preg_match in loop returning impossible results - php

Question

I'm sure I'm missing something. I know just enough to be dangerous.
In my php code I use file_get_contents() to put a file into a variable.
I then loop through an array and use preg_match to search the same variable many times. The file is a tab-delimited txt file. It does fine 800 times but one time randomly in the middle it does something very odd.
$current = file_get_contents($file);
foreach($blahs as $blah){
$image = 'somefile.jpg';
$pattern = '/https:\/\/www\.example\.com\/media(.*)\/' . preg_quote($image) . '/';
preg_match($pattern, $current, $matches);
echo $matches[0];
}
For some reason that one time it turns two URL's with a tab between them. When I look at the txt file the image i'm looking for is listed first then followed by the second iamge but echo $matches[0] returns it in reverse order. it does not exist like echo $matches[0] returns it. It would be like if you searched the string 'one two' and $matches returned 'two one'.

Answer 1

The regex engine is trying to do you a favor and capture the longest match. The \t tab between the two urls is being matched by the . (dot / any character).
Demonstration: (Link)
$blah='test case: https://www.example.com/media/foo/bar.jpg https://www.example.com/media/cat/fish.jpg some text';
$image = 'fish.jpg';
$your_pattern = '/https:\/\/www\.example\.com\/media(.*)\/'.preg_quote($image).'/';
echo preg_match($your_pattern,$blah,$matches)?$matches[0]:'fail';
echo "\n----\n";
$my_pattern='~https://www\.example\.com/media(?:[^/\s]*/)+'.preg_quote($image).'~';
echo preg_match($my_pattern,$blah,$out)?$out[0]:'fail';
Output:
https://www.example.com/media/foo/bar.jpg https://www.example.com/media/cat/fish.jpg
----
https://www.example.com/media/cat/fish.jpg
To crystallize...
test case: https://www.example.com/media/foo/bar.jpg https://www.example.com/media/cat/fish.jpg some text
// your (.*) is matching ---------------^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
My suggested pattern (I may be able to refine the pattern if you provide smoe sample strings) uses (?:[^/\s]*/)+ instead of the (.*).
My non-capturing group breaks down like this:
(?: #start non-capturing group
[^/\s]* #greedily match zero or more non-slash, non-whitespace characters
/ #match a slash
) #end non-capturing group
+ #allow the group to repeat one or more times
*note1: You can use \t where I use \s if you want to be more literal, I am using \s because a valid url shouldn't contain a space anyhow. You may make this adjustment in your project without any loss of accuracy.
*note2: Notice that I changed the pattern delimiters to ~ so that / doesn't need to be escaped inside the pattern.