# simple-bloom-filter
[]()
[]()
A simple implementation of Bloom Filter and Scalable Bloom Filter for Python 3.
## Installation
You can install this package from PyPI using [pip](http://www.pip-installer.org):
```
$ [sudo] pip install simplebloomfilter
```
## Example Usage
```python
#!/usr/bin/python
# -*- coding: utf-8 -*-
from bloomfilter import BloomFilter, ScalableBloomFilter, SizeGrowthRate
animals = [
"dog",
"cat",
"giraffe",
"fly",
"mosquito",
"horse",
"eagle",
"bird",
"bison",
"boar",
"butterfly",
"ant",
"anaconda",
"bear",
"chicken",
"dolphin",
"donkey",
"crow",
"crocodile",
]
other_animals = [
"badger",
"cow",
"pig",
"sheep",
"bee",
"wolf",
"fox",
"whale",
"shark",
"fish",
"turkey",
"duck",
"dove",
"deer",
"elephant",
"frog",
"falcon",
"goat",
"gorilla",
"hawk",
]
def bloom_filter_example():
print("========== Bloom Filter Example ==========")
bloom_filter = BloomFilter(size=1000, fp_prob=1e-6)
# Insert items into Bloom filter
for animal in animals:
bloom_filter.add(animal)
# Print several statistics of the filter
print(
"+ Capacity: {} item(s)".format(bloom_filter.size),
"+ Number of inserted items: {}".format(len(bloom_filter)),
"+ Filter size: {} bit(s)".format(bloom_filter.filter_size),
"+ False Positive probability: {}".format(bloom_filter.fp_prob),
"+ Number of hash functions: {}".format(bloom_filter.num_hashes),
sep="\n",
end="\n\n",
)
# Check whether an item is in the filter or not
for animal in animals + other_animals:
if animal in bloom_filter:
if animal in other_animals:
print(
f'"{animal}" is a FALSE POSITIVE case (please adjust fp_prob to a smaller value).'
)
else:
print(f'"{animal}" is PROBABLY IN the filter.')
else:
print(f'"{animal}" is DEFINITELY NOT IN the filter as expected.')
# Save to file
with open("bloom_filter.bin", "wb") as fp:
bloom_filter.save(fp)
# Load from file
with open("bloom_filter.bin", "rb") as fp:
bloom_filter = BloomFilter.load(fp)
def scalable_bloom_filter_example():
print("========== Bloom Filter Example ==========")
scalable_bloom_filter = ScalableBloomFilter(
initial_size=100,
initial_fp_prob=1e-7,
size_growth_rate=SizeGrowthRate.LARGE,
fp_prob_rate=0.9,
)
# Insert items into Bloom filter
for animal in animals:
scalable_bloom_filter.add(animal)
# Print several statistics of the filter
print(
"+ Capacity: {} item(s)".format(scalable_bloom_filter.size),
"+ Number of inserted items: {}".format(len(scalable_bloom_filter)),
"+ Number of Bloom filters: {}".format(scalable_bloom_filter.num_filters),
"+ Total size of filters: {} bit(s)".format(scalable_bloom_filter.filter_size),
"+ False Positive probability: {}".format(scalable_bloom_filter.fp_prob),
sep="\n",
end="\n\n",
)
# Check whether an item is in the filter or not
for animal in animals + other_animals:
if animal in scalable_bloom_filter:
if animal in other_animals:
print(
f'"{animal}" is a FALSE POSITIVE case (please adjust fp_prob to a smaller value).'
)
else:
print(f'"{animal}" is PROBABLY IN the filter.')
else:
print(f'"{animal}" is DEFINITELY NOT IN the filter as expected.')
# Save to file
with open("scalable_bloom_filter.bin", "wb") as fp:
scalable_bloom_filter.save(fp)
# Load from file
with open("scalable_bloom_filter.bin", "rb") as fp:
scalable_bloom_filter = ScalableBloomFilter.load(fp)
if __name__ == "__main__":
bloom_filter_example()
scalable_bloom_filter_example()
```
```
========== Bloom Filter Example ==========
+ Capacity: 1000 item(s)
+ Number of inserted items: 19
+ Filter size: 28756 bit(s)
+ False Positive probability: 1e-06
+ Number of hash functions: 20
"dog" is PROBABLY IN the filter.
"cat" is PROBABLY IN the filter.
"giraffe" is PROBABLY IN the filter.
"fly" is PROBABLY IN the filter.
"mosquito" is PROBABLY IN the filter.
"horse" is PROBABLY IN the filter.
"eagle" is PROBABLY IN the filter.
"bird" is PROBABLY IN the filter.
