Parse and match calculation string

Question

How do I parse and match calculation strings?

This is my code:

$pattern = '/(?:\d+(?:\.\d+)?\D)+\d+(?:\.\d+)/';
$input = [
    '1.0+2.5*5.4',
    '5*8-4'
];
foreach($input as $string){
    preg_match($pattern, $string, $match);
    print_r($match);
}

I can't figure out what I'm doing wrong.

`\D` matches that `-` minus in your formular. And you have two spots for matching the float. But there's no second placeholder for the `/` division nor the last number. — mario, Apr 01 '13 at 14:18

score 2 · Accepted Answer · edited Mar 09 '16 at 21:24

It's because of the greediness of "?", try "+" as in :

$pattern = '/(?:\d+(?:\.\d+)?\D)+\d+(?:\.\d+)/';
$input = '34.27-15.44/8.44';
echo $input."\n";
preg_match($pattern, $input, $match);
print_r($match);

As for your edit:

$pattern = '/(?:\d+(?:\.?\d+)?\D)+\d+/';
$input = [
    '1.0+2.5*5.4',
    '5*8-4',
    '5*8-4string',
    'string5*8-4',
    'string'
];
foreach($input as $string){
    preg_match($pattern, $string, $match);
    print_r($match);
}

It gives:

Array
(
    [0] => 1.0+2.5*5.4
)
Array
(
    [0] => 5*8-4
)
Array
(
    [0] => 5*8-4
)
Array
(
    [0] => 5*8-4
)
Array
(
)

as I suspect you want it to.

http://codepad.viper-7.com/4FoQnB

Have made your change.. Now it works with the first input.. The second input with integers doesn't work.. Have updated the question — clarkk, Apr 01 '13 at 14:54

score 0 · Answer 2 · edited Mar 09 '16 at 21:25

0

The result is expected, given your expression.

Note how \D can only match once (or nonce). \D has matched the -, so nothing else can match /.

You might be tempted to solve this by putting a quantifier on larger part of your expression, but let me anticipate the problems you'll face going that way and suggest that you use a loop that substitutes a single operand-operator-operand triplet at a time, quitting when no substitutions can be made. This is because no single-pass regular expression can properly manage order of operations (e.g. PEMDAS).

edited Mar 09 '16 at 21:25

the Tin Man

158,662
42
215
303

answered Apr 01 '13 at 14:23

Andrew Cheong

29,362
15
90
145

However, PCRE can very well match complex formulas, including operator precedence and parens. One can define named subpatterns and `(?R` recurse on them. It's not single-pass anymore, if you will. – mario Apr 01 '13 at 14:27
@mario - Good point; forgot about those regex extras. I don't disagree that matching and parsing can be done via pattern (single or looping). I think my caveat was more about the computation step, if there will be one. However I may be mistaken. The last time I answered such a question, it was [regarding reverse polish notation expressions](http://stackoverflow.com/questions/12293395/ruby-regexp-issue-involving-reverse-polish-calculator/12293553#12293553), where it seemed imperative that substitutions occurred in a loop, layer by layer. This may not apply to infix notation. – Andrew Cheong Apr 01 '13 at 14:35

score 0 · Answer 3 · edited Mar 09 '16 at 21:25

Try this. It works for integers and floats:

 $pattern = '/(?:\d+(?:(\.|)\d*)?\D)*\d*(?:(\.|)\d+)/';
 $input =array(
   '1.0+2.5*5.4',
   '5*8-4',
   '5*8-4string'
 );
 foreach($input as $string){
    echo "input:".$string."\n";
    preg_match($pattern, $string, $match);
    print_r("output:".$match[0]."\n");
  }

Output:

 input:1.0+2.5*5.4
 output:1.0+2.5*5.4
 input:5*8-4
 output:5*8-4
 input:5*8-4string
 output:5*8-4

Parse and match calculation string

3 Answers3