780079a · typeable/pandoc

git.iko.soy/typeable/pandoc

Merge branch 'lierdakil-docx-reader-styles'

commit780079aaec6b82d3d235e20afda06cdfc8b486d5 authorJohn MacFarlane <jgm@berkeley.edu> dateSat 21 Sep 2019, 11:39 (7 years ago) parente3a6648 9dbfd23

17 files changed+407−312

compare with master

unifiedsplit

.gitignore+1 MANUAL.txt+3−3 src/Text/Pandoc/Readers/Docx.hs+99−113 src/Text/Pandoc/Readers/Docx/Lists.hs+15−10 src/Text/Pandoc/Readers/Docx/Parse.hs+202−113 src/Text/Pandoc/Writers/Docx.hs+61−66 test/Tests/Readers/Docx.hs+9 test/docx/0_level_headers.native+2−2 test/docx/adjacent_codeblocks.docx test/docx/compact-style-removal.docx test/docx/compact-style-removal.native+5 test/docx/custom-style-with-styles.native+3−3 test/docx/lists-compact.docx test/docx/lists-compact.native+5 test/docx/lists.docx test/docx/lists.native+1−1 test/docx/nested_anchors_in_header.native+1−1

.gitignore+1

@@ -23,3 +23,4 @@ windows/*.wixobj

2323data/reference.docx

2424data/reference.odt

2525.stack-work

26+cabal.project.local

MANUAL.txt+3−3

@@ -5321,17 +5321,17 @@ And with the extension:

53215321

53225322 $ pandoc test/docx/custom-style-reference.docx -f docx+styles -t markdown

53235323

5324− ::: {custom-style="FirstParagraph"}

5324+ ::: {custom-style="First Paragraph"}

53255325 This is some text.

53265326 :::

53275327

5328− ::: {custom-style="BodyText"}

5328+ ::: {custom-style="Body Text"}

53295329 This is text with an [emphasized]{custom-style="Emphatic"} text style.

53305330 And this is text with a [strengthened]{custom-style="Strengthened"}

53315331 text style.

53325332 :::

53335333

5334− ::: {custom-style="MyBlockStyle"}

5334+ ::: {custom-style="My Block Style"}

53355335 > Here is a styled paragraph that inherits from Block Text.

53365336 :::

53375337

src/Text/Pandoc/Readers/Docx.hs+99−113

@@ -2,6 +2,8 @@

22{-# LANGUAGE CPP #-}

33{-# LANGUAGE OverloadedStrings #-}

44{-# LANGUAGE PatternGuards #-}

5+{-# LANGUAGE MultiWayIf #-}

6+{-# LANGUAGE FlexibleContexts #-}

57{- |

68 Module : Text.Pandoc.Readers.Docx

@@ -64,6 +66,7 @@ import Control.Monad.State.Strict

6466import qualified Data.ByteString.Lazy as B

6567import Data.Default (Default)

6668import Data.List (delete, intersect)

69+import Data.Char (isSpace)

6770import qualified Data.Map as M

6871import Data.Maybe (isJust, fromMaybe)

6972import Data.Sequence (ViewL (..), viewl)

@@ -129,16 +132,16 @@ instance Default DEnv where

129132type DocxContext m = ReaderT DEnv (StateT DState m)

130133

131134evalDocxContext :: PandocMonad m => DocxContext m a -> DEnv -> DState -> m a

132−evalDocxContext ctx env st = flip evalStateT st $flip runReaderT env ctx

135+evalDocxContext ctx env st = flip evalStateT st $ runReaderT ctx env

133136

134137-- This is empty, but we put it in for future-proofing.

135−spansToKeep :: [String]

138+spansToKeep :: [CharStyleName]

136139spansToKeep = []

137140

138−divsToKeep :: [String]

139−divsToKeep = ["list-item", "Definition", "DefinitionTerm"]

141+divsToKeep :: [ParaStyleName]

142+divsToKeep = ["Definition", "Definition Term"]

140143

141−metaStyles :: M.Map String String

144+metaStyles :: M.Map ParaStyleName String

142145metaStyles = M.fromList [ ("Title", "title")

143146 , ("Subtitle", "subtitle")

144147 , ("Author", "author")

@@ -150,7 +153,7 @@ sepBodyParts = span (\bp -> isMetaPar bp || isEmptyPar bp)

150153

151154isMetaPar :: BodyPart -> Bool

152155isMetaPar (Paragraph pPr _) =

153− not $ null $ intersect (pStyle pPr) (M.keys metaStyles)

156+ not $ null $ intersect (getStyleNames $ pStyle pPr) (M.keys metaStyles)

154157isMetaPar _ = False

155158

156159isEmptyPar :: BodyPart -> Bool

@@ -167,7 +170,7 @@ bodyPartsToMeta' :: PandocMonad m => [BodyPart] -> DocxContext m (M.Map String M

167170bodyPartsToMeta' [] = return M.empty

168171bodyPartsToMeta' (bp : bps)

169172 | (Paragraph pPr parParts) <- bp

170− , (c : _)<- (pStyle pPr) `intersect` (M.keys metaStyles)

173+ , (c : _)<- getStyleNames (pStyle pPr) `intersect` M.keys metaStyles

171174 , (Just metaField) <- M.lookup c metaStyles = do

172175 inlines <- smushInlines <$> mapM parPartToInlines parParts

173176 remaining <- bodyPartsToMeta' bps

@@ -197,11 +200,29 @@ fixAuthors (MetaBlocks blks) =

197200 g _ = MetaInlines []

198201fixAuthors mv = mv

199202

200−codeStyles :: [String]

201−codeStyles = ["VerbatimChar"]

203+isInheritedFromStyles :: (Eq (StyleName s), HasStyleName s, HasParentStyle s) => [StyleName s] -> s -> Bool

204+isInheritedFromStyles names sty

205+ | getStyleName sty `elem` names = True

206+ | Just psty <- getParentStyle sty = isInheritedFromStyles names psty

207+ | otherwise = False

202208

203−codeDivs :: [String]

204−codeDivs = ["SourceCode"]

209+hasStylesInheritedFrom :: [ParaStyleName] -> ParagraphStyle -> Bool

210+hasStylesInheritedFrom ns s = any (isInheritedFromStyles ns) $ pStyle s

211+

212+removeStyleNamed :: ParaStyleName -> ParagraphStyle -> ParagraphStyle

213+removeStyleNamed sn ps = ps{pStyle = filter (\psd -> getStyleName psd /= sn) $ pStyle ps}

214+

215+isCodeCharStyle :: CharStyle -> Bool

216+isCodeCharStyle = isInheritedFromStyles ["Verbatim Char"]

217+

218+isCodeDiv :: ParagraphStyle -> Bool

219+isCodeDiv = hasStylesInheritedFrom ["Source Code"]

220+

221+isBlockQuote :: ParStyle -> Bool

222+isBlockQuote =

223+ isInheritedFromStyles [

224+ "Quote", "Block Text", "Block Quote", "Block Quotation"

225+ ]

205226

206227runElemToInlines :: RunElem -> Inlines

207228runElemToInlines (TextRun s) = text s

@@ -227,57 +248,31 @@ parPartToString (InternalHyperLink _ runs) = concatMap runToString runs

227248parPartToString (ExternalHyperLink _ runs) = concatMap runToString runs

228249parPartToString _ = ""

229250

230−blacklistedCharStyles :: [String]

251+blacklistedCharStyles :: [CharStyleName]

231252blacklistedCharStyles = ["Hyperlink"]

232253

233254resolveDependentRunStyle :: PandocMonad m => RunStyle -> DocxContext m RunStyle

234255resolveDependentRunStyle rPr

235− | Just (s, _) <- rStyle rPr, s `elem` blacklistedCharStyles =

256+ | Just s <- rParentStyle rPr

257+ , getStyleName s `elem` blacklistedCharStyles =

236258 return rPr

237− | Just (_, cs) <- rStyle rPr = do

259+ | Just s <- rParentStyle rPr = do

238260 opts <- asks docxOptions

239261 if isEnabled Ext_styles opts

240262 then return rPr

241− else do rPr' <- resolveDependentRunStyle cs

242− return $

243− RunStyle { isBold = case isBold rPr of

244− Just bool -> Just bool

245− Nothing -> isBold rPr'

246− , isItalic = case isItalic rPr of

247− Just bool -> Just bool

248− Nothing -> isItalic rPr'

249− , isSmallCaps = case isSmallCaps rPr of

250− Just bool -> Just bool

251− Nothing -> isSmallCaps rPr'

252− , isStrike = case isStrike rPr of

253− Just bool -> Just bool

254− Nothing -> isStrike rPr'

255− , isRTL = case isRTL rPr of

256− Just bool -> Just bool

257− Nothing -> isRTL rPr'

258− , rVertAlign = case rVertAlign rPr of

259− Just valign -> Just valign

260− Nothing -> rVertAlign rPr'

261− , rUnderline = case rUnderline rPr of

262− Just ulstyle -> Just ulstyle

263− Nothing -> rUnderline rPr'

264− , rStyle = rStyle rPr

265− }

263+ else leftBiasedMergeRunStyle rPr <$> resolveDependentRunStyle (cStyleData s)

266264 | otherwise = return rPr

267265

268266runStyleToTransform :: PandocMonad m => RunStyle -> DocxContext m (Inlines -> Inlines)

269267runStyleToTransform rPr

270− | Just (s, _) <- rStyle rPr

271− , s `elem` spansToKeep = do

272− transform <- runStyleToTransform rPr{rStyle = Nothing}

273− return $ spanWith ("", [s], []) . transform

274− | Just (s, _) <- rStyle rPr = do

275− opts <- asks docxOptions

276− let extraInfo = if isEnabled Ext_styles opts

277− then spanWith ("", [], [("custom-style", s)])

