void
idn_testsuite_assertucs4string(idn_testsuite_t ctx,
const unsigned long *gotten,
const unsigned long *expected,
const char *file, int lineno) {
char *expected_hex = NULL;
char *gotten_hex = NULL;
char *msg;
assert(ctx != NULL && gotten != NULL && expected != NULL &&
file != NULL);
if (idn_testsuite_getstatus(ctx) != idn_teststatus_pass)
return;
if (idn_ucs4_strcmp(expected, gotten) == 0)
return;
idn_testsuite_setstatus(ctx, idn_teststatus_fail);
msg = (char *)malloc(idn_ucs4_strlen(expected) * 8 +
idn_ucs4_strlen(gotten) * 8 + 32);
expected_hex = make_hex_ucs4string(expected);
gotten_hex = make_hex_ucs4string(gotten);
if (msg == NULL || expected_hex == NULL || gotten_hex == NULL) {
msg = "";
} else {
sprintf(msg, "`%s' expected, but got `%s'",
expected_hex, gotten_hex);
}
put_failure_message(ctx, msg, file, lineno);
free(msg);
free(expected_hex);
free(gotten_hex);
}
static void
labellist_undo(labellist_t label) {
size_t length;
length = idn_ucs4_strlen(label->undo_name) + 1;
memcpy(label->name, label->undo_name, sizeof(long) * length);
}
static idn_result_t
label_localmap(idn_resconf_t ctx, labellist_t label) {
const unsigned long *from;
const unsigned long *tld;
unsigned long *to = NULL;
size_t to_length;
idn_mapselector_t local_mapper;
idn_result_t r;
from = labellist_getname(label);
tld = labellist_gettldname(label);
TRACE(("res localmap(label=\"%s\", tld=\"%s\")\n",
idn__debug_ucs4xstring(from, 50),
idn__debug_ucs4xstring(tld, 50)));
local_mapper = idn_resconf_getlocalmapselector(ctx);
if (local_mapper == NULL) {
r = idn_success;
goto ret;
}
if (tld == from)
tld = idn_mapselector_getdefaulttld();
to_length = idn_ucs4_strlen(from) + 1 + 15; /* 15 for margin */
for (;;) {
unsigned long *new_buffer;
new_buffer = (unsigned long *)
realloc(to, sizeof(long) * to_length);
if (new_buffer == NULL) {
r = idn_nomemory;
goto ret;
}
to = new_buffer;
r = idn_mapselector_map2(local_mapper, from, tld, to,
to_length);
if (r == idn_success)
break;
else if (r != idn_buffer_overflow)
goto ret;
to_length *= 2;
}
r = labellist_setname(label, to);
ret:
if (r == idn_success) {
TRACE(("res localmap(): success (label=\"%s\")\n",
idn__debug_ucs4xstring(labellist_getname(label),
50)));
} else {
TRACE(("res localmap(): %s\n", idn_result_tostring(r)));
}
if (local_mapper != NULL)
idn_mapselector_destroy(local_mapper);
free(to);
return (r);
}
static idn_result_t
label_lencheck_nonace(idn_resconf_t ctx, labellist_t label) {
idn_converter_t idn_converter;
const unsigned long *from;
size_t to_length;
idn_result_t r;
char *buffer = NULL;
size_t buffer_length;
from = labellist_getname(label);
TRACE(("res lencheck(label=\"%s\")\n",
idn__debug_ucs4xstring(from, 50)));
buffer_length = idn_ucs4_strlen(from) * 4 + 16; /* 16 for margin */
idn_converter = idn_resconf_getidnconverter(ctx);
for (;;) {
void *new_buffer;
new_buffer = realloc(buffer, sizeof(*buffer) * buffer_length);
if (new_buffer == NULL) {
r = idn_nomemory;
goto ret;
}
buffer = (char *)new_buffer;
if (idn_converter != NULL) {
r = idn_converter_convfromucs4(idn_converter, from,
buffer, buffer_length);
} else {
r = idn_ucs4_ucs4toutf8(from, buffer, buffer_length);
}
if (r == idn_success)
break;
else if (r != idn_buffer_overflow)
goto ret;
buffer_length *= 2;
}
to_length = strlen(buffer);
if (to_length == 0 || to_length > MAX_LABEL_LENGTH) {
r = idn_invalid_length;
goto ret;
}
r = idn_success;
ret:
TRACE(("res lencheck(): %s\n", idn_result_tostring(r)));
if (idn_converter != NULL)
idn_converter_destroy(idn_converter);
free(buffer);
return (r);
}
unsigned long *
idn_ucs4_strdup(const unsigned long *str) {
size_t length = idn_ucs4_strlen(str);
unsigned long *dupstr;
dupstr = (unsigned long *)malloc(sizeof(*str) * (length + 1));
if (dupstr == NULL)
return NULL;
memcpy(dupstr, str, sizeof(*str) * (length + 1));
return dupstr;
}
static idn_result_t
label_normalize(idn_resconf_t ctx, labellist_t label) {
const unsigned long *from;
unsigned long *to = NULL;
size_t to_length;
idn_normalizer_t normalizer;
idn_result_t r;
from = labellist_getname(label);
TRACE(("res normalzie(label=\"%s\")\n",
idn__debug_ucs4xstring(from, 50)));
normalizer = idn_resconf_getnormalizer(ctx);
if (normalizer == NULL) {
r = idn_success;
goto ret;
}
to_length = idn_ucs4_strlen(from) + 1 + 15; /* 15 for margin */
for (;;) {
