For a petl table, how do I replace empty values with zeros?
I would expect something like the following:
tb_probii = etl.fromcsv("data.csv").fill("score", "", 0)
Looked for a similar function here: http://petl.readthedocs.io/en/latest/_modules/petl/transform/fills.html
But no luck :/
I don't know if this is the best way or not. I'm actually grateful to you for bringing the existence of petl to my attention.
>>> import petl
>>> tb_probii = petl.fromcsv('trial.csv')
>>> tb_probii
+------+-------+
| team | score |
+======+=======+
| 'A' | '' |
+------+-------+
| 'B' | '25' |
+------+-------+
| 'C' | '35' |
+------+-------+
>>> from collections import OrderedDict
>>> mappings = OrderedDict()
>>> def f(s):
... if s == '':
... return '0'
... else:
... return s
...
>>> mappings['team'] = 'team'
>>> mappings['score'] = 'score', lambda s: f(s)
>>> tb_probii = petl.fieldmap(tb_probii, mappings)
>>> tb_probii
+-------+------+
| score | team |
+=======+======+
| '0' | 'A' |
+-------+------+
| '25' | 'B' |
+-------+------+
| '35' | 'C' |
+-------+------+
Some explanation:
fieldmap executes the collection of mappings contained in an OrderedDict. When I tried this out I did the mapping to a new table. That's why team is mapped identically to itself. This may be unnecessary if you are keeping the same table, although I somehow doubt it. Each mapping is a tuple. The one for score indicates that score is to be mapped to itself with a transformation. It appears to be necessary to use a lambda; however, lambdas can't include if statements. For that reason I created the function f for the lambda to call. I think the columns are re-ordered because the container is an OrderedDict and it's ordered lexicographically on the names of the columns. Perhaps it doesn't have to be an OrderedDict but this is what I found in the doc.
I emailed the help group python-etl@googlegroups.com and the creator himself responded with a function that worked perfectly:
tb_probii = etl.fromcsv("data.csv").replace("score", "", 0)