int
main ()
{
static enum iconv_ilseq_handler handlers[] =
{ iconveh_error, iconveh_question_mark, iconveh_escape_sequence };
size_t h;
size_t o;
size_t i;
#if HAVE_ICONV
/* Assume that iconv() supports at least the encodings ASCII, ISO-8859-1,
ISO-8859-2, and UTF-8. */
/* Test conversion from UTF-8 to ISO-8859-1 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const uint8_t input[] = "\303\204rger mit b\303\266sen B\303\274bchen ohne Augenma\303\237";
static const char expected[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (u8_strlen (input)) : NULL);
size_t length;
char *result = u8_conv_to_encoding ("ISO-8859-1", handler,
input, u8_strlen (input),
offsets,
NULL, &length);
ASSERT (result != NULL);
ASSERT (length == strlen (expected));
ASSERT (memcmp (result, expected, length) == 0);
if (o)
{
for (i = 0; i < 41; i++)
ASSERT (offsets[i] == (i < 1 ? i :
i == 1 ? (size_t)(-1) :
i < 13 ? i - 1 :
i == 13 ? (size_t)(-1) :
i < 20 ? i - 2 :
i == 20 ? (size_t)(-1) :
i < 40 ? i - 3 :
i == 40 ? (size_t)(-1) :
i - 4));
ASSERT (offsets[41] == MAGIC);
free (offsets);
}
free (result);
}
}
/* Test conversion from UTF-8 to ISO-8859-1 with EILSEQ. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const uint8_t input[] = "Rafa\305\202 Maszkowski"; /* Rafał Maszkowski */
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (u8_strlen (input)) : NULL);
size_t length = 0xdead;
char *result = u8_conv_to_encoding ("ISO-8859-1", handler,
input, u8_strlen (input),
offsets,
NULL, &length);
switch (handler)
{
case iconveh_error:
ASSERT (result == NULL);
ASSERT (errno == EILSEQ);
ASSERT (length == 0xdead);
break;
case iconveh_question_mark:
{
static const char expected[] = "Rafa? Maszkowski";
static const char expected_translit[] = "Rafal Maszkowski";
ASSERT (result != NULL);
ASSERT (length == strlen (expected));
ASSERT (memcmp (result, expected, length) == 0
|| memcmp (result, expected_translit, length) == 0);
if (o)
{
for (i = 0; i < 17; i++)
ASSERT (offsets[i] == (i < 5 ? i :
i == 5 ? (size_t)(-1) :
i - 1));
ASSERT (offsets[17] == MAGIC);
free (offsets);
}
free (result);
}
break;
case iconveh_escape_sequence:
{
static const char expected[] = "Rafa\\u0142 Maszkowski";
ASSERT (result != NULL);
ASSERT (length == strlen (expected));
ASSERT (memcmp (result, expected, length) == 0);
if (o)
{
for (i = 0; i < 17; i++)
ASSERT (offsets[i] == (i < 5 ? i :
i == 5 ? (size_t)(-1) :
i + 4));
ASSERT (offsets[17] == MAGIC);
free (offsets);
}
free (result);
}
break;
}
}
}
/* Test conversion from UTF-8 to ISO-8859-1 with EINVAL. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const uint8_t input[] = "\342";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (u8_strlen (input)) : NULL);
size_t length;
char *result = u8_conv_to_encoding ("ISO-8859-1", handler,
input, u8_strlen (input),
offsets,
NULL, &length);
ASSERT (result != NULL);
ASSERT (length == strlen (""));
if (o)
{
ASSERT (offsets[0] == 0);
ASSERT (offsets[1] == MAGIC);
free (offsets);
}
free (result);
}
}
#endif
return 0;
}
int
main ()
{
static enum iconv_ilseq_handler handlers[] =
{ iconveh_error, iconveh_question_mark, iconveh_escape_sequence };
size_t h;
size_t o;
size_t i;
#if HAVE_ICONV
/* Assume that iconv() supports at least the encodings ASCII, ISO-8859-1,
ISO-8859-2, and UTF-8. */
/* Test conversion from UTF-16 to ISO-8859-1 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const uint16_t input[] = /* Ärger mit bösen Bübchen ohne Augenmaß */
{
0xC4, 'r', 'g', 'e', 'r', ' ', 'm', 'i', 't', ' ', 'b', 0xF6, 's',
'e', 'n', ' ', 'B', 0xFC, 'b', 'c', 'h', 'e', 'n', ' ', 'o', 'h',
'n', 'e', ' ', 'A', 'u', 'g', 'e', 'n', 'm', 'a', 0xDF
};
static const char expected[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (SIZEOF (input)) : NULL);
size_t length;
char *result = u16_conv_to_encoding ("ISO-8859-1", handler,
input, SIZEOF (input),
offsets,
NULL, &length);
ASSERT (result != NULL);
ASSERT (length == strlen (expected));
ASSERT (memcmp (result, expected, length) == 0);
if (o)
{
for (i = 0; i < 37; i++)
ASSERT (offsets[i] == i);
ASSERT (offsets[37] == MAGIC);
free (offsets);
}
free (result);
}
}
/* Test conversion from UTF-16 to ISO-8859-1 with EILSEQ. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const uint16_t input[] = /* Rafał Maszkowski */
{
'R', 'a', 'f', 'a', 0x0142, ' ', 'M', 'a', 's', 'z', 'k', 'o', 'w',
's', 'k', 'i'
};
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (SIZEOF (input)) : NULL);
size_t length = 0xdead;
char *result = u16_conv_to_encoding ("ISO-8859-1", handler,
input, SIZEOF (input),
offsets,
NULL, &length);
switch (handler)
{
case iconveh_error:
ASSERT (result == NULL);
ASSERT (errno == EILSEQ);
ASSERT (length == 0xdead);
break;
case iconveh_question_mark:
{
static const char expected[] = "Rafa? Maszkowski";
static const char expected_translit[] = "Rafal Maszkowski";
ASSERT (result != NULL);
ASSERT (length == strlen (expected));
ASSERT (memcmp (result, expected, length) == 0
|| memcmp (result, expected_translit, length) == 0);
if (o)
{
for (i = 0; i < 16; i++)
ASSERT (offsets[i] == i);
ASSERT (offsets[16] == MAGIC);
free (offsets);
}
free (result);
}
break;
case iconveh_escape_sequence:
{
static const char expected[] = "Rafa\\u0142 Maszkowski";
ASSERT (result != NULL);
ASSERT (length == strlen (expected));
ASSERT (memcmp (result, expected, length) == 0);
if (o)
{
for (i = 0; i < 16; i++)
ASSERT (offsets[i] == (i < 5 ? i : i + 5));
ASSERT (offsets[16] == MAGIC);
free (offsets);
}
free (result);
}
break;
}
}
}
/* Test conversion from UTF-16 to ISO-8859-1 with EINVAL. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const uint16_t input[] = { 0xD845 };
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (SIZEOF (input)) : NULL);
size_t length;
char *result = u16_conv_to_encoding ("ISO-8859-1", handler,
input, SIZEOF (input),
offsets,
NULL, &length);
ASSERT (result != NULL);
ASSERT (length == strlen (""));