278− else id

279− transform <- runStyleToTransform rPr{rStyle = Nothing}

280− return $ extraInfo . transform

268+ | Just sn <- getStyleName <$> rParentStyle rPr

269+ , sn `elem` spansToKeep = do

270+ transform <- runStyleToTransform rPr{rParentStyle = Nothing}

271+ return $ spanWith ("", [normalizeToClassName sn], []) . transform

272+ | Just s <- rParentStyle rPr = do

273+ ei <- extraInfo spanWith s

274+ transform <- runStyleToTransform rPr{rParentStyle = Nothing}

275+ return $ ei . transform

281276 | Just True <- isItalic rPr = do

282277 transform <- runStyleToTransform rPr{isItalic = Nothing}

283278 return $ emph . transform

@@ -309,8 +304,7 @@ runStyleToTransform rPr

309304

310305runToInlines :: PandocMonad m => Run -> DocxContext m Inlines

311306runToInlines (Run rs runElems)

312− | Just (s, _) <- rStyle rs

313− , s `elem` codeStyles = do

307+ | maybe False isCodeCharStyle $ rParentStyle rs = do

314308 rPr <- resolveDependentRunStyle rs

315309 let codeString = code $ concatMap runElemToString runElems

316310 return $ case rVertAlign rPr of

@@ -525,55 +519,49 @@ trimSps (Many ils) = Many $ Seq.dropWhileL isSp $Seq.dropWhileR isSp ils

525519 isSp LineBreak = True

526520 isSp _ = False

527521

522+extraInfo :: (Eq (StyleName a), PandocMonad m, HasStyleName a)

523+ => (Attr -> i -> i) -> a -> DocxContext m (i -> i)

524+extraInfo f s = do

525+ opts <- asks docxOptions

526+ return $ if | isEnabled Ext_styles opts

527+ -> f ("", [], [("custom-style", fromStyleName $ getStyleName s)])

528+ | otherwise -> id

529+

528530parStyleToTransform :: PandocMonad m => ParagraphStyle -> DocxContext m (Blocks -> Blocks)

529531parStyleToTransform pPr

530532 | (c:cs) <- pStyle pPr

531− , c `elem` divsToKeep = do

533+ , getStyleName c `elem` divsToKeep = do

532534 let pPr' = pPr { pStyle = cs }

533535 transform <- parStyleToTransform pPr'

534− return $ divWith ("", [c], []) . transform

536+ return $ divWith ("", [normalizeToClassName $ getStyleName c], []) . transform

535537 | (c:cs) <- pStyle pPr,

536− c `elem` listParagraphDivs = do

538+ getStyleName c `elem` listParagraphStyles = do

537539 let pPr' = pPr { pStyle = cs, indentation = Nothing}

538540 transform <- parStyleToTransform pPr'

539− return $ divWith ("", [c], []) . transform

540− | (c:cs) <- pStyle pPr

541− , Just True <- pBlockQuote pPr = do

542− opts <- asks docxOptions

543− let pPr' = pPr { pStyle = cs }

544− transform <- parStyleToTransform pPr'

545− let extraInfo = if isEnabled Ext_styles opts

546− then divWith ("", [], [("custom-style", c)])

547− else id

548− return $ extraInfo . blockQuote . transform

541+ return $ divWith ("", [normalizeToClassName $ getStyleName c], []) . transform

549542 | (c:cs) <- pStyle pPr = do

550− opts <- asks docxOptions

551− let pPr' = pPr { pStyle = cs}

543+ let pPr' = pPr { pStyle = cs }

552544 transform <- parStyleToTransform pPr'

553− let extraInfo = if isEnabled Ext_styles opts

554− then divWith ("", [], [("custom-style", c)])

555− else id

556− return $ extraInfo . transform

545+ ei <- extraInfo divWith c

546+ return $ ei . (if isBlockQuote c then blockQuote else id) . transform

557547 | null (pStyle pPr)

558− , Just left <- indentation pPr >>= leftParIndent

559− , Just hang <- indentation pPr >>= hangingParIndent = do

548+ , Just left <- indentation pPr >>= leftParIndent = do

560549 let pPr' = pPr { indentation = Nothing }

550+ hang = fromMaybe 0 $ indentation pPr >>= hangingParIndent

561551 transform <- parStyleToTransform pPr'

562− return $ case (left - hang) > 0 of

563− True -> blockQuote . transform

564− False -> transform

565− | null (pStyle pPr),

566− Just left <- indentation pPr >>= leftParIndent = do

567− let pPr' = pPr { indentation = Nothing }

568− transform <- parStyleToTransform pPr'

569− return $ case left > 0 of

570− True -> blockQuote . transform

571− False -> transform

552+ return $ if (left - hang) > 0

553+ then blockQuote . transform

554+ else transform

572555parStyleToTransform _ = return id

573556

557+normalizeToClassName :: (FromStyleName a) => a -> String

558+normalizeToClassName = map go . fromStyleName

559+ where go c | isSpace c = '-'

560+ | otherwise = c

561+

574562bodyPartToBlocks :: PandocMonad m => BodyPart -> DocxContext m Blocks

575563bodyPartToBlocks (Paragraph pPr parparts)

576− | not $ null $ codeDivs `intersect` (pStyle pPr) = do

564+ | isCodeDiv pPr = do

577565 transform <- parStyleToTransform pPr

578566 return $

579567 transform $

@@ -583,62 +571,60 @@ bodyPartToBlocks (Paragraph pPr parparts)

583571 ils <-local (\s-> s{docxInHeaderBlock=True})

584572 (smushInlines <$> mapM parPartToInlines parparts)

585573 makeHeaderAnchor $

586− headerWith ("", delete style (pStyle pPr), []) n ils

574+ headerWith ("", map normalizeToClassName . delete style $ getStyleNames (pStyle pPr), []) n ils

587575 | otherwise = do

588− ils <- (trimSps . smushInlines) <$> mapM parPartToInlines parparts

576+ ils <- trimSps . smushInlines <$> mapM parPartToInlines parparts

589577 prevParaIls <- gets docxPrevPara

590578 dropIls <- gets docxDropCap

591579 let ils' = dropIls <> ils

592− if dropCap pPr

580+ let (paraOrPlain, pPr')

581+ | hasStylesInheritedFrom ["Compact"] pPr = (plain, removeStyleNamed "Compact" pPr)

582+ | otherwise = (para, pPr)

583+ if dropCap pPr'

593584 then do modify $ \s -> s { docxDropCap = ils' }

594585 return mempty

595586 else do modify $ \s -> s { docxDropCap = mempty }

596587 let ils'' = prevParaIls <>

597588 (if isNull prevParaIls then mempty else space) <>

598589 ils'

590+ handleInsertion = do

591+ modify $ \s -> s {docxPrevPara = mempty}

592+ transform <- parStyleToTransform pPr'

593+ return $ transform $ paraOrPlain ils''

599594 opts <- asks docxOptions

600− case () of

601−

602− _ | isNull ils'' && not (isEnabled Ext_empty_paragraphs opts) ->

595+ if | isNull ils'' && not (isEnabled Ext_empty_paragraphs opts) ->

603596 return mempty

604− _ | Just (TrackedChange Insertion _) <- pChange pPr

605− , AcceptChanges <- readerTrackChanges opts -> do

606− modify $ \s -> s {docxPrevPara = mempty}

607− transform <- parStyleToTransform pPr

608− return $ transform $ para ils''

609− _ | Just (TrackedChange Insertion _) <- pChange pPr

597+ | Just (TrackedChange Insertion _) <- pChange pPr'

598+ , AcceptChanges <- readerTrackChanges opts ->

599+ handleInsertion

600+ | Just (TrackedChange Insertion _) <- pChange pPr'

610601 , RejectChanges <- readerTrackChanges opts -> do

611602 modify $ \s -> s {docxPrevPara = ils''}

612603 return mempty

613− _ | Just (TrackedChange Insertion cInfo) <- pChange pPr

604+ | Just (TrackedChange Insertion cInfo) <- pChange pPr'

614605 , AllChanges <- readerTrackChanges opts

615606 , ChangeInfo _ cAuthor cDate <- cInfo -> do

616607 let attr = ("", ["paragraph-insertion"], [("author", cAuthor), ("date", cDate)])

617608 insertMark = spanWith attr mempty

618− transform <- parStyleToTransform pPr

609+ transform <- parStyleToTransform pPr'

619610 return $ transform $

620− para $ ils'' <> insertMark

621− _ | Just (TrackedChange Deletion _) <- pChange pPr

611+ paraOrPlain $ ils'' <> insertMark

612+ | Just (TrackedChange Deletion _) <- pChange pPr'

622613 , AcceptChanges <- readerTrackChanges opts -> do

623614 modify $ \s -> s {docxPrevPara = ils''}

624615 return mempty

625− _ | Just (TrackedChange Deletion _) <- pChange pPr

626− , RejectChanges <- readerTrackChanges opts -> do

627− modify $ \s -> s {docxPrevPara = mempty}

628− transform <- parStyleToTransform pPr

629− return $ transform $ para ils''

630− _ | Just (TrackedChange Deletion cInfo) <- pChange pPr

616+ | Just (TrackedChange Deletion _) <- pChange pPr'

617+ , RejectChanges <- readerTrackChanges opts ->

618+ handleInsertion

619+ | Just (TrackedChange Deletion cInfo) <- pChange pPr'

631620 , AllChanges <- readerTrackChanges opts

632621 , ChangeInfo _ cAuthor cDate <- cInfo -> do

633622 let attr = ("", ["paragraph-deletion"], [("author", cAuthor), ("date", cDate)])

634623 insertMark = spanWith attr mempty

635− transform <- parStyleToTransform pPr

624+ transform <- parStyleToTransform pPr'

636625 return $ transform $

637− para $ ils'' <> insertMark

638− _ | otherwise -> do

639− modify $ \s -> s {docxPrevPara = mempty}

640− transform <- parStyleToTransform pPr

641− return $ transform $ para ils''

626+ paraOrPlain $ ils'' <> insertMark

627+ | otherwise -> handleInsertion

642628bodyPartToBlocks (ListItem pPr numId lvl (Just levelInfo) parparts) = do

643629 -- We check whether this current numId has previously been used,

644630 -- since Docx expects us to pick up where we left off.

