Add HTML repr for groupby dataframe and series

pandas/core/config_init.py

@@ -109,6 +109,12 @@ def use_numba_cb(key):
  correct auto-detection.
 """
 
+pc_max_groups_doc = """
+: int
+ If max_groups is exceeded, switch to truncate groupby view. 'None' value
+ means unlimited.
+"""
+
 pc_min_rows_doc = """
 : int
  The numbers of rows to show in a truncated view (when `max_rows` is
@@ -354,6 +360,9 @@ def is_terminal() -> bool:
  validator=is_instance_factory((int, type(None))),
  )
  cf.register_option("max_rows", 60, pc_max_rows_doc, validator=is_nonnegative_int)
+ cf.register_option(
+ "max_groups", 10, pc_max_groups_doc, validator=is_nonnegative_int
+ )
  cf.register_option(
  "min_rows",
  10,

pandas/core/groupby/groupby.py

@@ -76,6 +76,8 @@ class providing the base-class of operations.
 from pandas.core.sorting import get_group_index_sorter
 from pandas.core.util.numba_ import NUMBA_FUNC_CACHE
 
+from pandas.io.formats.format import repr_html_groupby
+
 _common_see_also = """
  See Also
  --------
@@ -570,6 +572,9 @@ def __repr__(self) -> str:
  # TODO: Better repr for GroupBy object
  return object.__repr__(self)
 
+ def _repr_html_(self) -> str:
+ return repr_html_groupby(self)
+
  def _assure_grouper(self) -> None:
  """
  We create the grouper on instantiation sub-classes may have a

pandas/io/formats/format.py

@@ -2020,3 +2020,39 @@ def buffer_put_lines(buf: IO[str], lines: List[str]) -> None:
  if any(isinstance(x, str) for x in lines):
  lines = [str(x) for x in lines]
  buf.write("\n".join(lines))
+
+
+def repr_html_groupby(group_obj) -> str:
+ """
+ Create an HTML representation for a grouped DataFrame or Series.
+
+ Parameters
+ ----------
+ group_obj : [DataFrameGroupBy, SeriesGroupBy]
+ Object to make HTML representation of.
+ Returns
+ -------
+ str :
+ HTML representation of the input object.
+ """
+ max_groups = get_option("display.max_groups")
+ max_rows = max(
+ 1, get_option("display.max_rows") // min(max_groups, group_obj.ngroups)
+ )
+ group_names = list(group_obj.groups.keys())
+ truncated = max_groups < group_obj.ngroups
+ if truncated:
+ n_start = (max_groups + 1) // 2
+ n_end = max_groups - n_start
+ group_names = group_names[:n_start] + group_names[-n_end:]
+ repr_html_list = list()
+ for group_name in group_names:
+ group = group_obj.get_group(group_name)
+ if not hasattr(group, "to_html"):
+ group = group.to_frame()
+ repr_html_list.append(
+ f"<H3>Group Key: {group_name}<H3/>\n{group.to_html(max_rows=max_rows)}"
+ )
+ if truncated:
+ repr_html_list.insert(max_groups // 2, "<H3>...<H3/>")
+ return "\n".join(repr_html_list)

pandas/tests/groupby/test_groupby.py

@@ -6,6 +6,7 @@
 import pytest
 
 from pandas.errors import PerformanceWarning
+import pandas.util._test_decorators as td
 
 import pandas as pd
 from pandas import DataFrame, Index, MultiIndex, Series, Timestamp, date_range, read_csv
@@ -2059,6 +2060,49 @@ def test_groups_repr_truncates(max_seq_items, expected):
  assert result == expected
 
 
+@td.skip_if_no("lxml")
+@pytest.mark.parametrize(
+ "n_groups,n_rows,check_n_groups,check_n_rows",
+ [
+ (10, 60, 5, 3), # All groups and all rows in the groups are shown
+ (25, 100, 5, 2), # Not all groups are shown
+ (4, 400, 2, 7), # Not all rows are shown in the groups
+ (20, 400, 5, 3), # Not all groups and not all rows in the groups are shown
+ ],
+)
+def test_groupby_repr_not_all_groups(n_groups, n_rows, check_n_groups, check_n_rows):
+ # GH 34926
+ df = DataFrame(
+ {
+ "A": range(n_rows),
+ "B": range(0, n_rows * 2, 2),
+ "C": list(range(n_groups)) * (n_rows // n_groups),
+ }
+ )
+
+ df_groupby = df.groupby("C")
+ html_groupby = df_groupby._repr_html_()
+
+ df_from_html = pd.concat(pd.read_html(StringIO(html_groupby), index_col=0))
+
+ # Drop "..." rows and convert index and data to int
+ df_from_html = df_from_html[df_from_html.index != "..."].astype(int)
+ df_from_html.index = df_from_html.index.astype(int)
+
+ # Iterate over the first and last "check_n_groups" groups
+ df_group_iter = (
+ list(df_groupby)[:check_n_groups] + list(df_groupby)[-check_n_groups:]
+ )
+ for group_name, df_group in df_group_iter:
+ # Iterate over the first and last "check_n_rows" of every group
+ df_iter = pd.concat(
+ [df_group.iloc[:check_n_rows], df_group.iloc[-check_n_rows:]]
+ ).iterrows()
+ for index, row in df_iter:
+ print(group_name, index)
+ tm.assert_series_equal(row, df_from_html.loc[index])
+
+
 def test_group_on_two_row_multiindex_returns_one_tuple_key():
  # GH 18451
  df = DataFrame([{"a": 1, "b": 2, "c": 99}, {"a": 1, "b": 2, "c": 88}])