utf8_test.c File Reference

UTF-8 Unicode encoding tests. More...

#include <string.h>
#include <ipxe/utf8.h>
#include <ipxe/test.h>

Go to the source code of this file.

Data Structures
struct	utf8_accumulate_test
	A UTF-8 accumulation test. More...

Macros
#define	DATA(...)
	Define inline data.
#define	UTF8_ACCUMULATE(name, BYTES, EXPECTED)
	Define a UTF-8 accumulation test.
#define	utf8_accumulate_ok(test)

Functions
	FILE_LICENCE (GPL2_OR_LATER_OR_UBDL)
	UTF8_ACCUMULATE (ascii, "Hello world!", DATA( 'H', 'e', 'l', 'l', 'o', ' ', 'w', 'o', 'r', 'l', 'd', '!'))
	Basic ASCII test.
	UTF8_ACCUMULATE (multibyte, "Héllô wörld 🥳", DATA( 'H', 0, L 'é', 'l', 'l', 0, L 'ô', ' ', 'w', 0, L 'ö', 'r', 'l', 'd', ' ', 0, 0, 0, 0x1f973))
	Multi-byte character test.
	UTF8_ACCUMULATE (stray_continuation, DATA( 'a', 0x81, 'b', 0xc3, 0x82, 0x83, 'c'), DATA( 'a', 0xfffd, 'b', 0, 0xc2, 0xfffd, 'c'))
	Stray continuation byte test.
	UTF8_ACCUMULATE (missing_continuation, DATA( 'a', 0xc3, 'b', 0xe1, 0x86, 0xc3, 0x89, 'c'), DATA( 'a', 0, 'b', 0, 0, 0, 0xc9, 'c'))
	Missing continuation byte test.
	UTF8_ACCUMULATE (illegal_two, DATA( 'a', 0xc2, 0x80, 'b', 0xc1, 0xbf, 'c', 0xc0, 0x80, 'd'), DATA( 'a', 0, 0x80, 'b', 0, 0xfffd, 'c', 0, 0xfffd, 'd'))
	Illegal two-byte sequence test.
	UTF8_ACCUMULATE (illegal_three, DATA( 'a', 0xe0, 0xa0, 0x80, 'b', 0xe0, 0x9f, 0xbf, 'c', 0xe0, 0x80, 0x80, 'd'), DATA( 'a', 0, 0, 0x800, 'b', 0, 0, 0xfffd, 'c', 0, 0, 0xfffd, 'd'))
	Illegal three-byte sequence test.
	UTF8_ACCUMULATE (illegal_four, DATA( 'a', 0xf0, 0x90, 0x80, 0x80, 'b', 0xf0, 0x8f, 0xbf, 0xbf, 'c', 0xf0, 0x80, 0x80, 0x80, 'd'), DATA( 'a', 0, 0, 0, 0x10000, 'b', 0, 0, 0, 0xfffd, 'c', 0, 0, 0, 0xfffd, 'd'))
	Illegal four-byte sequence test.
	UTF8_ACCUMULATE (illegal_length, DATA( 'a', 0xf8, 0xbf, 0xbf, 0xbf, 0xbf, 'b', 0xfc, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 'c', 0xfe, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 'd', 0xff, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 'e'), DATA( 'a', 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 'b', 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 'c', 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 'd', 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 'e'))
	Illegal overlength sequence test.
static void	utf8_accumulate_okx (struct utf8_accumulate_test *test, const char *file, unsigned int line)
	Report UTF-8 accumulation test result.
static void	utf8_test_exec (void)
	Perform UTF-8 self-test.

Variables
struct self_test utf8_test	__self_test
	UTF-8 self-test.

Detailed Description

UTF-8 Unicode encoding tests.

Definition in file utf8_test.c.

Macro Definition Documentation

DATA

#define DATA

(

...

)

Value:

{ __VA_ARGS__ }

Define inline data.

Definition at line 50 of file utf8_test.c.

UTF8_ACCUMULATE

#define UTF8_ACCUMULATE	(		name,
			BYTES,
			EXPECTED )

Value:

        static const char name ## _bytes[] = BYTES;                 \
        static const unsigned int name ## _expected[] = EXPECTED;   \
        static struct utf8_accumulate_test name = {                     \
                .bytes = name ## _bytes,                            \
                .expected = name ## _expected,                              \
                .len = ( sizeof ( name ## _expected ) /                     \
                         sizeof ( name ## _expected[0] ) ),         \
        };

Define a UTF-8 accumulation test.

Definition at line 53 of file utf8_test.c.

#define UTF8_ACCUMULATE( name, BYTES, EXPECTED )                        \
        static const char name ## _bytes[] = BYTES;                     \
        static const unsigned int name ## _expected[] = EXPECTED;       \
        static struct utf8_accumulate_test name = {                     \
                .bytes = name ## _bytes,                                \
                .expected = name ## _expected,                          \
                .len = ( sizeof ( name ## _expected ) /                 \
                         sizeof ( name ## _expected[0] ) ),             \
        };

utf8_accumulate_ok

#define utf8_accumulate_ok

(

test

)

Value:

        utf8_accumulate_okx ( test, __FILE__, __LINE__ )

Definition at line 140 of file utf8_test.c.

#define utf8_accumulate_ok( test ) \
        utf8_accumulate_okx ( test, __FILE__, __LINE__ )

Referenced by utf8_test_exec().