if (o)
{
ASSERT (offsets[0] == 0);
ASSERT (offsets[1] == MAGIC);
free (offsets);
}
free (result);
}
}
#endif
return 0;
}
int
main ()
{
static enum iconv_ilseq_handler handlers[] =
{ iconveh_error, iconveh_question_mark, iconveh_escape_sequence };
size_t indirect;
size_t h;
size_t o;
size_t i;
#if HAVE_ICONV
/* Assume that iconv() supports at least the encodings ASCII, ISO-8859-1,
ISO-8859-2, and UTF-8. */
iconv_t cd_ascii_to_88591 = iconv_open ("ISO-8859-1", "ASCII");
iconv_t cd_88591_to_88592 = iconv_open ("ISO-8859-2", "ISO-8859-1");
iconv_t cd_88592_to_88591 = iconv_open ("ISO-8859-1", "ISO-8859-2");
iconv_t cd_ascii_to_utf8 = iconv_open ("UTF-8", "ASCII");
iconv_t cd_88591_to_utf8 = iconv_open ("UTF-8", "ISO-8859-1");
iconv_t cd_utf8_to_88591 = iconv_open ("ISO-8859-1", "UTF-8");
iconv_t cd_88592_to_utf8 = iconv_open ("UTF-8", "ISO-8859-2");
iconv_t cd_utf8_to_88592 = iconv_open ("ISO-8859-2", "UTF-8");
iconv_t cd_utf7_to_utf8 = iconv_open ("UTF-8", "UTF-7");
iconveh_t cdeh_ascii_to_88591;
iconveh_t cdeh_ascii_to_88591_indirectly;
iconveh_t cdeh_88592_to_88591;
iconveh_t cdeh_88592_to_88591_indirectly;
iconveh_t cdeh_ascii_to_utf8;
iconveh_t cdeh_88591_to_utf8;
iconveh_t cdeh_utf8_to_88591;
iconveh_t cdeh_utf7_to_utf8;
ASSERT (cd_ascii_to_utf8 != (iconv_t)(-1));
ASSERT (cd_88591_to_utf8 != (iconv_t)(-1));
ASSERT (cd_utf8_to_88591 != (iconv_t)(-1));
ASSERT (cd_88592_to_utf8 != (iconv_t)(-1));
ASSERT (cd_utf8_to_88592 != (iconv_t)(-1));
cdeh_ascii_to_88591.cd = cd_ascii_to_88591;
cdeh_ascii_to_88591.cd1 = cd_ascii_to_utf8;
cdeh_ascii_to_88591.cd2 = cd_utf8_to_88591;
cdeh_ascii_to_88591_indirectly.cd = (iconv_t)(-1);
cdeh_ascii_to_88591_indirectly.cd1 = cd_ascii_to_utf8;
cdeh_ascii_to_88591_indirectly.cd2 = cd_utf8_to_88591;
cdeh_88592_to_88591.cd = cd_88592_to_88591;
cdeh_88592_to_88591.cd1 = cd_88592_to_utf8;
cdeh_88592_to_88591.cd2 = cd_utf8_to_88591;
cdeh_88592_to_88591_indirectly.cd = (iconv_t)(-1);
cdeh_88592_to_88591_indirectly.cd1 = cd_88592_to_utf8;
cdeh_88592_to_88591_indirectly.cd2 = cd_utf8_to_88591;
cdeh_ascii_to_utf8.cd = cd_ascii_to_utf8;
cdeh_ascii_to_utf8.cd1 = cd_ascii_to_utf8;
cdeh_ascii_to_utf8.cd2 = (iconv_t)(-1);
cdeh_88591_to_utf8.cd = cd_88591_to_utf8;
cdeh_88591_to_utf8.cd1 = cd_88591_to_utf8;
cdeh_88591_to_utf8.cd2 = (iconv_t)(-1);
cdeh_utf8_to_88591.cd = cd_utf8_to_88591;
cdeh_utf8_to_88591.cd1 = (iconv_t)(-1);
cdeh_utf8_to_88591.cd2 = cd_utf8_to_88591;
cdeh_utf7_to_utf8.cd = cd_utf7_to_utf8;
cdeh_utf7_to_utf8.cd1 = cd_utf7_to_utf8;
cdeh_utf7_to_utf8.cd2 = (iconv_t)(-1);
/* ------------------------ Test mem_cd_iconveh() ------------------------ */
/* Test conversion from ISO-8859-2 to ISO-8859-1 with no errors. */
for (indirect = 0; indirect <= 1; indirect++)
{
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
static const char expected[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_cd_iconveh (input, strlen (input),
(indirect
? &cdeh_88592_to_88591_indirectly
: &cdeh_88592_to_88591),
handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 37; i++)
ASSERT (offsets[i] == i);
ASSERT (offsets[37] == MAGIC);
free (offsets);
}
free (result);
}
}
}
/* Test conversion from ASCII to ISO-8859-1 with invalid input (EILSEQ). */
for (indirect = 0; indirect <= 1; indirect++)
{
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Rafa\263 Maszkowski"; /* Rafa? Maszkowski */
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_cd_iconveh (input, strlen (input),
(indirect
? &cdeh_ascii_to_88591_indirectly
: &cdeh_ascii_to_88591),
handler,
offsets,
&result, &length);
switch (handler)
{
case iconveh_error:
ASSERT (retval == -1 && errno == EILSEQ);
ASSERT (result == NULL);
if (o)
free (offsets);
break;
case iconveh_question_mark:
case iconveh_escape_sequence:
{
static const char expected[] = "Rafa? Maszkowski";
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 16; i++)
ASSERT (offsets[i] == i);
ASSERT (offsets[16] == MAGIC);
free (offsets);
}
free (result);
}
break;
}
}
}
}
/* Test conversion from ISO-8859-2 to ISO-8859-1 with EILSEQ. */
for (indirect = 0; indirect <= 1; indirect++)
{
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Rafa\263 Maszkowski"; /* Rafał Maszkowski */
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_cd_iconveh (input, strlen (input),
(indirect
? &cdeh_88592_to_88591_indirectly
: &cdeh_88592_to_88591),
handler,
offsets,
&result, &length);
switch (handler)
{
case iconveh_error:
ASSERT (retval == -1 && errno == EILSEQ);
ASSERT (result == NULL);
if (o)
free (offsets);
break;
case iconveh_question_mark:
{
static const char expected[] = "Rafa? Maszkowski";
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 16; i++)
ASSERT (offsets[i] == i);
ASSERT (offsets[16] == MAGIC);
free (offsets);
}
free (result);
}
break;
case iconveh_escape_sequence:
{
static const char expected[] = "Rafa\\u0142 Maszkowski";
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 16; i++)
ASSERT (offsets[i] == (i < 5 ? i :
i + 5));
ASSERT (offsets[16] == MAGIC);
free (offsets);
}
free (result);
}
break;
}
}
}
}
/* Test conversion from ISO-8859-1 to UTF-8 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
static const char expected[] = "\303\204rger mit b\303\266sen B\303\274bchen ohne Augenma\303\237";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_cd_iconveh (input, strlen (input),
&cdeh_88591_to_utf8,
handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 37; i++)
ASSERT (offsets[i] == (i < 1 ? i :
i < 12 ? i + 1 :
i < 18 ? i + 2 :
i + 3));
ASSERT (offsets[37] == MAGIC);
free (offsets);
}
free (result);
}
}
/* Test conversion from UTF-8 to ISO-8859-1 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\303\204rger mit b\303\266sen B\303\274bchen ohne Augenma\303\237";
static const char expected[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_cd_iconveh (input, strlen (input),
&cdeh_utf8_to_88591,
handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 41; i++)
