bdec07b · typeable/pandoc

git.iko.soy/typeable/pandoc

Shared: Removed unescapeURI, modified escapeURI.

escapeURI now only escapes space characters, leaving unicode characters as they are, instead of converting them to octets and URL-encoding them, as before. This gives more readable URIs. User agents now do the percent-encoding themselves. URIs are no longer unescaped at all on conversion to markdown, asciidoc, rst, org. Closes #349.

commitbdec07bac9efb737dbc60e6985d8ba1500a5eeb9 authorJohn MacFarlane <fiddlosopher@gmail.com> dateFri 2 Dec 2011, 19:39 (15 years ago) parentc6456ef

6 files changed+18−33

compare with master

unifiedsplit

src/Text/Pandoc/Shared.hs+8−17 src/Text/Pandoc/Writers/AsciiDoc.hs+2−4 src/Text/Pandoc/Writers/Markdown.hs+1−2 src/Text/Pandoc/Writers/Org.hs+1−2 src/Text/Pandoc/Writers/RST.hs+3−5 tests/markdown-reader-more.native+3−3

src/Text/Pandoc/Shared.hs+8−17

@@ -44,7 +44,6 @@ module Text.Pandoc.Shared (

4444 camelCaseToHyphenated,

4545 toRomanNumeral,

4646 escapeURI,

47− unescapeURI,

4847 tabFilter,

4948 -- * Pandoc block and inline list processing

5049 orderedListMarkers,

@@ -73,11 +72,10 @@ module Text.Pandoc.Shared (

7372import Text.Pandoc.Definition

7473import Text.Pandoc.Generic

7574import qualified Text.Pandoc.UTF8 as UTF8 (readFile)

76−import Data.Char ( toLower, isLower, isUpper, isAlpha, isAscii,

77− isLetter, isDigit )

75+import Data.Char ( toLower, isLower, isUpper, isAlpha,

76+ isLetter, isDigit, isSpace )

7877import Data.List ( find, isPrefixOf, intercalate )

79−import Network.URI ( isAllowedInURI, escapeURIString, unEscapeString )

80−import Codec.Binary.UTF8.String ( encodeString, decodeString )

78+import Network.URI ( escapeURIString )

8179import System.Directory

8280import System.FilePath ( (</>) )

8381import Data.Generics (Typeable, Data)

@@ -181,16 +179,9 @@ toRomanNumeral x =

181179 _ | x >= 1 -> "I" ++ toRomanNumeral (x - 1)

182180 _ -> ""

183181

184−-- | Escape unicode characters in a URI. Characters that are

185−-- already valid in a URI, including % and ?, are left alone.

182+-- | Escape whitespace in URI.

186183escapeURI :: String -> String

187−escapeURI = escapeURIString isAllowedInURI . encodeString

188−

189−-- | Unescape unicode and some special characters in a URI, but

190−-- without introducing spaces.

191−unescapeURI :: String -> String

192−unescapeURI = escapeURIString (\c -> isAllowedInURI c || not (isAscii c)) .

193− decodeString . unEscapeString

184+escapeURI = escapeURIString (not . isSpace)

194185

195186-- | Convert tabs to spaces and filter out DOS line endings.

196187-- Tabs will be preserved if tab stop is set to 0.

@@ -304,9 +295,9 @@ consolidateInlines (Str x : ys) =

304295 fromStr (Str z) = z

305296 fromStr _ = error "consolidateInlines - fromStr - not a Str"

306297consolidateInlines (Space : ys) = Space : rest

307− where isSpace Space = True

308− isSpace _ = False

309− rest = consolidateInlines $ dropWhile isSpace ys

298+ where isSp Space = True

299+ isSp _ = False

300+ rest = consolidateInlines $ dropWhile isSp ys

310301consolidateInlines (Emph xs : Emph ys : zs) = consolidateInlines $

311302 Emph (xs ++ ys) : zs

312303consolidateInlines (Strong xs : Strong ys : zs) = consolidateInlines $

src/Text/Pandoc/Writers/AsciiDoc.hs+2−4

@@ -336,12 +336,11 @@ inlineToAsciiDoc _ (RawInline _ _) = return empty

336336inlineToAsciiDoc _ (LineBreak) = return $ " +" <> cr

337337inlineToAsciiDoc _ Space = return space

338338inlineToAsciiDoc opts (Cite _ lst) = inlineListToAsciiDoc opts lst

339−inlineToAsciiDoc opts (Link txt (src', _tit)) = do

339+inlineToAsciiDoc opts (Link txt (src, _tit)) = do

340340-- relative: link:downloads/foo.zip[download foo.zip]

341341-- abs: http://google.cod[Google]

342342-- or my@email.com[email john]

343343 linktext <- inlineListToAsciiDoc opts txt

344− let src = unescapeURI src'

345344 let isRelative = ':' `notElem` src

346345 let prefix = if isRelative

347346 then text "link:"

@@ -353,7 +352,7 @@ inlineToAsciiDoc opts (Link txt (src', _tit)) = do

353352 return $ if useAuto

354353 then text srcSuffix

355354 else prefix <> text src <> "[" <> linktext <> "]"

