minor test fixes for ipyplotly_integration

[Kully]

• fixes minor test failures in [WIP] Plotly 3.0.0 - Deep Jupyter Integration, Validation, Performance, and More #942
  cc. @chriddyp @jmmease

Minor Test Fixes - Status

NB Tests that are not minor and require changing something in a fundamental way are still checked if all other minor tests in the module are fixed. A description/error message is provided just underneath the bullet point to specify the error message for the unsolved problem

Core

• test_tools
  • test_make_subplots
    1 F w/ test_two_row_bottom_left
  • test_get_subplots
    same issue as test_make_subplots
  • test_configuration
• test_plotly
  • test_plot
    __init__() got an unexpected keyword argument '_raise'
  • test_credentials
• test_stream
• test_offline
• test_graph_objs
  • test_annotations
    'Annotations' object has no attribute 'validate'
  • test_append_trace
    same issue as 'test_make_subplots'
  • test_data
    'Data' object has no attribute 'validate'
  • test_error_bars
  • test_figure
    __init__() missing 1 required positional argument: 'class_strs_map'
  • test_frames
    AttributeError: 'Frames' object has no attribute 'to_string'
  • test_get_figure
    AttributeError: 'Figure' object has no attribute 'get_data'
  • test_graph_objs
  • test_graph_objs_tools
    ImportError: cannot import name 'graph_objs_tools'
  • test_plotly_base_classes
    from plotly.graph_objs.graph_objs import PlotlyDict, PlotlyList
  • test_strip_style
    AttributeError: 'Figure' object has no attribute 'strip_style'
  • test_to_string
    AttributeError: 'Figure' object has no attribute 'to_string'
  • test_update
    AttributeError: 'Data' object has no attribute 'update'
• test_grid
  The 'xsrc' property is a string and must be specified as: - A string
• test_get_figure
  TypeError: __init__() got an unexpected keyword argument '_raise'
• test_api

Optional

• test_tools
  • TestQuiver
  • TestFinanceCharts
  • TestAnnotatedHeatmap
  • TestTable
  • TestGantt
  • Test2D_Density
    ('ncontours', 'colorscale', 'reversescale', 'showscale') not valid in Scatter
• test_matplotlylib
  • test_annotations
    AssertionError: ['line']['width'] = 1.5 should be 1.0 (I think the issue is just locally due to matplotlib ver ==2.2.2; our tests use matplotlib1.3.1)
  • test_axis_scales
    from plotly.tests.test_optional.test_matplotlylib.data.axis_scales import * causes the error
    

ValueError: 
    Invalid value of type 'builtins.bool' received for the 'tickmode' property of layout.xaxis
        Received value: False

    The 'tickmode' property is an enumeration that may be specified as:
      - One of the following enumeration values:
            ['auto', 'linear', 'array']

• test_figure_factory
  • TestDistplot
  • TestStreamline
  • TestDendrogram
    X = np.array([[1, 2, 3, 4], [1, 1, 3, 4], [1, 2, 1, 4], [1, 2, 3, 1]]) dendro = ff.create_dendrogram(X) fails
  • TestTrisurf
  • TestScatterPlotMatrix
  • TestGantt
  • TestViolin (deprecated messages pop up)
  • TestFacetGrid
  • TestBullet
  • TestChoropleth (not installing geopandas, pyshp and shapely)

---

Current Issues/Test failures from April 12, 2018:

/Users/adamkulidjian/Desktop/Adam/Plotly/plotly.py/plotly/tests/test_optional/test_figure_factory/test_figure_factory.py
AttributeError: 'Scatter' object has no attribute 'pop'

• no .to_dataframe (test_optional/test_graph_objs)

