Explicit column dtype specification in read_* functions

[wesm]

e.g. columns with values like 01001 are getting converted to int

example from mailing list:

df = read_csv('test_data.csv')
df.head()
     oid   did mode             ox             oy      dx      dy
0  1001  1001   01  272311.659358  176751.822655  272675  176375
1  1001  1001   01  272311.659358  176751.822655  272375  176375
2  1001  1001   01  272311.659358  176751.822655  272125  176675
3  1001  1001   06  272311.659358  176751.822655  272675  177125
4  1001  1001   06  272311.659358  176751.822655  272675  176375

df.oid = df.oid.apply(lambda x: str(x).zfill(5))
df.head()
     oid   did mode             ox             oy      dx      dy
0  01001  1001   01  272311.659358  176751.822655  272675  176375
1  01001  1001   01  272311.659358  176751.822655  272375  176375
2  01001  1001   01  272311.659358  176751.822655  272125  176675
3  01001  1001   06  272311.659358  176751.822655  272675  177125
4  01001  1001   06  272311.659358  176751.822655  272675  176375

[wesm]

This is done in c-parser (dtype={'oid': object}) but needs a unit test

[wesm]

This works now:

In [11]: df = read_clipboard(delim_whitespace=True, dtype={'oid': 'O', 'did': 'O', 'mode': 'O'}); df
Out[11]: 
     oid    did mode             ox             oy      dx      dy
0  01001  01001   01  272311.659358  176751.822655  272675  176375
1  01001  01001   01  272311.659358  176751.822655  272375  176375
2  01001  01001   01  272311.659358  176751.822655  272125  176675
3  01001  01001   06  272311.659358  176751.822655  272675  177125
4  01001  01001   06  272311.659358  176751.822655  272675  176375

This needs to be able to accept more than just format strings though (e.g. 'f8'). I'll do that then close this issue.