@@ -658,7 +644,7 @@ bodyPartToBlocks (ListItem pPr numId lvl (Just levelInfo) parparts) = do

658644 blks <- bodyPartToBlocks (Paragraph pPr parparts)

659645 return $ divWith ("", ["list-item"], kvs) blks

660646bodyPartToBlocks (ListItem pPr _ _ _ parparts) =

661− let pPr' = pPr {pStyle = "ListParagraph": pStyle pPr}

647+ let pPr' = pPr {pStyle = constructBogusParStyleData "list-paragraph": pStyle pPr}

662648 in

663649 bodyPartToBlocks $ Paragraph pPr' parparts

664650bodyPartToBlocks (Tbl _ _ _ []) =

src/Text/Pandoc/Readers/Docx/Lists.hs+15−10

@@ -1,4 +1,5 @@

11{-# LANGUAGE NoImplicitPrelude #-}

2+{-# LANGUAGE OverloadedStrings #-}

23{- |

34 Module : Text.Pandoc.Readers.Docx.Lists

@@ -14,13 +15,16 @@ Functions for converting flat docx paragraphs into nested lists.

1415module Text.Pandoc.Readers.Docx.Lists ( blocksToBullets

1516 , blocksToDefinitions

1617 , listParagraphDivs

18+ , listParagraphStyles

1719 ) where

1820

1921import Prelude

2022import Data.List

2123import Data.Maybe

24+import Data.String (fromString)

2225import Text.Pandoc.Generic (bottomUp)

2326import Text.Pandoc.JSON

27+import Text.Pandoc.Readers.Docx.Parse (ParaStyleName)

2428import Text.Pandoc.Shared (trim, safeRead)

2529

2630isListItem :: Block -> Bool

@@ -79,7 +83,10 @@ getListType b@(Div (_, _, kvs) _) | isListItem b =

7983getListType _ = Nothing

8084

8185listParagraphDivs :: [String]

82−listParagraphDivs = ["ListParagraph"]

86+listParagraphDivs = ["list-paragraph"]

87+

88+listParagraphStyles :: [ParaStyleName]

89+listParagraphStyles = map fromString listParagraphDivs

8390

8491-- This is a first stab at going through and attaching meaning to list

8592-- paragraphs, without an item marker, following a list item. We

@@ -160,7 +167,7 @@ blocksToDefinitions' defAcc acc [] =

160167 reverse $ DefinitionList (reverse defAcc) : acc

161168blocksToDefinitions' defAcc acc

162169 (Div (_, classes1, _) blks1 : Div (ident2, classes2, kvs2) blks2 : blks)

163− | "DefinitionTerm" `elem` classes1 && "Definition" `elem` classes2 =

170+ | "Definition-Term" `elem` classes1 && "Definition" `elem` classes2 =

164171 let remainingAttr2 = (ident2, delete "Definition" classes2, kvs2)

165172 pair = if remainingAttr2 == ("", [], []) then (concatMap plainParaInlines blks1, [blks2]) else (concatMap plainParaInlines blks1, [[Div remainingAttr2 blks2]])

166173 in

@@ -169,12 +176,12 @@ blocksToDefinitions' ((defTerm, defItems):defs) acc

169176 (Div (ident2, classes2, kvs2) blks2 : blks)

170177 | "Definition" `elem` classes2 =

171178 let remainingAttr2 = (ident2, delete "Definition" classes2, kvs2)

172− defItems2 = case remainingAttr2 == ("", [], []) of

173− True -> blks2

174− False -> [Div remainingAttr2 blks2]

175− defAcc' = case null defItems of

176− True -> (defTerm, [defItems2]) : defs

177− False -> (defTerm, init defItems ++ [last defItems ++ defItems2]) : defs

179+ defItems2 = if remainingAttr2 == ("", [], [])

180+ then blks2

181+ else [Div remainingAttr2 blks2]

182+ defAcc' = if null defItems

183+ then (defTerm, [defItems2]) : defs

184+ else (defTerm, init defItems ++ [last defItems ++ defItems2]) : defs

178185 in

179186 blocksToDefinitions' defAcc' acc blks

180187blocksToDefinitions' [] acc (b:blks) =

@@ -198,7 +205,5 @@ removeListDivs' blk = [blk]

198205removeListDivs :: [Block] -> [Block]

199206removeListDivs = concatMap removeListDivs'

200207

201−

202−

203208blocksToDefinitions :: [Block] -> [Block]

204209blocksToDefinitions = blocksToDefinitions' [] []

src/Text/Pandoc/Readers/Docx/Parse.hs+202−113

@@ -1,7 +1,11 @@

11{-# LANGUAGE NoImplicitPrelude #-}

22{-# LANGUAGE FlexibleInstances #-}

3+{-# LANGUAGE FlexibleContexts #-}

34{-# LANGUAGE PatternGuards #-}

45{-# LANGUAGE ViewPatterns #-}

6+{-# LANGUAGE OverloadedStrings #-}

7+{-# LANGUAGE TypeFamilies #-}

8+{-# LANGUAGE GeneralizedNewtypeDeriving #-}

59{- |

610 Module : Text.Pandoc.Readers.Docx.Parse

@@ -31,6 +35,8 @@ module Text.Pandoc.Readers.Docx.Parse ( Docx(..)

3135 , VertAlign(..)

3236 , ParIndentation(..)

3337 , ParagraphStyle(..)

38+ , ParStyle

39+ , CharStyle(cStyleData)

3440 , Row(..)

3541 , Cell(..)

3642 , TrackedChange(..)

@@ -38,8 +44,17 @@ module Text.Pandoc.Readers.Docx.Parse ( Docx(..)

3844 , ChangeInfo(..)

3945 , FieldInfo(..)

4046 , Level(..)

47+ , ParaStyleName

48+ , CharStyleName

49+ , FromStyleName(..)

50+ , HasStyleName(..)

51+ , HasParentStyle(..)

4152 , archiveToDocx

4253 , archiveToDocxWithWarnings

54+ , getStyleNames

55+ , pHeading

56+ , constructBogusParStyleData

57+ , leftBiasedMergeRunStyle

4358 ) where

4459import Prelude

4560import Codec.Archive.Zip

@@ -49,10 +64,13 @@ import Control.Monad.Reader

4964import Control.Monad.State.Strict

5065import Data.Bits ((.|.))

5166import qualified Data.ByteString.Lazy as B

52−import Data.Char (chr, ord, readLitChar)

67+import Data.Char (chr, ord, readLitChar, toLower)

5368import Data.List

69+import Data.Function (on)

70+import Data.String (IsString(..))

5471import qualified Data.Map as M

5572import Data.Maybe

73+import Data.Coerce

5674import System.FilePath

5775import Text.Pandoc.Readers.Docx.Util

5876import Text.Pandoc.Readers.Docx.Fields

@@ -121,9 +139,9 @@ unwrap :: NameSpaces -> Content -> [Content]

121139unwrap ns (Elem element)

122140 | isElem ns "w" "sdt" element

123141 , Just sdtContent <- findChildByName ns "w" "sdtContent" element

124− = concatMap ((unwrap ns) . Elem) (elChildren sdtContent)

142+ = concatMap (unwrap ns . Elem) (elChildren sdtContent)

125143 | isElem ns "w" "smartTag" element

126− = concatMap ((unwrap ns) . Elem) (elChildren element)

144+ = concatMap (unwrap ns . Elem) (elChildren element)

127145unwrap _ content = [content]

128146

129147unwrapChild :: NameSpaces -> Content -> Content

@@ -149,24 +167,20 @@ walkDocument ns element =

149167 _ -> Nothing

150168

151169

152−data Docx = Docx Document

170+newtype Docx = Docx Document

153171 deriving Show

154172

155173data Document = Document NameSpaces Body

156174 deriving Show

157175

158−data Body = Body [BodyPart]

176+newtype Body = Body [BodyPart]

159177 deriving Show

160178

161179type Media = [(FilePath, B.ByteString)]

162180

163−type CharStyle = (String, RunStyle)

181+type CharStyleMap = M.Map CharStyleId CharStyle

164182

165−type ParStyle = (String, ParStyleData)

166−

167−type CharStyleMap = M.Map String RunStyle

168−

169−type ParStyleMap = M.Map String ParStyleData

183+type ParStyleMap = M.Map ParaStyleId ParStyle

170184

171185data Numbering = Numbering NameSpaces [Numb] [AbstractNumb]

172186 deriving Show

@@ -213,12 +227,9 @@ data ChangeInfo = ChangeInfo ChangeId Author ChangeDate

213227data TrackedChange = TrackedChange ChangeType ChangeInfo

214228 deriving Show

215229

216−data ParagraphStyle = ParagraphStyle { pStyle :: [String]

230+data ParagraphStyle = ParagraphStyle { pStyle :: [ParStyle]

217231 , indentation :: Maybe ParIndentation

218232 , dropCap :: Bool

219− , pHeading :: Maybe (String, Int)

