8aa6172 · typeable/pandoc

README+4−3

@@ -13,8 +13,8 @@ Description

1313Pandoc is a [Haskell] library for converting from one markup format to

1414another, and a command-line tool that uses this library. It can read

1515[markdown] and (subsets of) [Textile], [reStructuredText], [HTML],

16−[LaTeX], [MediaWiki markup], and [DocBook XML]; and it can write plain

17−text, [markdown], [reStructuredText], [XHTML], [HTML 5], [LaTeX]

16+[LaTeX], [MediaWiki markup], [OPML], and [DocBook XML]; and it can write

17+plain text, [markdown], [reStructuredText], [XHTML], [HTML 5], [LaTeX]

1818(including [beamer] slide shows), [ConTeXt], [RTF], [DocBook XML],

1919[OpenDocument XML], [ODT], [Word docx], [GNU Texinfo], [MediaWiki

2020markup], [EPUB] (v2 or v3), [FictionBook2], [Textile], [groff man] pages, [Emacs

@@ -143,7 +143,7 @@ General options

143143 `markdown_phpextra` (PHP Markdown Extra extended markdown),

144144 `markdown_github` (github extended markdown),

145145 `textile` (Textile), `rst` (reStructuredText), `html` (HTML),

146− `docbook` (DocBook XML), `mediawiki` (MediaWiki markup),

146+ `docbook` (DocBook XML), `opml` (OPML), `mediawiki` (MediaWiki markup),

147147 or `latex` (LaTeX). If `+lhs` is appended to `markdown`, `rst`,

148148 `latex`, the input will be treated as literate Haskell source:

149149 see [Literate Haskell support](#literate-haskell-support), below.

@@ -2624,6 +2624,7 @@ Sergey Astanin, Arlo O'Keeffe, Denis Laxalde, Brent Yorgey.

26242624[ConTeXt]: http://www.pragma-ade.nl/

26252625[RTF]: http://en.wikipedia.org/wiki/Rich_Text_Format

26262626[DocBook XML]: http://www.docbook.org/

2627+[OPML]: http://dev.opml.org/spec2.html

26272628[OpenDocument XML]: http://opendocument.xml.org/

26282629[ODT]: http://en.wikipedia.org/wiki/OpenDocument

26292630[Textile]: http://redcloth.org/textile

pandoc.cabal+4−1

@@ -16,7 +16,7 @@ Synopsis: Conversion between markup formats

1616Description: Pandoc is a Haskell library for converting from one markup

1717 format to another, and a command-line tool that uses

1818 this library. It can read markdown and (subsets of) HTML,

19− reStructuredText, LaTeX, DocBook, MediaWiki markup,

19+ reStructuredText, LaTeX, DocBook, MediaWiki markup, OPML,

2020 and Textile, and it can write markdown, reStructuredText,

2121 HTML, LaTeX, ConTeXt, Docbook, OpenDocument, ODT,

2222 Word docx, RTF, MediaWiki, Textile, groff man pages,

@@ -105,6 +105,8 @@ Extra-Source-Files:

105105 tests/docbook-reader.native

106106 tests/html-reader.html,

107107 tests/html-reader.native,

108+ tests/opml-reader.html,

109+ tests/opml-reader.native,

108110 tests/insert,

109111 tests/lalune.jpg,

110112 tests/movie.jpg,

@@ -278,6 +280,7 @@ Library

278280 Text.Pandoc.Readers.MediaWiki,

279281 Text.Pandoc.Readers.RST,

280282 Text.Pandoc.Readers.DocBook,

283+ Text.Pandoc.Readers.OPML,

281284 Text.Pandoc.Readers.TeXMath,

282285 Text.Pandoc.Readers.Textile,

283286 Text.Pandoc.Readers.Native,

pandoc.hs+1

@@ -758,6 +758,7 @@ defaultReaderName fallback (x:xs) =

758758 ".rst" -> "rst"

759759 ".lhs" -> "markdown+lhs"

760760 ".db" -> "docbook"

761+ ".opml" -> "opml"

761762 ".wiki" -> "mediawiki"

762763 ".textile" -> "textile"

763764 ".native" -> "native"

src/Text/Pandoc.hs+3

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

7272 , readHtml

7373 , readTextile

7474 , readDocBook

75+ , readOPML

7576 , readNative

7677 -- * Writers: converting /from/ Pandoc format

7778 , Writer (..)

