如何编写正则表达式匹配黑名单词汇的各类变形？

Hey there! Let's fix this regex problem you're having. The issue with your current patterns is they don't account for the separators (_,-,., etc.) between the letters of your blacklisted words. Let's break down how to build a regex that catches all those deformed versions.

Why Your Current Regexes Fall Short #

• Your first pattern \W*((?i)ask(?-i))\W* only matches the exact word "ask" (case-insensitive) surrounded by non-word characters—it won't catch a_ks or as-k because the separators are between the letters, not around the whole word.
• The second pattern ^((?!ask)[\s\S])*$ is a negative lookahead that blocks any input containing the exact string "ask", but again, it ignores any separators between the letters, so a.s.k would slip right through.

The Solution: Regex That Allows Separators Between Letters #

The key idea is to split each blacklisted word into individual characters, and between each character, allow zero or more of your target separators (_, -, ., etc.). We'll also make the whole pattern case-insensitive so it catches Ask, ASK, aSk, etc.

Example Regex for Your Blacklisted Words #

For the words ask, stackoverflow, and hello, here's a regex that will match all their deformed versions:

$blacklistRegex = '/(?i)(?:a[_\-.]*s[_\-.]*k|s[_\-.]*t[_\-.]*a[_\-.]*c[_\-.]*k[_\-.]*o[_\-.]*v[_\-.]*e[_\-.]*r[_\-.]*f[_\-.]*l[_\-.]*o[_\-.]*w|h[_\-.]*e[_\-.]*l[_\-.]*l[_\-.]*o)/';

Let's Break It Down #

• (?i): Enables case-insensitive matching so we don't have to worry about uppercase/lowercase variations.
• (?:...): A non-capturing group to wrap our alternated patterns (keeps the regex efficient).
• For each word: Split into characters, and insert [_\-.]* between each pair. This means "zero or more of underscore, hyphen, or dot" can appear between any two letters of the blacklisted word.
  • a[_\-.]*s[_\-.]*k: Matches ask, a_sk, as-k, a.s.k, A-S-K, etc.
  • The same logic applies to stackoverflow and hello.

Adjusting Separators #

If you want to allow more separators (like spaces, commas, or other symbols), just add them to the character set [_\-.]. For example, to include spaces and commas: [_\-.,\s].

Using It in PHP #

To check if a form input contains any of these deformed blacklisted words, you can use preg_match:

$input = $_POST['user_input'];
$blacklistRegex = '/(?i)(?:a[_\-.]*s[_\-.]*k|s[_\-.]*t[_\-.]*a[_\-.]*c[_\-.]*k[_\-.]*o[_\-.]*v[_\-.]*e[_\-.]*r[_\-.]*f[_\-.]*l[_\-.]*o[_\-.]*w|h[_\-.]*e[_\-.]*l[_\-.]*l[_\-.]*o)/';

if (preg_match($blacklistRegex, $input)) {
    // Input contains a blacklisted word (deformed or not)
    echo "Invalid input: Contains restricted content.";
} else {
    // Input is clean
    echo "Input is valid.";
}

Automating Regex Generation #

If you have a long list of blacklisted words, manually writing the regex for each one is tedious. You can generate the regex programmatically in PHP:

$blacklistedWords = ['ask', 'stackoverflow', 'hello'];
$separatorPattern = '[_\-.]*';
$regexParts = [];

foreach ($blacklistedWords as $word) {
    // Split the word into individual characters
    $chars = str_split($word);
    // Join with the separator pattern
    $regexParts[] = implode($separatorPattern, $chars);
}

// Combine into a single regex
$blacklistRegex = '/(?i)(?:' . implode('|', $regexParts) . ')/';

This will dynamically build the regex for any number of blacklisted words, saving you time.

Bonus: Avoid False Positives #

If you want to ensure the match isn't part of a longer alphanumeric word (e.g., don't match "task" as a deformed "ask"), you can add lookarounds to check that the characters before and after aren't letters/numbers:

$blacklistRegex = '/(?i)(?:(?<![\p{L}\p{N}])a[_\-.]*s[_\-.]*k(?![\p{L}\p{N}])|(?<![\p{L}\p{N}])s[_\-.]*t[_\-.]*a[_\-.]*c[_\-.]*k[_\-.]*o[_\-.]*v[_\-.]*e[_\-.]*r[_\-.]*f[_\-.]*l[_\-.]*o[_\-.]*w(?![\p{L}\p{N}])|(?<![\p{L}\p{N}])h[_\-.]*e[_\-.]*l[_\-.]*l[_\-.]*o(?![\p{L}\p{N}]))/';

The (?<![\p{L}\p{N}]) is a negative lookbehind (no letter/number before), and (?![\p{L}\p{N}]) is a negative lookahead (no letter/number after).

That should cover all the deformed versions you mentioned. Let me know if you need further tweaks!

内容的提问来源于stack exchange，提问作者DolDurma