downstream

1.22.0 verified Fri May 01 auth: no python

downstream provides efficient, constant-space algorithms for stream curation, including deduplication, filtering, and counting. Current version 1.22.0 supports Python >=3.10. Releases are regular, with breaking changes documented in changelog.

pip install downstream

Common errors

error ImportError: cannot import name 'BloomFilter' from 'downstream' ↓

cause Trying to import from submodule (downstream.bloom) which is removed in v1.0+

fix

Use 'from downstream import BloomFilter'

error TypeError: BloomFilter.__init__() got an unexpected keyword argument 'size' ↓

cause Using old constructor signature (size, num_hashes)

fix

Use 'BloomFilter(capacity=..., error_rate=...)'

Warnings

breaking In v1.0, BloomFilter constructor changed from (size, num_hashes) to (capacity, error_rate). Old code using positional arguments will break. ↓

fix Use keyword arguments: BloomFilter(capacity=..., error_rate=...)

gotcha BloomFilter does not support deletion. Attempting to remove items silently does nothing. ↓

fix Use CountingBloomFilter if deletion is needed.

deprecated The `downstream.utils` module is deprecated since v1.15 and will be removed in v2.0. ↓

fix Import utilities directly from downstream: e.g. from downstream import hash_functions

Imports

BloomFilter

wrong

from downstream.bloom import BloomFilter

correct

from downstream import BloomFilter

BloomFilter was moved to top-level in v1.0

CountMinSketch
wrong
```
import downstream
```
correct
```
from downstream import CountMinSketch
```
Must import explicitly; module top-level exposes classes

Quickstart

Create a Bloom filter, add an item, and test membership.

from downstream import BloomFilter
bf = BloomFilter(capacity=1000, error_rate=0.01)
bf.add('item1')
print('item1' in bf)  # True