Need help with regular expressions

numerical25

Hey guys, I need to sharpen my regular expression skills up. But until I need you guys help.

I need to make a php regular express to match 2 patterns in a single string.

5.75 per $1,000

I need pull out any value that is a number x.xx . I also Need to pull out any value that follows after a $ and is a number and may have a possible comma $1,000

So in a nutshell I need 5.75 or any number like it pulled out and 1,000 pulled out or any thing like it

Imperialoutpost

Is every row you need to match the same format? Eg. #.## per $#,###

johanafm

If that is the case I'd ditch regexp since they are often problematic to get right and go with

$numbers = explode(' per $', $string);
echo $numbers[0];  # 5.75
echo $numbers[1];  # 1,000

Doing this with regexp might get tricky, and you'd really need to be a lot more specific about what input you will have as well as what you will never have. Examples of conditions that may complicate things, on its own or in combination with other conditions and what should match and not.

$s = '5.75 per $1000.50';
$s = '5.75 per $1000. Next year we expect it to be 6.00 per $1000';
$s = '5.75 per $1000,000';
$s = '5.75 per $10.000,00';

Resort to regexp if there really is no easier way of doing things. And if you do need it, be very specific about what input will appear, will never appear, and exactly what should match and shouldn't match.

bradgrafelman

johanafm wrote:
If that is the case I'd ditch regexp since they are often problematic to get right and go with

They're only as problematic as you make them. 😉 Besides, a string of PHP code (i.e. a .php file) is parsed into tokens using regular expressions, so they can't be all that bad.

@: Here's one way of doing it:

$string =  '5.75 per $1,000';
$pattern = '/(?:[0-9]+\.[0-9]+)|(?:(?<=\$)[0-9]+(?:,[0-9]+)?)/';

preg_match_all($pattern, $string, $matches);

print_r($matches);

output is:

Array
(
    [0] => Array
        (
            [0] => 5.75
            [1] => 1,000
        )

)

johanafm

bradgrafelman;10966560 wrote:
They're only as problematic as you make them. 😉

Regexps are a great tool, but depending on the situation they are not always easy to get right, and sometimes they are easy to get wrong. And either way, you must really really know what input can and cannot appear. I'd rather say that "they are as complicated as the input require them to be".

bradgrafelman;10966560 wrote:

$string =  '5.75 per $1,000';
$pattern = '/(?:[0-9]+\.[0-9]+)|(?:(?<=\$)[0-9]+(?:,[0-9]+)?)/';

For example

$string = '5.75 per $1,000.20';

This would match, except that it would match 5.75 and 1,000, not 1,000.20. Slight difference in value, but it's there. Should the regexp match 1,000.20, 1,000 or not match at all?
And

'5.75 per $1,000,123.20'

would match

0 => 5.75
1 => 1,000
2 => 123.20

bradgrafelman

johanafm wrote:
Should the regexp match 1,000.20, 1,000 or not match at all?

According to the OP, it should only match 1,000 since the numerical value following the dollar sign shouldn't contain a decimal point:

numerical25 wrote:
after a $ and is a number and may have a possible comma

As for the 1,000/123.20 business, that's an excellent point and can be remedied by changing the '?' modifier to a '*' modifier, e.g.:

$pattern = '/(?:[0-9]+\.[0-9]+)|(?:(?<=\$)[0-9]+(?:,[0-9]+)*)/';

EDIT: Hmm... on second thought, perhaps my mind was thinking "numeral" instead of "number." If it can indeed contain a decimal point, then the pattern could be changed to:

$pattern = '/(?:[0-9]+\.[0-9]+)|(?:(?<=\$)[0-9]+(?:,[0-9]+)*(?:\.[0-9]+)?)/';