Function Documentation

FILE_LICENCE()

FILE_LICENCE

(

GPL2_OR_LATER_OR_UBDL

)

UTF8_ACCUMULATE() [1/8]

UTF8_ACCUMULATE	(	ascii	,
		"Hello world!"	,
		DATA('H', 'e', 'l', 'l', 'o', ' ', 'w', 'o', 'r', 'l', 'd', '!')	)

Basic ASCII test.

References DATA.

UTF8_ACCUMULATE() [2/8]

UTF8_ACCUMULATE	(	multibyte	,
		"Héllô wörld 🥳"	,
		DATA('H', 0, L 'é', 'l', 'l', 0, L 'ô', ' ', 'w', 0, L 'ö', 'r', 'l', 'd', ' ', 0, 0, 0, 0x1f973)	)

Multi-byte character test.

References DATA.

UTF8_ACCUMULATE() [3/8]

UTF8_ACCUMULATE	(	stray_continuation	,
		DATA('a', 0x81, 'b', 0xc3, 0x82, 0x83, 'c')	,
		DATA('a', 0xfffd, 'b', 0, 0xc2, 0xfffd, 'c')	)

Stray continuation byte test.

References DATA.

UTF8_ACCUMULATE() [4/8]

UTF8_ACCUMULATE	(	missing_continuation	,
		DATA('a', 0xc3, 'b', 0xe1, 0x86, 0xc3, 0x89, 'c')	,
		DATA('a', 0, 'b', 0, 0, 0, 0xc9, 'c')	)

Missing continuation byte test.

References DATA.

UTF8_ACCUMULATE() [5/8]

UTF8_ACCUMULATE	(	illegal_two	,
		DATA('a', 0xc2, 0x80, 'b', 0xc1, 0xbf, 'c', 0xc0, 0x80, 'd')	,
		DATA('a', 0, 0x80, 'b', 0, 0xfffd, 'c', 0, 0xfffd, 'd')	)

Illegal two-byte sequence test.

References DATA.

UTF8_ACCUMULATE() [6/8]

UTF8_ACCUMULATE	(	illegal_three	,
		DATA('a', 0xe0, 0xa0, 0x80, 'b', 0xe0, 0x9f, 0xbf, 'c', 0xe0, 0x80, 0x80, 'd')	,
		DATA('a', 0, 0, 0x800, 'b', 0, 0, 0xfffd, 'c', 0, 0, 0xfffd, 'd')	)

Illegal three-byte sequence test.

References DATA.

UTF8_ACCUMULATE() [7/8]

UTF8_ACCUMULATE	(	illegal_four	,
		DATA('a', 0xf0, 0x90, 0x80, 0x80, 'b', 0xf0, 0x8f, 0xbf, 0xbf, 'c', 0xf0, 0x80, 0x80, 0x80, 'd')	,
		DATA('a', 0, 0, 0, 0x10000, 'b', 0, 0, 0, 0xfffd, 'c', 0, 0, 0, 0xfffd, 'd')	)

Illegal four-byte sequence test.

References DATA.

UTF8_ACCUMULATE() [8/8]

UTF8_ACCUMULATE	(	illegal_length	,
		DATA('a', 0xf8, 0xbf, 0xbf, 0xbf, 0xbf, 'b', 0xfc, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 'c', 0xfe, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 'd', 0xff, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 'e')	,
		DATA('a', 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 'b', 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 'c', 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 'd', 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 0xfffd, 'e')	)

Illegal overlength sequence test.

References DATA.

utf8_accumulate_okx()

void utf8_accumulate_okx ( struct utf8_accumulate_test * test, const char * file, unsigned int line )

static

Report UTF-8 accumulation test result.

Parameters

test	UTF-8 accumulation test
file	Test code file
line	Test code line

Definition at line 123 of file utf8_test.c.

                                                                        {
        struct utf8_accumulator utf8;
        unsigned int character;
        unsigned int i;
 
        /* Initialise accumulator */
        memset ( &utf8, 0, sizeof ( utf8 ) );
 
        /* Test each byte in turn */
        for ( i = 0 ; i < test->len ; i++ ) {
                character = utf8_accumulate ( &utf8, test->bytes[i] );
                DBGC ( test, "UTF8 byte %02x character %02x\n",
                       test->bytes[i], character );
                okx ( character == test->expected[i], file, line );
        }
}

References utf8_accumulator::character, DBGC, memset(), okx, test, and utf8_accumulate().

utf8_test_exec()

void utf8_test_exec ( void )

static

Perform UTF-8 self-test.

Definition at line 147 of file utf8_test.c.

                                    {
 
        /* Accumulation tests */
        utf8_accumulate_ok ( &ascii );
        utf8_accumulate_ok ( &multibyte );
        utf8_accumulate_ok ( &stray_continuation );
        utf8_accumulate_ok ( &missing_continuation );
        utf8_accumulate_ok ( &illegal_two );
        utf8_accumulate_ok ( &illegal_three );
        utf8_accumulate_ok ( &illegal_four );
        utf8_accumulate_ok ( &illegal_length );
}

References utf8_accumulate_ok.

Variable Documentation

__self_test

struct self_test utf8_test __self_test

Initial value:

= {
        .name = "utf8",
        .exec = utf8_test_exec,
}

UTF-8 self-test.

Definition at line 161 of file utf8_test.c.

                                         {
        .name = "utf8",
        .exec = utf8_test_exec,
};