fix errors in pandas 1.5 and address comments

Author: chelsea-lin

bigframes/dataframe.py

@@ -59,7 +59,7 @@
 import bigframes.formatting_helpers as formatter
 import bigframes.operations as ops
 import bigframes.operations.aggregations as agg_ops
-import bigframes.operations.plot as plot
+import bigframes.operations.plotting as plotting
 import bigframes.series
 import bigframes.series as bf_series
 import bigframes.session._io.bigquery
@@ -3196,6 +3196,6 @@ def get_right_id(id):
 
     @property
     def plot(self):
-        return plot.PlotAccessor(self)
+        return plotting.PlotAccessor(self)
 
     __matmul__ = dot

bigframes/operations/_matplotlib/core.py

@@ -12,13 +12,12 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from abc import ABC, abstractmethod
-
+import abc
 import matplotlib.pyplot as plt
 
 
-class MPLPlot(ABC):
-    @abstractmethod
+class MPLPlot(abc.ABC):
+    @abc.abstractmethod
     def generate(self):
         pass
 

bigframes/operations/_matplotlib/hist.py

@@ -19,10 +19,10 @@
 import pandas as pd
 
 import bigframes.constants as constants
-from bigframes.operations._matplotlib.core import MPLPlot
+import bigframes.operations._matplotlib.core as bfplt
 
 
-class HistPlot(MPLPlot):
+class HistPlot(bfplt.MPLPlot):
     @property
     def _kind(self) -> Literal["hist"]:
         return "hist"
@@ -54,44 +54,53 @@ def __init__(
         self.data = self._compute_plot_data(data)
 
     def generate(self) -> None:
-        hist_bars = self._calculate_hist_bar(self.data, self.bins)
-
-        bin_edges = None
-        hist_x = {}
-        weights = {}
-        for col_name, hist_bar in hist_bars.items():
-            left = hist_bar.index.get_level_values("left_exclusive")
-            right = hist_bar.index.get_level_values("right_inclusive")
-
-            hist_x[col_name] = pd.Series((left + right) / 2.0)
-            weights[col_name] = hist_bar.values
-            if bin_edges is None:
-                bin_edges = left.union(right)
-            else:
-                bin_edges = left.union(right).union(bin_edges)
-
-        bins = None
-        if bin_edges is not None:
-            _, bins = np.histogram(
-                bin_edges, bins=self.bins, range=self.kwargs.get("range", None)
+        """
+        Calculates weighted histograms through BigQuery and plots them through pandas
+        native histogram plot.
+        """
+        hist_bars = self._calculate_hist_bars(self.data, self.bins)
+        bin_edges = self._calculate_bin_edges(
+            hist_bars, self.bins, self.kwargs.get("range", None)
+        )
+
+        weights = {
+            col_name: hist_bar.values for col_name, hist_bar in hist_bars.items()
+        }
+        hist_x = {
+            col_name: pd.Series(
+                (
+                    hist_bar.index.get_level_values("left_exclusive")
+                    + hist_bar.index.get_level_values("right_inclusive")
+                )
+                / 2.0
             )
+            for col_name, hist_bar in hist_bars.items()
+        }
 
-        # Fills with NA values when items have different lengths.
-        ordered_columns = self.data.columns.values
+        # Align DataFrames for plotting despite potential differences in column
+        # lengths, filling shorter columns with zeros.
         hist_x_pd = pd.DataFrame(
             list(itertools.zip_longest(*hist_x.values())), columns=list(hist_x.keys())
-        ).sort_index(axis=1)
+        ).sort_index(axis=1)[self.data.columns.values]
         weights_pd = pd.DataFrame(
             list(itertools.zip_longest(*weights.values())), columns=list(weights.keys())
-        ).sort_index(axis=1)
+        ).sort_index(axis=1)[self.data.columns.values]
+        hist_x_pd.fillna(0, inplace=True)
+        weights_pd.fillna(0, inplace=True)
 
-        self.axes = hist_x_pd[ordered_columns].plot.hist(
-            bins=bins,
-            weights=np.array(weights_pd[ordered_columns].values),
+        self.axes = hist_x_pd.plot.hist(
+            bins=bin_edges,
+            weights=np.array(weights_pd.values),
             **self.kwargs,
         )  # type: ignore
 
     def _compute_plot_data(self, data):
+        """
+        Prepares data for plotting, focusing on numeric data types.
+
+        Raises:
+            TypeError: If the input data contains no numeric columns.
+        """
         # Importing at the top of the file causes a circular import.
         import bigframes.series as series
 
