Weird behavior of `grapheme_stripos()` with grapheme `\u{34F}`

[nikophil]

Description

The grapheme \u{34F} has a weird behavior with grapheme_stripos():

<?php

var_dump(grapheme_stripos("abcdef", "\u{34F}"));
var_dump(grapheme_stripos("abc\u{34F}def", "\u{34F}"));

Resulted in this output:

int(0)
int(0)

But I expected this output instead:

bool(false)
int(3)

Note that on the other hand the following code works as expected:

echo "abc\u{34F}def\n"; // "abcdef"
echo grapheme_strlen("abc\u{34F}def"); // "6"

see https://3v4l.org/vRVCR/rfc#vgit.master

PHP Version

used git.master in 3v4l.org

Operating System

No response

[youkidearitai]

U+034F is COMBINING GRAPHEME JOINER(CGJ). And grapheme_stripos 's strength uses secondary level.

https://unicode.org/reports/tr10/tr10-34.html#DUCET_Order_Table

secondary collation elements
Some letters and other combining marks

I think include CGJ.
Therefore, this behavior is correct.

[nikophil]

Hi, thanks for your answer

I don't understand how it is a valid behavior that grapheme_stripos() says the grapheme is present in the string although it is not

[youkidearitai]

Please read at the further

https://unicode.org/reports/tr10/tr10-34.html#DUCET_Order_Table

Some letters and other combining marks

"other combining marks" means ZWJ, ZWNJ and CGJ etc. For example.

$ ~/php85/bin/php -r 'var_dump(grapheme_stripos("abcd", "\u{200D}"));' # ZWJ(U+200D)
int(0)

Anyway, this behavior is not PHP's bug. If you still says a bug, please ask to the ICU.

[youkidearitai]

I think also not ICU's bug.

[youkidearitai]

I'll close this issue. If you say not a bug, I suggest ask to the ICU