dequedict 0.1.3

Fast ordered dict with deque-like operations at both ends

DequeDict

Ordered dictionary with O(1) deque operations at both ends.

Features

• O(1) key lookup — like dict
• O(1) pop/peek from both ends — like deque
• O(1) appendleft, move_to_end, delete by key
• Full typing support with .pyi stubs

Installation

pip install dequedict

Usage

from dequedict import DequeDict, DefaultDequeDict

dd = DequeDict([("a", 1), ("b", 2), ("c", 3)])

dd["d"] = 4              # Set
value = dd["a"]          # Get
del dd["b"]              # Delete

dd.peekleft()            # First value
dd.popleft()             # Remove first
dd.appendleft("z", 0)    # Insert at front
dd.move_to_end("a")      # Move to back

# DefaultDequeDict: auto-create values like defaultdict
groups = DefaultDequeDict(list)
groups["a"].append(1)
groups["a"].append(2)

API

Method	Description
peekleft() / peek()	First/last value
peekleftitem() / peekitem()	First/last (key, value)
popleft() / pop()	Remove and return first/last
popleftitem() / popitem()	Remove and return first/last pair
appendleft(key, value)	Insert at front
move_to_end(key, last=True)	Move to front or back
get, keys, values, items, clear, copy, update, setdefault	Standard dict ops

Performance

Mac M1, Python 3.11, C extension:

Operation	DequeDict	dict	deque	OrderedDict
Key lookup	32 ns	23 ns	O(n)	24 ns
Peek left	22 ns	N/A	23 ns	59 ns
Peek right	22 ns	N/A	23 ns	70 ns
Pop left ×100	3.7 µs	N/A	2.1 µs	7.7 µs
Delete by key ×100	4.3 µs	2.8 µs	O(n)	6.0 µs
Insert ×100	6.6 µs	9.5 µs	N/A	6.8 µs

Pop is ~1.8x slower than deque due to dict maintenance, but provides O(1) key lookup.

Benchmarks

python benchmarks/benchmark.py

Tests

pip install pytest
python -m pytest tests/ -v

License

MIT