PHP: Text Spinner Class - Nested spinning supported.

spintax.php

<?PHP

/**
 * Spintax - A helper class to process Spintax strings.
 */
class Spintax
{
    /**
     * Set seed to make the spinner predictable. 
     */
    public function seed(mixed $seed)
    {
        mt_srand(crc32($seed));
    }

    public function process(string $text)
    {
        return preg_replace_callback(
            '/\{(((?>[^\{\}]+)|(?R))*?)\}/x',
            function ($match) {
                $text = $this->process($match[1]);
                $parts = explode('|', $text);

                return $parts[mt_rand(0, count($parts) - 1)];
            },
            $text
        );
    }

    /**
     * If you want highest randomness,
     * this is the method to use.
     * Note: Seed has no effect.
     *
     * Last updated: 2023-10-26
     */
    public function spin(string $text)
    {
        $pattern = '/{([^{}]+)}/';
        while (preg_match($pattern, $text)) {
            $text = preg_replace_callback($pattern, function ($matches) {
                $options = explode('|', $matches[1]);

                return $options[ random_int(0, count($options) - 1) ];
            }, $text);
        }

        return $text;
    }
}


/* EXAMPLE USAGE */
$spintax = new Spintax();
$string = '{Hello|Howdy|Hola} to you, {Mr.|Mrs.|Ms.} {Smith|Williams|Davis}!';
echo $spintax->process($string);

/* NESTED SPINNING EXAMPLE */
echo $spintax->process('{Hello|Howdy|Hola} to you, {Mr.|Mrs.|Ms.} {{Jason|Malina|Sara}|Williams|Davis}');

Perfectly working. Thank you so much for the explanation about the regexp.
Regexp are really cool, but not so easy to understand/explain, also has a really HUGE syntax :)

Simple but powerful :)

Thanks

[ask] it is possible to spin the title also ?

If you want to use the [square|brackets|syntax] use this line in the process function:

'/\[(((?>[^\[\]]+)|(?R))*)\]/x',

thanks @lury

Thank for very helpful code.
How can I use PHP to find match text in data then replace
example my content:
$content = 'The speedy black wolf bounded over the lazy hound';
$patten = array('{the|a|}',{black|red|yellow}....);

Expect result after replace : $content = '{the|a|} speedy {black|red|yellow} wolf bounded over the lazy hound';
they will find 1 of some word in {} and if 1 of some word by | match, they will replace to the content.

thank you for this

Is there way to ignore spin for "first capital alphabet word" (e,g. names, country, etc.)?

hi
i love this scripts. is any ideas on how can it be implement with mysql data? like storing the synonyms in a database

more power

hi, if i want to calculate maximum number of my spintax combination can generate unique article how to do that?

Is there a way I could show all the possible variations?

The problem with this kind of spinners is that spuns do not have the same probability to be chosen.
For example, something like '{A|{B|C|D|E|F}}' will have 50% of 'A'.
I've coded another one with Python where every possible spuns have the same probability to be chosen, also you can retrieve the number of every possible spuns that can be generated from a spintax string.
You can have a look here : https://gist.github.com/fbparis/738c52e253639edb3cb41f90ea0b490c

Very interesting. Pretty cool. As it is very slim in size and heavy in working.

great! very useful - thanks!

is posible? {
Mensaje001
|Mensaje 002
{
Mensaje 002-1
|
Mensaje 002-2
}
}
To infinity data spintax.

Hello! Generally class working great even with 3rd or 4th nesting level. But i think there is some mysterious bug.

Spintax works fine with short texts, but when text is longer than 8.000 character (whole text including synonims in brackets {}) then class cause HTTP errors (connecion reset).

I make a test in separated file where was only class definition, one object and variable with text.

Can somone help me with this?

In php ini i have declared 2GB of ram and unlimited time of execution.

Regards!

I guess we could make the asterisk lazy for a little performance boost '/{(((?>[^\{\}]+)|(?R))*?)}/x'.

/\{(((?>[^\{\}]+)|(?R))*)\}/x 816 matches, 11424 steps (~30ms)

/\{(((?>[^\{\}]+)|(?R))*?)\}/x 816 matches, 9792 steps (~13ms)

I guess we could make the asterisk lazy for a little performance boost '/{(((?>[^{}]+)|(?R))*?)}/x'.

I forked this optimized solution and added seed rand in order to fix the spin text with srand.

Hello!

Great job. It is helping me a lot!

I'm trying to get the parts of the spintax text separately, but I only get the ones between {}

For example, in this text:
{Hi|Hello}{ Mr.|} Irfaq Syed. {How are you?|How about the day?}

I get:
{Hi | Hello}
{Mr. |}
{How are you? | How about the day?}

But not the text in the middle, that is, I don't get:

Irfaq Syed.

There is any Regex to get all, the spintax groups and the plain text?

Thanks!

Very interesting!
How to make the reverse?
Input: Hello to you, Mr. Smith!
Output: {Hello|Howdy|Hola} to you, {Mr.|Mrs.|Ms.} {Smith|Williams|Davis}!

how to output an array of all possible combinations?