Series.searchsorted with different timezones

[kenahoo]

Code Sample, a copy-pastable example if possible

DatetimeIndex.searchsorted works fine with mixed timezones:

>>> series_dt = pd.date_range('2015-02-19', periods=2, freq='1d', tz='UTC')
>>> dt = pd.to_datetime('2015-02-19T12:34:56').tz_localize('America/New_York')
>>> series_dt.searchsorted(dt)
1

However, Series.searchsorted does not:

>>> from io import StringIO
>>> df = pd.read_csv(StringIO("datetime\n2015-02-19T00:00:00Z\n2015-02-20T00:00:00Z"), parse_dates=['datetime'])
>>> df.datetime.searchsorted(dt)
Traceback (most recent call last):
  File "/Users/kwilliams/Library/Application Support/IntelliJIdea2019.2/python/helpers/pydev/_pydevd_bundle/pydevd_exec2.py", line 3, in Exec
    exec(exp, global_vars, local_vars)
  File "<input>", line 1, in <module>
  File "venv/lib/python3.7/site-packages/pandas/core/series.py", line 2694, in searchsorted
    return algorithms.searchsorted(self._values, value, side=side, sorter=sorter)
  File "venv/lib/python3.7/site-packages/pandas/core/algorithms.py", line 1887, in searchsorted
    result = arr.searchsorted(value, side=side, sorter=sorter)
  File "venv/lib/python3.7/site-packages/pandas/core/arrays/datetimelike.py", line 666, in searchsorted
    self._check_compatible_with(value)
  File "venv/lib/python3.7/site-packages/pandas/core/arrays/datetimes.py", line 591, in _check_compatible_with
    own=self.tz, other=other.tz
ValueError: Timezones don't match. 'UTC != America/New_York'

Ostensibly, the underlying data vector is the same in both cases:

>>> series_dt.dtype
datetime64[ns, UTC]
>>> df.datetime.dtype
datetime64[ns, UTC]

Problem description

I believe the DatetimeIndex is correct (or at least more useful), because even if timezones don't agree, the underlying instants are well-ordered and compare fine. In fact, both versions compare fine using a simple > comparison:

>>> dt > series_dt
array([ True, False])
>>> dt > df.datetime
0     True
1    False
Name: datetime, dtype: bool

Expected Output

>>> df.datetime.searchsorted(dt)
1

A workaround is to wrap the column using pd.DatetimeIndex():

>>> pd.DatetimeIndex(df.datetime).searchsorted(dt)
1

Output of pd.show_versions()

INSTALLED VERSIONS
------------------
commit           : None
python           : 3.7.3.final.0
python-bits      : 64
OS               : Darwin
OS-release       : 18.0.0
machine          : x86_64
processor        : i386
byteorder        : little
LC_ALL           : None
LANG             : None
LOCALE           : en_US.UTF-8
pandas           : 0.25.1
numpy            : 1.17.2
pytz             : 2019.3
dateutil         : 2.8.0
pip              : 19.3.1
setuptools       : 41.4.0
Cython           : None
pytest           : None
hypothesis       : None
sphinx           : None
blosc            : None
##teamcity[testStdOut timestamp='2019-12-05T12:38:05.901' flowId='test.test_driver.MyTestCase.test_driver' locationHint='python://test.test_driver.MyTestCase.test_driver' name='test_driver' nodeId='75' out='feather          : None|n' parentNodeId='74']
xlsxwriter       : None
lxml.etree       : None
html5lib         : None
pymysql          : None
psycopg2         : None
jinja2           : 2.10.3
IPython          : None
pandas_datareader: None
bs4              : None
bottleneck       : None
fastparquet      : None
gcsfs            : None
lxml.etree       : None
matplotlib       : None
numexpr          : None
odfpy            : None
openpyxl         : None
pandas_gbq       : None
pyarrow          : None
pytables         : None
s3fs             : 0.3.5
scipy            : 1.3.1
sqlalchemy       : None
tables           : None
xarray           : None
##teamcity[testStdOut timestamp='2019-12-05T12:38:05.904' flowId='test.test_driver.MyTestCase.test_driver' locationHint='python://test.test_driver.MyTestCase.test_driver' name='test_driver' nodeId='75' out='xlrd             : None|n' parentNodeId='74']
xlwt             : None
xlsxwriter       : None

[jbrockmendel]

This now works on master for me, can you confirm

[mroeschke]

Confirmed. Could use a test