/*[nonstandard]*/
/* Test whether character ranges work with UTF-8 encoding. */

#include "suite.h"

#include <locale.h>

void run(void)
{
	match("[æ-ø]", "æ");
	match("[æ-ø]", "ø");
	reject("[æ-ø]", "Æ");
	reject("[æ-ø]", "Ø");
	reject("[æ-ø]", "a");

	reject("[!æ-ø]", "æ");
	reject("[!æ-ø]", "ø");
	match("[!æ-ø]", "Æ");
	match("[!æ-ø]", "Ø");
	match("[!æ-ø]", "a");

	// Off by one UTF-8 decoding bug in musl ranges where it processed the last
	// byte in the range twice, where an UTF-8 continuation byte is an invalid
	// start of a UTF-8 codepoint. Basically [x-y] is parsed as [x-yy] for ASCII
	// and the multibyte sequence [XYZ-XYZ] is parsed as [XYZ-XYZZ].
	match("[æ-øa]", "æ");
	match("[æ-øa]", "ø");
	match("[æ-øa]", "a");
	reject("[æ-øa]", "A");
	reject("[æ-øa]", "x");
	reject("[æ-øa]", "]");

	reject("[^æ-øa]", "æ");
	reject("[^æ-øa]", "ø");
	reject("[^æ-øa]", "a");
	match("[^æ-øa]", "A");
	match("[^æ-øa]", "x");
	match("[^æ-øa]", "]");
}

int main(void)
{
	const char* utf8_locale = getenv("OS_TEST_UTF8_LOCALE");
	if ( !utf8_locale )
		utf8_locale = "C.UTF-8";
	if ( !setlocale(LC_CTYPE, utf8_locale) &&
	     !setlocale(LC_CTYPE, "C.UTF-8") &&
	     !setlocale(LC_CTYPE, "POSIX.UTF-8") &&
	     !setlocale(LC_CTYPE, "en_US.UTF-8") )
		errx(1, "setlocale LC_CTYPE failed: %s", utf8_locale);
	if ( !setlocale(LC_CTYPE, utf8_locale) &&
	     !setlocale(LC_COLLATE, "C.UTF-8") &&
	     !setlocale(LC_COLLATE, "POSIX.UTF-8") &&
	     !setlocale(LC_CTYPE, "en_US.UTF-8") )
		errx(1, "setlocale LC_COLLATE failed: %s", utf8_locale);

	flags = 0;
	run();
	return 0;
}