Parsing of time series text files with date/time split across multiple columns

[wesm]

No description provided.

[timmie]

Reference:
https://groups.google.com/forum/?fromgroups#!topic/pydata/pZjQMX_avmY

Example from tsfromtxt (http://pytseries.sourceforge.net/core.timeseries.io.html):

data = "year, month, a, b\n 2001, 01, 0.0, 10.\n 2001, 02, 1.1, 11."
datecols = (1,2,3)
dateconverter = lambda y, m, d: ts.Date('D', year=int(y), month=int(m),
day=int(d))
dateconverter = lambda y, m: Date('M', year=int(y), month=int(m))
data_ts = ts.tsfromtxt('file.csv', skiprows=1, missing='999.9',
delimiter=';', datecols=datecols, dateconverter=dateconverter)

[wesm]

This isn't addressed by #1186 and #1227 yet, need to think about parsing to periods and all that, the tsfromtxt functionality is not bad here, post #1186, the strings are concatenated together and parsed, which isn't exactly what you'd want here

[changhiskhan]

In #1186, string concatenation and then parsing is just the really naive default behavior. If you specify a date parsing function it first passes the separate columns in for custom handling and then only do concat+parse if it fails.

@timmie if you're still interested this issue might be a good way to form the start of the pandas.io.converters module. Otherwise we would also appreciate more examples on common examples you've encountered that splits datetime across multiple columns