02a125d · typeable/pandoc

git.iko.soy/typeable/pandoc

Use walk, walkM in place of bottomUp, bottomUpM when possible.

They are significantly faster.

commit02a125d0aa8becd258c99b27c5e30116f0cbacb4 authorJohn MacFarlane <fiddlosopher@gmail.com> dateSat 10 Aug 2013, 18:45 (13 years ago) parent9152fa1

11 files changed+31−27

compare with master

unifiedsplit

src/Text/Pandoc/PDF.hs+2−2 src/Text/Pandoc/Readers/LaTeX.hs+2−2 src/Text/Pandoc/Readers/MediaWiki.hs+2−2 src/Text/Pandoc/Shared.hs+1−1 src/Text/Pandoc/Writers/Docx.hs+6−5 src/Text/Pandoc/Writers/EPUB.hs+3−3 src/Text/Pandoc/Writers/FB2.hs+6−2 src/Text/Pandoc/Writers/LaTeX.hs+1−2 src/Text/Pandoc/Writers/Markdown.hs+4−4 src/Text/Pandoc/Writers/ODT.hs+2−2 src/Text/Pandoc/Writers/RTF.hs+2−2

src/Text/Pandoc/PDF.hs+2−2

@@ -44,7 +44,7 @@ import Data.List (isInfixOf)

4444import qualified Data.ByteString.Base64 as B64

4545import qualified Text.Pandoc.UTF8 as UTF8

4646import Text.Pandoc.Definition

47−import Text.Pandoc.Generic (bottomUpM)

47+import Text.Pandoc.Walk (walkM)

4848import Text.Pandoc.Shared (fetchItem, warn)

4949import Text.Pandoc.Options (WriterOptions(..))

5050import Text.Pandoc.MIME (extensionFromMimeType)

@@ -73,7 +73,7 @@ handleImages :: String -- ^ source directory/base URL

7373 -> FilePath -- ^ temp dir to store images

7474 -> Pandoc -- ^ document

7575 -> IO Pandoc

76−handleImages baseURL tmpdir = bottomUpM (handleImage' baseURL tmpdir)

76+handleImages baseURL tmpdir = walkM (handleImage' baseURL tmpdir)

7777

7878handleImage' :: String

7979 -> FilePath

src/Text/Pandoc/Readers/LaTeX.hs+2−2

@@ -35,7 +35,7 @@ module Text.Pandoc.Readers.LaTeX ( readLaTeX,

3535 ) where

3636

3737import Text.Pandoc.Definition

38−import Text.Pandoc.Generic

38+import Text.Pandoc.Walk

3939import Text.Pandoc.Shared

4040import Text.Pandoc.Options

4141import Text.Pandoc.Biblio (processBiblio)

@@ -815,7 +815,7 @@ keyvals :: LP [(String, String)]

815815keyvals = try $ char '[' *> manyTill keyval (char ']')

816816

817817alltt :: String -> LP Blocks

818−alltt t = bottomUp strToCode <$> parseFromString blocks

818+alltt t = walk strToCode <$> parseFromString blocks

819819 (substitute " " "\\ " $ substitute "%" "\\%" $

820820 concat $ intersperse "\\\\\n" $ lines t)

821821 where strToCode (Str s) = Code nullAttr s

src/Text/Pandoc/Readers/MediaWiki.hs+2−2

@@ -42,7 +42,7 @@ import Text.Pandoc.Options

4242import Text.Pandoc.Readers.HTML ( htmlTag, isBlockTag, isCommentTag )

4343import Text.Pandoc.XML ( fromEntities )

4444import Text.Pandoc.Parsing hiding ( nested )

45−import Text.Pandoc.Generic ( bottomUp )

45+import Text.Pandoc.Walk ( walk )

4646import Text.Pandoc.Shared ( stripTrailingNewlines, safeRead )

4747import Data.Monoid (mconcat, mempty)

4848import Control.Applicative ((<$>), (<*), (*>), (<$))

@@ -342,7 +342,7 @@ preformatted = try $ do

342342 spacesStr _ = False

343343 if F.all spacesStr contents

344344 then return mempty

345− else return $ B.para $ bottomUp strToCode contents

345+ else return $ B.para $ walk strToCode contents

346346

347347header :: MWParser Blocks

348348header = try $ do

src/Text/Pandoc/Shared.hs+1−1

@@ -518,7 +518,7 @@ isHeaderBlock _ = False

518518

519519-- | Shift header levels up or down.

