How to create a text classifier?

Question

Machine learning

In general, the essence is quite simple. There is a set of texts in different languages.
There are two stages, the first is the definition of the quality of the text, the second is the division into categories.
First stage. The task is to divide texts into groups within the language according to the principle whether I like it or not.
The second stage is the division of the text into groups based on commonality or similarity. The simplest example is to separate poetry from prose.
Therefore, a number of questions:
1. Where to start?
2. How are these things generally done in practice? What free solutions exist for such tasks?
If someone offers solutions in PHP, I will be grateful. Python solutions are some kind of dependency hell trying to fix a piece of the OS.

Answer 1

From the Bayesian classifier:
https://github.com/loguntsov/bayes

Answer 2

https://habrahabr.ru/company/dca/blog/274027/