873f342 · typeable/pandoc

git.iko.soy/typeable/pandoc

Ipynb reader/writer: better handling of cell metadata.

We now handle even complex cell metadata in the Div's attributes. Simple metadata fields are rendered as a plain string, and complex ones as JSON.

commit873f342f1144111f79f9415abd3dc00d44a7a3c6 authorJohn MacFarlane <jgm@berkeley.edu> dateSat 9 Mar 2019, 14:17 (8 years ago) parentbf4a164

5 files changed+24−17

compare with master

unifiedsplit

src/Text/Pandoc/Readers/Ipynb.hs+10−7 src/Text/Pandoc/Writers/Ipynb.hs+6−7 test/ipynb/simple.in.native+1−1 test/ipynb/simple.ipynb+6−1 test/ipynb/simple.out.native+1−1

src/Text/Pandoc/Readers/Ipynb.hs+10−7

@@ -18,6 +18,7 @@ Ipynb (Jupyter notebook JSON format) reader for pandoc.

1818module Text.Pandoc.Readers.Ipynb ( readIpynb )

1919where

2020import Prelude

21+import Data.Char (isDigit)

2122import Data.List (isPrefixOf)

2223import Data.Maybe (fromMaybe)

2324import Data.Digest.Pure.SHA (sha1, showDigest)

@@ -39,6 +40,7 @@ import qualified Data.ByteString.Lazy as BL

3940import Data.Aeson as Aeson

4041import Control.Monad.Except (throwError)

4142import Text.Pandoc.Readers.Markdown (readMarkdown)

43+import qualified Text.Pandoc.UTF8 as UTF8

4244

4345readIpynb :: PandocMonad m => ReaderOptions -> Text -> m Pandoc

4446readIpynb opts t = do

@@ -217,10 +219,11 @@ jsonMetaToMeta = M.mapKeys T.unpack . M.map valueToMetaValue

217219 valueToMetaValue Aeson.Null = MetaString ""

218220

219221jsonMetaToPairs :: JSONMeta -> [(String, String)]

220−jsonMetaToPairs = M.toList . M.mapMaybe

221− (\case

222− MetaString s -> Just s

223− MetaBool True -> Just "true"

224− MetaBool False -> Just "false"

225− -- for now we skip complex cell metadata:

226− _ -> Nothing) . jsonMetaToMeta

222+jsonMetaToPairs = M.toList . M.mapKeys T.unpack . M.map

223+ (\case

224+ String t

225+ | not (T.all isDigit t)

226+ , t /= "true"

227+ , t /= "false"

228+ -> T.unpack t

229+ x -> UTF8.toStringLazy $ Aeson.encode x)

src/Text/Pandoc/Writers/Ipynb.hs+6−7

@@ -213,10 +213,9 @@ extractData bs = do

213213

214214pairsToJSONMeta :: [(String, String)] -> JSONMeta

215215pairsToJSONMeta kvs =

216− M.fromList [(T.pack k, case v of

217− "true" -> Bool True

218− "false" -> Bool False

219− _ -> case safeRead v of

220− Just n -> Number n

221− _ -> String (T.pack v))

222− | (k,v) <- kvs , k /= "execution_count" ]

216+ M.fromList [(T.pack k, case Aeson.decode (UTF8.fromStringLazy v) of

217+ Just val -> val

218+ Nothing -> String (T.pack v))

219+ | (k,v) <- kvs

220+ , k /= "execution_count"

221+ ]

test/ipynb/simple.in.native+1−1

@@ -8,6 +8,6 @@ Pandoc (Meta {unMeta = fromList [("jupyter",MetaMap (fromList [("nbformat",MetaI

88 [Header 2 ("pyout",[],[]) [Str "Pyout"]]

99,Div ("",["cell","code"],[])

1010 [CodeBlock ("",["python"],[]) "from IPython.display import HTML\nHTML(\"\"\"\n<script>\nconsole.log(\"hello\");\n</script>\n<b>HTML</b>\n\"\"\")"]

11−,Div ("",["cell","markdown"],[])

11+,Div ("",["cell","markdown"],[("tags","[\"foo\",\"bar\"]")])

1212 [Header 2 ("image",[],[]) [Str "Image"]

1313 ,Para [Str "This",Space,Str "image",Space,Image ("",[],[]) [Str "the",Space,Str "moon"] ("lalune.jpg",""),Space,Str "will",Space,Str "be",Space,Str "included",Space,Str "as",Space,Str "a",Space,Str "cell",SoftBreak,Str "attachment."]]]

test/ipynb/simple.ipynb+6−1

@@ -45,7 +45,12 @@

4545 },

4646 {

4747 "cell_type": "markdown",

48− "metadata": {},

48+ "metadata": {

49+ "tags": [

50+ "foo",

51+ "bar"

52+ ]

53+ },

4954 "source": [

5055 "Image\n",

5156 "-----\n",