Calculate with regex group matches? - php

Question

Is it possible to calculate with regex group matches?
String:
(00) Bananas
...
(02) Apples (red ones)
...
(05) Oranges
...
(11) Some Other Fruit
...
If the difference between the numbers at the beginning of each row is 3 or less, remove the "..." inbetween. So the string should be returned like this:
(00) Bananas
(02) Apples (red ones)
(05) Oranges
...
(11) Some Other Fruit
Regex:
$match = '/(*ANYCRLF)\((\d+)\) (.+)$
\.{3}
\((\d+)\) (.+)/m';
Now the tricky part is how to grab the matches and add some to a condition like
if($3-$1 >= 3) {
//replace
}
Test: http://codepad.viper-7.com/f6iI4m
Thanks!

Answer 1

Here's how you could do it with preg_replace_callback().
$callback = function ($match) {
if ($match[3] <= $match[2] + 3) {
return $match[1];
} else {
return $match[0];
}
};
$newtxt = preg_replace_callback('/(^\((\d+)\).+$)\s+^\.{3}$(?=\s+^\((\d+)\))/m', $callback, $txt);
/(^\((\d+)\).+$)\s+^\.{3}$(?=\s+^\((\d+)\))/m
Here's the pattern in pieces:
(^\((\d+)\).+$) # subpattern 1, first line; subpattern 2, the number
\s+^\.{3}$ # newline(s) and second line ("...")
(?=\s+^\((\d+)\)) # lookahead that matches another numbered line
# without consuming it; contains subpattern 3, next number
Thus, the entire pattern's match is the first two lines (i.e., numbered line and '...' line).
If the difference in numbers is greater than 3, replace with original text in $match[0] (i.e., no change). If difference is less than or equal to 3, replace with first line only (found in $match1]).

Answer 2

You could employ preg_replace_callback and use any php code to return the replacement string, the callback receives the captures. However for you output you would have to get overlapping matches for replacement:
compare (00) Bananas vs (02) Apples -> 2-0=2 replace
compare (02) Apples vs (05) Oranges -> 5-2=3 replace
...
But since the input's (02) Apples part have been used for a previous match you it won't be picked up for the second time.
edit:
Here's a regexp based solutinon with lookaheads, credit goes to Wiseguy:
$s = "(00) Bananas
...
(02) Apples (red ones)
...
(05) Oranges
...
(11) Some Other Fruit
...";
$match = '/(*ANYCRLF)\((\d+)\) (.+)$
\.{3}
(?=\((\d+)\) (.+))/m';
// php5.3 anonymous function syntax
$s = preg_replace_callback($match, function($m){
if ($m[3] - $m[1] <= 3) {
print preg_replace("/[\r\n]+.../", '', $m[0]);
} else {
print $m[0];
}
}, $s);
echo $s;
Here is my first take, based on the logic "find the dots then see the previous / next lines":
$s = "(00) Bananas
...
(02) Apples (red ones)
...
(05) Oranges
...
(11) Some Other Fruit
...
(18) Some Other Fruit
...
(19) Some Other Fruit
...
";
$s = preg_replace("/[\r\n]{2}/", "\n", $s);
$num_pattern = '/^\((?<num>\d+)\)/';
$dots_removed = 0;
preg_match_all('/\.{3}/', $s, $m, PREG_OFFSET_CAPTURE);
foreach ($m[0] as $i => $dots) {
$offset = $dots[1] - ($dots_removed * 4); // fix offset of changing input
$prev_line_end = $offset - 2; // -2 since the offset is pointing to the first '.', prev char is "\n"
$prev_line_start = $prev_line_end; // start the search for the prev line's start from its end
while ($prev_line_start > 0 && $s[$prev_line_start] != "\n") {
--$prev_line_start;
}
$next_line_start = $offset + strlen($dots[0]) + 1;
$next_line_end = strpos($s, "\n", $next_line_start);
$next_line_end or $next_line_end = strlen($s);
$prev_line = trim(substr($s, $prev_line_start, $prev_line_end - $prev_line_start));
$next_line = trim(substr($s, $next_line_start, $next_line_end - $next_line_start));
if (!$next_line) {
break;
}
// get the numbers
preg_match($num_pattern, $prev_line, $prev);
preg_match($num_pattern, $next_line, $next);
if (intval($next['num']) - intval($prev['num']) <= 3) {
// delete the "..." line
$s = substr_replace($s, '', $offset-1, strlen($dots[0]) + 1);
++$dots_removed;
}
}
print $s;