如何使用PHP正则表达式在字符串中搜索包含重复单词的单词序列？

Question

I am using PHP to count the number of occurrences of a word sequence in a string. In the following example cases, I am not getting the result I would like to see.

$subject1 = " [word1 [word1 [word1 [word1 [word3 ";
$pattern1 = preg_quote("[word1 [word1", '/');
echo "count of '[word1 [word1'=". preg_match_all("/(\s|^|\W)" . $pattern1 . "(?=\s|$|\W)/", $subject1, $dummy) . "<br/>"; 

$subject2 = " [word1 [word2 [word1 [word2 [word1 [helloagain ";
$pattern2 = preg_quote("[word1 [word2 [word1", '/');
echo "count of '[word1 [word2 [word1'=". preg_match_all("/(\s|^|\W)" . $pattern2 . "(?=\s|$|\W)/", $subject2, $dummy) . "<br/>";

the above returns:

count of '[word1 [word1'=2
count of '[word1 [word2 [word1'=1

I would like the result to be:

count of '[word1 [word1'=3 // there are 3  instances of ‘[word1 [word1’ in $subject1
count of '[word1 [word2 [word1'=2 // // there are 2  instances of [word1 [word2 [word1’ in $subject2

One way to achieve this is each time the pattern is found in subject the next search should start from the second word in the matched substring. Can such a regular expression be constructed? Thank you.

Answer 1

Use mb_substr_count

substr_count does not count overlapped values, but i dont know why, mb_substr_count does

$subject1 = " [word1 [word1 [word1 [word1 [word3 ";
echo mb_substr_count($subject1, "[word1 [word1"); // 3
echo mb_substr_count($subject1, "[word1 [word1 [word1"); // 2

EDIT:

For future reference,

Apparently mb_substr_count acts differently on php 5.2 than php 5.3 . I suppose the right behavior of this function should be same as substr_count , only for multibyte support, and since substr_count doesn't support overlapping, so should mb_substr_count .

So, though this answer works on php 5.2.6, do not use it, or you may have problems when you update your php version.

Answer 2

Instead of preg_match_all, I'd use a while loop on preg_match with offset:

$subject1 = " [word1 [word1 [word1 [word1 [word3 ";
$pattern1 = preg_quote("[word1 [word1", '/');
$offset=0;
$total=0;
while($count = preg_match("/(?:\s|^|\W)$pattern1(?=\s|$|\W)/", $subject1, $matches, PREG_OFFSET_CAPTURE, $offset)) {
    // summ all matches
    $total  += $count;
    // valorisation of offset with the position of the match + 1
    // the next preg_match will start at this position
    $offset  = $matches[0][1]+1;
}
echo "total=$total\n";

output:

total=3

The result for the second example is : total=2