ASSERT (offsets[i] == (i < 1 ? i :
i == 1 ? (size_t)(-1) :
i < 13 ? i - 1 :
i == 13 ? (size_t)(-1) :
i < 20 ? i - 2 :
i == 20 ? (size_t)(-1) :
i < 40 ? i - 3 :
(size_t)(-1)));
ASSERT (offsets[41] == MAGIC);
free (offsets);
}
free (result);
}
}
/* Test conversion from ASCII to UTF-8 with invalid input (EILSEQ). */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Rafa\263 Maszkowski"; /* Rafa? Maszkowski */
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_cd_iconveh (input, strlen (input),
&cdeh_ascii_to_utf8,
handler,
offsets,
&result, &length);
switch (handler)
{
case iconveh_error:
ASSERT (retval == -1 && errno == EILSEQ);
ASSERT (result == NULL);
if (o)
free (offsets);
break;
case iconveh_question_mark:
case iconveh_escape_sequence:
{
static const char expected[] = "Rafa? Maszkowski";
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 16; i++)
ASSERT (offsets[i] == i);
ASSERT (offsets[16] == MAGIC);
free (offsets);
}
free (result);
}
break;
}
}
}
/* Test conversion from UTF-8 to ISO-8859-1 with EILSEQ. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Rafa\305\202 Maszkowski"; /* Rafał Maszkowski */
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_cd_iconveh (input, strlen (input),
&cdeh_utf8_to_88591,
handler,
offsets,
&result, &length);
switch (handler)
{
case iconveh_error:
ASSERT (retval == -1 && errno == EILSEQ);
ASSERT (result == NULL);
if (o)
free (offsets);
break;
case iconveh_question_mark:
{
static const char expected[] = "Rafa? Maszkowski";
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 17; i++)
ASSERT (offsets[i] == (i < 5 ? i :
i == 5 ? (size_t)(-1) :
i - 1));
ASSERT (offsets[17] == MAGIC);
free (offsets);
}
free (result);
}
break;
case iconveh_escape_sequence:
{
static const char expected[] = "Rafa\\u0142 Maszkowski";
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 17; i++)
ASSERT (offsets[i] == (i < 5 ? i :
i == 5 ? (size_t)(-1) :
i + 4));
ASSERT (offsets[17] == MAGIC);
free (offsets);
}
free (result);
}
break;
}
}
}
/* Test conversion from UTF-8 to ISO-8859-1 with EINVAL. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\342";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_cd_iconveh (input, strlen (input),
&cdeh_utf8_to_88591,
handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == 0);
if (o)
{
ASSERT (offsets[0] == 0);
ASSERT (offsets[1] == MAGIC);
free (offsets);
}
free (result);
}
}
if (cd_utf7_to_utf8 != (iconv_t)(-1))
{
/* Disabled on Solaris, because Solaris 9 iconv() is buggy: it returns
-1 / EILSEQ when converting the 7th byte of the input "+VDLYP9hA". */
# if !(defined __sun && !defined _LIBICONV_VERSION)
/* Test conversion from UTF-7 to UTF-8 with EINVAL. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
/* This is base64 encoded 0x54 0x32 0xD8 0x3F 0xD8 0x40. It would
convert to U+5432 U+D83F U+D840 but these are Unicode surrogates. */
static const char input[] = "+VDLYP9hA";
static const char expected1[] = "\345\220\262"; /* 吲 glibc */
static const char expected2[] = ""; /* libiconv */
char *result = NULL;
size_t length = 0;
int retval = mem_cd_iconveh (input, 7,
&cdeh_utf7_to_utf8,
handler,
NULL,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == strlen (expected1) || length == strlen (expected2));
ASSERT (result != NULL);
if (length == strlen (expected1))
ASSERT (memcmp (result, expected1, strlen (expected1)) == 0);
else
ASSERT (memcmp (result, expected2, strlen (expected2)) == 0);
free (result);
}
/* Disabled on NetBSD, because NetBSD 5.0 iconv() is buggy: it converts
the input "+2D/YQNhB" to U+1FED8 U+3FD8 U+40D8. */
# if !(defined __NetBSD__ && !defined _LIBICONV_VERSION)
/* Test conversion from UTF-7 to UTF-8 with EILSEQ. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
/* This is base64 encoded 0xD8 0x3F 0xD8 0x40 0xD8 0x41. It would
convert to U+D83F U+D840 U+D841 but these are Unicode surrogates. */
static const char input[] = "+2D/YQNhB";
char *result = NULL;
size_t length = 0;
int retval = mem_cd_iconveh (input, strlen (input),
&cdeh_utf7_to_utf8,
handler,
NULL,
&result, &length);
switch (handler)
{
case iconveh_error:
ASSERT (retval == -1 && errno == EILSEQ);
ASSERT (result == NULL);
break;
case iconveh_question_mark:
case iconveh_escape_sequence:
{
/* glibc result */
static const char expected1[] = "?????";
/* libiconv <= 1.12 result */
static const char expected2[] = "?2D/YQNhB";
/* libiconv behaviour changed in version 1.13: the result is
'?' U+0FF6 U+1036; this is U+D83F U+D840 U+D841 shifted left
by 6 bits. */
static const char expected3[] = "?\340\277\266\341\200\266";
ASSERT (retval == 0);
ASSERT (length == strlen (expected1)
|| length == strlen (expected2)
|| length == strlen (expected3));
ASSERT (result != NULL);
if (length == strlen (expected1))
ASSERT (memcmp (result, expected1, strlen (expected1)) == 0);
else if (length == strlen (expected2))
ASSERT (memcmp (result, expected2, strlen (expected2)) == 0);
else
ASSERT (memcmp (result, expected3, strlen (expected3)) == 0);
free (result);
}
break;
}
}
# endif
# endif
}
/* ------------------------ Test str_cd_iconveh() ------------------------ */
/* Test conversion from ISO-8859-2 to ISO-8859-1 with no errors. */
for (indirect = 0; indirect <= 1; indirect++)
{
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
static const char expected[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
char *result = str_cd_iconveh (input,
(indirect
? &cdeh_88592_to_88591_indirectly
: &cdeh_88592_to_88591),
handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
}
/* Test conversion from ASCII to ISO-8859-1 with invalid input (EILSEQ). */
for (indirect = 0; indirect <= 1; indirect++)
{
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Rafa\263 Maszkowski"; /* Rafa? Maszkowski */
char *result = str_cd_iconveh (input,
(indirect
? &cdeh_ascii_to_88591_indirectly
: &cdeh_ascii_to_88591),
handler);
switch (handler)
{
