I love itertools; it's one of the most beautiful, composable standard libs. "Aha! I have an iteration problem here; I'm sure there is an itertools routine that fits it perfectly" oft passes my lips. Often, my confidence is well-placed, but sometimes, neither itertools nor the recipes included in its docs do quite what I need.
Here I've collected several routines I've reached for but not found. Since these are deceptively tricky to get right, I thought I'd wrap them up into a library. Enjoy! Any additions are welcome; just file a pull request.
- chunked(iterable, n)
Break an iterable into tuples of a given length.
chunked([1, 2, 3, 4, 5, 6, 7], 3) --> [(1, 2, 3), (4, 5, 6), (7,)]
If the length of iterable is not evenly divisible by n, the last returned tuple will be shorter.
Wrapper for an iterator to allow 1-item lookahead
peekable(iterator).peek() returns the value that will next pop out of next().
- collate(*iterables[, key=<key function>, reverse=<bool>])
Return an iterable ordered collation of the already-sorted items from each of iterables, compared by kwarg key.
If reverse=True is passed, iterables must return their results in descending order rather than ascending.
More Itertools is under the MIT License. See the LICENSE file.
This uses nose for tests. To run the tests, do:
It should also be possible to say python setup.py test. However, some part of the test runner throws an error after the tests pass.
- Initial release, with collate, peekable, and chunked. Could really use better docs.