Recently I came across a very strange behaviour in a regular expression I wrote years ago.
The webserver upgraded to PHP 5.3.3 and it suddenly stopped working.
Case as follows: I use PHP to read a text-file and with the aid of PCRE (preg_match & preg_split) I look for certain matches in that text-file. The program splits different paragraphs into seperate records. But suddenly preg_split stopped working:
1 2 3 | <?php$paragraphs = preg_split("/\n\n/",$contents_of_file);?> |
Multiple newlines could not be detected, whereas
1 2 3 | <?php$lines = preg_split("/\n/",$contents_of_file);?> |
generated no problem at all. At first I thought I had to use the multiline modifier /m but that didn’t work out either.
Although a HEX editor didn’t indicate any carriage returns in the text-file I managed to get things working by doing this:
1 2 3 | <?php$paragrapgs = preg_split("/\r?\n\r?\n/",$contents_of_file);?> |
Curious thing: previous and newer versions of PHP (e.g. 5.3.8 in a XAMPP test environment) did not require to look for possible occurrences of carriage returns (\r).