@@ -113,6 +114,7 @@ import Text.Pandoc.Readers.Markdown

113114import Text.Pandoc.Readers.MediaWiki

114115import Text.Pandoc.Readers.RST

115116import Text.Pandoc.Readers.DocBook

117+import Text.Pandoc.Readers.OPML

116118import Text.Pandoc.Readers.LaTeX

117119import Text.Pandoc.Readers.HTML

118120import Text.Pandoc.Readers.Textile

@@ -192,6 +194,7 @@ readers = [("native" , \_ s -> return $ readNative s)

192194 ,("rst" , \o s -> return $ readRST o s)

193195 ,("mediawiki" , \o s -> return $ readMediaWiki o s)

194196 ,("docbook" , \o s -> return $ readDocBook o s)

197+ ,("opml" , \o s -> return $ readOPML o s)

195198 ,("textile" , \o s -> return $ readTextile o s) -- TODO : textile+lhs

196199 ,("html" , \o s -> return $ readHtml o s)

197200 ,("latex" , \o s -> return $ readLaTeX o s)

src/Text/Pandoc/Readers/OPML.hsnew file+95

@@ -0,0 +1,95 @@

1+module Text.Pandoc.Readers.OPML ( readOPML ) where

2+import Data.Char (toUpper)

3+import Text.Pandoc.Options

4+import Text.Pandoc.Definition

5+import Text.Pandoc.Builder

6+import Text.Pandoc.Readers.HTML (readHtml)

7+import Text.Pandoc.Readers.Markdown (readMarkdown)

8+import Text.XML.Light

9+import Text.HTML.TagSoup.Entity (lookupEntity)

10+import Data.Generics

11+import Data.Monoid

12+import Control.Monad.State

13+import Control.Applicative ((<$>), (<$))

14+

15+type OPML = State OPMLState

16+

17+data OPMLState = OPMLState{

18+ opmlSectionLevel :: Int

19+ , opmlDocTitle :: Inlines

20+ , opmlDocAuthors :: [Inlines]

21+ , opmlDocDate :: Inlines

22+ } deriving Show

23+

24+readOPML :: ReaderOptions -> String -> Pandoc

25+readOPML _ inp = setTitle (opmlDocTitle st')

26+ $ setAuthors (opmlDocAuthors st')

27+ $ setDate (opmlDocDate st')

28+ $ doc $ mconcat bs

29+ where (bs, st') = runState (mapM parseBlock $ normalizeTree $ parseXML inp)

30+ OPMLState{ opmlSectionLevel = 0

31+ , opmlDocTitle = mempty

32+ , opmlDocAuthors = []

33+ , opmlDocDate = mempty

34+ }

35+

36+-- normalize input, consolidating adjacent Text and CRef elements

37+normalizeTree :: [Content] -> [Content]

38+normalizeTree = everywhere (mkT go)

39+ where go :: [Content] -> [Content]

40+ go (Text (CData CDataRaw _ _):xs) = xs

41+ go (Text (CData CDataText s1 z):Text (CData CDataText s2 _):xs) =

42+ Text (CData CDataText (s1 ++ s2) z):xs

43+ go (Text (CData CDataText s1 z):CRef r:xs) =

44+ Text (CData CDataText (s1 ++ convertEntity r) z):xs

45+ go (CRef r:Text (CData CDataText s1 z):xs) =

46+ Text (CData CDataText (convertEntity r ++ s1) z):xs

47+ go (CRef r1:CRef r2:xs) =

48+ Text (CData CDataText (convertEntity r1 ++ convertEntity r2) Nothing):xs

49+ go xs = xs

50+

51+convertEntity :: String -> String

52+convertEntity e = maybe (map toUpper e) (:[]) (lookupEntity e)