case iconveh_error:
ASSERT (result == NULL && errno == EILSEQ);
break;
case iconveh_question_mark:
case iconveh_escape_sequence:
{
static const char expected[] = "Rafa? Maszkowski";
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
break;
}
}
}
/* Test conversion from ISO-8859-2 to ISO-8859-1 with EILSEQ. */
for (indirect = 0; indirect <= 1; indirect++)
{
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Rafa\263 Maszkowski"; /* Rafał Maszkowski */
char *result = str_cd_iconveh (input,
(indirect
? &cdeh_88592_to_88591_indirectly
: &cdeh_88592_to_88591),
handler);
switch (handler)
{
case iconveh_error:
ASSERT (result == NULL && errno == EILSEQ);
break;
case iconveh_question_mark:
{
static const char expected[] = "Rafa? Maszkowski";
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
break;
case iconveh_escape_sequence:
{
static const char expected[] = "Rafa\\u0142 Maszkowski";
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
break;
}
}
}
/* Test conversion from ISO-8859-1 to UTF-8 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
static const char expected[] = "\303\204rger mit b\303\266sen B\303\274bchen ohne Augenma\303\237";
char *result = str_cd_iconveh (input,
&cdeh_88591_to_utf8,
handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
/* Test conversion from UTF-8 to ISO-8859-1 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\303\204rger mit b\303\266sen B\303\274bchen ohne Augenma\303\237";
static const char expected[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
char *result = str_cd_iconveh (input,
&cdeh_utf8_to_88591,
handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
/* Test conversion from ASCII to UTF-8 with invalid input (EILSEQ). */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Rafa\263 Maszkowski"; /* Rafa? Maszkowski */
char *result = str_cd_iconveh (input,
&cdeh_ascii_to_utf8,
handler);
switch (handler)
{
case iconveh_error:
ASSERT (result == NULL && errno == EILSEQ);
break;
case iconveh_question_mark:
case iconveh_escape_sequence:
{
static const char expected[] = "Rafa? Maszkowski";
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
break;
}
}
/* Test conversion from UTF-8 to ISO-8859-1 with EILSEQ. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Costs: 27 \342\202\254"; /* EURO SIGN */
char *result = str_cd_iconveh (input,
&cdeh_utf8_to_88591,
handler);
switch (handler)
{
case iconveh_error:
ASSERT (result == NULL && errno == EILSEQ);
break;
case iconveh_question_mark:
{
static const char expected[] = "Costs: 27 ?";
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
break;
case iconveh_escape_sequence:
{
static const char expected[] = "Costs: 27 \\u20AC";
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
break;
}
}
/* Test conversion from UTF-8 to ISO-8859-1 with EINVAL. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\342";
char *result = str_cd_iconveh (input,
&cdeh_utf8_to_88591,
handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, "") == 0);
free (result);
}
if (cd_88591_to_88592 != (iconv_t)(-1))
iconv_close (cd_88591_to_88592);
if (cd_88592_to_88591 != (iconv_t)(-1))
iconv_close (cd_88592_to_88591);
iconv_close (cd_88591_to_utf8);
iconv_close (cd_utf8_to_88591);
iconv_close (cd_88592_to_utf8);
iconv_close (cd_utf8_to_88592);
/* ------------------------- Test mem_iconveh() ------------------------- */
/* Test conversion from ISO-8859-2 to ISO-8859-1 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
static const char expected[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveh (input, strlen (input),
"ISO-8859-2", "ISO-8859-1",
handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 37; i++)
ASSERT (offsets[i] == i);
ASSERT (offsets[37] == MAGIC);
free (offsets);
}
free (result);
}
}
/* Test conversion from ISO-8859-2 to ISO-8859-1 with EILSEQ. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Rafa\263 Maszkowski"; /* Rafał Maszkowski */
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveh (input, strlen (input),
"ISO-8859-2", "ISO-8859-1",
handler,
offsets,
&result, &length);
switch (handler)
{
case iconveh_error:
ASSERT (retval == -1 && errno == EILSEQ);
ASSERT (result == NULL);
if (o)
free (offsets);
break;
case iconveh_question_mark:
{
static const char expected[] = "Rafa? Maszkowski";
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 16; i++)
ASSERT (offsets[i] == i);
ASSERT (offsets[16] == MAGIC);
free (offsets);
}
free (result);
}
break;
case iconveh_escape_sequence:
{
static const char expected[] = "Rafa\\u0142 Maszkowski";
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 16; i++)
ASSERT (offsets[i] == (i < 5 ? i :
i + 5));
ASSERT (offsets[16] == MAGIC);
free (offsets);
}
free (result);
}
break;
}
}
}
/* Test conversion from ISO-8859-1 to UTF-8 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
static const char expected[] = "\303\204rger mit b\303\266sen B\303\274bchen ohne Augenma\303\237";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveh (input, strlen (input),
"ISO-8859-1", "UTF-8",
handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 37; i++)
ASSERT (offsets[i] == (i < 1 ? i :
i < 12 ? i + 1 :
i < 18 ? i + 2 :
i + 3));
ASSERT (offsets[37] == MAGIC);
free (offsets);
}
free (result);
}
}
/* Test conversion from UTF-8 to ISO-8859-1 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\303\204rger mit b\303\266sen B\303\274bchen ohne Augenma\303\237";
static const char expected[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveh (input, strlen (input),
"UTF-8", "ISO-8859-1",
handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 41; i++)
ASSERT (offsets[i] == (i < 1 ? i :
i == 1 ? (size_t)(-1) :
i < 13 ? i - 1 :
i == 13 ? (size_t)(-1) :
i < 20 ? i - 2 :
i == 20 ? (size_t)(-1) :
i < 40 ? i - 3 :
(size_t)(-1)));
ASSERT (offsets[41] == MAGIC);
free (offsets);
}
free (result);
}
}
