An RNG-core-based random data generator written in a cryptographic style — but lacking any cryptographic proof — which creates an information asymmetry between the program and /dev/urandom.
senol-randomizer is a Python library that wraps /dev/urandom entropy with a modular exponentiation layer using large 256-bit prime numbers. The result is a pseudorandom stream that passes all 15 NIST SP 800-22 Rev. 1a statistical tests, while deliberately not claiming cryptographic security — the transformation adds complexity at the cost of provable properties.
pip install senol-randomizer- Entropy source: Reads 256-bit raw values from
/dev/urandom. - Transformation: Applies
powmod(a, b, n)via thegmpy2library — performing modular exponentiation for each value of n. - Prime moduli: The set of n values consists exclusively of 256-bit prime numbers.
- Derived types: All higher-level data types (
str,float,bool, etc.) are derived from this core output.
The design intentionally introduces an asymmetry: the consumer of the output has less information about the internal state than /dev/urandom alone would expose, but this property is structural, not mathematically proven.
| Function | Description | Arguments | Returns |
|---|---|---|---|
RNG() |
Core function. Generates a random large integer with no range constraint. | — | int |
newint() |
Random integer within a given range. | min_val: int, max_val: int |
int |
newfloat() |
Random float within a given range. | min_val: int, max_val: int |
float |
newbool() |
Random boolean. | — | bool |
newbyte() |
Random bytes of a given length. | num: int |
bytes |
newstr() |
Random string sampled from the full Unicode range. | num: int, min_val: int, max_val: int |
str |
newtoken() |
Random URL-safe string. | length: int |
str |
choice() |
Selects a random element from a list. | lst: list |
any |
shuffle() |
Shuffles a list in-place and returns it. O(n). | lst: list |
list |
compress() |
Maps a value into [min, max] while minimizing modulo bias. | value, min_val: int, max_val: int |
int |
Results from the NIST SP 800-22 Rev. 1a test suite, run against 1 Mibibit (33,554,432 bits) of RNG() output.
| Test | P-Value | Result |
|---|---|---|
monobit_test |
0.046613138703915244 | ✅ PASS |
frequency_within_block_test |
0.6787903492122158 | ✅ PASS |
runs_test |
0.33232036052397196 | ✅ PASS |
longest_run_ones_in_a_block_test |
0.21020703185565603 | ✅ PASS |
binary_matrix_rank_test |
0.2571877713800565 | ✅ PASS |
dft_test |
0.8685773273998374 | ✅ PASS |
non_overlapping_template_matching_test |
0.893926430778068 | ✅ PASS |
overlapping_template_matching_test |
0.7796009868841686 | ✅ PASS |
maurers_universal_test |
0.9802530560443485 | ✅ PASS |
linear_complexity_test |
0.4685290247356243 | ✅ PASS |
serial_test |
0.5630465232357214 | ✅ PASS |
approximate_entropy_test |
0.5631130245829565 | ✅ PASS |
cumulative_sums_test |
0.039939937090386124 | ✅ PASS |
random_excursion_test |
0.5178616373360564 | ✅ PASS |
random_excursion_variant_test |
0.14694060874229203 | ✅ PASS |
15 / 15 PASS — Work in progress.
GitHub Repository · PyPI Package
Author on GitHub · Author on PyPI
— Batuhan Şenol