Integrate _normalize_data with existing infrastructure

Author: kdorr

mplaltair/_convert.py

@@ -1,6 +1,6 @@
 import matplotlib.dates as mdates
 import numpy as np
-from ._data import _locate_channel_data, _locate_channel_dtype, _convert_to_mpl_date
+from ._data import _locate_channel_data, _locate_channel_dtype, _convert_to_mpl_date, _normalize_data
 
 def _allowed_ranged_marks(enc_channel, mark):
     """TODO: DOCS
@@ -110,6 +110,8 @@ def _convert(chart):
     """
     mapping = {}
 
+    _normalize_data(chart)
+
     if not chart.to_dict().get('encoding'):
         raise ValueError("Encoding not provided with the chart specification")
 

mplaltair/_data.py

@@ -8,18 +8,17 @@
 
 from ._utils import _fetch
 
-def _normalize_data(spec):
+def _normalize_data(chart):
     """Converts the data to a Pandas dataframe
 
     Parameters
     ----------
-    spec : dict
+    chart : altair.Chart
     The vega-lite specification in json format
 
     Returns
     -------
-    dict
-    The vega-lite specification with the data format fixed to a Pandas dataframe
+    None
 
     Raises
     ------
@@ -30,20 +29,19 @@ def _normalize_data(spec):
     Raised when the data specification has an unsupported data source
     """
 
+    spec = chart.to_dict()
+
     if not spec.get('data'):
         raise ValidationError('Please specify a data source.')
 
     if spec['data'].get('url'):
         df = pd.DataFrame(_fetch(spec['data']['url']))
     elif spec['data'].get('values'):
-        df = pd.DataFrame(spec['data']['values'])
+        return
     else:
         raise NotImplementedError('Given data specification is unsupported at the moment.')
 
-    del spec['data']
-    spec['data'] = df
-
-    return spec
+    chart.data = df
 
 def _locate_channel_dtype(chart, channel):
     """Locates dtype used for each channel

mplaltair/tests/test_data.py

@@ -18,35 +18,14 @@
 })
 
 def test_data_list():
-    spec = {
-        "data": {
-            "values": [{"a": 1, "b": 2}, {"c": 3, "d": 4}]
-
-        }
-    }
-    assert type(_normalize_data(spec)["data"]) == pd.DataFrame
+    chart = alt.Chart(pd.DataFrame({'a': [1], 'b': [2], 'c': [3]})).mark_point()
+    _normalize_data(chart)
+    assert type(chart.data) == pd.DataFrame
 
 def test_data_url():
-    spec = {
-        "data": {
-            "url": data.cars.url
-        }
-    }
-    assert type(_normalize_data(spec)["data"]) == pd.DataFrame
-
-def test_data_no_pass():
-    spec = {}
-    with pytest.raises(ValidationError):
-        _normalize_data(spec)
-
-def test_data_invalid():
-    spec = {
-        "data": {
-            "source": "path"
-        }
-    }
-    with pytest.raises(NotImplementedError):
-        _normalize_data(spec)
+    chart = alt.Chart(data.cars.url).mark_point()
+    _normalize_data(chart)
+    assert type(chart.data) == pd.DataFrame
 
 # _locate_channel_data() tests