7b8c2b6 · typeable/pandoc

git.iko.soy/typeable/pandoc

Add --strip-empty-paragraphs option.

This works for any input format.

commit7b8c2b6691e3816ba52ee07ee7f63573d4ae7253 authorJohn MacFarlane <jgm@berkeley.edu> dateSat 2 Dec 2017, 15:21 (9 years ago) parente09e6a6

3 files changed+30−5

compare with master

unifiedsplit

MANUAL.txt+6 src/Text/Pandoc/App.hs+15−5 src/Text/Pandoc/Shared.hs+9

MANUAL.txt+6

@@ -427,6 +427,12 @@ Reader options

427427

428428: Specify the base level for headers (defaults to 1).

429429

430+`--strip-empty-paragraphs`

431+

432+: Ignore paragraphs with non content. This option is useful

433+ for converting word processing documents where users have

434+ used empty paragraphs to create inter-paragraph space.

435+

430436`--indented-code-classes=`*CLASSES*

431437

432438: Specify classes to use for indented code blocks--for example,

src/Text/Pandoc/App.hs+15−5

@@ -86,8 +86,8 @@ import Text.Pandoc.Lua (LuaException (..), runLuaFilter)

8686import Text.Pandoc.PDF (makePDF)

8787import Text.Pandoc.Process (pipeProcess)

8888import Text.Pandoc.SelfContained (makeDataURI, makeSelfContained)

89−import Text.Pandoc.Shared (eastAsianLineBreakFilter, headerShift, isURI, ordNub,

90− safeRead, tabFilter)

89+import Text.Pandoc.Shared (eastAsianLineBreakFilter, stripEmptyParagraphs,

90+ headerShift, isURI, ordNub, safeRead, tabFilter)

9191import qualified Text.Pandoc.UTF8 as UTF8

9292import Text.Pandoc.Writers.Math (defaultKaTeXURL, defaultMathJaxURL)

9393import Text.Pandoc.XML (toEntities)

@@ -461,14 +461,17 @@ convertWithOpts opts = do

461461

462462 let transforms = (case optBaseHeaderLevel opts of

463463 x | x > 1 -> (headerShift (x - 1) :)

464− | otherwise -> id) $

464+ | otherwise -> id) .

465+ (if optStripEmptyParagraphs opts

466+ then (stripEmptyParagraphs :)

467+ else id) .

465468 (if extensionEnabled Ext_east_asian_line_breaks

466469 readerExts &&

467470 not (extensionEnabled Ext_east_asian_line_breaks

468471 writerExts &&

469472 writerWrapText writerOptions == WrapPreserve)

470473 then (eastAsianLineBreakFilter :)

471− else id)

474+ else id) $

472475 []

473476

474477 let sourceToDoc :: [FilePath] -> PandocIO Pandoc

@@ -622,6 +625,7 @@ data Opt = Opt

622625 , optLuaFilters :: [FilePath] -- ^ Lua filters to apply

623626 , optEmailObfuscation :: ObfuscationMethod

624627 , optIdentifierPrefix :: String

628+ , optStripEmptyParagraphs :: Bool -- ^ Strip empty paragraphs

625629 , optIndentedCodeClasses :: [String] -- ^ Default classes for indented code blocks

626630 , optDataDir :: Maybe FilePath

627631 , optCiteMethod :: CiteMethod -- ^ Method to output cites

@@ -694,6 +698,7 @@ defaultOpts = Opt

694698 , optLuaFilters = []

695699 , optEmailObfuscation = NoObfuscation

696700 , optIdentifierPrefix = ""

701+ , optStripEmptyParagraphs = False

697702 , optIndentedCodeClasses = []

698703 , optDataDir = Nothing

699704 , optCiteMethod = Citeproc

@@ -940,7 +945,12 @@ options =

940945 "NUMBER")

941946 "" -- "Headers base level"

942947

943− , Option "" ["indented-code-classes"]

948+ , Option "" ["strip-empty-paragraphs"]

949+ (NoArg

950+ (\opt -> return opt{ optStripEmptyParagraphs = True }))

951+ "" -- "Strip empty paragraphs"

952+

953+ , Option "" ["indented-code-classes"]

944954 (ReqArg

945955 (\arg opt -> return opt { optIndentedCodeClasses = words $

946956 map (\c -> if c == ',' then ' ' else c) arg })

src/Text/Pandoc/Shared.hs+9

@@ -72,6 +72,7 @@ module Text.Pandoc.Shared (

7272 inlineListToIdentifier,

7373 isHeaderBlock,

7474 headerShift,

75+ stripEmptyParagraphs,

7576 isTightList,

7677 addMetaField,

7778 makeMeta,

@@ -529,6 +530,14 @@ headerShift n = walk shift

529530 shift (Header level attr inner) = Header (level + n) attr inner

530531 shift x = x

531532

533+-- | Remove empty paragraphs.

534+stripEmptyParagraphs :: Pandoc -> Pandoc

535+stripEmptyParagraphs = walk go

536+ where go :: [Block] -> [Block]

537+ go = filter (not . isEmptyParagraph)

538+ isEmptyParagraph (Para []) = True

539+ isEmptyParagraph _ = False

540+

532541-- | Detect if a list is tight.

533542isTightList :: [[Block]] -> Bool

534543isTightList = all firstIsPlain