/* Test conversion from UTF-8 to ISO-8859-1 with EILSEQ. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Rafa\305\202 Maszkowski"; /* Rafał Maszkowski */
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveh (input, strlen (input),
"UTF-8", "ISO-8859-1",
handler,
offsets,
&result, &length);
switch (handler)
{
case iconveh_error:
ASSERT (retval == -1 && errno == EILSEQ);
ASSERT (result == NULL);
if (o)
free (offsets);
break;
case iconveh_question_mark:
{
static const char expected[] = "Rafa? Maszkowski";
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 17; i++)
ASSERT (offsets[i] == (i < 5 ? i :
i == 5 ? (size_t)(-1) :
i - 1));
ASSERT (offsets[17] == MAGIC);
free (offsets);
}
free (result);
}
break;
case iconveh_escape_sequence:
{
static const char expected[] = "Rafa\\u0142 Maszkowski";
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 17; i++)
ASSERT (offsets[i] == (i < 5 ? i :
i == 5 ? (size_t)(-1) :
i + 4));
ASSERT (offsets[17] == MAGIC);
free (offsets);
}
free (result);
}
break;
}
}
}
/* Test conversion from UTF-8 to ISO-8859-1 with EINVAL. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\342";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveh (input, strlen (input),
"UTF-8", "ISO-8859-1",
handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == 0);
if (o)
{
ASSERT (offsets[0] == 0);
ASSERT (offsets[1] == MAGIC);
free (offsets);
}
free (result);
}
}
/* ------------------------- Test str_iconveh() ------------------------- */
/* Test conversion from ISO-8859-2 to ISO-8859-1 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
static const char expected[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
char *result = str_iconveh (input, "ISO-8859-2", "ISO-8859-1", handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
/* Test conversion from ISO-8859-2 to ISO-8859-1 with EILSEQ. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Rafa\263 Maszkowski"; /* Rafał Maszkowski */
char *result = str_iconveh (input, "ISO-8859-2", "ISO-8859-1", handler);
switch (handler)
{
case iconveh_error:
ASSERT (result == NULL && errno == EILSEQ);
break;
case iconveh_question_mark:
{
static const char expected[] = "Rafa? Maszkowski";
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
break;
case iconveh_escape_sequence:
{
static const char expected[] = "Rafa\\u0142 Maszkowski";
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
break;
}
}
/* Test conversion from ISO-8859-1 to UTF-8 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
static const char expected[] = "\303\204rger mit b\303\266sen B\303\274bchen ohne Augenma\303\237";
char *result = str_iconveh (input, "ISO-8859-1", "UTF-8", handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
/* Test conversion from UTF-8 to ISO-8859-1 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\303\204rger mit b\303\266sen B\303\274bchen ohne Augenma\303\237";
static const char expected[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
char *result = str_iconveh (input, "UTF-8", "ISO-8859-1", handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
/* Test conversion from UTF-8 to ISO-8859-1 with EILSEQ. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Costs: 27 \342\202\254"; /* EURO SIGN */
char *result = str_iconveh (input, "UTF-8", "ISO-8859-1", handler);
switch (handler)
{
case iconveh_error:
ASSERT (result == NULL && errno == EILSEQ);
break;
case iconveh_question_mark:
{
static const char expected[] = "Costs: 27 ?";
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
break;
case iconveh_escape_sequence:
{
static const char expected[] = "Costs: 27 \\u20AC";
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
break;
}
}
/* Test conversion from UTF-8 to ISO-8859-1 with EINVAL. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\342";
char *result = str_iconveh (input, "UTF-8", "ISO-8859-1", handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, "") == 0);
free (result);
}
#endif
return 0;
}
int
main ()
{
static enum iconv_ilseq_handler handlers[] =
{ iconveh_error, iconveh_question_mark, iconveh_escape_sequence };
size_t h;
size_t o;
size_t i;
#if HAVE_ICONV
/* Assume that iconv() supports at least the encodings ASCII, ISO-8859-1,
ISO-8859-2, and UTF-8. */
/* Test conversion from ISO-8859-1 to UTF-8 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
static const uint8_t expected[] = "\303\204rger mit b\303\266sen B\303\274bchen ohne Augenma\303\237";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
size_t length;
uint8_t *result = u8_conv_from_encoding ("ISO-8859-1", handler,
input, strlen (input),
offsets,
NULL, &length);
ASSERT (result != NULL);
ASSERT (length == u8_strlen (expected));
ASSERT (u8_cmp (result, expected, u8_strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 37; i++)
ASSERT (offsets[i] == (i < 1 ? i :
i < 12 ? i + 1 :
i < 18 ? i + 2 :
i + 3));
ASSERT (offsets[37] == MAGIC);
free (offsets);
}
free (result);
}
}
/* Test conversion from ISO-8859-2 to UTF-8 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Rafa\263 Maszkowski"; /* Rafał Maszkowski */
static const uint8_t expected[] = "Rafa\305\202 Maszkowski";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
size_t length;
uint8_t *result = u8_conv_from_encoding ("ISO-8859-2", handler,
input, strlen (input),
offsets,
NULL, &length);
ASSERT (result != NULL);
ASSERT (length == u8_strlen (expected));
ASSERT (u8_cmp (result, expected, u8_strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 16; i++)
ASSERT (offsets[i] == (i < 5 ? i :
i + 1));
ASSERT (offsets[16] == MAGIC);
free (offsets);
}
free (result);
}
}
/* autodetect_jp is only supported when iconv() support ISO-2022-JP-2. */
# if defined _LIBICONV_VERSION || !(defined _AIX || defined __sgi || defined __hpux || defined __osf__ || defined __sun)
/* Test conversions from autodetect_jp to UTF-8. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\244\263\244\363\244\313\244\301\244\317"; /* こんにちは in EUC-JP */
