I now know a couple of ways: this, if a simple captcha, cut the captcha into pieces so that each piece has a symbol, then write the md5 hash of each symbol into an array along with the symbol itself. An array would look something like this:
$cap['c93d3bf7a7c4afe94b64e30c2ce39f4f'] = "2";
And then, when you need to recognize a captcha, you will need to cut it into characters and recognize the md5 of each character, compare it with the array and substitute it. But with such useless captcha sites of the unit. Therefore, there is a more horrible way: download all the captcha project and get their md5 hash. Then compare. And the values ​​of captcha will be hammered by users who want to "earn some extra money", they are given a picture, they enter, the project from which captcha, md5 captcha and meaning is written into the database. And for the introduced captcha they are paid a penny. Example: kolotibablo.com.
Now the questions:
But how on such services as "KolotiBablo" check whether the user entered the captcha correctly? Are there any other ways to recognize a captcha?
How do services like antigate.com work?
Where can I practice in cURL, otherwise strong protections are everywhere, in no way can I find a suitable training ground, and for what purpose, besides sending forms, does this library use?