520520headerShift :: Int -> Pandoc -> Pandoc

521−headerShift n = bottomUp shift

521+headerShift n = walk shift

522522 where shift :: Block -> Block

523523 shift (Header level attr inner) = Header (level + n) attr inner

524524 shift x = x

src/Text/Pandoc/Writers/Docx.hs+6−5

@@ -45,6 +45,7 @@ import Text.Pandoc.Shared hiding (Element)

4545import Text.Pandoc.Options

4646import Text.Pandoc.Readers.TeXMath

4747import Text.Pandoc.Highlighting ( highlight )

48+import Text.Pandoc.Walk

4849import Text.Highlighting.Kate.Types ()

4950import Text.XML.Light

5051import Text.TeXMath

@@ -108,7 +109,7 @@ writeDocx :: WriterOptions -- ^ Writer options

108109 -> IO BL.ByteString

109110writeDocx opts doc@(Pandoc meta _) = do

110111 let datadir = writerUserDataDir opts

111− let doc' = bottomUp (concatMap fixDisplayMath) doc

112+ let doc' = walk fixDisplayMath doc

112113 refArchive <- liftM (toArchive . toLazy) $

113114 case writerReferenceDocx opts of

114115 Just f -> B.readFile f

@@ -810,17 +811,17 @@ stripLeadingTrailingSpace = go . reverse . go . reverse

810811 where go (Space:xs) = xs

811812 go xs = xs

812813

813−fixDisplayMath :: Block -> [Block]

814+fixDisplayMath :: Block -> Block

814815fixDisplayMath (Plain lst)

815816 | any isDisplayMath lst && not (all isDisplayMath lst) =

816817 -- chop into several paragraphs so each displaymath is its own

817− map (Plain . stripLeadingTrailingSpace) $

818+ Div ("",["math"],[]) $ map (Plain . stripLeadingTrailingSpace) $

818819 groupBy (\x y -> (isDisplayMath x && isDisplayMath y) ||

819820 not (isDisplayMath x || isDisplayMath y)) lst

820821fixDisplayMath (Para lst)

821822 | any isDisplayMath lst && not (all isDisplayMath lst) =

822823 -- chop into several paragraphs so each displaymath is its own

823− map (Para . stripLeadingTrailingSpace) $

824+ Div ("",["math"],[]) $ map (Para . stripLeadingTrailingSpace) $

824825 groupBy (\x y -> (isDisplayMath x && isDisplayMath y) ||

825826 not (isDisplayMath x || isDisplayMath y)) lst

826−fixDisplayMath x = [x]

827+fixDisplayMath x = x

src/Text/Pandoc/Writers/EPUB.hs+3−3

@@ -48,7 +48,7 @@ import qualified Text.Pandoc.Shared as Shared

4848import Text.Pandoc.Builder (fromList, setMeta)

4949import Text.Pandoc.Options

5050import Text.Pandoc.Definition

51−import Text.Pandoc.Generic

51+import Text.Pandoc.Walk

5252import Control.Monad.State

5353import Text.XML.Light hiding (ppTopElement)

5454import Text.Pandoc.UUID

@@ -116,7 +116,7 @@ writeEPUB opts doc@(Pandoc meta _) = do

116116

117117 -- handle pictures

118118 picsRef <- newIORef []

119− Pandoc _ blocks <- bottomUpM

119+ Pandoc _ blocks <- walkM

120120 (transformInline opts' sourceDir picsRef) doc

121121 pics <- readIORef picsRef

122122 let readPicEntry entries (oldsrc, newsrc) = do

@@ -520,7 +520,7 @@ correlateRefs chapterHeaderLevel bs =

520520-- Replace internal link references using the table produced

521521-- by correlateRefs.