220− , pNumInfo :: Maybe (String, String)

221− , pBlockQuote :: Maybe Bool

222233 , pChange :: Maybe TrackedChange

223234 }

224235 deriving Show

@@ -227,9 +238,6 @@ defaultParagraphStyle :: ParagraphStyle

227238defaultParagraphStyle = ParagraphStyle { pStyle = []

228239 , indentation = Nothing

229240 , dropCap = False

230− , pHeading = Nothing

231− , pNumInfo = Nothing

232− , pBlockQuote = Nothing

233241 , pChange = Nothing

234242 }

235243

@@ -242,18 +250,61 @@ data BodyPart = Paragraph ParagraphStyle [ParPart]

242250

243251type TblGrid = [Integer]

244252

245−data TblLook = TblLook {firstRowFormatting::Bool}

253+newtype TblLook = TblLook {firstRowFormatting::Bool}

246254 deriving Show

247255

248256defaultTblLook :: TblLook

249257defaultTblLook = TblLook{firstRowFormatting = False}

250258

251−data Row = Row [Cell]

259+newtype Row = Row [Cell]

252260 deriving Show

253261

254−data Cell = Cell [BodyPart]

262+newtype Cell = Cell [BodyPart]

255263 deriving Show

256264

265+newtype CharStyleId = CharStyleId { fromCharStyleId :: String }

266+ deriving (Show, Eq, Ord, FromStyleId)

267+newtype ParaStyleId = ParaStyleId { fromParaStyleId :: String }

268+ deriving (Show, Eq, Ord, FromStyleId)

269+

270+newtype CharStyleName = CharStyleName { fromCharStyleName :: CIString }

271+ deriving (Show, Eq, Ord, IsString, FromStyleName)

272+newtype ParaStyleName = ParaStyleName { fromParaStyleName :: CIString }

273+ deriving (Show, Eq, Ord, IsString, FromStyleName)

274+

275+-- Case-insensitive comparisons

276+newtype CIString = CIString String deriving (Show, IsString, FromStyleName)

277+

278+class FromStyleName a where

279+ fromStyleName :: a -> String

280+

281+instance FromStyleName String where

282+ fromStyleName = id

283+

284+class FromStyleId a where

285+ fromStyleId :: a -> String

286+

287+instance FromStyleId String where

288+ fromStyleId = id

289+

290+instance Eq CIString where

291+ (==) = (==) `on` map toLower . coerce

292+

293+instance Ord CIString where

294+ compare = compare `on` map toLower . coerce

295+

296+leftBiasedMergeRunStyle :: RunStyle -> RunStyle -> RunStyle

297+leftBiasedMergeRunStyle a b = RunStyle

298+ { isBold = isBold a <|> isBold b

299+ , isItalic = isItalic a <|> isItalic b

300+ , isSmallCaps = isSmallCaps a <|> isSmallCaps b

301+ , isStrike = isStrike a <|> isStrike b

302+ , isRTL = isRTL a <|> isRTL b

303+ , rVertAlign = rVertAlign a <|> rVertAlign b

304+ , rUnderline = rUnderline a <|> rUnderline b

305+ , rParentStyle = rParentStyle a

306+ }

307+

257308-- (width, height) in EMUs

258309type Extent = Maybe (Double, Double)

259310

@@ -285,21 +336,28 @@ data RunElem = TextRun String | LnBrk | Tab | SoftHyphen | NoBreakHyphen

285336data VertAlign = BaseLn | SupScrpt | SubScrpt

286337 deriving Show

287338

288−data RunStyle = RunStyle { isBold :: Maybe Bool

289− , isItalic :: Maybe Bool

290− , isSmallCaps :: Maybe Bool

291− , isStrike :: Maybe Bool

292− , isRTL :: Maybe Bool

293− , rVertAlign :: Maybe VertAlign

294− , rUnderline :: Maybe String

295− , rStyle :: Maybe CharStyle

339+data CharStyle = CharStyle { cStyleId :: CharStyleId

340+ , cStyleName :: CharStyleName

341+ , cStyleData :: RunStyle

342+ } deriving (Show)

343+

344+data RunStyle = RunStyle { isBold :: Maybe Bool

345+ , isItalic :: Maybe Bool

346+ , isSmallCaps :: Maybe Bool

347+ , isStrike :: Maybe Bool

348+ , isRTL :: Maybe Bool

349+ , rVertAlign :: Maybe VertAlign

350+ , rUnderline :: Maybe String

351+ , rParentStyle :: Maybe CharStyle

296352 }

297353 deriving Show

298354

299−data ParStyleData = ParStyleData { headingLev :: Maybe (String, Int)

300− , isBlockQuote :: Maybe Bool

301− , numInfo :: Maybe (String, String)

302− , psStyle :: Maybe ParStyle}

355+data ParStyle = ParStyle { headingLev :: Maybe (ParaStyleName, Int)

356+ , numInfo :: Maybe (String, String)

357+ , psParentStyle :: Maybe ParStyle

358+ , pStyleName :: ParaStyleName

359+ , pStyleId :: ParaStyleId

360+ }

303361 deriving Show

304362

305363defaultRunStyle :: RunStyle

@@ -310,7 +368,7 @@ defaultRunStyle = RunStyle { isBold = Nothing

310368 , isRTL = Nothing

311369 , rVertAlign = Nothing

312370 , rUnderline = Nothing

313− , rStyle = Nothing

371+ , rParentStyle = Nothing

314372 }

315373

316374type Target = String

@@ -390,7 +448,10 @@ elemToBody ns element | isElem ns "w" "body" element =

390448elemToBody _ _ = throwError WrongElem

391449

392450archiveToStyles :: Archive -> (CharStyleMap, ParStyleMap)

393−archiveToStyles zf =

451+archiveToStyles = archiveToStyles' getStyleId getStyleId

452+archiveToStyles' :: (Ord k1, Ord k2, ElemToStyle a1, ElemToStyle a2) =>

453+ (a1 -> k1) -> (a2 -> k2) -> Archive -> (M.Map k1 a1, M.Map k2 a2)

454+archiveToStyles' conv1 conv2 zf =

394455 let stylesElem = findEntryByPath "word/styles.xml" zf >>=

395456 (parseXMLDoc . UTF8.toStringLazy . fromEntry)

396457 in

@@ -399,19 +460,17 @@ archiveToStyles zf =

399460 Just styElem ->

400461 let namespaces = elemToNameSpaces styElem

401462 in

402− ( M.fromList $ buildBasedOnList namespaces styElem

403− (Nothing :: Maybe CharStyle),

404− M.fromList $ buildBasedOnList namespaces styElem

405− (Nothing :: Maybe ParStyle) )

463+ ( M.fromList $ map (\r -> (conv1 r, r)) $ buildBasedOnList namespaces styElem Nothing,

464+ M.fromList $ map (\p -> (conv2 p, p)) $ buildBasedOnList namespaces styElem Nothing)

406465

407−isBasedOnStyle :: (ElemToStyle a) => NameSpaces -> Element -> Maybe a -> Bool

466+isBasedOnStyle :: (ElemToStyle a, FromStyleId (StyleId a)) => NameSpaces -> Element -> Maybe a -> Bool

408467isBasedOnStyle ns element parentStyle

409468 | isElem ns "w" "style" element

410469 , Just styleType <- findAttrByName ns "w" "type" element

411470 , styleType == cStyleType parentStyle

412471 , Just basedOnVal <- findChildByName ns "w" "basedOn" element >>=

413472 findAttrByName ns "w" "val"

414− , Just ps <- parentStyle = basedOnVal == getStyleId ps

473+ , Just ps <- parentStyle = basedOnVal == fromStyleId (getStyleId ps)

415474 | isElem ns "w" "style" element

416475 , Just styleType <- findAttrByName ns "w" "type" element

417476 , styleType == cStyleType parentStyle

@@ -419,30 +478,70 @@ isBasedOnStyle ns element parentStyle

419478 , Nothing <- parentStyle = True

420479 | otherwise = False

421480

422−class ElemToStyle a where

481+class HasStyleId a => ElemToStyle a where

423482 cStyleType :: Maybe a -> String

424483 elemToStyle :: NameSpaces -> Element -> Maybe a -> Maybe a

425− getStyleId :: a -> String

484+

485+class FromStyleId (StyleId a) => HasStyleId a where

486+ type StyleId a

487+ getStyleId :: a -> StyleId a

488+

489+class FromStyleName (StyleName a) => HasStyleName a where

490+ type StyleName a

491+ getStyleName :: a -> StyleName a

492+

493+class HasParentStyle a where

494+ getParentStyle :: a -> Maybe a

495+

496+instance HasParentStyle CharStyle where

497+ getParentStyle = rParentStyle . cStyleData

498+

499+instance HasParentStyle ParStyle where

500+ getParentStyle = psParentStyle

501+

502+getStyleNames :: (Functor t, HasStyleName a) => t a -> t (StyleName a)

503+getStyleNames = fmap getStyleName

504+

505+constructBogusParStyleData :: ParaStyleName -> ParStyle

506+constructBogusParStyleData stName = ParStyle

507+ { headingLev = Nothing

508+ , numInfo = Nothing

509+ , psParentStyle = Nothing

510+ , pStyleName = stName

511+ , pStyleId = ParaStyleId . filter (/=' ') . fromStyleName $ stName

512+ }

426513

427514instance ElemToStyle CharStyle where

428515 cStyleType _ = "character"

429516 elemToStyle ns element parentStyle

430517 | isElem ns "w" "style" element

431− , Just "character" <- findAttrByName ns "w" "type" element

432− , Just styleId <- findAttrByName ns "w" "styleId" element =

433− Just (styleId, elemToRunStyle ns element parentStyle)