"bison" is PROBABLY IN the filter.
"boar" is PROBABLY IN the filter.
"butterfly" is PROBABLY IN the filter.
"ant" is PROBABLY IN the filter.
"anaconda" is PROBABLY IN the filter.
"bear" is PROBABLY IN the filter.
"chicken" is PROBABLY IN the filter.
"dolphin" is PROBABLY IN the filter.
"donkey" is PROBABLY IN the filter.
"crow" is PROBABLY IN the filter.
"crocodile" is PROBABLY IN the filter.
"badger" is DEFINITELY NOT IN the filter as expected.
"cow" is DEFINITELY NOT IN the filter as expected.
"pig" is DEFINITELY NOT IN the filter as expected.
"sheep" is DEFINITELY NOT IN the filter as expected.
"bee" is DEFINITELY NOT IN the filter as expected.
"wolf" is DEFINITELY NOT IN the filter as expected.
"fox" is DEFINITELY NOT IN the filter as expected.
"whale" is DEFINITELY NOT IN the filter as expected.
"shark" is DEFINITELY NOT IN the filter as expected.
"fish" is DEFINITELY NOT IN the filter as expected.
"turkey" is DEFINITELY NOT IN the filter as expected.
"duck" is DEFINITELY NOT IN the filter as expected.
"dove" is DEFINITELY NOT IN the filter as expected.
"deer" is DEFINITELY NOT IN the filter as expected.
"elephant" is DEFINITELY NOT IN the filter as expected.
"frog" is DEFINITELY NOT IN the filter as expected.
"falcon" is DEFINITELY NOT IN the filter as expected.
"goat" is DEFINITELY NOT IN the filter as expected.
"gorilla" is DEFINITELY NOT IN the filter as expected.
"hawk" is DEFINITELY NOT IN the filter as expected.
========== Bloom Filter Example ==========
+ Capacity: 100 item(s)
+ Number of inserted items: 19
+ Number of Bloom filters: 1
+ Total size of filters: 3355 bit(s)
+ False Positive probability: 9.999999994736442e-08
"dog" is PROBABLY IN the filter.
"cat" is PROBABLY IN the filter.
"giraffe" is PROBABLY IN the filter.
"fly" is PROBABLY IN the filter.
"mosquito" is PROBABLY IN the filter.
"horse" is PROBABLY IN the filter.
"eagle" is PROBABLY IN the filter.
"bird" is PROBABLY IN the filter.
"bison" is PROBABLY IN the filter.
"boar" is PROBABLY IN the filter.
"butterfly" is PROBABLY IN the filter.
"ant" is PROBABLY IN the filter.
"anaconda" is PROBABLY IN the filter.
"bear" is PROBABLY IN the filter.
"chicken" is PROBABLY IN the filter.
"dolphin" is PROBABLY IN the filter.
"donkey" is PROBABLY IN the filter.
"crow" is PROBABLY IN the filter.
"crocodile" is PROBABLY IN the filter.
"badger" is DEFINITELY NOT IN the filter as expected.
"cow" is DEFINITELY NOT IN the filter as expected.
"pig" is DEFINITELY NOT IN the filter as expected.
"sheep" is DEFINITELY NOT IN the filter as expected.
"bee" is DEFINITELY NOT IN the filter as expected.
"wolf" is DEFINITELY NOT IN the filter as expected.
"fox" is DEFINITELY NOT IN the filter as expected.
"whale" is DEFINITELY NOT IN the filter as expected.
"shark" is DEFINITELY NOT IN the filter as expected.
"fish" is DEFINITELY NOT IN the filter as expected.
"turkey" is DEFINITELY NOT IN the filter as expected.
"duck" is DEFINITELY NOT IN the filter as expected.
"dove" is DEFINITELY NOT IN the filter as expected.
"deer" is DEFINITELY NOT IN the filter as expected.
"elephant" is DEFINITELY NOT IN the filter as expected.
"frog" is DEFINITELY NOT IN the filter as expected.
"falcon" is DEFINITELY NOT IN the filter as expected.
"goat" is DEFINITELY NOT IN the filter as expected.
"gorilla" is DEFINITELY NOT IN the filter as expected.
"hawk" is DEFINITELY NOT IN the filter as expected.
```
## License
MIT
[]()
[]()
A simple implementation of Bloom Filter and Scalable Bloom Filter for Python 3.