(venv3.6) adamkulidjian@Adams-Air:~/desktop/adam/plotly/plotly.py/plotly/tests/test_core/test_to$ clenosetests test_make_subplots.py
.......................................................F.
======================================================================
FAIL: test_two_row_bottom_left (plotly.tests.test_core.test_tools.test_make_subplots.TestMakeSubplots)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/adamkulidjian/Desktop/Adam/Plotly/plotly.py/plotly/tests/test_core/test_tools/test_make_subplots.py", line 166, in test_two_row_bottom_left
    self.assertEqual(fig.to_plotly_json(), expected.to_plotly_json())
AssertionError: {'dat[156 chars][0.0, 1.0]}, 'yaxis': {'anchor': 'x', 'domain': [0.0, 0.425]}}} != {'dat[156 chars][0.0, 1.0]}, 'yaxis': {'anchor': 'x', 'domain': [0.575, 1.0]}}}
  {'data': [],
   'layout': {'xaxis': {'anchor': 'y', 'domain': [0.0, 1.0]},
              'xaxis2': {'anchor': 'y2', 'domain': [0.0, 1.0]},
-             'yaxis': {'anchor': 'x', 'domain': [0.0, 0.425]},
?                                                    -------

+             'yaxis': {'anchor': 'x', 'domain': [0.575, 1.0]},
?                                                   +++++++

              'yaxis2': {'anchor': 'x2', 'domain': [0.575, 1.0]}}}
-------------------- >> begin captured stdout << ---------------------
This is the format of your plot grid:
[ (2,1) x2,y2 ]
[ (1,1) x1,y1 ]

-removing the to_dataframe to_string, strip_styles, get_data, and validate methods?

[Kully]

to_dataframe to_string, strip_styles, get_data, and validate methods.

Is _raise one of these methods as well?

[jonmmease]

I don't think @chriddyp and I discussed _raise at all. I don't think we're going to be able to support the _raise=False option directly because validation is pretty baked into the new object hierarchy. Could you try to understand what the use cases are for the _raise=False path? If we understand that we can make an informed decision on whether we can simply remove it, or if there is anything we need to add to continue to support these use cases.

[Kully]

@chriddyp @jmmease
The purpose of this commit is to refactor get_valid_graph_obj function to not remove invalid properties of a figure when .get_figure() is called. I think this is important because:

• it is consistent with Jon's preference to validation over coercing to default attributes as a default (which I agree with)
• if a user does want to bypass validation errors, setting raw=True is still an option in the get_figure function
• one step on the road to finishing Remove graph_objs dependency from tools.py #283 fwiw

What do y'all think?

[jonmmease]

I like this. The user will either get a dict with whatever was on the server (raw=True) or a fully valid Figure.

Do you think the stripping of invalid properties is something that regular users are going to need? I'm not clear on why we end up with figures with invalid properties on the server in the first place.

[Kully]

I wondered the same thing. In refactoring these tests I noticed that a few of the example figures I pulled from were 4 years old and contained params that are no longer part of the valid params.

[jonmmease]

I'd be inclined to update the test data, but I'll leave it to you and @chriddyp to decide whether we need to support some kind of legacy property removal.

[jonmmease]

I think these changes from xaxis1 -> xaxis in the constructors are no longer neccesary after the fix in the following commit:

Rework handling of unknown kwargs in datatype classes (a89dcb5)
@Kully

[jonmmease]

@Kully The Key error 'type' exception raised in figure_factory/_scatterplot.py should be fixed by 4fb462f

[Kully]

The Key error 'type' exception raised in figure_factory/_scatterplot.py should be fixed

it is

[Kully]

@jmmease from plotly.graph_objs.graph_objs import PlotlyDict, PlotlyList doesn't work anymore. Is there a way to import PlotlyDict and PlotlyList?

[jonmmease]

@Kully No, PlotlyDict and PlotlyList don't exist anymore

[Kully]

@Kully No, PlotlyDict and PlotlyList don't exist anymore

ok, I will remove their associated test

[jonmmease]

@Kully @chriddyp I'm ready to start working on adding some new tests against the new functionality of graph_objs. So I'm going to go ahead and merge this branch into ipyplotly_integration so I can use some of the new testing utilities.

[chriddyp]

@chriddyp and I chatted about that briefly and I think he was fine with removing the to_dataframe to_string, strip_styles, get_data, and validate methods.

Yup, that sounds right to me 👍

[Kully]

Yup, that sounds right to me 👍

Does the same go for .update?

[jonmmease]

No, .update is still supported on datatypes. I think the currently failing tests are uses of .update on the deprecated list types.