Merge #207

207: Dequantify duplicate cols r=andrewgsavage a=MichaelTiemannOSC

Use enumeration to wade through duplicate column names.  This does not preserve default column names when dequantifying, but it doesn't break.  Should `pint.dequantify()` also preserve duplicated column names?

- [x] Closes #202 
- [x] Executed `pre-commit run --all-files` with no errors
- [x] The change is fully covered by automated unit tests
- [x] Documented in docs/ as appropriate
- [x] Added an entry to the CHANGES file


Co-authored-by: Michael Tiemann <72577720+MichaelTiemannOSC@users.noreply.github.com>

CHANGES

@@ -4,6 +4,7 @@ pint-pandas Changelog
 0.6 (unreleased)
 ----------------
 
+- Fix dequantify duplicate column failure #202
 - Fix astype issue #196
 
 

pint_pandas/pint_array.py

@@ -1,7 +1,6 @@
 import copy
 import re
 import warnings
-from collections import OrderedDict
 from importlib.metadata import version
 
 import numpy as np
@@ -981,23 +980,35 @@ def formatter_func(dtype):
 
  df_columns = df.columns.to_frame()
  df_columns["units"] = [
- formatter_func(df[col].dtype)
- if isinstance(df[col].dtype, PintType)
+ formatter_func(df.dtypes.iloc[i])
+ if isinstance(df.dtypes.iloc[i], PintType)
  else NO_UNIT
- for col in df.columns
+ for i, col in enumerate(df.columns)
  ]
 
- data_for_df = OrderedDict()
+ data_for_df = []
  for i, col in enumerate(df.columns):
- if isinstance(df[col].dtype, PintType):
- data_for_df[tuple(df_columns.iloc[i])] = df[col].values.data
+ if isinstance(df.dtypes.iloc[i], PintType):
+ data_for_df.append(
+ pd.Series(
+ data=df.iloc[:, i].values.data,
+ name=tuple(df_columns.iloc[i]),
+ index=df.index,
+ copy=False,
+ )
+ )
  else:
- data_for_df[tuple(df_columns.iloc[i])] = df[col].values
-
- df_new = DataFrame(data_for_df, columns=data_for_df.keys())
+ data_for_df.append(
+ pd.Series(
+ data=df.iloc[:, i].values,
+ name=tuple(df_columns.iloc[i]),
+ index=df.index,
+ copy=False,
+ )
+ )
 
+ df_new = pd.concat(data_for_df, axis=1, copy=False)
  df_new.columns.names = df.columns.names + ["unit"]
- df_new.index = df.index
 
  return df_new
 

pint_pandas/testsuite/test_issues.py

@@ -194,3 +194,38 @@ def test_issue_194(dtype):
  s2 = s1.astype(dtype)
 
  tm.assert_series_equal(s0, s2)
+
+
+class TestIssue202(BaseExtensionTests):
+ def test_dequantify(self):
+ df = pd.DataFrame()
+ df["test"] = pd.Series([1, 2, 3], dtype="pint[kN]")
+ df.insert(0, "test", df["test"], allow_duplicates=True)
+
+ expected = pd.DataFrame.from_dict(
+ data={
+ "index": [0, 1, 2],
+ "columns": [("test", "kilonewton")],
+ "data": [[1], [2], [3]],
+ "index_names": [None],
+ "column_names": [None, "unit"],
+ },
+ orient="tight",
+ dtype="Int64",
+ )
+ result = df.iloc[:, 1:].pint.dequantify()
+ tm.assert_frame_equal(expected, result)
+
+ expected = pd.DataFrame.from_dict(
+ data={
+ "index": [0, 1, 2],
+ "columns": [("test", "kilonewton"), ("test", "kilonewton")],
+ "data": [[1, 1], [2, 2], [3, 3]],
+ "index_names": [None],
+ "column_names": [None, "unit"],
+ },
+ orient="tight",
+ dtype="Int64",
+ )
+ result = df.pint.dequantify()
+ tm.assert_frame_equal(expected, result)