## Installation
You can install this package from PyPI using [pip](http://www.pip-installer.org):
```
$ [sudo] pip install simplebloomfilter
```
## Example Usage
```python
#!/usr/bin/python
# -*- coding: utf-8 -*-
from bloomfilter import BloomFilter, ScalableBloomFilter, SizeGrowthRate
animals = [
"dog",
"cat",
"giraffe",
"fly",
"mosquito",
"horse",
"eagle",
"bird",
"bison",
"boar",
"butterfly",
"ant",
"anaconda",
"bear",
"chicken",
"dolphin",
"donkey",
"crow",
"crocodile",
]
other_animals = [
"badger",
"cow",
"pig",
"sheep",
"bee",
"wolf",
"fox",
"whale",
"shark",
"fish",
"turkey",
"duck",
"dove",
"deer",
"elephant",
"frog",
"falcon",
"goat",
"gorilla",
"hawk",
]
def bloom_filter_example():
print("========== Bloom Filter Example ==========")
bloom_filter = BloomFilter(size=1000, fp_prob=1e-6)
# Insert items into Bloom filter
for animal in animals:
bloom_filter.add(animal)
# Print several statistics of the filter
print(
"+ Capacity: {} item(s)".format(bloom_filter.size),
"+ Number of inserted items: {}".format(len(bloom_filter)),
"+ Filter size: {} bit(s)".format(bloom_filter.filter_size),
"+ False Positive probability: {}".format(bloom_filter.fp_prob),
"+ Number of hash functions: {}".format(bloom_filter.num_hashes),
sep="\n",
end="\n\n",
)
# Check whether an item is in the filter or not
for animal in animals + other_animals:
if animal in bloom_filter:
if animal in other_animals:
print(
f'"{animal}" is a FALSE POSITIVE case (please adjust fp_prob to a smaller value).'
)
else:
print(f'"{animal}" is PROBABLY IN the filter.')
else:
print(f'"{animal}" is DEFINITELY NOT IN the filter as expected.')
# Save to file
with open("bloom_filter.bin", "wb") as fp:
bloom_filter.save(fp)
# Load from file
with open("bloom_filter.bin", "rb") as fp:
bloom_filter = BloomFilter.load(fp)
def scalable_bloom_filter_example():
print("========== Bloom Filter Example ==========")
scalable_bloom_filter = ScalableBloomFilter(
initial_size=100,
initial_fp_prob=1e-7,
size_growth_rate=SizeGrowthRate.LARGE,
fp_prob_rate=0.9,
)
# Insert items into Bloom filter
for animal in animals:
scalable_bloom_filter.add(animal)
# Print several statistics of the filter
print(
"+ Capacity: {} item(s)".format(scalable_bloom_filter.size),
"+ Number of inserted items: {}".format(len(scalable_bloom_filter)),
"+ Number of Bloom filters: {}".format(scalable_bloom_filter.num_filters),
"+ Total size of filters: {} bit(s)".format(scalable_bloom_filter.filter_size),
"+ False Positive probability: {}".format(scalable_bloom_filter.fp_prob),
sep="\n",
end="\n\n",
)
# Check whether an item is in the filter or not
for animal in animals + other_animals:
if animal in scalable_bloom_filter:
if animal in other_animals:
print(
f'"{animal}" is a FALSE POSITIVE case (please adjust fp_prob to a smaller value).'
)
else:
print(f'"{animal}" is PROBABLY IN the filter.')
else:
print(f'"{animal}" is DEFINITELY NOT IN the filter as expected.')
# Save to file
with open("scalable_bloom_filter.bin", "wb") as fp:
scalable_bloom_filter.save(fp)
# Load from file
with open("scalable_bloom_filter.bin", "rb") as fp:
scalable_bloom_filter = ScalableBloomFilter.load(fp)
if __name__ == "__main__":
bloom_filter_example()
scalable_bloom_filter_example()
```
```
========== Bloom Filter Example ==========
+ Capacity: 1000 item(s)
+ Number of inserted items: 19
+ Filter size: 28756 bit(s)
+ False Positive probability: 1e-06
+ Number of hash functions: 20
"dog" is PROBABLY IN the filter.
"cat" is PROBABLY IN the filter.
"giraffe" is PROBABLY IN the filter.
"fly" is PROBABLY IN the filter.
"mosquito" is PROBABLY IN the filter.
"horse" is PROBABLY IN the filter.
"eagle" is PROBABLY IN the filter.
"bird" is PROBABLY IN the filter.
"bison" is PROBABLY IN the filter.
"boar" is PROBABLY IN the filter.
"butterfly" is PROBABLY IN the filter.
"ant" is PROBABLY IN the filter.
"anaconda" is PROBABLY IN the filter.
"bear" is PROBABLY IN the filter.
"chicken" is PROBABLY IN the filter.
"dolphin" is PROBABLY IN the filter.
