[MAINTENANCE] Improve Altair plotting extensibility

great_expectations/rule_based_profiler/types/altair/__init__.py

@@ -1,2 +1,2 @@
-from .default_configuration import ALTAIR_DEFAULT_CONFIGURATION
 from .encodings import AltairDataTypes
+from .themes import AltairThemes

great_expectations/rule_based_profiler/types/altair/encodings.py

@@ -4,13 +4,15 @@
 
 
 class AltairDataTypes(Enum):
+    # available data types: https://altair-viz.github.io/user_guide/encoding.html#encoding-data-types
     QUANTITATIVE = alt.StandardType("quantitative")
     ORDINAL = alt.StandardType("ordinal")
     NOMINAL = alt.StandardType("nominal")
     TEMPORAL = alt.StandardType("temporal")
 
 
 class AltairAggregates(Enum):
+    # available aggregates: https://altair-viz.github.io/user_guide/encoding.html#encoding-channel-options
     MEAN = alt.Aggregate("mean")
     MEDIAN = alt.Aggregate("median")
     MIN = alt.Aggregate("min")

great_expectations/rule_based_profiler/types/altair/themes.py

@@ -0,0 +1,114 @@
+from enum import Enum
+from typing import List
+
+from great_expectations.types import ColorPalettes, Colors
+
+# Size
+chart_width: int = 800
+chart_height: int = 250
+
+# Font
+font: str = "Verdana"
+
+#
+# Chart Components
+#
+
+# Title
+title_align: str = "center"
+title_font_size: int = 15
+title_color: str = Colors.PURPLE.value
+
+# Both Axes
+axis_title_color: str = Colors.PURPLE.value
+axis_title_font_size: int = 14
+axis_title_padding: int = 10
+axis_label_color: str = Colors.BLUE_1.value
+axis_label_font_size: int = 12
+
+# X-Axis Only
+x_axis_label_angle: int = 0
+x_axis_label_flush: bool = True
+x_axis_grid: bool = True
+# Known vega-lite bug: https://github.com/vega/vega-lite/issues/5732
+# forces us to choose between features "interactive scaling" (below) and "tooltips"
+# x_axis_selection_type: str = "interval"
+# x_axis_selection_bind: str = "scales"
+
+# Y-Axis Only
+
+#
+# Color Palettes
+#
+
+category_color_scheme: List[str] = ColorPalettes.CATEGORY.value
+diverging_color_scheme: List[str] = ColorPalettes.DIVERGING.value
+heatmap_color_scheme: List[str] = ColorPalettes.HEATMAP.value
+ordinal_color_scheme: List[str] = ColorPalettes.ORDINAL.value
+
+#
+# Chart Types
+#
+
+# Area
+fill_opacity = 0.5
+
+# Line Chart
+line_color: str = Colors.BLUE_1.value
+line_stroke_width: int = 3
+line_opacity: float = 0.9
+# Known vega-lite bug: https://github.com/vega/vega-lite/issues/5732
+# forces us to choose between features "interactive scaling" and "tooltips" (below)
+line_tooltip_content: str = "data"
+
+# Point
+point_size: int = 70
+point_color: str = Colors.GREEN.value
+point_filled: bool = True
+point_opacity: float = 1.0
+point_tooltip_content: str = "data"
+
+
+class AltairThemes(Enum):
+    DEFAULT_THEME = {
+        "view": {"width": chart_width, "height": chart_height},
+        "font": font,
+        "title": {
+            "align": title_align,
+            "color": title_color,
+            "fontSize": title_font_size,
+        },
+        "axis": {
+            "titleFontSize": axis_title_font_size,
+            "titleColor": axis_title_color,
+            "titlePadding": axis_title_padding,
+            "labelFontSize": axis_label_font_size,
+            "labelColor": axis_label_color,
+        },
+        "axisX": {
+            "labelAngle": x_axis_label_angle,
+            "labelFlush": x_axis_label_flush,
+            "grid": x_axis_grid,
+        },
+        "range": {
+            "category": category_color_scheme,
+            "diverging": diverging_color_scheme,
+            "heatmap": heatmap_color_scheme,
+            "ordinal": ordinal_color_scheme,
+        },
+        "area": {
+            "fillOpacity": fill_opacity,
+        },
+        "line": {
+            "color": line_color,
+            "strokeWidth": line_stroke_width,
+            "tooltip": {"content": line_tooltip_content},
+        },
+        "point": {
+            "size": point_size,
+            "color": point_color,
+            "filled": point_filled,
+            "opacity": point_opacity,
+            "tooltip": {"content": point_tooltip_content},
+        },
+    }

great_expectations/rule_based_profiler/types/data_assistant_result/data_assistant_result.py

@@ -1,3 +1,4 @@
+import copy
 from abc import abstractmethod
 from dataclasses import asdict, dataclass
 from typing import Any, Dict, List, Optional
@@ -6,16 +7,16 @@
 import pandas as pd
 
 from great_expectations.core import ExpectationSuite
-from great_expectations.core.util import convert_to_json_serializable
+from great_expectations.core.util import convert_to_json_serializable, nested_update
 from great_expectations.rule_based_profiler.types import (
     FULLY_QUALIFIED_PARAMETER_NAME_ATTRIBUTED_VALUE_KEY,
     FULLY_QUALIFIED_PARAMETER_NAME_METADATA_KEY,
     Domain,
     ParameterNode,
 )
 from great_expectations.rule_based_profiler.types.altair import (
-    ALTAIR_DEFAULT_CONFIGURATION,
     AltairDataTypes,
+    AltairThemes,
 )
 from great_expectations.types import ColorPalettes, Colors, SerializableDictDot
 