518+ , Just "character" <- findAttrByName ns "w" "type" element =

519+ elemToCharStyle ns element parentStyle

434520 | otherwise = Nothing

435− getStyleId s = fst s

521+

522+instance HasStyleId CharStyle where

523+ type StyleId CharStyle = CharStyleId

524+ getStyleId = cStyleId

525+

526+instance HasStyleName CharStyle where

527+ type StyleName CharStyle = CharStyleName

528+ getStyleName = cStyleName

436529

437530instance ElemToStyle ParStyle where

438531 cStyleType _ = "paragraph"

439532 elemToStyle ns element parentStyle

440533 | isElem ns "w" "style" element

441534 , Just "paragraph" <- findAttrByName ns "w" "type" element

442− , Just styleId <- findAttrByName ns "w" "styleId" element =

443− Just (styleId, elemToParStyleData ns element parentStyle)

535+ = elemToParStyleData ns element parentStyle

444536 | otherwise = Nothing

445− getStyleId s = fst s

537+

538+instance HasStyleId ParStyle where

539+ type StyleId ParStyle = ParaStyleId

540+ getStyleId = pStyleId

541+

542+instance HasStyleName ParStyle where

543+ type StyleName ParStyle = ParaStyleName

544+ getStyleName = pStyleName

446545

447546getStyleChildren :: (ElemToStyle a) => NameSpaces -> Element -> Maybe a -> [a]

448547getStyleChildren ns element parentStyle

@@ -495,7 +594,7 @@ filePathToRelType "word/_rels/endnotes.xml.rels" _ = Just InEndnote

495594-- -- to see if it's a documentPath, we have to check against the dynamic

496595-- -- docPath specified in "_rels/.rels"

497596filePathToRelType path docXmlPath =

498− if path == "word/_rels/" ++ (takeFileName docXmlPath) ++ ".rels"

597+ if path == "word/_rels/" ++ takeFileName docXmlPath ++ ".rels"

499598 then Just InDocument

500599 else Nothing

501600

@@ -537,7 +636,7 @@ lookupLevel numId ilvl (Numbering _ numbs absNumbs) = do

537636 case lvlOverride of

538637 Just (LevelOverride _ _ (Just lvl')) -> Just lvl'

539638 Just (LevelOverride _ (Just strt) _) ->

540− lookup ilvl $ map (\(Level i fmt s _) -> (i, (Level i fmt s (Just strt)))) lvls

639+ lookup ilvl $ map (\(Level i fmt s _) -> (i, Level i fmt s (Just strt))) lvls

541640 _ ->

542641 lookup ilvl $ map (\l@(Level i _ _ _) -> (i, l)) lvls

543642

@@ -693,6 +792,12 @@ testBitMask bitMaskS n =

693792stringToInteger :: String -> Maybe Integer

694793stringToInteger s = listToMaybe $ map fst (reads s :: [(Integer, String)])

695794

795+pHeading :: ParagraphStyle -> Maybe (ParaStyleName, Int)

796+pHeading = getParStyleField headingLev . pStyle

797+

798+pNumInfo :: ParagraphStyle -> Maybe (String, String)

799+pNumInfo = getParStyleField numInfo . pStyle

800+

696801elemToBodyPart :: NameSpaces -> Element -> D BodyPart

697802elemToBodyPart ns element

698803 | isElem ns "w" "p" element

@@ -703,23 +808,19 @@ elemToBodyPart ns element

703808elemToBodyPart ns element

704809 | isElem ns "w" "p" element

705810 , Just (numId, lvl) <- getNumInfo ns element = do

706− sty <- asks envParStyles

707− let parstyle = elemToParagraphStyle ns element sty

811+ parstyle <- elemToParagraphStyle ns element <$> asks envParStyles

708812 parparts <- mapD (elemToParPart ns) (elChildren element)

709− num <- asks envNumbering

710− let levelInfo = lookupLevel numId lvl num

813+ levelInfo <- lookupLevel numId lvl <$> asks envNumbering

711814 return $ ListItem parstyle numId lvl levelInfo parparts

712815elemToBodyPart ns element

713816 | isElem ns "w" "p" element = do

714− sty <- asks envParStyles

715− let parstyle = elemToParagraphStyle ns element sty

817+ parstyle <- elemToParagraphStyle ns element <$> asks envParStyles

716818 parparts <- mapD (elemToParPart ns) (elChildren element)

717819 -- Word uses list enumeration for numbered headings, so we only

718820 -- want to infer a list from the styles if it is NOT a heading.

719821 case pHeading parstyle of

720822 Nothing | Just (numId, lvl) <- pNumInfo parstyle -> do

721− num <- asks envNumbering

722− let levelInfo = lookupLevel numId lvl num

823+ levelInfo <- lookupLevel numId lvl <$> asks envNumbering

723824 return $ ListItem parstyle numId lvl levelInfo parparts

724825 _ -> return $ Paragraph parstyle parparts

725826elemToBodyPart ns element

@@ -727,7 +828,7 @@ elemToBodyPart ns element

727828 let caption' = findChildByName ns "w" "tblPr" element

728829 >>= findChildByName ns "w" "tblCaption"

729830 >>= findAttrByName ns "w" "val"

730− caption = (fromMaybe "" caption')

831+ caption = fromMaybe "" caption'

731832 grid' = case findChildByName ns "w" "tblGrid" element of

732833 Just g -> elemToTblGrid ns g

733834 Nothing -> return []

@@ -1007,20 +1108,18 @@ elemToRun ns element

10071108 return $ Run runStyle runElems

10081109elemToRun _ _ = throwError WrongElem

10091110

1010−getParentStyleValue :: (ParStyleData -> Maybe a) -> ParStyleData -> Maybe a

1111+getParentStyleValue :: (ParStyle -> Maybe a) -> ParStyle -> Maybe a

10111112getParentStyleValue field style

10121113 | Just value <- field style = Just value

1013− | Just parentStyle <- psStyle style

1014− = getParentStyleValue field (snd parentStyle)

1114+ | Just parentStyle <- psParentStyle style

1115+ = getParentStyleValue field parentStyle

10151116getParentStyleValue _ _ = Nothing

10161117

1017−getParStyleField :: (ParStyleData -> Maybe a) -> ParStyleMap -> [String] ->

1018− Maybe a

1019−getParStyleField field stylemap styles

1020− | x <- mapMaybe (\x -> M.lookup x stylemap) styles

1021− , (y:_) <- mapMaybe (getParentStyleValue field) x

1118+getParStyleField :: (ParStyle -> Maybe a) -> [ParStyle] -> Maybe a

1119+getParStyleField field styles

1120+ | (y:_) <- mapMaybe (getParentStyleValue field) styles

10221121 = Just y

1023−getParStyleField _ _ _ = Nothing

1122+getParStyleField _ _ = Nothing

10241123

10251124getTrackedChange :: NameSpaces -> Element -> Maybe TrackedChange

10261125getTrackedChange ns element

@@ -1042,10 +1141,10 @@ elemToParagraphStyle ns element sty

10421141 | Just pPr <- findChildByName ns "w" "pPr" element =

10431142 let style =

10441143 mapMaybe

1045− (findAttrByName ns "w" "val")

1144+ (fmap ParaStyleId . findAttrByName ns "w" "val")

10461145 (findChildrenByName ns "w" "pStyle" pPr)

10471146 in ParagraphStyle

1048− {pStyle = style

1147+ {pStyle = mapMaybe (`M.lookup` sty) style

10491148 , indentation =

10501149 findChildByName ns "w" "ind" pPr >>=

10511150 elemToParIndentation ns

@@ -1057,9 +1156,6 @@ elemToParagraphStyle ns element sty

10571156 Just "none" -> False

10581157 Just _ -> True

10591158 Nothing -> False

1060− , pHeading = getParStyleField headingLev sty style

1061− , pNumInfo = getParStyleField numInfo sty style

1062− , pBlockQuote = getParStyleField isBlockQuote sty style

10631159 , pChange = findChildByName ns "w" "rPr" pPr >>=

10641160 filterChild (\e -> isElem ns "w" "ins" e ||

10651161 isElem ns "w" "moveTo" e ||

@@ -1089,16 +1185,20 @@ elemToRunStyleD :: NameSpaces -> Element -> D RunStyle

10891185elemToRunStyleD ns element

10901186 | Just rPr <- findChildByName ns "w" "rPr" element = do

10911187 charStyles <- asks envCharStyles

1092− let parentSty = case

1188+ let parentSty =

10931189 findChildByName ns "w" "rStyle" rPr >>=

1094− findAttrByName ns "w" "val"

1095− of

1096− Just styName | Just style <- M.lookup styName charStyles ->

1097− Just (styName, style)

1098− _ -> Nothing

1190+ findAttrByName ns "w" "val" >>=

1191+ flip M.lookup charStyles . CharStyleId

10991192 return $ elemToRunStyle ns element parentSty

11001193elemToRunStyleD _ _ = return defaultRunStyle

11011194

1195+elemToCharStyle :: NameSpaces

1196+ -> Element -> Maybe CharStyle -> Maybe CharStyle

1197+elemToCharStyle ns element parentStyle

1198+ = CharStyle <$> (CharStyleId <$> findAttrByName ns "w" "styleId" element)

1199+ <*> getElementStyleName ns element

1200+ <*> (Just $ elemToRunStyle ns element parentStyle)

1201+

11021202elemToRunStyle :: NameSpaces -> Element -> Maybe CharStyle -> RunStyle

11031203elemToRunStyle ns element parentStyle

11041204 | Just rPr <- findChildByName ns "w" "rPr" element =

@@ -1121,38 +1221,23 @@ elemToRunStyle ns element parentStyle

11211221 , rUnderline =

11221222 findChildByName ns "w" "u" rPr >>=

11231223 findAttrByName ns "w" "val"

