ENH: bad directive in to_datetime format - this uses std. strptime zone offset

[ghost]

this is really bad for me since almost all linux logfiles contain zone offset.

As far as I can see there is not workaround without parsing the timestamp twice.

Code Sample, a copy-pastable example if possible

parse common Apache access log timestamp

pd.to_datetime(['28/Jul/2006:10:22:04 -0300'], format='%d/%b/%Y:%H:%M:%S %z')

...
/usr/lib/python2.7/dist-packages/pandas/tseries/tools.pyc in _convert_listlike(arg, box, format, name)
    381                 return DatetimeIndex._simple_new(values, name=name, tz=tz)
    382             except (ValueError, TypeError):
--> 383                 raise e
    384 
    385     if arg is None:

ValueError: 'z' is a bad directive in format '%d/%b/%Y:%H:%M:%S %z'

Expected Output

should parse timestamps with zone offset

output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 2.7.11.final.0
python-bits: 64
OS: Linux
OS-release: 4.4.0-24-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8

pandas: 0.17.1
nose: 1.3.7
pip: 8.1.2
setuptools: 23.0.0
Cython: None
numpy: 1.11.0
scipy: 0.17.0
statsmodels: 0.6.1
IPython: 4.2.0
sphinx: 1.3.6
patsy: 0.4.1
dateutil: 2.4.2
pytz: 2014.10
blosc: None
bottleneck: None
tables: 3.2.2
numexpr: 2.4.3
matplotlib: 1.5.1
openpyxl: 2.3.0
xlrd: 0.9.4
xlwt: 0.7.5
xlsxwriter: None
lxml: 3.5.0
bs4: 4.4.1
html5lib: 0.999
httplib2: None
apiclient: None
sqlalchemy: None
pymysql: None
psycopg2: None
Jinja2: None

[jorisvandenbossche]

Timezones %z directive are indeed not supported at the moment.

Two possible workarounds (without parsing the date in two parts, the offset separately):

• use python 3 and there datetime.datetime.strptime() has gained %z support:

  In [2]: import datetime
  
  In [3]: datetime.datetime.strptime('28/Jul/2006:10:22:04 -0300','%d/%b/%Y:%H:%M:%S %z')
  Out[3]: datetime.datetime(2006, 7, 28, 10, 22, 4, tzinfo=datetime.timezone(datetime.timedelta(-1, 75600)))
  
  In [4]: import sys
  
  In [5]: sys.version
  Out[5]: '3.5.1 |Anaconda 2.4.0 (64-bit)| (default, Feb 16 2016, 09:49:46) [MSC v.1900 64 bit (AMD64)]'
  

• use dateutil. This is able to parse timezone offsets, but apparantly not the format as you have it. If you replace the first colon with a space, it works:

  In [10]: from dateutil import parser
  
  In [11]: parser.parse('28/Jul/2006:10:22:04 -0300')
  ...
  ValueError: Unknown string format
  
  In [12]: parser.parse('28/Jul/2006:10:22:04 -0300'.replace(':', ' ', 1))
  Out[12]: datetime.datetime(2006, 7, 28, 10, 22, 4, tzinfo=tzoffset(None, -10800))
  

Both methods you could apply on the values in your datetime column. But that will not necessarily be faster than parsing the timezone offset separately (if performance is a bottleneck)

See also http://stackoverflow.com/questions/1101508/how-to-parse-dates-with-0400-timezone-string-in-python

[jreback]

actually I don't think this would be very hard to add here (and should have the single whitespace be optional)

its a pretty regular regex, and we could just be passed in directly as the tz arg (which already knows how to handle fixed offsets).

[ghost]

@jorisvandenbossche thanks for the great advice! Obviously the python 2 docu is wrong since %z is actually not supported. I learned a lot! At least for now I go with the dateutil workaround.
@jreback I am not into pandas performance but I assume this would be the fastest option. I just assumed that pandas implements the strptime format which is obviously not the case. I can imagine that others bump into that, too.

[jreback]

@markfink

just assumed that pandas implements the strptime format which is obviously not the case.

pandas DOES implement the strptime format parsing, the %z is missing