SQL-style joins for Python iterables.
>>> from join import join, merge, tuple_join
>>> dogs = [
... Dog('gatsby', 'Rruff!', 16),
... Dog('ein', 'wruf!', 9),
... ]
>>> cats = [
... Cat('pleo', 'mreeeoww', 16),
... Cat('xena', 'mreow', 12),
... Cat('gatsby', 'rowr', 15),
... ]
>>> catdogs = merge(cats, dogs, key='name')
>>> catdogs
[CatDog({'right': Dog(name='gatsby', woof='Rruff!', weight=16), 'name': 'gatsby', 'weight': 15, 'meow': 'rowr', 'woof': 'Rruff!', 'left': Cat(name='gatsby', meow='rowr', weight=15)})]
>>> catdogs[0].meow
'rowr'
>>> catdogs[0].woof
'Rruff!'
Install using pip:
pip install join
join
does the work of associating iterable items together, but gives you all the power for customization, letting you supply your own join function, separate keys for left and right iterables, and even letting you use functions instead of attribute names.
merge
used above, for example, is join
using an object union to join matched objects. You can use a tuple join, which is default for join
:
>>> join(cats, dogs, key='name', join_fn=tuple_join)
[(Cat(name='gatsby', meow='rowr', weight=15), Dog(name='gatsby', woof='Rruff!', weight=16))]
Supplying your own join function is easy:
>>> def weight_sum(left, right):
... return left.weight + right.weight
...
>>> join(cats, dogs, key='name', join_fn=weight_sum)
[31]
Using separate key functions is easy too:
>>> def cat_key(cat):
... return cat.weight % 3 == 0 # weight divisible by 3
...
>>> def dog_key(dog):
... return dog.weight % 4 == 0 # weight divisible by 4
...
>>> def name_join(left, right):
... return left.name + '-' + right.name
...
>>> join(cats, dogs, left_key=cat_key, right_key=dog_key, join_fn=name_join)
['pleo-ein', 'xena-gatsby', 'gatsby-gatsby']
The much larger library Toolz contains a similar, but less flexible, join
function.
Pull requests and issues welcome! Run tests from project root with sh run_tests.sh
.