PHP mb_substr() not working correctly?

Question

This code

print mb_substr('éxxx', 0, 1);

prints an empty space :(

It is supposed to print the first character, é. This seems to work however:

print mb_substr('éxxx', 0, 2);

But it's not right, because (0, 2) means 2 characters...

Solution 1

Try passing the encoding parameter to mb_substr, as such:

print mb_substr('éxxx', 0, 1, 'utf-8');

The encoding is never detected automatically.

Solution 2

In practice I've found that, in some systems, multi-byte functions default to ISO-8859-1 for internal encoding. That effectively ruins their ability to handle multi-byte text.

Setting a good default will probably fix this and some other issues:

mb_internal_encoding('UTF-8');

Comments

[Gromski]

The encoding is never detected automatically, it just always defaults to something.

[Alvin Wong]

Could it be a better idea if you use mb_detect_encoding to actually try to detect the encoding?

[Gromski]

@AlvinWong No. Know what encoding you're working with, there's no other way.

[povilasp]

@Alvin Wong, that would be more correct, yes, but I could also say that using anything but utf-8 can be considered adventurous and marginal :)

[povilasp]

@deceze, wasn't sure, but thanks for the clarification, I updated the answer.

[Alex]

tx that works. Can mb_substr work like substr($string, 1) without giving it the mb_strlen() argument ?

[povilasp]

@Alex, that I think is another question, but my guess would be that yes - because the parameter is optional as it is in substr.

[Alex]

yes, but that UTF-8 thing has to go after that argument. Anyway nvm, I`ll just use mb_strlen ..

[Alvin Wong]

OK, then how about mb_internal_encoding instead of passing "utf-8" to all mb_* functions? Just like Álvaro G. Vicario has pointed out

[povilasp]

@AlvinWong is right, it's better to look to mb_internal_encoding if this is not only function usage and you are planning to use a lot of mb_* functions through out your code.