way we escape characters depending on whether

src/Text/Pandoc/Writers/Org.hs+9−6

@@ -84,12 +84,15 @@ noteToOrg num note = do

8484

8585-- | Escape special characters for Org.

8686escapeString :: Text -> Text

87−escapeString = escapeStringUsing

88− [ ('\x2014',"---")

89− , ('\x2013',"--")

90− , ('\x2019',"'")

91− , ('\x2026',"...")

92− ]

87+escapeString t

88+ | T.all (\c -> c < '\x2013' || c > '\x2026') t = t

89+ | otherwise = T.concatMap escChar t

90+ where

91+ escChar '\x2013' = "--"

92+ escChar '\x2014' = "---"

93+ escChar '\x2019' = "'"

94+ escChar '\x2026' = "..."

95+ escChar c = T.singleton c

9396

9497isRawFormat :: Format -> Bool

9598isRawFormat f =

src/Text/Pandoc/Writers/RTF.hs+16−10

@@ -16,7 +16,7 @@ module Text.Pandoc.Writers.RTF ( writeRTF

1616import Control.Monad.Except (catchError, throwError)

1717import Control.Monad

1818import qualified Data.ByteString as B

19−import Data.Char (chr, isDigit, ord)

19+import Data.Char (chr, isDigit, ord, isAlphaNum)

2020import qualified Data.Map as M

2121import Data.Text (Text)

2222import qualified Data.Text as T

@@ -137,15 +137,21 @@ handleUnicode = T.concatMap $ \c ->

137137

138138-- | Escape special characters.

139139escapeSpecial :: Text -> Text

140−escapeSpecial = escapeStringUsing $

141− [ ('\t',"\\tab ")

142− , ('\8216',"\\u8216'")

143− , ('\8217',"\\u8217'")

144− , ('\8220',"\\u8220\"")

145− , ('\8221',"\\u8221\"")

146− , ('\8211',"\\u8211-")

147− , ('\8212',"\\u8212-")

148− ] <> backslashEscapes "{\\}"

140+escapeSpecial t

141+ | T.all isAlphaNum t = t

142+ | otherwise = T.concatMap escChar t

143+ where

144+ escChar '\t' = "\\tab "

145+ escChar '\8216' = "\\u8216'"

146+ escChar '\8217' = "\\u8217'"

147+ escChar '\8220' = "\\u8220\""

148+ escChar '\8221' = "\\u8221\""

149+ escChar '\8211' = "\\u8211-"

150+ escChar '\8212' = "\\u8212-"

151+ escChar '{' = "\\{"

152+ escChar '}' = "\\}"

153+ escChar '\\' = "\\\\"

154+ escChar c = T.singleton c

149155

150156-- | Escape strings as needed for rich text format.

151157stringToRTF :: Text -> Text

src/Text/Pandoc/Writers/Texinfo.hs+13−11

@@ -14,7 +14,7 @@ Conversion of 'Pandoc' format into Texinfo.

1414module Text.Pandoc.Writers.Texinfo ( writeTexinfo ) where

1515import Control.Monad.Except (throwError)

1616import Control.Monad.State.Strict

17−import Data.Char (chr, ord)

17+import Data.Char (chr, ord, isAlphaNum)

1818import Data.List (maximumBy, transpose, foldl')

1919import Data.List.NonEmpty (nonEmpty)

2020import Data.Ord (comparing)

@@ -85,16 +85,18 @@ pandocToTexinfo options (Pandoc meta blocks) = do

8585

8686-- | Escape things as needed for Texinfo.

8787stringToTexinfo :: Text -> Text

88−stringToTexinfo = escapeStringUsing texinfoEscapes

89− where texinfoEscapes = [ ('{', "@{")

90− , ('}', "@}")

91− , ('@', "@@")

92− , ('\160', "@ ")

93− , ('\x2014', "---")

94− , ('\x2013', "--")

95− , ('\x2026', "@dots{}")

96− , ('\x2019', "'")

97− ]

88+stringToTexinfo t

89+ | T.all isAlphaNum t = t

90+ | otherwise = T.concatMap escChar t

91+ where escChar '{' = "@{"

92+ escChar '}' = "@}"

93+ escChar '@' = "@@"

94+ escChar '\160' = "@ "

95+ escChar '\x2014' = "---"

96+ escChar '\x2013' = "--"

97+ escChar '\x2026' = "@dots{}"

98+ escChar '\x2019' = "'"

99+ escChar c = T.singleton c

98100

99101escapeCommas :: PandocMonad m => TI m (Doc Text) -> TI m (Doc Text)

100102escapeCommas parser = do