1124− , rStyle = parentStyle

1224+ , rParentStyle = parentStyle

11251225 }

11261226elemToRunStyle _ _ _ = defaultRunStyle

11271227

1128−getHeaderLevel :: NameSpaces -> Element -> Maybe (String,Int)

1228+getHeaderLevel :: NameSpaces -> Element -> Maybe (ParaStyleName, Int)

11291229getHeaderLevel ns element

1130− | Just styleId <- findAttrByName ns "w" "styleId" element

1131− , Just index <- stripPrefix "Heading" styleId

1132− , Just n <- stringToInteger index

1133− , n > 0 = Just (styleId, fromInteger n)

1134− | Just styleId <- findAttrByName ns "w" "styleId" element

1135− , Just index <- findChildByName ns "w" "name" element >>=

1136− findAttrByName ns "w" "val" >>=

1137− stripPrefix "heading "

1138− , Just n <- stringToInteger index

1139− , n > 0 = Just (styleId, fromInteger n)

1230+ | Just styleName <- getElementStyleName ns element

1231+ , Just n <- stringToInteger =<<

1232+ (stripPrefix "heading " . map toLower $

1233+ fromStyleName styleName)

1234+ , n > 0 = Just (styleName, fromInteger n)

11401235getHeaderLevel _ _ = Nothing

11411236

1142−blockQuoteStyleIds :: [String]

1143−blockQuoteStyleIds = ["Quote", "BlockQuote", "BlockQuotation"]

1144−

1145−blockQuoteStyleNames :: [String]

1146−blockQuoteStyleNames = ["Quote", "Block Text"]

1147−

1148−getBlockQuote :: NameSpaces -> Element -> Maybe Bool

1149−getBlockQuote ns element

1150− | Just styleId <- findAttrByName ns "w" "styleId" element

1151− , styleId `elem` blockQuoteStyleIds = Just True

1152− | Just styleName <- findChildByName ns "w" "name" element >>=

1153− findAttrByName ns "w" "val"

1154− , styleName `elem` blockQuoteStyleNames = Just True

1155−getBlockQuote _ _ = Nothing

1237+getElementStyleName :: Coercible String a => NameSpaces -> Element -> Maybe a

1238+getElementStyleName ns el = coerce <$>

1239+ ((findChildByName ns "w" "name" el >>= findAttrByName ns "w" "val")

1240+ <|> findAttrByName ns "w" "styleId" el)

11561241

11571242getNumInfo :: NameSpaces -> Element -> Maybe (String, String)

11581243getNumInfo ns element = do

@@ -1167,15 +1252,19 @@ getNumInfo ns element = do

11671252 return (numId, lvl)

11681253

11691254

1170−elemToParStyleData :: NameSpaces -> Element -> Maybe ParStyle -> ParStyleData

1171−elemToParStyleData ns element parentStyle =

1172− ParStyleData

1255+elemToParStyleData :: NameSpaces -> Element -> Maybe ParStyle -> Maybe ParStyle

1256+elemToParStyleData ns element parentStyle

1257+ | Just styleId <- findAttrByName ns "w" "styleId" element

1258+ , Just styleName <- getElementStyleName ns element

1259+ = Just $ ParStyle

11731260 {

11741261 headingLev = getHeaderLevel ns element

1175− , isBlockQuote = getBlockQuote ns element

11761262 , numInfo = getNumInfo ns element

1177− , psStyle = parentStyle

1178− }

1263+ , psParentStyle = parentStyle

1264+ , pStyleName = styleName

1265+ , pStyleId = ParaStyleId styleId

1266+ }

1267+elemToParStyleData _ _ _ = Nothing

11791268

11801269elemToRunElem :: NameSpaces -> Element -> D RunElem

11811270elemToRunElem ns element

src/Text/Pandoc/Writers/Docx.hs+61−66

@@ -208,7 +208,7 @@ writeDocx opts doc@(Pandoc meta _) = do

208208 let doc' = walk fixDisplayMath doc

209209 username <- P.lookupEnv "USERNAME"

210210 utctime <- P.getCurrentTime

211− distArchive <- (toArchive . BL.fromStrict) <$> do

211+ distArchive <- toArchive . BL.fromStrict <$> do

212212 oldUserDataDir <- P.getUserDataDir

213213 P.setUserDataDir Nothing

214214 res <- P.readDefaultDataFile "reference.docx"

@@ -216,7 +216,7 @@ writeDocx opts doc@(Pandoc meta _) = do

216216 return res

217217 refArchive <- case writerReferenceDoc opts of

218218 Just f -> toArchive <$> P.readFileLazy f

219− Nothing -> (toArchive . BL.fromStrict) <$>

219+ Nothing -> toArchive . BL.fromStrict <$>

220220 P.readDataFile "reference.docx"

221221

222222 parsedDoc <- parseXml refArchive distArchive "word/document.xml"

@@ -237,7 +237,7 @@ writeDocx opts doc@(Pandoc meta _) = do

237237 >>= subtrct mbAttrMarRight

238238 >>= subtrct mbAttrMarLeft

239239 where

240− subtrct mbStr = \x -> mbStr >>= safeRead >>= (\y -> Just $ x - y)

240+ subtrct mbStr x = mbStr >>= safeRead >>= (\y -> Just $ x - y)

241241

242242 -- styles

243243 mblang <- toLang $ getLang opts meta

@@ -285,7 +285,7 @@ writeDocx opts doc@(Pandoc meta _) = do

285285 envRTL = isRTLmeta

286286 , envChangesAuthor = fromMaybe "unknown" username

287287 , envChangesDate = formatTime defaultTimeLocale "%FT%XZ" utctime

288− , envPrintWidth = maybe 420 (\x -> quot x 20) pgContentWidth

288+ , envPrintWidth = maybe 420 (`quot` 20) pgContentWidth

289289 }

290290

291291

@@ -366,7 +366,7 @@ writeDocx opts doc@(Pandoc meta _) = do

366366 map (\x -> (maybe "" ("/word/" ++) $ extractTarget x,

367367 "application/vnd.openxmlformats-officedocument.wordprocessingml.footer+xml")) footers) ++

368368 map mkImageOverride imgs ++

369− map mkMediaOverride [ eRelativePath e | e <- zEntries refArchive

369+ [ mkMediaOverride (eRelativePath e) | e <- zEntries refArchive

370370 , "word/media/" `isPrefixOf` eRelativePath e ]

371371

372372 let defaultnodes = [mknode "Default"

@@ -589,8 +589,8 @@ writeDocx opts doc@(Pandoc meta _) = do

589589 mapMaybe (fmap ("word/" ++) . extractTarget)

590590 (headers ++ footers)

591591 let miscRelEntries = [ e | e <- zEntries refArchive

592− , "word/_rels/" `isPrefixOf` (eRelativePath e)

593− , ".xml.rels" `isSuffixOf` (eRelativePath e)

592+ , "word/_rels/" `isPrefixOf` eRelativePath e

593+ , ".xml.rels" `isSuffixOf` eRelativePath e

594594 , eRelativePath e /= "word/_rels/document.xml.rels"

595595 , eRelativePath e /= "word/_rels/footnotes.xml.rels" ]

596596 let otherMediaEntries = [ e | e <- zEntries refArchive

@@ -778,24 +778,24 @@ makeTOC opts = do

778778 tocTitle <- gets stTocTitle

779779 title <- withParaPropM (pStyleM "TOC Heading") (blocksToOpenXML opts [Para tocTitle])

780780 return

781− [mknode "w:sdt" [] ([

781+ [mknode "w:sdt" [] [

782782 mknode "w:sdtPr" [] (

783− mknode "w:docPartObj" [] (

783+ mknode "w:docPartObj" []

784784 [mknode "w:docPartGallery" [("w:val","Table of Contents")] (),

785785 mknode "w:docPartUnique" [] ()]

786− ) -- w:docPartObj

786+ -- w:docPartObj

787787 ), -- w:sdtPr

788788 mknode "w:sdtContent" [] (title++[

789789 mknode "w:p" [] (

790− mknode "w:r" [] ([

790+ mknode "w:r" [] [

791791 mknode "w:fldChar" [("w:fldCharType","begin"),("w:dirty","true")] (),

792792 mknode "w:instrText" [("xml:space","preserve")] tocCmd,

793793 mknode "w:fldChar" [("w:fldCharType","separate")] (),

794794 mknode "w:fldChar" [("w:fldCharType","end")] ()

795− ]) -- w:r

795+ ] -- w:r

796796 ) -- w:p

797797 ])

798− ])] -- w:sdt

798+ ]] -- w:sdt

799799

800800-- | Convert Pandoc document to two lists of

801801-- OpenXML elements (the main document and footnotes).

