aboutsummaryrefslogtreecommitdiff
path: root/src/Text/Pandoc/Readers/Ipynb.hs
blob: 9d46111c33c8bc043c32890e352a48ddc6e401e7 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
{-# LANGUAGE LambdaCase #-}
{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE FlexibleContexts  #-}
{-# LANGUAGE ScopedTypeVariables #-}
{- |
   Module      : Text.Pandoc.Readers.Ipynb
   Copyright   : Copyright (C) 2019-2021 John MacFarlane
   License     : GNU GPL, version 2 or above

   Maintainer  : John MacFarlane <jgm@berkeley.edu>
   Stability   : alpha
   Portability : portable

Ipynb (Jupyter notebook JSON format) reader for pandoc.
-}
module Text.Pandoc.Readers.Ipynb ( readIpynb )
where
import Data.Char (isDigit)
import Data.Maybe (fromMaybe)
import Data.Digest.Pure.SHA (sha1, showDigest)
import Text.Pandoc.Options
import qualified Data.Scientific as Scientific
import qualified Text.Pandoc.Builder as B
import Text.Pandoc.Logging
import Text.Pandoc.Definition
import Data.Ipynb as Ipynb
import Text.Pandoc.Class.PandocMonad
import Text.Pandoc.MIME (extensionFromMimeType)
import Text.Pandoc.Shared (tshow)
import Text.Pandoc.UTF8
import Text.Pandoc.Walk (walk)
import Text.Pandoc.Error
import Data.Text (Text)
import qualified Data.Map as M
import qualified Data.Text as T
import qualified Data.Text.Encoding as TE
import qualified Data.ByteString.Lazy as BL
import Data.Aeson as Aeson
import Control.Monad.Except (throwError)
import Text.Pandoc.Readers.Markdown (readMarkdown)
import qualified Text.Pandoc.UTF8 as UTF8
import Text.Pandoc.Sources (ToSources(..), sourcesToText)

readIpynb :: (PandocMonad m, ToSources a)
          => ReaderOptions -> a -> m Pandoc
readIpynb opts x = do
  let src = BL.fromStrict . TE.encodeUtf8 . sourcesToText $ toSources x
  case eitherDecode src of
    Right (notebook4 :: Notebook NbV4) -> notebookToPandoc opts notebook4
    Left _ ->
      case eitherDecode src of
        Right (notebook3 :: Notebook NbV3) -> notebookToPandoc opts notebook3
        Left err -> throwError $ PandocIpynbDecodingError $ T.pack err

notebookToPandoc :: PandocMonad m
                 => ReaderOptions -> Notebook a -> m Pandoc
notebookToPandoc opts notebook = do
  let cells = notebookCells notebook
  let (fmt,fmtminor) = notebookFormat notebook
  let m = M.insert "nbformat" (MetaString $ tshow fmt) $
          M.insert "nbformat_minor" (MetaString $ tshow fmtminor) $
          jsonMetaToMeta (notebookMetadata notebook)
  let lang = case M.lookup "kernelspec" m of
                   Just (MetaMap ks) ->
                      case M.lookup "language" ks of
                         Just (MetaString l) -> l
                         _ -> "python"
                   _ -> "python"
  bs <- mconcat <$> mapM (cellToBlocks opts lang) cells
  let Pandoc _ blocks = B.doc bs
  return $ Pandoc (Meta $ M.insert "jupyter" (MetaMap m) mempty) blocks

cellToBlocks :: PandocMonad m
             => ReaderOptions -> Text -> Ipynb.Cell a -> m B.Blocks
cellToBlocks opts lang c = do
  let Source ts = cellSource c
  let source = mconcat ts
  let kvs = jsonMetaToPairs (cellMetadata c)
  let attachments = maybe mempty M.toList $ cellAttachments c
  mapM_ addAttachment attachments
  case cellType c of
    Ipynb.Markdown -> do
      bs <- if isEnabled Ext_raw_markdown opts
               then return [RawBlock (Format "markdown") source]
               else do
                 Pandoc _ bs <- walk fixImage <$> readMarkdown opts source
                 return bs
      return $ B.divWith ("",["cell","markdown"],kvs)
             $ B.fromList bs
    Ipynb.Heading lev -> do
      Pandoc _ bs <- readMarkdown opts
        (T.replicate lev "#" <> " " <> source)
      return $ B.divWith ("",["cell","markdown"],kvs)
             $ B.fromList bs
    Ipynb.Raw -> do
      -- we use ipynb to indicate no format given (a wildcard in nbformat)
      let format = fromMaybe "ipynb" $ lookup "format" kvs
      let format' =
            case format of
              "text/html"       -> "html"
              "text/latex"      -> "latex"
              "application/pdf" -> "latex"
              "text/markdown"   -> "markdown"
              "text/x-rsrt"     -> "rst"
              _                 -> format
      return $ B.divWith ("",["cell","raw"],kvs) $ B.rawBlock format' source
    Ipynb.Code{ codeOutputs = outputs, codeExecutionCount = ec } -> do
      outputBlocks <- mconcat <$> mapM outputToBlock outputs
      let kvs' = maybe kvs (\x -> ("execution_count", tshow x):kvs) ec
      return $ B.divWith ("",["cell","code"],kvs') $
        B.codeBlockWith ("",[lang],[]) source
        <> outputBlocks

