How to programmatically catch which command failed on a try block

Question

I'm trying to get some data from an JSON API. I don't want all the data that the API returns so I wrote a method that reads all the data and returns a dictionary with the relevant fields. Sometimes though, some data are missing and I would like to replace the fields that are missing with an underscore. A sample of the method is like that;

return {
    'foo': data['foo'],
    'bar': data['bar']
 }

If a field is missing from the data, this throughs a KeyError. Is it possible to catch programmatically which field produced the error, in a single try-except block and not write a try-except block for every field?

try:
    ret_dict =  {
        'foo': data['foo'],
        'bar': data['bar']
    }
 except KeyError:
    ret_dict[thefailurekey] = '_'

instead of

ret_dict = {}
try:
    ret_dict['foo'] = data['foo']
except KeyError:
    ret_dict['foo'] = '_'
try:
    ret_dict['bar'] = data['bar']
except:
    ret_dict['bar'] = '_'

Thank you

score 3 · Answer 1 · answered Oct 02 '16 at 18:55

3

You can probably get that information from the members of the KeyError exception object, but a simpler way would be to just use get() that will return a default value if the key is not there.

return {
    'foo': data.get('foo', '_'),
    'bar': data.get('bar', '_'),
 }

Another reason this is better than handling an exception is that you can only handle one exception. What happens if two keys are missing? And on top of that, ret_dict will not even be defined in your example because the code failed.

answered Oct 02 '16 at 18:55

kichik

33,220
7
94
114

Thank you for your answer. Some fields are themselves dictionaries, which means that I need to do something like data['foo']['bar']. Is there a better solution from data.get('foo',{}).get('bar','_')? – jzlas Oct 02 '16 at 19:03
I haven't seen anything better than that. – kichik Oct 02 '16 at 19:07

score 2 · Answer 2 · answered Oct 02 '16 at 18:58

2

Instead of using try block, you can use dict.get(key, default_val)

For example:

ret_dict = dict(
    foo=data.get('foo', '-'),
    bar=data.get('bar', '-')
)

return ret_dict

answered Oct 02 '16 at 18:58

Anant Bhardwaj

131
5

Thank you for your answer. If the field is itself a dictionary, is this good practice; data.get('foo',{}).get('bar','_')? – jzlas Oct 02 '16 at 19:06
yes, that would work but a better approach is to use `defaultdict` for that – Anant Bhardwaj Oct 02 '16 at 19:10
I didn't know about this, I'll check it out. Thanks again! – jzlas Oct 02 '16 at 19:14

skovorodkin · Accepted Answer · 2016-10-02T19:06:32.483

2

Use .get method of dict:

def get_data(data):
    return {
        # If you want to accept falsy values from API:
        'foo': data.get('foo', '_'),

        # If you want to override falsy values from API:
        'bar': data.get('bar') or '_',
    }

.get returns its second argument (None by default) if a dict doesn't have requested key, so it is always safe to use it in uncertain situations.

Example:

>>> data = {'foo': False, 'bar': False}
>>> get_data(data)
{'bar': '_', 'foo': False}

edited Oct 02 '16 at 19:06

answered Oct 02 '16 at 19:00

skovorodkin

9,394
1
39
30

Thank you for your answer. If the field is itself a dictionary, should I use get like that; data.get('foo',{}).get('bar','_') or is there a better-more pythonic way? – jzlas Oct 02 '16 at 19:05
Yes, that's a way to go. Though if you want a more generic method, look at this answer http://stackoverflow.com/a/36131992/847552. – skovorodkin Oct 02 '16 at 19:10
Thank you, that's great. Doesn't get more pythonic than that! – jzlas Oct 02 '16 at 19:11

score 1 · Answer 4 · answered Oct 02 '16 at 19:26

If you want to avoid the repetitiveness of typing .get(attr, '_') for each key, you can use a defaultdict, setting it to return _ when a key is trying to be accessed but missing.

from collections import defaultdict

data = {
    'foo': 'foo_value',
}

ret_dict = defaultdict(lambda: '_')
ret_dict.update(data)

print(ret_dict['foo'])  # 'foo_value'
print(ret_dict['bar'])  # '_'

How to programmatically catch which command failed on a try block

4 Answers4