@@ -64,15 +65,21 @@ def get_attributed_metrics_by_domain(
         return metrics_attributed_values_by_domain
 
     @staticmethod
-    def display(charts: List[alt.Chart]) -> None:
+    def display(charts: List[alt.Chart], theme: Optional[Dict[str, Any]]) -> None:
         """
         Display each chart passed by DataAssistantResult.plot()
 
         Args:
             charts: A list of altair chart objects to display
+            theme: Altair top-level chart configuration dictionary
         """
+        altair_configuration: Dict[str, Any] = copy.deepcopy(
+            AltairThemes.DEFAULT_THEME.value
+        )
+        if theme is not None:
+            nested_update(altair_configuration, theme)
+
         chart: alt.Chart
-        altair_configuration: Dict[str, Any] = ALTAIR_DEFAULT_CONFIGURATION
         for chart in charts:
             chart.configure(**altair_configuration).display()
 
@@ -83,8 +90,6 @@ def get_line_chart(
         metric_type: alt.StandardType,
         domain_name: str,
         domain_type: alt.StandardType,
-        line_color: Optional[str] = Colors.BLUE_2.value,
-        point_color: Optional[str] = Colors.GREEN.value,
         point_color_condition: Optional[alt.condition] = None,
         tooltip: Optional[List[alt.Tooltip]] = None,
     ) -> alt.Chart:
@@ -95,8 +100,6 @@ def get_line_chart(
             metric_type: The altair data type for the metric being plotted
             domain_name: The name of the domain as it exists in the pandas dataframe
             domain_type: The altair data type for the domain being plotted
-            line_color: Hex code for the line color
-            point_color: Hex code for the point color
             point_color_condition: Altair condition for changing the point color
             tooltip: Altair tooltip for displaying relevant information on the chart
 
@@ -118,7 +121,7 @@ def get_line_chart(
 
         line: alt.Chart = (
             alt.Chart(data=df, title=title)
-            .mark_line(color=line_color)
+            .mark_line()
             .encode(
                 x=alt.X(
                     domain_name,
@@ -133,23 +136,22 @@ def get_line_chart(
         if point_color_condition is not None:
             points: alt.Chart = (
                 alt.Chart(data=df, title=title)
-                .mark_point(opacity=1.0)
+                .mark_point()
                 .encode(
                     x=alt.X(
                         domain_name,
                         type=domain_type,
                         title=domain_title,
                     ),
                     y=alt.Y(metric_name, type=metric_type, title=metric_title),
-                    stroke=point_color_condition,
-                    fill=point_color_condition,
+                    color=point_color_condition,
                     tooltip=tooltip,
                 )
             )
         else:
             points: alt.Chart = (
                 alt.Chart(data=df, title=title)
-                .mark_point(stroke=point_color, fill=point_color, opacity=1.0)
+                .mark_point()
                 .encode(
                     x=alt.X(
                         domain_name,
@@ -182,9 +184,7 @@ def get_expect_domain_values_to_be_between_chart(
         Returns:
             An altair line chart with confidence intervals corresponding to "between" expectations
         """
-        line_opacity: float = 0.9
         line_color: alt.HexColor = alt.HexColor(ColorPalettes.HEATMAP.value[4])
-        fill_opacity: float = 0.5
         fill_color: alt.HexColor = alt.HexColor(ColorPalettes.HEATMAP.value[5])
 
         metric_title: str = metric_name.replace("_", " ").title()
@@ -206,7 +206,7 @@ def get_expect_domain_values_to_be_between_chart(
 
         lower_limit: alt.Chart = (
             alt.Chart(data=df)
-            .mark_line(color=line_color, opacity=line_opacity)
+            .mark_line(color=line_color)
             .encode(
                 x=alt.X(
                     domain_name,
@@ -220,7 +220,7 @@ def get_expect_domain_values_to_be_between_chart(
 
         upper_limit: alt.Chart = (
             alt.Chart(data=df)
-            .mark_line(color=line_color, opacity=line_opacity)
+            .mark_line(color=line_color)
             .encode(
                 x=alt.X(
                     domain_name,
@@ -234,7 +234,7 @@ def get_expect_domain_values_to_be_between_chart(
 
         band: alt.Chart = (
             alt.Chart(data=df)
-            .mark_area(fill=fill_color, fillOpacity=fill_opacity)
+            .mark_area(fill=fill_color)
             .encode(
                 x=alt.X(
                     domain_name,
@@ -246,7 +246,7 @@ def get_expect_domain_values_to_be_between_chart(
             )
         )
 
-        predicate = (
+        predicate: alt.expr.core.BinaryExpression = (
             (alt.datum.min_value > alt.datum.table_row_count)
             & (alt.datum.max_value > alt.datum.table_row_count)
         ) | (
@@ -273,12 +273,14 @@ def get_expect_domain_values_to_be_between_chart(
     @abstractmethod
     def plot(
         self,
-        prescriptive: bool = False,
+        prescriptive: Optional[bool] = False,
+        theme: Optional[Dict[str, Any]] = None,
     ) -> None:
         """
         Use contents of "DataAssistantResult" object to display mentrics and other detail for visualization purposes.
 
         Args:
-            prescriptive: Type of plot to generate.
+            prescriptive: Type of plot to generate, prescriptive if True, descriptive if False
+            theme: Altair top-level chart configuration dictionary
         """
         pass