# PyLCG
> Linear Congruential Generator for IP Sharding

PyLCG is a Python implementation of a memory-efficient IP address sharding system using Linear Congruential Generators *(LCG)* for deterministic random number generation. This tool aids in distributed scanning & network reconnaissance by efficiently dividing IP ranges across multiple machines while being in a pseudo-random order.

___

## Table of Contents

- [Overview](#overview)
- [How It Works](#how-it-works)
    - [Understanding IP Addresses](#understanding-ip-addresses)
    - [The Magic of Linear Congruential Generators](#the-magic-of-linear-congruential-generators)
    - [Sharding: Dividing the Work](#sharding-dividing-the-work)
    - [Memory-Efficient Processing](#memory-efficient-processing)
- [Real-World Applications](#real-world-applications)
    - [Network Security Testing](#network-security-testing)
    - [Cloud-Based Scanning](#cloud-based-scanning)

___

## Overview

When performing network reconnaissance or scanning large IP ranges, it's often necessary to split the work across multiple machines. However, this presents several challenges:

1. You want to ensure each machine works on a different part of the network *(no overlap)*
2. You want to avoid scanning IPs in sequence *(which can trigger security alerts)*
3. You need a way to resume scans if a machine fails
4. You can't load millions of IPs into memory at once

PyLCG solves these challenges through clever mathematics & efficient algorithms.

___

## How It Works

### Understanding IP Addresses

First, let's understand how IP addresses work in our system:

- An IP address like `192.168.1.1` is really just a 32-bit number equal to `3232235777` or `0xC0A80101` in hexadecimal
- A CIDR range like `192.168.0.0/16` represents a continuous range of these numbers
    - For example, `192.168.0.0/16` includes all IPs from `192.168.0.0` to `192.168.255.255` *(65,536 addresses)*
    - The 32-bit number can be represented as `0xC0A80000` in hexadecimal & its from `3232235520` to `3232239103` in decimal

### The Magic of Linear Congruential Generators

At the heart of PyLCG is something called a Linear Congruential Generator *(LCG)*. Think of it as a mathematical recipe that generates a sequence of numbers that appear random but are actually predictable if you know the starting point *(seed)*.

Here's how it works:

1. Start with a number *(called the seed, which can be random)*
2. Multiply it by `1664525` & add `1013904223`
3. Take the remainder when divided by `2^32` *(the modulo operando)*
4. Repeat the process to continue the sequence

###### Mathematical notation:
```math
Next_Number = (1664525 * Current_Number + 1013904223) mod 2^32
```

###### Why these specific numbers?
The numbers `1664525` and `1013904223` are the multiplier and increment values used in a Linear Congruential Generator *(LCG)* for random number generation. This specific combination was featured in "Numerical Recipes in C" and became widely known through its use in glibc's rand() implementation.

### Sharding: Dividing the Work

PyLCG uses an interleaved sharding approach to ensure truly distributed scanning. Here's how it works:

1. **Interleaved Distribution**: Instead of dividing the IP range into sequential blocks, PyLCG distributes IPs across shards using an offset pattern:
   - For 4 shards scanning a network:
     - Shard 0 handles IPs at indices: 0, 4, 8, 12, ...
     - Shard 1 handles IPs at indices: 1, 5, 9, 13, ...
     - Shard 2 handles IPs at indices: 2, 6, 10, 14, ...
     - Shard 3 handles IPs at indices: 3, 7, 11, 15, ...

2. **Randomization**: Within each shard, the LCG randomizes the order of IPs:
   - Each index is fed through the LCG to generate a random value
   - IPs are scanned in order of these random values
   - The same seed ensures consistent ordering across runs

This approach ensures:
- Even distribution across the entire IP space
- No sequential scanning patterns that could trigger alerts
- Perfect distribution of work across shards
- Deterministic results that can be reproduced

### Memory-Efficient Processing

To handle large IP ranges without consuming too much memory, PyLCG uses several techniques:

1. **Chunked Processing**
   Instead of loading all IPs at once, it processes them in chunks.

2. **Lazy Generation**
   - IPs are generated only when needed using Python's async generators
   - The system yields one IP at a time rather than creating huge lists
   - This keeps memory usage constant regardless of IP range size

3. **Direct Calculation**
   - The LCG can jump directly to any position in its sequence
   - No need to generate all previous numbers
   - Enables efficient random access to any part of the sequence

___

## Roadmap

- [ ] Add support for IPv6
- [ ] Add support for custom LCG parameters like adding port numbers
- [ ] Add support for custom chunk sizes & auto-tuning based on available system resources
- [ ] Add support for resuming from a specific point in the sequence
- [ ] Add support for saving the state of the LCG to a file so you can resume later
- [ ] Add support for sharding line-based input files locally, from as s3 bucket, or from a URL by reading it in chunks.
- [ ] Update the unit tests to include benchmarks & better coverage for future efficiency improvements & validation.

___

###### Mirrors for this repository: [acid.vegas](https://git.acid.vegas/pylcg) • [SuperNETs](https://git.supernets.org/acidvegas/pylcg) • [GitHub](https://github.com/acidvegas/pylcg) • [GitLab](https://gitlab.com/acidvegas/pylcg) • [Codeberg](https://codeberg.org/acidvegas/pylcg)