-- Remove attachment: prefix from images...
fixImage :: Inline -> Inline
fixImage (Image attr lab (src,tit))
  | "attachment:" `T.isPrefixOf` src = Image attr lab (T.drop 11 src, tit)
fixImage x = x

addAttachment :: PandocMonad m => (Text, MimeBundle) -> m ()
addAttachment (fname, mimeBundle) = do
  let fp = T.unpack fname
  case M.toList (unMimeBundle mimeBundle) of
    (mimeType, BinaryData bs):_ ->
      insertMedia fp (Just mimeType) (BL.fromStrict bs)
    (mimeType, TextualData t):_ ->
      insertMedia fp (Just mimeType)
          (BL.fromStrict $ TE.encodeUtf8 t)
    (mimeType, JsonData v):_ ->
      insertMedia fp (Just mimeType) (encode v)
    [] -> report $ CouldNotFetchResource fname "no attachment"

outputToBlock :: PandocMonad m => Output a -> m B.Blocks
outputToBlock Stream{ streamName = sName,
                      streamText = Source text } =
  return $ B.divWith ("",["output","stream",sName],[])
         $ B.codeBlock $ T.concat text
outputToBlock DisplayData{ displayData = data',
                            displayMetadata = metadata' } =
  B.divWith ("",["output", "display_data"],[]) <$>
    handleData metadata' data'
outputToBlock ExecuteResult{ executeCount = ec,
                              executeData = data',
                              executeMetadata = metadata' } =
  B.divWith ("",["output", "execute_result"],[("execution_count",tshow ec)])
    <$> handleData metadata' data'
outputToBlock Err{ errName = ename,
                   errValue = evalue,
                   errTraceback = traceback } =
  return $ B.divWith ("",["output","error"],
                         [("ename",ename),
                          ("evalue",evalue)])
         $ B.codeBlock $ T.unlines traceback

-- We want to display the richest output possible given
-- the output format.
handleData :: PandocMonad m
           => JSONMeta -> MimeBundle -> m B.Blocks
handleData metadata (MimeBundle mb) =
  mconcat <$> mapM dataBlock (M.toList mb)

  where

    dataBlock :: PandocMonad m => (MimeType, MimeData) -> m B.Blocks
    dataBlock (mt, d)
     | "image/" `T.isPrefixOf` mt || mt == "application/pdf"
      = do
      -- normally metadata maps from mime types to key-value map;
      -- but not always...
      let meta = case M.lookup mt metadata of
                   Just v@Object{} ->
                     case fromJSON v of
                       Success m' -> m'
                       Error _   -> mempty
                   _ -> mempty
      let metaPairs = jsonMetaToPairs meta
      let bl = case d of
                 BinaryData bs  -> BL.fromStrict bs
                 TextualData t  -> BL.fromStrict $ UTF8.fromText t
                 JsonData v     -> encode v
      -- SHA1 hash for filename
      let fname = T.pack (showDigest (sha1 bl)) <>
            case extensionFromMimeType mt of
              Nothing  -> ""
              Just ext -> "." <> ext
      insertMedia (T.unpack fname) (Just mt) bl
      return $ B.para $ B.imageWith ("",[],metaPairs) fname "" mempty

    dataBlock ("text/html", TextualData t)
      = return $ B.rawBlock "html" t

    dataBlock ("text/latex", TextualData t)
      = return $ B.rawBlock "latex" t

    dataBlock ("text/markdown", TextualData t)
      = return $ B.rawBlock "markdown" t

    dataBlock ("text/plain", TextualData t) =
      return $ B.codeBlock t

    dataBlock (_, JsonData v) =
      return $ B.codeBlockWith ("",["json"],[]) $ T.pack $ toStringLazy $ encode v

    dataBlock _ = return mempty

jsonMetaToMeta :: JSONMeta -> M.Map Text MetaValue
jsonMetaToMeta = M.map valueToMetaValue
  where
    valueToMetaValue :: Value -> MetaValue
    valueToMetaValue x@Object{} =
      case fromJSON x of
        Error s -> MetaString $ T.pack s
        Success jm' -> MetaMap $ jsonMetaToMeta jm'
    valueToMetaValue x@Array{} =
      case fromJSON x of
        Error s -> MetaString $ T.pack s
        Success xs -> MetaList $ map valueToMetaValue xs
    valueToMetaValue (Bool b) = MetaBool b
    valueToMetaValue (String t) = MetaString t
    valueToMetaValue (Number n)
      | Scientific.isInteger n = MetaString (tshow (floor n :: Integer))
      | otherwise              = MetaString (tshow n)
    valueToMetaValue Aeson.Null = MetaString ""

jsonMetaToPairs :: JSONMeta -> [(Text, Text)]
jsonMetaToPairs = M.toList . M.map
  (\case
      String t
        | not (T.all isDigit t)
        , t /= "true"
        , t /= "false"
                 -> t
      x          -> T.pack $ UTF8.toStringLazy $ Aeson.encode x)