53+

54+-- convenience function to get an attribute value, defaulting to ""

55+attrValue :: String -> Element -> String

56+attrValue attr elt =

57+ case lookupAttrBy (\x -> qName x == attr) (elAttribs elt) of

58+ Just z -> z

59+ Nothing -> ""

60+

61+asHtml :: String -> Inlines

62+asHtml s = case readHtml def s of

63+ Pandoc _ [Plain ils] -> fromList ils

64+ _ -> mempty

65+

66+asMarkdown :: String -> Blocks

67+asMarkdown s = fromList bs

68+ where Pandoc _ bs = readMarkdown def s

69+

70+getBlocks :: Element -> OPML Blocks

71+getBlocks e = mconcat <$> (mapM parseBlock $ elContent e)

72+

73+parseBlock :: Content -> OPML Blocks

74+parseBlock (Elem e) =

75+ case qName (elName e) of

76+ "ownerName" -> mempty <$ modify (\st ->

77+ st{opmlDocAuthors = [text $ strContent e]})

78+ "dateModified" -> mempty <$ modify (\st ->

79+ st{opmlDocDate = text $ strContent e})

80+ "title" -> mempty <$ modify (\st ->

81+ st{opmlDocTitle = text $ strContent e})

82+ "outline" -> gets opmlSectionLevel >>= sect . (+1)

83+ "?xml" -> return mempty

84+ _ -> getBlocks e

85+ where sect n = do let headerText = asHtml $ attrValue "text" e

86+ let noteBlocks = asMarkdown $ attrValue "_note" e

87+ modify $ \st -> st{ opmlSectionLevel = n }

88+ bs <- getBlocks e

89+ modify $ \st -> st{ opmlSectionLevel = n - 1 }

90+ let headerText' = case attrValue "type" e of

91+ "link" -> link

92+ (attrValue "url" e) "" headerText

93+ _ -> headerText

94+ return $ header n headerText' <> noteBlocks <> bs

95+parseBlock _ = return mempty

tests/Tests/Old.hs+4

@@ -124,6 +124,10 @@ tests = [ testGroup "markdown"

124124 , test "reader" ["-r", "mediawiki", "-w", "native", "-s"]

125125 "mediawiki-reader.wiki" "mediawiki-reader.native"

126126 ]

127+ , testGroup "opml"

128+ [ test "reader" ["-r", "opml", "-w", "native", "-s"]

129+ "opml-reader.opml" "opml-reader.native"

130+ ]

127131 , testGroup "other writers" $ map (\f -> testGroup f $ writerTests f)

128132 [ "opendocument" , "context" , "texinfo"

129133 , "man" , "plain" , "rtf", "org", "asciidoc"

tests/opml-reader.nativenew file+66

@@ -0,0 +1,66 @@

1+Pandoc (Meta {docTitle = [Str "states.opml"], docAuthors = [[Str "Dave",Space,Str "Winer"]], docDate = [Str "Thu,",Space,Str "14",Space,Str "Jul",Space,Str "2005",Space,Str "23:41:05",Space,Str "GMT"]})

2+[Header 1 ("",[],[]) [Str "United",Space,Str "States"]

3+,Header 2 ("",[],[]) [Str "Far",Space,Str "West"]

4+,Header 3 ("",[],[]) [Str "Alaska"]

5+,Header 3 ("",[],[]) [Str "California"]

6+,Header 3 ("",[],[]) [Str "Hawaii"]

7+,Header 3 ("",[],[]) [Strong [Str "Nevada"]]

8+,Para [Str "I",Space,Str "lived",Space,Str "here",Space,Emph [Str "once"],Str "."]

9+,Para [Str "Loved",Space,Str "it."]