static const uint8_t expected[] = "\343\201\223\343\202\223\343\201\253\343\201\241\343\201\257"; /* こんにちは */
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
size_t length;
uint8_t *result = u8_conv_from_encoding ("autodetect_jp", handler,
input, strlen (input),
offsets,
NULL, &length);
ASSERT (result != NULL);
ASSERT (length == u8_strlen (expected));
ASSERT (u8_cmp (result, expected, u8_strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 10; i++)
ASSERT (offsets[i] == ((i % 2) == 0 ? (i / 2) * 3 : (size_t)(-1)));
ASSERT (offsets[10] == MAGIC);
free (offsets);
}
free (result);
}
}
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\202\261\202\361\202\311\202\277\202\315"; /* こんにちは in Shift_JIS */
static const uint8_t expected[] = "\343\201\223\343\202\223\343\201\253\343\201\241\343\201\257"; /* こんにちは */
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
size_t length;
uint8_t *result = u8_conv_from_encoding ("autodetect_jp", handler,
input, strlen (input),
offsets,
NULL, &length);
ASSERT (result != NULL);
ASSERT (length == u8_strlen (expected));
ASSERT (u8_cmp (result, expected, u8_strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 10; i++)
ASSERT (offsets[i] == ((i % 2) == 0 ? (i / 2) * 3 : (size_t)(-1)));
ASSERT (offsets[10] == MAGIC);
free (offsets);
}
free (result);
}
}
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\033$B$3$s$K$A$O\033(B"; /* こんにちは in ISO-2022-JP-2 */
static const uint8_t expected[] = "\343\201\223\343\202\223\343\201\253\343\201\241\343\201\257"; /* こんにちは */
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
size_t length;
uint8_t *result = u8_conv_from_encoding ("autodetect_jp", handler,
input, strlen (input),
offsets,
NULL, &length);
ASSERT (result != NULL);
ASSERT (length == u8_strlen (expected));
ASSERT (u8_cmp (result, expected, u8_strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 16; i++)
ASSERT (offsets[i] == (i == 0 ? 0 :
i == 5 ? 3 :
i == 7 ? 6 :
i == 9 ? 9 :
i == 11 ? 12 :
i == 13 ? 15 :
(size_t)(-1)));
ASSERT (offsets[16] == MAGIC);
free (offsets);
}
free (result);
}
}
# endif
#endif
return 0;
}
int
main ()
{
static enum iconv_ilseq_handler handlers[] =
{ iconveh_error, iconveh_question_mark, iconveh_escape_sequence };
size_t h;
size_t o;
size_t i;
#if HAVE_ICONV
/* Assume that iconv() supports at least the encodings ASCII, ISO-8859-1,
ISO-8859-2, and UTF-8. */
/* Test conversion from ISO-8859-1 to UTF-16 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
static const uint32_t expected[] = /* Ärger mit bösen Bübchen ohne Augenmaß */
{
0xC4, 'r', 'g', 'e', 'r', ' ', 'm', 'i', 't', ' ', 'b', 0xF6, 's',
'e', 'n', ' ', 'B', 0xFC, 'b', 'c', 'h', 'e', 'n', ' ', 'o', 'h',
'n', 'e', ' ', 'A', 'u', 'g', 'e', 'n', 'm', 'a', 0xDF
};
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
size_t length;
uint32_t *result = u32_conv_from_encoding ("ISO-8859-1", handler,
input, strlen (input),
offsets,
NULL, &length);
ASSERT (result != NULL);
ASSERT (length == SIZEOF (expected));
ASSERT (u32_cmp (result, expected, SIZEOF (expected)) == 0);
if (o)
{
for (i = 0; i < 37; i++)
ASSERT (offsets[i] == i);
ASSERT (offsets[37] == MAGIC);
free (offsets);
}
free (result);
}
}
/* Test conversion from ISO-8859-2 to UTF-16 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Rafa\263 Maszkowski"; /* Rafał Maszkowski */
static const uint32_t expected[] =
{
'R', 'a', 'f', 'a', 0x0142, ' ', 'M', 'a', 's', 'z', 'k', 'o', 'w',
's', 'k', 'i'
};
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
size_t length;
uint32_t *result = u32_conv_from_encoding ("ISO-8859-2", handler,
input, strlen (input),
offsets,
NULL, &length);
ASSERT (result != NULL);
ASSERT (length == SIZEOF (expected));
ASSERT (u32_cmp (result, expected, SIZEOF (expected)) == 0);
if (o)
{
for (i = 0; i < 16; i++)
ASSERT (offsets[i] == i);
ASSERT (offsets[16] == MAGIC);
free (offsets);
}
free (result);
}
}
/* autodetect_jp is only supported when iconv() support ISO-2022-JP-2. */
# if defined _LIBICONV_VERSION || !(defined _AIX || defined __sgi || defined __hpux || defined __osf__ || defined __sun)
/* Test conversions from autodetect_jp to UTF-16. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\244\263\244\363\244\313\244\301\244\317"; /* こんにちは in EUC-JP */
static const uint32_t expected[] = /* こんにちは */
{
0x3053, 0x3093, 0x306B, 0x3061, 0x306F
};
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
size_t length;
uint32_t *result = u32_conv_from_encoding ("autodetect_jp", handler,
input, strlen (input),
offsets,
NULL, &length);
ASSERT (result != NULL);
ASSERT (length == SIZEOF (expected));
ASSERT (u32_cmp (result, expected, SIZEOF (expected)) == 0);
if (o)
{
for (i = 0; i < 10; i++)
ASSERT (offsets[i] == ((i % 2) == 0 ? i / 2 : (size_t)(-1)));
ASSERT (offsets[10] == MAGIC);
free (offsets);
}
free (result);
}
}
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\202\261\202\361\202\311\202\277\202\315"; /* こんにちは in Shift_JIS */
static const uint32_t expected[] = /* こんにちは */
{
0x3053, 0x3093, 0x306B, 0x3061, 0x306F
};
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
size_t length;
uint32_t *result = u32_conv_from_encoding ("autodetect_jp", handler,
input, strlen (input),
offsets,
NULL, &length);
ASSERT (result != NULL);
ASSERT (length == SIZEOF (expected));
ASSERT (u32_cmp (result, expected, SIZEOF (expected)) == 0);
if (o)
{
for (i = 0; i < 10; i++)
ASSERT (offsets[i] == ((i % 2) == 0 ? i / 2 : (size_t)(-1)));
ASSERT (offsets[10] == MAGIC);
free (offsets);
}
free (result);
}
}
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\033$B$3$s$K$A$O\033(B"; /* こんにちは in ISO-2022-JP-2 */
static const uint32_t expected[] = /* こんにちは */
{
0x3053, 0x3093, 0x306B, 0x3061, 0x306F
};
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
size_t length;
uint32_t *result = u32_conv_from_encoding ("autodetect_jp", handler,
input, strlen (input),
offsets,
NULL, &length);
ASSERT (result != NULL);
