Fix CFCharacterSet equality detection

[sashabelonogov]

I was investigating why the test checking the CharacterSet equality was failing.

Firstly, I found out that the _hashValue of the CFCharacterSet is always 0. The method that checks equality(__CFCharacterSetEqual) is checking the _hashValue, which is correct and the method calculating hash(__CFCharacterSetHash) seemed to do it properly but it seemed that it was simply never called to set the _hashValue field.
So, as the solution for it, I decided to add this method call at the step of CFCharacterSet initialization when _hashValue should not be 0. After that the tests started to execute correctly, but I'm not sure if it's a correct solution, maybe there is a better way to do it.

Nevertheless, __CFCharacterSetEqual shouldn't rely only on hash but it also should check the contents, which it did, but most likely not correctly. I found out that here:

const UniChar *buf1 = __CFCSetStringBuffer((CFCharacterSetRef)cf1);
const UniChar *buf1End = buf1 + __CFCSetStringLength((CFCharacterSetRef)cf1);
const UniChar *buf2 = __CFCSetStringBuffer((CFCharacterSetRef)cf2);
const UniChar *buf2End = buf2 + __CFCSetStringLength((CFCharacterSetRef)cf2);

 while ((buf1 < buf1End) && (buf2 < buf2End)) {
                        UniChar char1 = *buf1;
                        UniChar char2 = *buf2;

                        if (char1 != char2) return false;

                        do { ++buf1; } while ((buf1 < buf1End) && (char1 == *buf1));
                        do { ++buf2; } while ((buf2 < buf2End) && (char2 == *buf2));
                    }

because of && operator in while ((buf1 < buf1End) && (buf2 < buf2End))
it was not comparing all the characters in both string buffers and replacing it with || fixes the issue.
Also, for some cases we can make the comparison here easier simply by checking the length of the strings beforehand, which is also implemented in this PR.

Please, let me know what you think

[alblue]

Some minor observations; plus, I'd like @phausler to have a look as well.

[alblue]

Please don't move the comments as it increases noise in the delta. Can you move that back here, and then it will be clearly seen you're just adding a new line in.

[sashabelonogov]

Yes, this comment is not related to the line that I was adding, so I decided to move it below, if you think it's still will be clear when it will "Check for surrogate" - I can leave it where it was.

[alblue]

As above, plus the indentation of the _CFCharacterSetHash doesn't appear to be indented correctly

[sashabelonogov]

Will be fixed, thanks for noticing

[alblue]

Was this comment copied and pasted from above? I thought the comment was regarding the condition rather than the action of the line, so I'm not sure it makes sense to add it here - but @phausler would know more, I suspect.

[sashabelonogov]

I simply added it for the consistency, in all the similar places the same comment exist.

[parkera]

I think that #1305 finally has the other character set fixes we needed. We should check the behavior of this bug vs that once it lands too.

[sashabelonogov]

Yes, the tests will execute successfully with the changes done in #1305, surely the comparison of buffers with memcmp idea is better than what is implemented in this PR. The simple comparison of lengths is also implemented in the other PR. I will check the situation with _hashValue later.

But anyways, it means that most (if not all of them) of the changes implemented in this PR are not relevant anymore.

[alblue]

Since #1305 has been merged (then reverted in #1328 and then re-reverted in #1338) I don't think this PR is necessary as is. However, it might be worth testing whether or not the hashValue change can be applied as a subsequent PR if needed.