Timestamp losing nanosecond accuracy when reading from string

[cyber42]

Hi,

I'm on pandas 0.14.1. Timestamp seems to lose nanoseconds when converting to string and back. Wonder if this is bug.

In [18]: pd.Timestamp('20010101 00:00:00')+pd.tseries.offsets.Nano()
Out[18]: Timestamp('2001-01-01 00:00:00.000000001')

In [19]: pd.Timestamp('2001-01-01 00:00:00.000000001')
Out[19]: Timestamp('2001-01-01 00:00:00')

Thanks,
Mark

INSTALLED VERSIONS

commit: None
python: 2.7.5.final.0
python-bits: 64
OS: Linux
OS-release: 3.11.0-26-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_GB.UTF-8

pandas: 0.14.1
nose: 1.3.0
Cython: 0.19.2
numpy: 1.8.1
scipy: 0.13.2
statsmodels: 0.5.0
IPython: 0.13.2
sphinx: None
patsy: 0.2.1
scikits.timeseries: None
dateutil: 2.2
pytz: 2014.2
bottleneck: 0.8.0
tables: 3.0.0
numexpr: 2.3
matplotlib: 1.3.1
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: None
html5lib: None
httplib2: 0.8
apiclient: None
rpy2: None
sqlalchemy: None
pymysql: None
psycopg2: None

[jreback]

hmm, #7610 should have fixed this

cc @sinhrks ?

[sinhrks]

@jreback #7610 is for Timestamp instance with nanosecond. This looks a parsing problem.

@cyber42 You can use pd.to_datetime as workaround.

pd.to_datetime('2001-01-01 00:00:00.000000001')
# 2001-01-01 00:00:00.000000001

[sinhrks]

Briefly checked, there seems to be no parse function which can handle both nanosecond and tz. It looks better to create new internal parse function, and merge others.

pd.to_datetime('2001-01-01 00:00:00.000000001')
# 2001-01-01 00:00:00.000000001
pd.to_datetime('2001-01-01 00:00:00.000000001+9:00')
# 2000-12-31 15:00:00

pd.tslib.parse_date('2001-01-01 00:00:00.000000001')
# 2001-01-01 00:00:00
pd.tslib.parse_date('2001-01-01 00:00:00.000000001+9:00')
# 2001-01-01 00:00:00+09:00

# internal parse funcs
pd.tslib.parse_datetime_string('2001-01-01 00:00:00.000000001')
# 2001-01-01 00:00:00
pd.tslib.parse_datetime_string('2001-01-01 00:00:00.000000001+9:00')
# 2001-01-01 00:00:00+09:00