I am trying to stop spam and a lot of it is recurring words, is there any way to check the string to see if there is a recurring phrase inside the string?
$string = 'Hello ! Hello ! Hello ! Hello !";
Thanks,
1
-
2What about using akismet instead of rolling your own anti-spam solution? I've written a spam-filter for generic texts once, and it's hard to filter without having false positives. Akismet is ready to be used.Berry Langerak– Berry Langerak2012-04-23 13:20:52 +00:00Commented Apr 23, 2012 at 13:20
Add a comment
|
3 Answers
substr_count is fine when you know what you're looking for. If you don't know what is the spam word, you can use str_word_count and array_count_values:
$string = 'Hello! Hello! Hello! Hello! Lorem Ipsum';
$words = str_word_count($string, 1);
$count = array_count_values($words);
print_r($count);
This will give you this:
Array
(
[Hello] => 4
[Lorem] => 1
[Ipsum] => 1
)
You can sort() this array and get a ranking of the most used words on the string. You should also check for stopwords (like "and", "or", "me" and such).
2 Comments
alganet
You'll need a list of stopwords in an array, then loop through the
$count list. I believe there is no PHP function that solves the stopword problem by its own. Perhaps array_filter can help you.A fast Google Search gave me that:
http://php.net/manual/en/function.substr-count.php
However there are better anti-spam ideas. Like Captchas. Human spammers are pretty difficult to catch. They'll find a way around your word counter. Maybe you should think about an approach of user-based spam report like Youtube does.
Comments
You could try substr_count(): http://php.net/manual/en/function.substr-count.php
$string = 'hello ! hello ! hello ! hello !';
echo substr_count($string, 'hello');
Of course if you want to check whether ANY of the words in your string occur multiple times... then this becomes a lot less efficient. You'd probably have to keep track of a list of 'checked words' and, for each not-yet-checked word in your string check whether it occurs multiple times.
LIke Binarious mentioned, a captcha would be a nicer way to stop spam ;-)
