正则表达式返回参考负前瞻原子团

Question

I want to match a single or double quote mark, followed by any amount of characters that are not the character just matched, followed by one of the character matched:

"--'__'--"

Should match by the double quotes at each end. However, I want the match to be possessive in that any characters that have already been tested should not be included in any future matches:

"--'__'--

Should not match because the double quote at the beginning is never followed by another one at the end. I have come up with:

(?P<q>['"])(?>((?!(?P=q)).)*)(?P=q)

But this still matches my second string example above by the single quotes in the middle. I don't understand why the atomic group doesn't accomplish this. I have not been able to accomplish this with any other arrangement of atomic grouping either.

Also, if it is possible at all to match only the characters in between the quotes while asserting that the quotes are present that would be excellent. Because lookbehind assertions are fixed width I can't use a back reference to assert that the captured group of either single or double quotes occurs prior to the negative lookahead.

Answer 1

Assuming there will be only one valid quoted substring per line, this may be a good starting point:

<?php // test.php Rev:20120105_1800
// Return array of valid quoted substrings, one per line.
function getArrayOfOnePerLineValidQuotedSubstrings($text) {
    $re = '%
        # Match line w/1 valid "single" or "double" substring.
        ^               # Anchor to start of line.
        [^\'"]*         # Everything up to first quote.
        (?|             # Branch reset group $1: Contents.
          "([^"]*)"     # Either $1.1 Double quoted,
        | \'([^\']*)\'  # or $1.2 Single quoted contents.
        )               # End $1: branch reset group.
        [^\'"]*         # Everything after quoted sub-string.
        $               # Anchor to end of line.
        %xm';
    if (preg_match_all($re, $text, $matches)) {
        return $matches[1];
    }
    return array();
}
// Fetch test data from file.
$data = file_get_contents('testdata.txt');
// Get array of valid quoted substrings, one per line.
$output = getArrayOfOnePerLineValidQuotedSubstrings($data);
// Display results.
$count = count($output);
printf("%d matches found.\n", $count);
for ($i = 0; $i < $count; ++$i) {
    printf("  match[%d] = {%s}\n", $i + 1, $output[$i]);
}
?>

This regex matches each line that contains one valid quoted substring and skips over lines that have invalid (ie "--'__'-- which has an unbalanced double quoted substring) or no quoted substrings. For lines which match, the contents of the valid quoted substring are returned in group $1 . The function returns an array of the matched substrings.

If your data will contain more than one substring per line, or if the quoted substrings or stuff between quoted substrings may contain escaped quotes, then a more complex solution may be formulated.