70405ef · typeable/pandoc

git.iko.soy/typeable/pandoc

normalize: Normalize spaces too.

In normal form, Space elements only occur to separate two non-Space elements. So, we never have [Space], or [, ..., Space].

commit70405ef98ab48e76e9cc8cdd827cb7580dacf4e6 authorJohn MacFarlane <jgm@berkeley.edu> dateFri 4 Feb 2011, 13:22 (16 years ago) parent8e81437

2 files changed+15−5

compare with master

unifiedsplit

src/Tests/Shared.hs+5 src/Text/Pandoc/Shared.hs+10−5

src/Tests/Shared.hs+5

@@ -10,6 +10,8 @@ tests :: [Test]

1010tests = [ testGroup "normalize"

1111 [ property "p_normalize_blocks_rt" p_normalize_blocks_rt

1212 , property "p_normalize_inlines_rt" p_normalize_inlines_rt

13+ , property "p_normalize_no_trailing_spaces"

14+ p_normalize_no_trailing_spaces

1315 ]

1416 ]

1517

@@ -19,3 +21,6 @@ p_normalize_blocks_rt bs = normalize bs == normalize (normalize bs)

1921p_normalize_inlines_rt :: [Inline] -> Bool

2022p_normalize_inlines_rt ils = normalize ils == normalize (normalize ils)

2123

24+p_normalize_no_trailing_spaces :: [Inline] -> Bool

25+p_normalize_no_trailing_spaces ils = null ils' || last ils' /= Space

26+ where ils' = normalize $ ils ++ [Space]

src/Text/Pandoc/Shared.hs+10−5

@@ -243,10 +243,7 @@ orderedListMarkers (start, numstyle, numdelim) =

243243-- remove empty Str elements.

244244normalizeSpaces :: [Inline] -> [Inline]

245245normalizeSpaces = cleanup . dropWhile isSpaceOrEmpty

246− where isSpaceOrEmpty Space = True

247− isSpaceOrEmpty (Str "") = True

248− isSpaceOrEmpty _ = False

249− cleanup [] = []

246+ where cleanup [] = []

250247 cleanup (Space:rest) = let rest' = dropWhile isSpaceOrEmpty rest

251248 in case rest' of

252249 [] -> []

@@ -254,13 +251,18 @@ normalizeSpaces = cleanup . dropWhile isSpaceOrEmpty

254251 cleanup ((Str ""):rest) = cleanup rest

255252 cleanup (x:rest) = x : cleanup rest

256253

254+isSpaceOrEmpty :: Inline -> Bool

255+isSpaceOrEmpty Space = True

256+isSpaceOrEmpty (Str "") = True

257+isSpaceOrEmpty _ = False

258+

257259-- | Normalize @Pandoc@ document, consolidating doubled 'Space's,

258260-- combining adjacent 'Str's and 'Emph's, remove 'Null's and

259261-- empty elements, etc.

260262normalize :: (Eq a, Data a) => a -> a

261263normalize = topDown removeEmptyBlocks .

262264 topDown consolidateInlines .

263− bottomUp removeEmptyInlines

265+ bottomUp (removeEmptyInlines . removeTrailingInlineSpaces)

264266

265267removeEmptyBlocks :: [Block] -> [Block]

266268removeEmptyBlocks (Null : xs) = removeEmptyBlocks xs

@@ -284,6 +286,9 @@ removeEmptyInlines (Str "" : zs) = removeEmptyInlines zs

284286removeEmptyInlines (x : xs) = x : removeEmptyInlines xs

285287removeEmptyInlines [] = []

286288

289+removeTrailingInlineSpaces :: [Inline] -> [Inline]

290+removeTrailingInlineSpaces = reverse . dropWhile isSpaceOrEmpty . reverse

291+

287292consolidateInlines :: [Inline] -> [Inline]

288293consolidateInlines (Str x : ys) =

289294 case concat (x : map fromStr strs) of