preg_replace explanation

[justinh]

I'm looking at this page about preg_replace:

http://php.net/manual/en/function.preg-replace.php

Which gives this example:

<?php
$string = 'April 15, 2003';
$pattern = '/(\w+) (\d+), (\d+)/i';
$replacement = '${1}1,$3';
echo preg_replace($pattern, $replacement, $string);
?>

The above example will output:

April1,2003

I don't understand the transformation, how they go the answer.  Can someone explain?

Thanks,

Justin

[justsomeguy]

This part is replaced by the first matched pattern:

${1}

Then there's a 1 and a comma:

1,

Then the third matched pattern: 

$3

The reason why the curly brackets are there is to show how to reference a numbered pattern when you immediately want another digit to follow in the replacement (in this case, they want a 1 in the replacement).

[justinh]

Okay, I'll need a much greater breakdown that that...

First, how to understand the value for $pattern.  Then, where is the second matched pattern?  (it looks like we are missing one)  Then, why are there no spaces in the output string?

It just looks way too complicated to replace only the day, as it looks like it is matching the entire string and carrying forward some of the original string instead of just what needs to be replaced.

Edited September 23, 2017 by justinh

[justsomeguy]

If you want to understand the pattern syntax, check here:

http://php.net/manual/en/pcre.pattern.php

The pattern looks for one or more word characters, then a space, then one or more digits, then a comma and a space, then one or more digits.  Each of those sub-patterns is captured because they have parentheses around them.

Then, where is the second matched pattern?

It's not used in the replacement.  This example is just to illustrate how you refer to a sub-pattern when you want to output a number after it.

Then, why are there no spaces in the output string?

Because the replacement pattern has no spaces in it, and the spaces aren't part of what was captured.

[justinh]

Thank you so much - I understand it now after finally putting all the pieces together.  I also figured out (from the link you provided) that the pattern string has to be enclosed by a delimiter, thus the two "/".  Then the ending "i" mean case-insensitive alpha match.  Just for anyone else that reads this post...

It looks to me like someone could specialize just in regex coding!

Thanks again.