10+,Header 4 ("",[],[]) [Link [Str "Reno"] ("http://www.reno.gov","")]

11+,Header 4 ("",[],[]) [Str "Las",Space,Str "Vegas"]

12+,Header 4 ("",[],[]) [Str "Ely"]

13+,Header 4 ("",[],[]) [Str "Gerlach"]

14+,Header 3 ("",[],[]) [Str "Oregon"]

15+,Header 3 ("",[],[]) [Str "Washington"]

16+,Header 2 ("",[],[]) [Str "Great",Space,Str "Plains"]

17+,Header 3 ("",[],[]) [Str "Kansas"]

18+,Header 3 ("",[],[]) [Str "Nebraska"]

19+,Header 3 ("",[],[]) [Str "North",Space,Str "Dakota"]

20+,Header 3 ("",[],[]) [Str "Oklahoma"]

21+,Header 3 ("",[],[]) [Str "South",Space,Str "Dakota"]

22+,Header 2 ("",[],[]) [Str "Mid",Str "-",Str "Atlantic"]

23+,Header 3 ("",[],[]) [Str "Delaware"]

24+,Header 3 ("",[],[]) [Str "Maryland"]

25+,Header 3 ("",[],[]) [Str "New",Space,Str "Jersey"]

26+,Header 3 ("",[],[]) [Str "New",Space,Str "York"]

27+,Header 3 ("",[],[]) [Str "Pennsylvania"]

28+,Header 2 ("",[],[]) [Str "Midwest"]

29+,Header 3 ("",[],[]) [Str "Illinois"]

30+,Header 3 ("",[],[]) [Str "Indiana"]

31+,Header 3 ("",[],[]) [Str "Iowa"]

32+,Header 3 ("",[],[]) [Str "Kentucky"]

33+,Header 3 ("",[],[]) [Str "Michigan"]

34+,Header 3 ("",[],[]) [Str "Minnesota"]

35+,Header 3 ("",[],[]) [Str "Missouri"]

36+,Header 3 ("",[],[]) [Str "Ohio"]

37+,Header 3 ("",[],[]) [Str "West",Space,Str "Virginia"]

38+,Header 3 ("",[],[]) [Str "Wisconsin"]

39+,Header 2 ("",[],[]) [Str "Mountains"]

40+,Header 3 ("",[],[]) [Str "Colorado"]

41+,Header 3 ("",[],[]) [Str "Idaho"]

42+,Header 3 ("",[],[]) [Str "Montana"]

43+,Header 3 ("",[],[]) [Str "Utah"]

44+,Header 3 ("",[],[]) [Str "Wyoming"]

45+,Header 2 ("",[],[]) [Str "New",Space,Str "England"]

46+,Header 3 ("",[],[]) [Str "Connecticut"]

47+,Header 3 ("",[],[]) [Str "Maine"]

48+,Header 3 ("",[],[]) [Str "Massachusetts"]

49+,Header 3 ("",[],[]) [Str "New",Space,Str "Hampshire"]

50+,Header 3 ("",[],[]) [Str "Rhode",Space,Str "Island"]

51+,Header 3 ("",[],[]) [Str "Vermont"]

52+,Header 2 ("",[],[]) [Str "South"]

53+,Header 3 ("",[],[]) [Str "Alabama"]

54+,Header 3 ("",[],[]) [Str "Arkansas"]

55+,Header 3 ("",[],[]) [Str "Florida"]

56+,Header 3 ("",[],[]) [Str "Georgia"]

57+,Header 3 ("",[],[]) [Str "Louisiana"]

58+,Header 3 ("",[],[]) [Str "Mississippi"]

59+,Header 3 ("",[],[]) [Str "North",Space,Str "Carolina"]

60+,Header 3 ("",[],[]) [Str "South",Space,Str "Carolina"]

61+,Header 3 ("",[],[]) [Str "Tennessee"]

62+,Header 3 ("",[],[]) [Str "Virginia"]

63+,Header 2 ("",[],[]) [Str "Southwest"]

64+,Header 3 ("",[],[]) [Str "Arizona"]

65+,Header 3 ("",[],[]) [Str "New",Space,Str "Mexico"]