522522replaceRefs :: [(String,String)] -> [Block] -> [Block]

523−replaceRefs refTable = bottomUp replaceOneRef

523+replaceRefs refTable = walk replaceOneRef

524524 where replaceOneRef x@(Link lab ('#':xs,tit)) =

525525 case lookup xs refTable of

526526 Just url -> Link lab (url,tit)

src/Text/Pandoc/Writers/FB2.hs+6−2

@@ -45,7 +45,7 @@ import qualified Text.XML.Light.Cursor as XC

4545import Text.Pandoc.Definition

4646import Text.Pandoc.Options (WriterOptions(..), HTMLMathMethod(..), def)

4747import Text.Pandoc.Shared (orderedListMarkers)

48−import Text.Pandoc.Generic (bottomUp)

48+import Text.Pandoc.Walk

4949

5050-- | Data to be written at the end of the document:

5151-- (foot)notes, URLs, references, images.

@@ -423,6 +423,10 @@ indent = indentBlock

423423 indentLines ins = let lns = split isLineBreak ins :: [[Inline]]

424424 in intercalate [LineBreak] $ map ((Str spacer):) lns

425425

426+capitalize :: Inline -> Inline

427+capitalize (Str xs) = Str $ map toUpper xs

428+capitalize x = x

429+

426430-- | Convert a Pandoc's Inline element to FictionBook XML representation.

427431toXml :: Inline -> FBM [Content]

428432toXml (Str s) = return [txt s]

@@ -432,7 +436,7 @@ toXml (Strong ss) = list `liftM` wrap "strong" ss

432436toXml (Strikeout ss) = list `liftM` wrap "strikethrough" ss

433437toXml (Superscript ss) = list `liftM` wrap "sup" ss

434438toXml (Subscript ss) = list `liftM` wrap "sub" ss

435−toXml (SmallCaps ss) = cMapM toXml $ bottomUp (map toUpper) ss

439+toXml (SmallCaps ss) = cMapM toXml $ walk capitalize ss

436440toXml (Quoted SingleQuote ss) = do -- FIXME: should be language-specific

437441 inner <- cMapM toXml ss

438442 return $ [txt "‘"] ++ inner ++ [txt "’"]

src/Text/Pandoc/Writers/LaTeX.hs+1−2

@@ -31,7 +31,6 @@ Conversion of 'Pandoc' format into LaTeX.

3131module Text.Pandoc.Writers.LaTeX ( writeLaTeX ) where

3232import Text.Pandoc.Definition

3333import Text.Pandoc.Walk

34−import Text.Pandoc.Generic

3534import Text.Pandoc.Shared

3635import Text.Pandoc.Writers.Shared

3736import Text.Pandoc.Options

@@ -498,7 +497,7 @@ sectionHeader unnumbered ref level lst = do

498497 txt <- inlineListToLaTeX lst

499498 let noNote (Note _) = Str ""

500499 noNote x = x

501− let lstNoNotes = bottomUp noNote lst

500+ let lstNoNotes = walk noNote lst

502501 let star = if unnumbered then text "*" else empty

503502 -- footnotes in sections don't work unless you specify an optional

504503 -- argument: \section[mysec]{mysec\footnote{blah}}

src/Text/Pandoc/Writers/Markdown.hs+4−4

@@ -32,7 +32,7 @@ Markdown: <http://daringfireball.net/projects/markdown/>

3232-}

3333module Text.Pandoc.Writers.Markdown (writeMarkdown, writePlain) where

3434import Text.Pandoc.Definition

35−import Text.Pandoc.Generic

35+import Text.Pandoc.Walk

3636import Text.Pandoc.Templates (renderTemplate')