ASSERT (length == SIZEOF (expected));
ASSERT (u32_cmp (result, expected, SIZEOF (expected)) == 0);
if (o)
{
for (i = 0; i < 16; i++)
ASSERT (offsets[i] == (i == 0 ? 0 :
i == 5 ? 1 :
i == 7 ? 2 :
i == 9 ? 3 :
i == 11 ? 4 :
i == 13 ? 5 :
(size_t)(-1)));
ASSERT (offsets[16] == MAGIC);
free (offsets);
}
free (result);
}
}
# endif
#endif
return 0;
}
int
main ()
{
static enum iconv_ilseq_handler handlers[] =
{ iconveh_error, iconveh_question_mark, iconveh_escape_sequence };
size_t h;
size_t o;
size_t i;
#if HAVE_ICONV
/* Assume that iconv() supports at least the encodings ASCII, ISO-8859-1,
ISO-8859-2, and UTF-8. */
/* ------------------------- Test mem_iconveha() ------------------------- */
/* Test conversion from ISO-8859-2 to ISO-8859-1 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
static const char expected[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveha (input, strlen (input),
"ISO-8859-2", "ISO-8859-1",
false, handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 37; i++)
ASSERT (offsets[i] == i);
ASSERT (offsets[37] == MAGIC);
free (offsets);
}
free (result);
}
}
/* Test conversion from ISO-8859-2 to ISO-8859-1 with EILSEQ. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Rafa\263 Maszkowski"; /* Rafał Maszkowski */
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveha (input, strlen (input),
"ISO-8859-2", "ISO-8859-1",
false, handler,
offsets,
&result, &length);
switch (handler)
{
case iconveh_error:
ASSERT (retval == -1 && errno == EILSEQ);
ASSERT (result == NULL);
if (o)
free (offsets);
break;
case iconveh_question_mark:
{
static const char expected[] = "Rafa? Maszkowski";
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 16; i++)
ASSERT (offsets[i] == i);
ASSERT (offsets[16] == MAGIC);
free (offsets);
}
free (result);
}
break;
case iconveh_escape_sequence:
{
static const char expected[] = "Rafa\\u0142 Maszkowski";
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 16; i++)
ASSERT (offsets[i] == (i < 5 ? i :
i + 5));
ASSERT (offsets[16] == MAGIC);
free (offsets);
}
free (result);
}
break;
}
}
}
/* Test conversion from ISO-8859-1 to UTF-8 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
static const char expected[] = "\303\204rger mit b\303\266sen B\303\274bchen ohne Augenma\303\237";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveha (input, strlen (input),
"ISO-8859-1", "UTF-8",
false, handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 37; i++)
ASSERT (offsets[i] == (i < 1 ? i :
i < 12 ? i + 1 :
i < 18 ? i + 2 :
i + 3));
ASSERT (offsets[37] == MAGIC);
free (offsets);
}
free (result);
}
}
/* Test conversion from UTF-8 to ISO-8859-1 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\303\204rger mit b\303\266sen B\303\274bchen ohne Augenma\303\237";
static const char expected[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveha (input, strlen (input),
"UTF-8", "ISO-8859-1",
false, handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 41; i++)
ASSERT (offsets[i] == (i < 1 ? i :
i == 1 ? (size_t)(-1) :
i < 13 ? i - 1 :
i == 13 ? (size_t)(-1) :
i < 20 ? i - 2 :
i == 20 ? (size_t)(-1) :
i < 40 ? i - 3 :
(size_t)(-1)));
ASSERT (offsets[41] == MAGIC);
free (offsets);
}
free (result);
}
}
/* Test conversion from UTF-8 to ISO-8859-1 with EILSEQ. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Rafa\305\202 Maszkowski"; /* Rafał Maszkowski */
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveha (input, strlen (input),
"UTF-8", "ISO-8859-1",
false, handler,
offsets,
&result, &length);
switch (handler)
{
case iconveh_error:
ASSERT (retval == -1 && errno == EILSEQ);
ASSERT (result == NULL);
if (o)
free (offsets);
break;
case iconveh_question_mark:
{
static const char expected[] = "Rafa? Maszkowski";
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 17; i++)
ASSERT (offsets[i] == (i < 5 ? i :
i == 5 ? (size_t)(-1) :
i - 1));
ASSERT (offsets[17] == MAGIC);
free (offsets);
}
free (result);
}
break;
case iconveh_escape_sequence:
{
static const char expected[] = "Rafa\\u0142 Maszkowski";
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 17; i++)
ASSERT (offsets[i] == (i < 5 ? i :
i == 5 ? (size_t)(-1) :
i + 4));
ASSERT (offsets[17] == MAGIC);
free (offsets);
}
free (result);
}
break;
}
}
}
/* Test conversion from UTF-8 to ISO-8859-1 with EINVAL. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\342";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveha (input, strlen (input),
"UTF-8", "ISO-8859-1",
false, handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == 0);
if (o)
{
ASSERT (offsets[0] == 0);
ASSERT (offsets[1] == MAGIC);
free (offsets);
}
free (result);
}
}
/* autodetect_jp is only supported when iconv() support ISO-2022-JP-2. */
# if defined _LIBICONV_VERSION || !(defined _AIX || defined __sgi || defined __hpux || defined __osf__ || defined __sun)
/* Test conversions from autodetect_jp to UTF-8. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\244\263\244\363\244\313\244\301\244\317"; /* こんにちは in EUC-JP */
static const char expected[] = "\343\201\223\343\202\223\343\201\253\343\201\241\343\201\257"; /* こんにちは */
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveha (input, strlen (input),
"autodetect_jp", "UTF-8",
false, handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 10; i++)
ASSERT (offsets[i] == ((i % 2) == 0 ? (i / 2) * 3 : (size_t)(-1)));
ASSERT (offsets[10] == MAGIC);
free (offsets);
}
free (result);
}
}
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\202\261\202\361\202\311\202\277\202\315"; /* こんにちは in Shift_JIS */
static const char expected[] = "\343\201\223\343\202\223\343\201\253\343\201\241\343\201\257"; /* こんにちは */
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveha (input, strlen (input),
"autodetect_jp", "UTF-8",
false, handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 10; i++)
ASSERT (offsets[i] == ((i % 2) == 0 ? (i / 2) * 3 : (size_t)(-1)));