@@ -118,7 +127,13 @@ def _compute_plot_data(self, data):
         return numeric_data
 
     @staticmethod
-    def _calculate_hist_bar(data, bins):
+    def _calculate_hist_bars(data, bins):
+        """
+        Calculates histogram bars for each column in a BigFrames DataFrame, and
+        returns a dictionary where keys are column names and values are pandas
+        Series. The series values are the histogram bins' heights with a
+        multi-index defining 'left_exclusive' and 'right_inclusive' bin edges.
+        """
         import bigframes.pandas as bpd
 
         # TODO: Optimize this by batching multiple jobs into one.
@@ -132,3 +147,23 @@ def _calculate_hist_bar(data, bins):
                 .sort_index(level="left_exclusive")
             )
         return hist_bar
+
+    @staticmethod
+    def _calculate_bin_edges(hist_bars, bins, range):
+        """
+        Calculate bin edges from the histogram bars.
+        """
+        bin_edges = None
+        for _, hist_bar in hist_bars.items():
+            left = hist_bar.index.get_level_values("left_exclusive")
+            right = hist_bar.index.get_level_values("right_inclusive")
+            if bin_edges is None:
+                bin_edges = left.union(right)
+            else:
+                bin_edges = left.union(right).union(bin_edges)
+
+        if bin_edges is None:
+            return None
+
+        _, bins = np.histogram(bin_edges, bins=bins, range=range)
+        return bins

bigframes/operations/plot.py renamed to bigframes/operations/plotting.py

@@ -18,7 +18,6 @@
 import bigframes.operations._matplotlib as plotbackend
 import third_party.bigframes_vendored.pandas.plotting._core as vendordt
 
-
 class PlotAccessor:
     __doc__ = vendordt.PlotAccessor.__doc__
 

bigframes/series.py

@@ -50,7 +50,7 @@
 import bigframes.operations.aggregations as agg_ops
 import bigframes.operations.base
 import bigframes.operations.datetimes as dt
-import bigframes.operations.plot as plot
+import bigframes.operations.plotting as plotting
 import bigframes.operations.strings as strings
 import bigframes.operations.structs as structs
 import third_party.bigframes_vendored.pandas.core.series as vendored_pandas_series
@@ -1560,7 +1560,7 @@ def str(self) -> strings.StringMethods:
 
     @property
     def plot(self):
-        return plot.PlotAccessor(self)
+        return plotting.PlotAccessor(self)
 
     def _slice(
         self,

docs/reference/bigframes.pandas/frame.rst

@@ -7,3 +7,14 @@ DataFrame
     :members:
     :inherited-members:
     :undoc-members:
+
+Accessors
+---------
+
+Plotting handling
+^^^^^^^^^^^^^^^
+
+.. automodule:: bigframes.operations.plotting
+    :members:
+    :inherited-members:
+    :undoc-members:

docs/reference/bigframes.pandas/series.rst

@@ -42,3 +42,11 @@ Struct handling
     :members:
     :inherited-members:
     :undoc-members:
+
+Plotting handling
+^^^^^^^^^^^^^^^
+
+.. automodule:: bigframes.operations.plotting
+    :members:
+    :inherited-members:
+    :undoc-members:

third_party/bigframes_vendored/pandas/core/frame.py

@@ -5231,7 +5231,7 @@ def plot(self):
         Make plots of Dataframes.
 
         Returns:
-            bigframes.operations.plot.PlotAccessor:
+            bigframes.operations.plotting.PlotAccessor:
                 An accessor making plots.
         """
         raise NotImplementedError(constants.ABSTRACT_METHOD_ERROR_MESSAGE)

third_party/bigframes_vendored/pandas/core/series.py

@@ -3117,7 +3117,7 @@ def plot(self):
         Make plots of Series.
 
         Returns:
-            bigframes.operations.plot.PlotAccessor:
+            bigframes.operations.plotting.PlotAccessor:
                 An accessor making plots.
         """
         raise NotImplementedError(constants.ABSTRACT_METHOD_ERROR_MESSAGE)