"donkey" is PROBABLY IN the filter.
"crow" is PROBABLY IN the filter.
"crocodile" is PROBABLY IN the filter.
"badger" is DEFINITELY NOT IN the filter as expected.
"cow" is DEFINITELY NOT IN the filter as expected.
"pig" is DEFINITELY NOT IN the filter as expected.
"sheep" is DEFINITELY NOT IN the filter as expected.
"bee" is DEFINITELY NOT IN the filter as expected.
"wolf" is DEFINITELY NOT IN the filter as expected.
"fox" is DEFINITELY NOT IN the filter as expected.
"whale" is DEFINITELY NOT IN the filter as expected.
"shark" is DEFINITELY NOT IN the filter as expected.
"fish" is DEFINITELY NOT IN the filter as expected.
"turkey" is DEFINITELY NOT IN the filter as expected.
"duck" is DEFINITELY NOT IN the filter as expected.
"dove" is DEFINITELY NOT IN the filter as expected.
"deer" is DEFINITELY NOT IN the filter as expected.
"elephant" is DEFINITELY NOT IN the filter as expected.
"frog" is DEFINITELY NOT IN the filter as expected.
"falcon" is DEFINITELY NOT IN the filter as expected.
"goat" is DEFINITELY NOT IN the filter as expected.
"gorilla" is DEFINITELY NOT IN the filter as expected.
"hawk" is DEFINITELY NOT IN the filter as expected.
========== Bloom Filter Example ==========
+ Capacity: 100 item(s)
+ Number of inserted items: 19
+ Number of Bloom filters: 1
+ Total size of filters: 3355 bit(s)
+ False Positive probability: 9.999999994736442e-08
"dog" is PROBABLY IN the filter.
"cat" is PROBABLY IN the filter.
"giraffe" is PROBABLY IN the filter.
"fly" is PROBABLY IN the filter.
"mosquito" is PROBABLY IN the filter.
"horse" is PROBABLY IN the filter.
"eagle" is PROBABLY IN the filter.
"bird" is PROBABLY IN the filter.
"bison" is PROBABLY IN the filter.
"boar" is PROBABLY IN the filter.
"butterfly" is PROBABLY IN the filter.
"ant" is PROBABLY IN the filter.
"anaconda" is PROBABLY IN the filter.
"bear" is PROBABLY IN the filter.
"chicken" is PROBABLY IN the filter.
"dolphin" is PROBABLY IN the filter.
"donkey" is PROBABLY IN the filter.
"crow" is PROBABLY IN the filter.
"crocodile" is PROBABLY IN the filter.
"badger" is DEFINITELY NOT IN the filter as expected.
"cow" is DEFINITELY NOT IN the filter as expected.
"pig" is DEFINITELY NOT IN the filter as expected.
"sheep" is DEFINITELY NOT IN the filter as expected.
"bee" is DEFINITELY NOT IN the filter as expected.
"wolf" is DEFINITELY NOT IN the filter as expected.
"fox" is DEFINITELY NOT IN the filter as expected.
"whale" is DEFINITELY NOT IN the filter as expected.
"shark" is DEFINITELY NOT IN the filter as expected.
"fish" is DEFINITELY NOT IN the filter as expected.
"turkey" is DEFINITELY NOT IN the filter as expected.
"duck" is DEFINITELY NOT IN the filter as expected.
"dove" is DEFINITELY NOT IN the filter as expected.
"deer" is DEFINITELY NOT IN the filter as expected.
"elephant" is DEFINITELY NOT IN the filter as expected.
"frog" is DEFINITELY NOT IN the filter as expected.
"falcon" is DEFINITELY NOT IN the filter as expected.
"goat" is DEFINITELY NOT IN the filter as expected.
"gorilla" is DEFINITELY NOT IN the filter as expected.
"hawk" is DEFINITELY NOT IN the filter as expected.
```
## License
MIT
Metadata
Release files for simplebloomfilter 1.0.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | |
|---|---|---|---|---|
| simplebloomfilter-1.0.0.tar.gz | 4.8 kB | Without provenance | Details |
Release files / simplebloomfilter-1.0.0.tar.gz
| Download URL | simplebloomfilter-1.0.0.tar.gz |
|---|---|
| Size | 4.8 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
3611236e6a1b6d16033e413c205d2a00be96061e02109679eb517144f8f09137
|
|
BLAKE2b-256 checksum How to use checksums |
dc6bec90026a1eb6f06f8d6e73c5a99e9710acf3949ef1a7770406c4925073b9
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
Python-urllib/3.6
|
Security for simplebloomfilter 1.0.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.