ASSERT (offsets[10] == MAGIC);
free (offsets);
}
free (result);
}
}
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\033$B$3$s$K$A$O\033(B"; /* こんにちは in ISO-2022-JP-2 */
static const char expected[] = "\343\201\223\343\202\223\343\201\253\343\201\241\343\201\257"; /* こんにちは */
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveha (input, strlen (input),
"autodetect_jp", "UTF-8",
false, handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 16; i++)
ASSERT (offsets[i] == (i == 0 ? 0 :
i == 5 ? 3 :
i == 7 ? 6 :
i == 9 ? 9 :
i == 11 ? 12 :
i == 13 ? 15 :
(size_t)(-1)));
ASSERT (offsets[16] == MAGIC);
free (offsets);
}
free (result);
}
}
# endif
# if (((__GLIBC__ == 2 && __GLIBC_MINOR__ >= 2) || __GLIBC__ > 2) && !defined __UCLIBC__) || _LIBICONV_VERSION >= 0x0105
/* Test conversion from UTF-8 to ISO-8859-1 with transliteration. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Costs: 27 \342\202\254"; /* EURO SIGN */
static const char expected[] = "Costs: 27 EUR";
for (o = 0; o < 2; o++)
{
size_t *offsets = (o ? new_offsets (strlen (input)) : NULL);
char *result = NULL;
size_t length = 0;
int retval = mem_iconveha (input, strlen (input),
"UTF-8", "ISO-8859-1",
true, handler,
offsets,
&result, &length);
ASSERT (retval == 0);
ASSERT (length == strlen (expected));
ASSERT (result != NULL && memcmp (result, expected, strlen (expected)) == 0);
if (o)
{
for (i = 0; i < 13; i++)
ASSERT (offsets[i] == (i < 11 ? i : (size_t)(-1)));
ASSERT (offsets[13] == MAGIC);
free (offsets);
}
free (result);
}
}
# endif
/* ------------------------- Test str_iconveha() ------------------------- */
/* Test conversion from ISO-8859-2 to ISO-8859-1 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
static const char expected[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
char *result = str_iconveha (input, "ISO-8859-2", "ISO-8859-1", false, handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
/* Test conversion from ISO-8859-2 to ISO-8859-1 with EILSEQ. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Rafa\263 Maszkowski"; /* Rafał Maszkowski */
char *result = str_iconveha (input, "ISO-8859-2", "ISO-8859-1", false, handler);
switch (handler)
{
case iconveh_error:
ASSERT (result == NULL && errno == EILSEQ);
break;
case iconveh_question_mark:
{
static const char expected[] = "Rafa? Maszkowski";
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
break;
case iconveh_escape_sequence:
{
static const char expected[] = "Rafa\\u0142 Maszkowski";
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
break;
}
}
/* Test conversion from ISO-8859-1 to UTF-8 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
static const char expected[] = "\303\204rger mit b\303\266sen B\303\274bchen ohne Augenma\303\237";
char *result = str_iconveha (input, "ISO-8859-1", "UTF-8", false, handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
/* Test conversion from UTF-8 to ISO-8859-1 with no errors. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\303\204rger mit b\303\266sen B\303\274bchen ohne Augenma\303\237";
static const char expected[] = "\304rger mit b\366sen B\374bchen ohne Augenma\337";
char *result = str_iconveha (input, "UTF-8", "ISO-8859-1", false, handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
/* Test conversion from UTF-8 to ISO-8859-1 with EILSEQ. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Costs: 27 \342\202\254"; /* EURO SIGN */
char *result = str_iconveha (input, "UTF-8", "ISO-8859-1", false, handler);
switch (handler)
{
case iconveh_error:
ASSERT (result == NULL && errno == EILSEQ);
break;
case iconveh_question_mark:
{
static const char expected[] = "Costs: 27 ?";
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
break;
case iconveh_escape_sequence:
{
static const char expected[] = "Costs: 27 \\u20AC";
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
break;
}
}
/* Test conversion from UTF-8 to ISO-8859-1 with EINVAL. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\342";
char *result = str_iconveha (input, "UTF-8", "ISO-8859-1", false, handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, "") == 0);
free (result);
}
/* autodetect_jp is only supported when iconv() support ISO-2022-JP-2. */
# if defined _LIBICONV_VERSION || !(defined _AIX || defined __sgi || defined __hpux || defined __osf__ || defined __sun)
/* Test conversions from autodetect_jp to UTF-8. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\244\263\244\363\244\313\244\301\244\317"; /* こんにちは in EUC-JP */
static const char expected[] = "\343\201\223\343\202\223\343\201\253\343\201\241\343\201\257"; /* こんにちは */
char *result = str_iconveha (input, "autodetect_jp", "UTF-8", false, handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\202\261\202\361\202\311\202\277\202\315"; /* こんにちは in Shift_JIS */
static const char expected[] = "\343\201\223\343\202\223\343\201\253\343\201\241\343\201\257"; /* こんにちは */
char *result = str_iconveha (input, "autodetect_jp", "UTF-8", false, handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "\033$B$3$s$K$A$O\033(B"; /* こんにちは in ISO-2022-JP-2 */
static const char expected[] = "\343\201\223\343\202\223\343\201\253\343\201\241\343\201\257"; /* こんにちは */
char *result = str_iconveha (input, "autodetect_jp", "UTF-8", false, handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
# endif
# if (((__GLIBC__ == 2 && __GLIBC_MINOR__ >= 2) || __GLIBC__ > 2) && !defined __UCLIBC__) || _LIBICONV_VERSION >= 0x0105
/* Test conversion from UTF-8 to ISO-8859-1 with transliteration. */
for (h = 0; h < SIZEOF (handlers); h++)
{
enum iconv_ilseq_handler handler = handlers[h];
static const char input[] = "Costs: 27 \342\202\254"; /* EURO SIGN */
static const char expected[] = "Costs: 27 EUR";
char *result = str_iconveha (input, "UTF-8", "ISO-8859-1", true, handler);
ASSERT (result != NULL);
ASSERT (strcmp (result, expected) == 0);
free (result);
}
# endif
#endif
return 0;
}