list of dictionaries processing - readability and complexity

Question

I have basic question about comprehension. There is list of dicts where values are lists, it looks like this:

listionary = [{'path': ['/tmp/folder/cat/number/letter', '/tmp/folder/hog/char/number/letter', '/tmp/folder/hog/number/letter', '/etc'], 
'mask': True, 
'name': 'dict-1'}, 
{'path': ['/tmp/folder/dog/number-2/letter-4', '/tmp/folder/hog-00/char/number-1/letter-5', '/tmp/folder/cow/number-2/letter-3'], 
'mask': True, 
'name': 'dict-2'}, 
{'path': ['/tmp/folder/dog_111/number/letter', '/tmp/folder/ant/char/number/letter', '/tmp/folder/hen/number/letter'], 
'mask': True, 
'name': 'dict-3'}]

What I need is to get from list-type values every unique animal. Animal is always between tmp/folder/ and next /. What I did:

import re
flat_list = [item for sublist in [i['path'] for i in listionary] for item in sublist]
animals = list(set([re.search('folder/([a-z]+)', elem).group(1) for elem in flat_list if 'tmp' in elem]))

It might be also compressed into one line, but it is quite complicated and unreadable:

animals = list(set([re.search('folder/([a-z]+)', elem).group(1) for elem in [item for sublist in [i['path'] for i in listionary] for item in sublist] if 'tmp' in elem]))

Is there any golden rule(e.g. zen of python) about size of comprehension? How can I make it better? Thank you in advance.

Answer 1

How can I make it better?

1. have someone else read it. ✓
2. use functions to encapsulate more complex operations
3. don't nest loops on the same line

Here's how I would break down the last two points..

def get_animals(d):
    animals = []
    for item in d['path']:
        if item.startswith('/tmp/folder/'):
            animals.append(item[12:item.find('/',12)])
    return animals

animals = set()
for d in dlist:
    animals.update(get_animals(d))
animals = list(animals)