356−inlineToAsciiDoc opts (Image alternate (src', tit)) = do

355+inlineToAsciiDoc opts (Image alternate (src, tit)) = do

357356-- image:images/logo.png[Company logo, title="blah"]

358357 let txt = if (null alternate) || (alternate == [Str ""])

359358 then [Str "image"]

@@ -362,7 +361,6 @@ inlineToAsciiDoc opts (Image alternate (src', tit)) = do

362361 let linktitle = if null tit

363362 then empty

364363 else text $ ",title=\"" ++ tit ++ "\""

365− let src = unescapeURI src'

366364 return $ "image:" <> text src <> "[" <> linktext <> linktitle <> "]"

367365inlineToAsciiDoc opts (Note [Para inlines]) =

368366 inlineToAsciiDoc opts (Note [Plain inlines])

src/Text/Pandoc/Writers/Markdown.hs+1−2

@@ -493,12 +493,11 @@ inlineToMarkdown opts (Cite (c:cs) lst)

493493 modekey SuppressAuthor = "-"

494494 modekey _ = ""

495495inlineToMarkdown _ (Cite _ _) = return $ text ""

496−inlineToMarkdown opts (Link txt (src', tit)) = do

496+inlineToMarkdown opts (Link txt (src, tit)) = do

497497 linktext <- inlineListToMarkdown opts txt

498498 let linktitle = if null tit

499499 then empty

500500 else text $ " \"" ++ tit ++ "\""

501− let src = unescapeURI src'

502501 let srcSuffix = if isPrefixOf "mailto:" src then drop 7 src else src

503502 let useRefLinks = writerReferenceLinks opts

504503 let useAuto = case (tit,txt) of

src/Text/Pandoc/Writers/Org.hs+1−2

@@ -272,8 +272,7 @@ inlineToOrg (Link txt (src, _)) = do

272272 _ -> do contents <- inlineListToOrg txt

273273 modify $ \s -> s{ stLinks = True }

274274 return $ "[[" <> text src <> "][" <> contents <> "]]"

275−inlineToOrg (Image _ (source', _)) = do

276− let source = unescapeURI source'

275+inlineToOrg (Image _ (source, _)) = do

277276 modify $ \s -> s{ stImages = True }

278277 return $ "[[" <> text source <> "]]"

279278inlineToOrg (Note contents) = do

src/Text/Pandoc/Writers/RST.hs+3−5

@@ -298,9 +298,8 @@ inlineToRST Space = return space

298298inlineToRST (Link [Code _ str] (src, _)) | src == str ||

299299 src == "mailto:" ++ str = do

300300 let srcSuffix = if isPrefixOf "mailto:" src then drop 7 src else src

301− return $ text $ unescapeURI srcSuffix

302−inlineToRST (Link txt (src', tit)) = do

303− let src = unescapeURI src'

301+ return $ text srcSuffix

302+inlineToRST (Link txt (src, tit)) = do

304303 useReferenceLinks <- get >>= return . writerReferenceLinks . stOptions

305304 linktext <- inlineListToRST $ normalizeSpaces txt

306305 if useReferenceLinks

@@ -311,8 +310,7 @@ inlineToRST (Link txt (src', tit)) = do

311310 modify $ \st -> st { stLinks = refs' }

312311 return $ "`" <> linktext <> "`_"

313312 else return $ "`" <> linktext <> " <" <> text src <> ">`_"

314−inlineToRST (Image alternate (source', tit)) = do

315− let source = unescapeURI source'

313+inlineToRST (Image alternate (source, tit)) = do

316314 pics <- get >>= return . stImages

317315 let labelsUsed = map fst pics

318316 let txt = if null alternate || alternate == [Str ""] ||

tests/markdown-reader-more.native+3−3

@@ -29,9 +29,9 @@

2929,Para [Str "`",Str "hi"]

3030,Para [Str "there",Str "`"]

3131,Header 2 [Str "Multilingual",Space,Str "URLs"]

32−,Para [Link [Code ("",["url"],[]) "http://\27979.com?\27979=\27979"] ("http://%E6%B5%8B.com?%E6%B5%8B=%E6%B5%8B","")]

33−,Para [Link [Str "foo"] ("/bar/%E6%B5%8B?x=%E6%B5%8B","title")]

34−,Para [Link [Code ("",["url"],[]) "\27979@foo.\27979.baz"] ("mailto:%E6%B5%8B@foo.%E6%B5%8B.baz","")]

32+,Plain [RawInline "html" "<http://\27979.com?\27979=\27979>"]

33+,Para [Link [Str "foo"] ("/bar/\27979?x=\27979","title")]

34+,Para [Link [Code ("",["url"],[]) "\27979@foo.\27979.baz"] ("mailto:\27979@foo.\27979.baz","")]

3535,Header 2 [Str "Numbered",Space,Str "examples"]

3636,OrderedList (1,Example,TwoParens)

3737 [[Plain [Str "First",Space,Str "example",Str "."]]