66+,Header 3 ("",[],[]) [Str "Texas"]]

tests/opml-reader.opmlnew file+91

@@ -0,0 +1,91 @@

1+<?xml version="1.0" encoding="ISO-8859-1"?>

2+<opml version="2.0">

3+ <head>

4+ <title>states.opml</title>

5+ <dateCreated>Tue, 15 Mar 2005 16:35:45 GMT</dateCreated>

6+ <dateModified>Thu, 14 Jul 2005 23:41:05 GMT</dateModified>

7+ <ownerName>Dave Winer</ownerName>

8+ <ownerEmail>dave@scripting.com</ownerEmail>

9+ <expansionState>1, 6, 13, 16, 18, 20</expansionState>

10+ <vertScrollState>1</vertScrollState>

11+ <windowTop>106</windowTop>

12+ <windowLeft>106</windowLeft>

13+ <windowBottom>558</windowBottom>

14+ <windowRight>479</windowRight>

15+ </head>

16+ <body>

17+ <outline text="United States">

18+ <outline text="Far West">

19+ <outline text="Alaska"/>

20+ <outline text="California"/>

21+ <outline text="Hawaii"/>

22+ <outline text="<strong>Nevada</strong>" _note="I lived here *once*.

Loved it.">

23+ <outline text="Reno" created="Tue, 12 Jul 2005 23:56:35 GMT" type="link" url="http://www.reno.gov"/>

24+ <outline text="Las Vegas" created="Tue, 12 Jul 2005 23:56:37 GMT"/>

25+ <outline text="Ely" created="Tue, 12 Jul 2005 23:56:39 GMT"/>

26+ <outline text="Gerlach" created="Tue, 12 Jul 2005 23:56:47 GMT"/>

27+ </outline>

28+ <outline text="Oregon"/>

29+ <outline text="Washington"/>

30+ </outline>

31+ <outline text="Great Plains">

32+ <outline text="Kansas"/>

33+ <outline text="Nebraska"/>

34+ <outline text="North Dakota"/>

35+ <outline text="Oklahoma"/>

36+ <outline text="South Dakota"/>

37+ </outline>

38+ <outline text="Mid-Atlantic">

39+ <outline text="Delaware"/>

40+ <outline text="Maryland"/>

41+ <outline text="New Jersey"/>

42+ <outline text="New York"/>

43+ <outline text="Pennsylvania"/>

44+ </outline>

45+ <outline text="Midwest">

46+ <outline text="Illinois"/>

47+ <outline text="Indiana"/>

48+ <outline text="Iowa"/>

49+ <outline text="Kentucky"/>

50+ <outline text="Michigan"/>

51+ <outline text="Minnesota"/>

52+ <outline text="Missouri"/>

53+ <outline text="Ohio"/>

54+ <outline text="West Virginia"/>

55+ <outline text="Wisconsin"/>

56+ </outline>

57+ <outline text="Mountains">

58+ <outline text="Colorado"/>

59+ <outline text="Idaho"/>

60+ <outline text="Montana"/>

61+ <outline text="Utah"/>

62+ <outline text="Wyoming"/>

63+ </outline>

64+ <outline text="New England">

65+ <outline text="Connecticut"/>

66+ <outline text="Maine"/>

67+ <outline text="Massachusetts"/>

68+ <outline text="New Hampshire"/>

69+ <outline text="Rhode Island"/>

70+ <outline text="Vermont"/>

71+ </outline>

72+ <outline text="South">

73+ <outline text="Alabama"/>

74+ <outline text="Arkansas"/>

75+ <outline text="Florida"/>

76+ <outline text="Georgia"/>

77+ <outline text="Louisiana"/>

78+ <outline text="Mississippi"/>

79+ <outline text="North Carolina"/>

80+ <outline text="South Carolina"/>

81+ <outline text="Tennessee"/>

82+ <outline text="Virginia"/>

83+ </outline>

84+ <outline text="Southwest">

85+ <outline text="Arizona"/>

86+ <outline text="New Mexico"/>

87+ <outline text="Texas"/>

88+ </outline>

89+ </outline>

90+ </body>

91+ </opml>