3737import Text.Pandoc.Shared

3838import Text.Pandoc.Writers.Shared

@@ -82,7 +82,7 @@ writePlain opts document =

8282 where document' = plainify document

8383

8484plainify :: Pandoc -> Pandoc

85−plainify = bottomUp go

85+plainify = walk go

8686 where go :: Inline -> Inline

8787 go (Emph xs) = SmallCaps xs

8888 go (Strong xs) = SmallCaps xs

@@ -643,13 +643,13 @@ inlineToMarkdown opts (Strikeout lst) = do

643643 then "~~" <> contents <> "~~"

644644 else "<s>" <> contents <> "</s>"

645645inlineToMarkdown opts (Superscript lst) = do

646− let lst' = bottomUp escapeSpaces lst

646+ let lst' = walk escapeSpaces lst

647647 contents <- inlineListToMarkdown opts lst'

648648 return $ if isEnabled Ext_superscript opts

649649 then "^" <> contents <> "^"

650650 else "<sup>" <> contents <> "</sup>"

651651inlineToMarkdown opts (Subscript lst) = do

652− let lst' = bottomUp escapeSpaces lst

652+ let lst' = walk escapeSpaces lst

653653 contents <- inlineListToMarkdown opts lst'

654654 return $ if isEnabled Ext_subscript opts

655655 then "~" <> contents <> "~"

src/Text/Pandoc/Writers/ODT.hs+2−2

@@ -39,7 +39,7 @@ import Text.Pandoc.Shared ( stringify, readDataFile, fetchItem, warn )

3939import Text.Pandoc.ImageSize ( imageSize, sizeInPoints )

4040import Text.Pandoc.MIME ( getMimeType )

4141import Text.Pandoc.Definition

42−import Text.Pandoc.Generic

42+import Text.Pandoc.Walk

4343import Text.Pandoc.Writers.OpenDocument ( writeOpenDocument )

4444import Control.Monad (liftM)

4545import Text.Pandoc.XML

@@ -63,7 +63,7 @@ writeODT opts doc@(Pandoc meta _) = do

6363 -- handle pictures

6464 picEntriesRef <- newIORef ([] :: [Entry])

6565 let sourceDir = writerSourceDirectory opts

66− doc' <- bottomUpM (transformPic sourceDir picEntriesRef) doc

66+ doc' <- walkM (transformPic sourceDir picEntriesRef) doc

6767 let newContents = writeOpenDocument opts{writerWrapText = False} doc'

6868 epochtime <- floor `fmap` getPOSIXTime

6969 let contentEntry = toEntry "content.xml" epochtime $ fromStringLazy newContents

src/Text/Pandoc/Writers/RTF.hs+2−2

@@ -34,7 +34,7 @@ import Text.Pandoc.Shared

3434import Text.Pandoc.Writers.Shared

3535import Text.Pandoc.Readers.TeXMath

3636import Text.Pandoc.Templates (renderTemplate')

37−import Text.Pandoc.Generic (bottomUpM)

37+import Text.Pandoc.Walk

3838import Data.List ( isSuffixOf, intercalate )

3939import Data.Char ( ord, chr, isDigit, toLower )

4040import System.FilePath ( takeExtension )

@@ -70,7 +70,7 @@ rtfEmbedImage x = return x

7070-- images embedded as encoded binary data.

7171writeRTFWithEmbeddedImages :: WriterOptions -> Pandoc -> IO String

7272writeRTFWithEmbeddedImages options doc =

73− writeRTF options `fmap` bottomUpM rtfEmbedImage doc

73+ writeRTF options `fmap` walkM rtfEmbedImage doc

7474

7575-- | Convert Pandoc to a string in rich text format.

7676writeRTF :: WriterOptions -> Pandoc -> String