My PHP function to spin text

[moomycow]

Hi guys,

Just spent 5 minutes knocking this up, you pass it some text in spinnable format, and it outputs a spun version:

<?php
	function get_spin($input)
	{
		$pos = strpos($input, '{');
		$output = substr($input, 0, $pos);
		
		while($pos)
		{
			$close_pos = strpos($input, '}', $pos);
			$diff = $close_pos - $pos - 1;
			
			$words = explode('|', substr($input, $pos + 1, $diff));
			$word_num = mt_rand(1, count($words)) -1;
			
			$pos = strpos($input, '{', $pos+1);
			$output .= $words[$word_num];
			
			if ($pos)
			{
				$output .= substr($input, $close_pos + 1, ($pos - $close_pos - 1));
			}
		}
		
		$output .= substr($input, $close_pos +1 );
		return $output;
	}
	
	$input = file_get_contents('input.txt');
	echo get_spin($input);
	echo '<br/>';
	echo get_spin($input);
?>

Sample input: I am a {little|small|tiny|big fat}, {lovely|nice|horrible|smelly} {cow|bovine|bull} and I {say|shout} {moo|moo moo moo} to you!

Hope someone gets some use from it

 

[dchuk]

I'm going to take this and make it a self submitting form to toss on a server, I'll post the code later. Nice submission

 

[moomycow]

You're welcome

My original plan was to have it look through a directory of spinnable articles and then output a file for each one containing many, many spins. I think both work as an idea

 

[moltar]

I've written a similar thing in Perl with regexes. Though I think your algo would be faster, because it uses substr(), the regex one would be easier to implement. I don't have PHP code, but the Perl code went something like this (did not test, going off memory):

$text = 'the {text|stuff} you want to spin';
$text =~ s/{([^}]+)}/spinit($1)/g;
sub spinit {
    my @words = split(/\|/, shift @_);
    return $words[rand @words];
}

The idea here is that you use a regex to find all text that is enclosed in {}, then you "remember" that text and replace it with the output of a function, into which you pass the "remembered" text. The function accepts the text, splits it by "|", as a result creating an array of variations. Then you return a random array member (variation). That returned value replaces everything in, and including {}.

 

[moomycow]

I've written a similar thing in Perl with regexes. Though I think your algo would be faster, because it uses substr(), the regex one would be easier to implement. I don't have PHP code, but the Perl code went something like this (did not test, going off memory):

$text = 'the {text|stuff} you want to spin';
$text =~ s/{([^}]+)}/spinit($1)/g;
sub spinit {
    my @words = split(/\|/, shift @_);
    return $words[rand @words];
}

The idea here is that you use a regex to find all text that is enclosed in {}, then you "remember" that text and replace it with the output of a function, into which you pass the "remembered" text. The function accepts the text, splits it by "|", as a result creating an array of variations. Then you return a random array member (variation). That returned value replaces everything in, and including {}.

I thought of doing something with regular expressions, but was worried about the following case:

This is {a|one} test, I want {a|one} meal.

I wasn't sure how to do a replace on the text so that I'd do two random choices, instead of doing a random choice between {a|one} and replacing the other instance with that. I'm not too hot on regular expressions, so accept I could be worrying needlessly here

 

[moltar]

{([^}]+)} -- that says { followed by anything other than }, which basically copies everything between {} into $1. You can also use a non-greedy quantifier:
{(.+?)}

I'm pretty sure PHP's regex were fully borrowed from Perl, so it should all work the same.

 

[kineticbug]

I'm pretty sure PHP's regex were fully borrowed from Perl, so it should all work the same.

PHP uses PCRE, which is the C regex library Perl uses, so yes, it'll work the same

Also, someone posted this a while ago, neat little spinner function that handles nested.

function spinText( $string ) {
    while ( preg_match( '/{([^{}]+)}/', $string, $matches ) ) {
        $parts = explode( '|', $matches[1] );
        $string = str_replace( $matches[0], $parts[ array_rand( $parts ) ], $string );
    }
    return $string;
}