@@ -809,12 +809,12 @@ writeOpenXML opts (Pandoc meta blocks) = do

809809 let includeTOC = writerTableOfContents opts || lookupMetaBool "toc" meta

810810 title <- withParaPropM (pStyleM "Title") $ blocksToOpenXML opts [Para tit | not (null tit)]

811811 subtitle <- withParaPropM (pStyleM "Subtitle") $ blocksToOpenXML opts [Para subtitle' | not (null subtitle')]

812− authors <- withParaProp (pCustomStyle "Author") $ blocksToOpenXML opts $

812+ authors <- withParaPropM (pStyleM "Author") $ blocksToOpenXML opts $

813813 map Para auths

814814 date <- withParaPropM (pStyleM "Date") $ blocksToOpenXML opts [Para dat | not (null dat)]

815815 abstract <- if null abstract'

816816 then return []

817− else withParaProp (pCustomStyle "Abstract") $ blocksToOpenXML opts abstract'

817+ else withParaPropM (pStyleM "Abstract") $ blocksToOpenXML opts abstract'

818818 let convertSpace (Str x : Space : Str y : xs) = Str (x ++ " " ++ y) : xs

819819 convertSpace (Str x : Str y : xs) = Str (x ++ y) : xs

820820 convertSpace xs = xs

@@ -848,18 +848,12 @@ writeOpenXML opts (Pandoc meta blocks) = do

848848blocksToOpenXML :: (PandocMonad m) => WriterOptions -> [Block] -> WS m [Element]

849849blocksToOpenXML opts bls = concat `fmap` mapM (blockToOpenXML opts) bls

850850

851−pCustomStyle :: String -> Element

852−pCustomStyle sty = mknode "w:pStyle" [("w:val",sty)] ()

853−

854851pStyleM :: (PandocMonad m) => String -> WS m XML.Element

855852pStyleM styleName = do

856853 styleMaps <- gets stStyleMaps

857854 let sty' = getStyleId styleName $ sParaStyleMap styleMaps

858855 return $ mknode "w:pStyle" [("w:val",sty')] ()

859856

860−rCustomStyle :: String -> Element

861−rCustomStyle sty = mknode "w:rStyle" [("w:val",sty)] ()

862−

863857rStyleM :: (PandocMonad m) => String -> WS m XML.Element

864858rStyleM styleName = do

865859 styleMaps <- gets stStyleMaps

@@ -921,19 +915,19 @@ blockToOpenXML' opts (Header lev (ident,_,_) lst) = do

921915blockToOpenXML' opts (Plain lst) = do

922916 isInTable <- gets stInTable

923917 let block = blockToOpenXML opts (Para lst)

924− para <- if isInTable then withParaProp (pCustomStyle "Compact") block else block

918+ prop <- pStyleM "Compact"

919+ para <- if isInTable then withParaProp prop block else block

925920 return $ para

926−

927921-- title beginning with fig: indicates that the image is a figure

928922blockToOpenXML' opts (Para [Image attr alt (src,'f':'i':'g':':':tit)]) = do

929923 setFirstPara

930− let prop = pCustomStyle $

924+ prop <- pStyleM $

931925 if null alt

932926 then "Figure"

933− else "CaptionedFigure"

927+ else "Captioned Figure"

934928 paraProps <- local (\env -> env { envParaProperties = EnvProps (Just prop) [] <> envParaProperties env }) (getParaProps False)

935929 contents <- inlinesToOpenXML opts [Image attr alt (src,tit)]

936− captionNode <- withParaProp (pCustomStyle "ImageCaption")

930+ captionNode <- withParaPropM (pStyleM "Image Caption")

937931 $ blockToOpenXML opts (Para alt)

938932 return $ mknode "w:p" [] (paraProps ++ contents) : captionNode

939933blockToOpenXML' opts (Para lst)

@@ -944,10 +938,10 @@ blockToOpenXML' opts (Para lst)

944938 [x] -> isDisplayMath x

945939 _ -> False

946940 paraProps <- getParaProps displayMathPara

947− bodyTextStyle <- pStyleM "Body Text"

941+ bodyTextStyle <- if isFirstPara

942+ then pStyleM "First Paragraph"

943+ else pStyleM "Body Text"

948944 let paraProps' = case paraProps of

949− [] | isFirstPara -> [mknode "w:pPr" []

950− [pCustomStyle "FirstParagraph"]]

951945 [] -> [mknode "w:pPr" [] [bodyTextStyle]]

952946 ps -> ps

953947 modify $ \s -> s { stFirstPara = False }

@@ -965,7 +959,7 @@ blockToOpenXML' opts (BlockQuote blocks) = do

965959 setFirstPara

966960 return p

967961blockToOpenXML' opts (CodeBlock attrs@(ident, _, _) str) = do

968− p <- withParaProp (pCustomStyle "SourceCode") (blockToOpenXML opts $ Para [Code attrs str])

962+ p <- withParaPropM (pStyleM "Source Code") (blockToOpenXML opts $ Para [Code attrs str])

969963 setFirstPara

970964 wrapBookmark ident p

971965blockToOpenXML' _ HorizontalRule = do

@@ -981,7 +975,7 @@ blockToOpenXML' opts (Table caption aligns widths headers rows) = do

981975 let captionStr = stringify caption

982976 caption' <- if null caption

983977 then return []

984− else withParaProp (pCustomStyle "TableCaption")

978+ else withParaPropM (pStyleM "Table Caption")

985979 $ blockToOpenXML opts (Para caption)

986980 let alignmentFor al = mknode "w:jc" [("w:val",alignmentToString al)] ()

987981 -- Table cells require a <w:p> element, even an empty one!

@@ -997,7 +991,8 @@ blockToOpenXML' opts (Table caption aligns widths headers rows) = do

997991 [ mknode "w:tcBorders" []

998992 $ mknode "w:bottom" [("w:val","single")] ()

999993 , mknode "w:vAlign" [("w:val","bottom")] () ]

1000− let emptyCell = [mknode "w:p" [] [mknode "w:pPr" [] [pCustomStyle "Compact"]]]

994+ compactStyle <- pStyleM "Compact"

995+ let emptyCell = [mknode "w:p" [] [mknode "w:pPr" [] [compactStyle]]]

1001996 let mkcell border contents = mknode "w:tc" []

1002997 $ [ borderProps | border ] ++

1003998 if null contents

@@ -1030,20 +1025,17 @@ blockToOpenXML' opts (Table caption aligns widths headers rows) = do

10301025 : [ mkrow True headers' | hasHeader ] ++

10311026 map (mkrow False) rows'

10321027 )]

1033−blockToOpenXML' opts (BulletList lst) = do

1034− let marker = BulletMarker

1035− addList marker

1036− numid <- getNumId

1037− l <- asList $ concat `fmap` mapM (listItemToOpenXML opts numid) lst

1038− setFirstPara

1039− return l

1040−blockToOpenXML' opts (OrderedList (start, numstyle, numdelim) lst) = do

1041− let marker = NumberMarker numstyle numdelim start

1042− addList marker

1043− numid <- getNumId

1044− l <- asList $ concat `fmap` mapM (listItemToOpenXML opts numid) lst

1045− setFirstPara

1046− return l

1028+blockToOpenXML' opts el

1029+ | BulletList lst <- el = addOpenXMLList BulletMarker lst

1030+ | OrderedList (start, numstyle, numdelim) lst <- el

1031+ = addOpenXMLList (NumberMarker numstyle numdelim start) lst

1032+ where

1033+ addOpenXMLList marker lst = do

1034+ addList marker

1035+ numid <- getNumId

1036+ l <- asList $ concat `fmap` mapM (listItemToOpenXML opts numid) lst

1037+ setFirstPara

1038+ return l

10471039blockToOpenXML' opts (DefinitionList items) = do

10481040 l <- concat `fmap` mapM (definitionListItemToOpenXML opts) items

10491041 setFirstPara

@@ -1051,9 +1043,9 @@ blockToOpenXML' opts (DefinitionList items) = do

10511043

10521044definitionListItemToOpenXML :: (PandocMonad m) => WriterOptions -> ([Inline],[[Block]]) -> WS m [Element]

10531045definitionListItemToOpenXML opts (term,defs) = do

1054− term' <- withParaProp (pCustomStyle "DefinitionTerm")

1046+ term' <- withParaPropM (pStyleM "Definition Term")

10551047 $ blockToOpenXML opts (Para term)

1056− defs' <- withParaProp (pCustomStyle "Definition")

1048+ defs' <- withParaPropM (pStyleM "Definition")

10571049 $ concat `fmap` mapM (blocksToOpenXML opts) defs

10581050 return $ term' ++ defs'

10591051

@@ -1159,7 +1151,7 @@ inlineToOpenXML' _ (Str str) =

11591151 formattedString str

11601152inlineToOpenXML' opts Space = inlineToOpenXML opts (Str " ")

11611153inlineToOpenXML' opts SoftBreak = inlineToOpenXML opts (Str " ")

1162−inlineToOpenXML' opts (Span (_,["underline"],_) ils) = do

1154+inlineToOpenXML' opts (Span (_,["underline"],_) ils) =

11631155 withTextProp (mknode "w:u" [("w:val","single")] ()) $

11641156 inlinesToOpenXML opts ils

11651157inlineToOpenXML' _ (Span (ident,["comment-start"],kvs) ils) = do

@@ -1192,18 +1184,21 @@ inlineToOpenXML' opts (Span (ident,classes,kvs) ils) = do

11921184 Just "rtl" -> local (\env -> env { envRTL = True })

11931185 Just "ltr" -> local (\env -> env { envRTL = False })

11941186 _ -> id

1195− let off x = withTextProp (mknode x [("w:val","0")] ())

1196− let pmod = (if "csl-no-emph" `elem` classes then off "w:i" else id) .

1187+ off x = withTextProp (mknode x [("w:val","0")] ())

1188+ pmod = (if "csl-no-emph" `elem` classes then off "w:i" else id) .

11971189 (if "csl-no-strong" `elem` classes then off "w:b" else id) .

