I have a database of hundreds of thousands of sentences which I translate from utf-8 using iconv.
In two of the sentences, I get the following error:
Notice: iconv(): Detected an illegal character in input string
I tried to check the input strings using the methods here: How to detect malformed utf-8 string in PHP?
$isUTF8 = mb_check_encoding($input, 'utf-8');
But, this function returns true (i.e. $input is a valid utf-8), and I still get those two error notices.
How can I detect which sentences are causing the problems?
It's very possible that your input is valid UTF-8, but that your target character encoding doesn't include one of the characters you're using. See utf8 to ISO-8859-1 not converting some characters correctly through Curl for an example.
If this is the case, you could perhaps use //TRANSLIT, if a substitution would be appropriate, or consider if it's possible to deliver content in UTF-8.