Include clearer warning regarding about preferred data input

As a new user of altair I’ve been regularly surprised/confused by the preferred data format being pandas. Gradually I’ve gotten the hang of it, but it would be really great if the warning message that describes altair not supporting JSON was a little bit clearer (or if json was supported (maybe it is I am a new user?)).

For instance this innocous code chunk

alt.Chart(
  [{"x": 1, "y": 2}, {"x": 2, "y": 3}, {"x": 3, "y": 1}]
).mark_circle(
  size=50, color='steelblue', opacity=1
).encode(
  x='x',
  y='y',
)

yields a pretty long error:

/usr/local/lib/python3.6/dist-packages/altair/vegalite/v3/api.py:97: UserWarning: data of type <class 'list'> not recognized
  warnings.warn("data of type {} not recognized".format(type(data)))
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
/usr/local/lib/python3.6/dist-packages/altair/vegalite/v3/api.py in to_dict(self, *args, **kwargs)
    367 
    368         try:
--> 369             dct = super(TopLevelMixin, copy).to_dict(*args, **kwargs)
    370         except jsonschema.ValidationError:
    371             dct = None

8 frames
/usr/local/lib/python3.6/dist-packages/altair/vegalite/v3/schema/channels.py in to_dict(self, validate, ignore, context)
     46                                      "the type cannot be automatically inferred because "
     47                                      "the data is not specified as a pandas.DataFrame."
---> 48                                      "".format(self.shorthand))
     49         else:
     50             # Shorthand is not a string; we pass the definition to field,

ValueError: x encoding field is specified without a type; the type cannot be automatically inferred because the data is not specified as a pandas.DataFrame.
Chart({
  data: [{'x': 1, 'y': 2}, {'x': 2, 'y': 3}, {'x': 3, 'y': 1}],
  encoding: FacetedEncoding({
    x: X({
      shorthand: 'x'
    }),
    y: Y({
      shorthand: 'y'
    })
  }),
  mark: MarkDef({
    color: 'steelblue',
    opacity: 1,
    size: 50,
    type: 'circle'
  })
})

While it is true that the error lists “list” as unsupported, it would be cool if the error also included a hint on how to fix it (eg "list is unsupported, please use a pandas dataframe).

[jakevdp]

You can use JSON directly if you specify that the data are values:

alt.Chart(
  {'values': [{"x": 1, "y": 2}, {"x": 2, "y": 3}, {"x": 3, "y": 1}]}
).mark_circle(
  size=50, color='steelblue', opacity=1
).encode(
  x='x:Q',
  y='y:Q',
)

If you can think of a way that the warning and error shown in the output above could be made clearer, I’d happily review a pull request!

[mcnuttandrew]

Amazing!!! That was so quick!