11981190 (if "csl-no-smallcaps" `elem` classes

11991191 then off "w:smallCaps"

12001192 else id)

1193+ getChangeAuthorDate = do

1194+ defaultAuthor <- asks envChangesAuthor

1195+ defaultDate <- asks envChangesDate

1196+ let author = fromMaybe defaultAuthor (lookup "author" kvs)

1197+ date = fromMaybe defaultDate (lookup "date" kvs)

1198+ return (author, date)

12011199 insmod <- if "insertion" `elem` classes

12021200 then do

1203− defaultAuthor <- asks envChangesAuthor

1204− defaultDate <- asks envChangesDate

1205− let author = fromMaybe defaultAuthor (lookup "author" kvs)

1206− date = fromMaybe defaultDate (lookup "date" kvs)

1201+ (author, date) <- getChangeAuthorDate

12071202 insId <- gets stInsId

12081203 modify $ \s -> s{stInsId = insId + 1}

12091204 return $ \f -> do

@@ -1215,10 +1210,7 @@ inlineToOpenXML' opts (Span (ident,classes,kvs) ils) = do

12151210 else return id

12161211 delmod <- if "deletion" `elem` classes

12171212 then do

1218− defaultAuthor <- asks envChangesAuthor

1219− defaultDate <- asks envChangesDate

1220− let author = fromMaybe defaultAuthor (lookup "author" kvs)

1221− date = fromMaybe defaultDate (lookup "date" kvs)

1213+ (author, date) <- getChangeAuthorDate

12221214 delId <- gets stDelId

12231215 modify $ \s -> s{stDelId = delId + 1}

12241216 return $ \f -> local (\env->env{envInDel=True}) $ do

@@ -1266,14 +1258,17 @@ inlineToOpenXML' opts (Math mathType str) = do

12661258 Left il -> inlineToOpenXML' opts il

12671259inlineToOpenXML' opts (Cite _ lst) = inlinesToOpenXML opts lst

12681260inlineToOpenXML' opts (Code attrs str) = do

1261+ let alltoktypes = [KeywordTok ..]

1262+ tokTypesMap <- mapM (\tt -> (,) tt <$> rStyleM (show tt)) alltoktypes

12691263 let unhighlighted = intercalate [br] `fmap`

12701264 mapM formattedString (lines str)

12711265 formatOpenXML _fmtOpts = intercalate [br] . map (map toHlTok)

1272− toHlTok (toktype,tok) = mknode "w:r" []

1273− [ mknode "w:rPr" []

1274− [ rCustomStyle (show toktype) ]

1275− , mknode "w:t" [("xml:space","preserve")] (T.unpack tok) ]

1276− withTextProp (rCustomStyle "VerbatimChar")

1266+ toHlTok (toktype,tok) =

1267+ mknode "w:r" []

1268+ [ mknode "w:rPr" [] $

1269+ maybeToList (lookup toktype tokTypesMap)

1270+ , mknode "w:t" [("xml:space","preserve")] (T.unpack tok) ]

1271+ withTextPropM (rStyleM "Verbatim Char")

12771272 $ if isNothing (writerHighlightStyle opts)

12781273 then unhighlighted

12791274 else case highlight (writerSyntaxMap opts)

@@ -1431,12 +1426,12 @@ defaultFootnotes :: [Element]

14311426defaultFootnotes = [ mknode "w:footnote"

14321427 [("w:type", "separator"), ("w:id", "-1")]

14331428 [ mknode "w:p" []

1434− [mknode "w:r" [] $

1429+ [mknode "w:r" []

14351430 [ mknode "w:separator" [] ()]]]

14361431 , mknode "w:footnote"

14371432 [("w:type", "continuationSeparator"), ("w:id", "0")]

14381433 [ mknode "w:p" []

1439− [ mknode "w:r" [] $

1434+ [ mknode "w:r" []

14401435 [ mknode "w:continuationSeparator" [] ()]]]]

14411436

14421437

test/Tests/Readers/Docx.hs+9

@@ -255,6 +255,10 @@ tests = [ testGroup "document"

255255 "lists"

256256 "docx/lists.docx"

257257 "docx/lists.native"

258+ , testCompare

259+ "compact lists"

260+ "docx/lists-compact.docx"

261+ "docx/lists-compact.native"

258262 , testCompare

259263 "lists with level overrides"

260264 "docx/lists_level_override.docx"

@@ -425,6 +429,11 @@ tests = [ testGroup "document"

425429 "custom styles (`+styles`) enabled"

426430 "docx/custom-style-reference.docx"

427431 "docx/custom-style-with-styles.native"

432+ , testCompareWithOpts

433+ def{readerExtensions=extensionsFromList [Ext_styles]}

434+ "custom styles (`+styles`): Compact style is removed from output"

435+ "docx/compact-style-removal.docx"

436+ "docx/compact-style-removal.native"

428437 ]

429438 , testGroup "metadata"

430439 [ testCompareWithOpts def{readerStandalone=True}

test/docx/0_level_headers.native+2−2

@@ -15,10 +15,10 @@

1515,Para [Str "FIGURES",Space,Str "iv"]

1616,Para [Str "TABLES",Space,Str "v"]

1717,Para [Str "SECTION",Space,Str "1",Space,Str "Introduction",Space,Str "2"]

18−,Header 1 ("figures",["Heading0"],[]) [Str "FIGURES"]

18+,Header 1 ("figures",["Heading-0"],[]) [Str "FIGURES"]

1919,Para [Strong [Str "Figure",Space,Str "Page"]]

2020,Para [Strong [Str "No",Space,Str "table",Space,Str "of",Space,Str "figures",Space,Str "entries",Space,Str "found."]]

21−,Header 1 ("tables",["Heading0"],[]) [Str "TABLES"]

21+,Header 1 ("tables",["Heading-0"],[]) [Str "TABLES"]

2222,Para [Strong [Str "Table",Space,Str "Page"]]

2323,Para [Strong [Str "No",Space,Str "table",Space,Str "of",Space,Str "figures",Space,Str "entries",Space,Str "found."]]

2424,Header 1 ("introduction",[],[]) [Str "Introduction"]

test/docx/adjacent_codeblocks.docx

Binary file not shown.

test/docx/compact-style-removal.docxnew file

Binary file not shown.

test/docx/compact-style-removal.nativenew file+5

@@ -0,0 +1,5 @@

1+[OrderedList (1,Decimal,Period)

2+ [[Plain [Str "One"]]

3+ ,[Plain [Str "Two"]]

4+ ,[Plain [Str "Three"]]

5+ ,[Plain [Str "Four"]]]]

test/docx/custom-style-with-styles.native+3−3

@@ -1,7 +1,7 @@

1−[Div ("",[],[("custom-style","FirstParagraph")])

1+[Div ("",[],[("custom-style","First Paragraph")])

22 [Para [Str "This",Space,Str "is",Space,Str "some",Space,Str "text."]]

3−,Div ("",[],[("custom-style","BodyText")])

3+,Div ("",[],[("custom-style","Body Text")])

44 [Para [Str "This",Space,Str "is",Space,Str "text",Space,Str "with",Space,Str "an",Space,Span ("",[],[("custom-style","Emphatic")]) [Str "emphasized"],Space,Str "text",Space,Str "style.",Space,Str "And",Space,Str "this",Space,Str "is",Space,Str "text",Space,Str "with",Space,Str "a",Space,Span ("",[],[("custom-style","Strengthened")]) [Str "strengthened"],Space,Str "text",Space,Str "style."]]

5−,Div ("",[],[("custom-style","MyBlockStyle")])

5+,Div ("",[],[("custom-style","My Block Style")])

66 [BlockQuote

77 [Para [Str "Here",Space,Str "is",Space,Str "a",Space,Str "styled",Space,Str "paragraph",Space,Str "that",Space,Str "inherits",Space,Str "from",Space,Str "Block",Space,Str "Text."]]]]

test/docx/lists-compact.docxnew file

Binary file not shown.

test/docx/lists-compact.nativenew file+5

@@ -0,0 +1,5 @@

1+[OrderedList (1,Decimal,Period)

2+ [[Plain [Str "One"]]

3+ ,[Plain [Str "Two"]]

4+ ,[Plain [Str "Three"]]

5+ ,[Plain [Str "Four"]]]]

test/docx/lists.docx

Binary file not shown.

test/docx/lists.native+1−1

@@ -15,4 +15,4 @@

1515 ,Para [Str "Sub",Space,Str "paragraph"]]]]]]

1616 ,[Para [Str "Same",Space,Str "list"]]]

1717,BulletList

18− [[Para [Str "Different",Space,Str "list",Space,Str "adjacent",Space,Str "to",Space,Str "the",Space,Str "one",Space,Str "above."]]]]

18+ [[Plain [Str "Different",Space,Str "list",Space,Str "adjacent",Space,Str "to",Space,Str "the",Space,Str "one",Space,Str "above."]]]]

test/docx/nested_anchors_in_header.native+1−1

@@ -1,4 +1,4 @@

1−[Header 1 ("\1086\1075\1083\1072\1074\1083\1077\1085\1080\1077",["TOCHeading"],[]) [Str "\1054\1075\1083\1072\1074\1083\1077\1085\1080\1077"]

1+[Header 1 ("\1086\1075\1083\1072\1074\1083\1077\1085\1080\1077",["TOC-Heading"],[]) [Str "\1054\1075\1083\1072\1074\1083\1077\1085\1080\1077"]

22,Para [Link ("",[],[]) [Str "Short",Space,Str "instructions",Space,Str "1"] ("#short-instructions","")]

33,Para [Link ("",[],[]) [Str "Some",Space,Str "instructions",Space,Str "1"] ("#some-instructions","")]

44,Para [Link ("",[],[]) [Str "Remote",Space,Str "folder",Space,Str "or",Space,Str "longlonglonglonglong",Space,Str "file",Space,Str "with",Space,Str "manymanymanymany",Space,Str "letters",Space,Str "inside",Space,Str "opening",Space,Str "2"] ("#remote-folder-or-longlonglonglonglong-file-with-manymanymanymany-letters-inside-opening","")]