unsigned long *new_buffer;
new_buffer = (unsigned long *)
realloc(to, sizeof(long) * to_length);
if (new_buffer == NULL) {
r = idn_nomemory;
goto ret;
}
to = new_buffer;
r = idn_normalizer_normalize(normalizer, from, to, to_length);
if (r == idn_success)
break;
else if (r != idn_buffer_overflow)
goto ret;
to_length *= 2;
}
r = labellist_setname(label, to);
ret:
if (r == idn_success) {
TRACE(("res normalize(): success (label=\"%s\")\n",
idn__debug_ucs4xstring(labellist_getname(label),
50)));
} else {
TRACE(("res normalize(): %s\n", idn_result_tostring(r)));
}
if (normalizer != NULL)
idn_normalizer_destroy(normalizer);
free(to);
return (r);
}
static idn_result_t
label_localdecodecheck(idn_resconf_t ctx, labellist_t label) {
idn_converter_t local_converter = NULL;
const unsigned long *from;
char *to = NULL;
size_t to_length;
idn_result_t r;
from = labellist_getname(label);
to_length = idn_ucs4_strlen(from) + 1 + 15; /* 15 for margin */
TRACE(("res ucs4tolocal_check(label=\"%s\")\n",
idn__debug_ucs4xstring(from, 50)));
local_converter = idn_resconf_getlocalconverter(ctx);
if (local_converter == NULL) {
r = idn_success;
goto ret;
}
for (;;) {
char *new_buffer;
new_buffer = (char *)realloc(to, to_length);
if (new_buffer == NULL) {
r = idn_nomemory;
goto ret;
}
to = new_buffer;
r = idn_converter_convfromucs4(local_converter, from, to,
to_length);
if (r == idn_success)
break;
else if (r == idn_nomapping) {
r = label_idnencode_ace(ctx, label);
if (r != idn_success)
goto ret;
break;
} else if (r != idn_buffer_overflow) {
goto ret;
}
to_length *= 2;
}
r = idn_success;
ret:
TRACE(("res ucs4tolocal_check(): %s\n", idn_result_tostring(r)));
if (local_converter != NULL)
idn_converter_destroy(local_converter);
free(to);
return (r);
}
static idn_result_t
labellist_getnamelist(labellist_t label, unsigned long *name,
size_t name_length) {
static const unsigned long dot_string[] = {0x002e, 0x0000}; /* "." */
size_t length;
labellist_t l;
for (l = label, length = 0; l != NULL; l = l->next)
length += idn_ucs4_strlen(l->name) + 1; /* name + `.' */
length++; /* for NUL */
if (name_length < length)
return (idn_buffer_overflow);
*name = '\0';
for (l = label; l != NULL; l = l->next) {
idn_ucs4_strcat(name, l->name);
name += idn_ucs4_strlen(name);
if (l->dot_followed)
idn_ucs4_strcat(name, dot_string);
}
return (idn_success);
}
static idn_result_t
labellist_setname(labellist_t label, const unsigned long *name) {
unsigned long *new_name;
size_t length, new_length;
length = idn_ucs4_strlen(name) + 1;
new_length = length + 15; /* add 15 for margin */
if (label->name_length < new_length) {
new_name = (unsigned long *)
realloc(label->name, sizeof(long) * new_length);
if (new_name == NULL)
return (idn_nomemory);
label->name = new_name;
label->name_length = new_length;
}
memcpy(label->name, name, sizeof(long) * length);
return (idn_success);
}
static idn_result_t
label_lencheck_ace(idn_resconf_t ctx, labellist_t label) {
const unsigned long *name;
size_t name_length;
idn_result_t r;
name = labellist_getname(label);
name_length = idn_ucs4_strlen(name);
TRACE(("res lencheck(label=\"%s\")\n",
idn__debug_ucs4xstring(name, 50)));
if (name_length == 0 || name_length > MAX_LABEL_LENGTH) {
r = idn_invalid_length;
goto ret;
}
r = idn_success;
ret:
TRACE(("res lencheck(): %s\n", idn_result_tostring(r)));
return (r);
}
static char *
make_hex_ucs4string(const unsigned long *string) {
static const char hex[] = {"0123456789abcdef"};
char *hex_string;
const unsigned long *src;
char *dst;
hex_string = (char *)malloc((idn_ucs4_strlen(string)) * 8 + 1);
if (hex_string == NULL)
return NULL;
for (src = string, dst = hex_string; *src != '\0'; src++) {
if (0x20 <= *src && *src <= 0x7e && *src != '\\') {
*dst++ = *src;
} else if (*src <= UCS4_MAX) {
*dst++ = '\\';
*dst++ = 'u';
if (*src >= 0x100000) {
*dst++ = hex[(*src >> 20) & 0x0f];
}
if (*src >= 0x10000) {
*dst++ = hex[(*src >> 16) & 0x0f];
}
idn_result_t
idn_normalizer_normalize(idn_normalizer_t ctx, const unsigned long *from,
unsigned long *to, size_t tolen) {
idn_result_t r;
unsigned long *src, *dst;
unsigned long *buffers[2] = {NULL, NULL};
size_t buflen[2] = {0, 0};
size_t dstlen;
int idx;
int i;
assert(scheme_hash != NULL);
assert(ctx != NULL && from != NULL && to != NULL);
TRACE(("idn_normalizer_normalize(from=\"%s\", tolen=%d)\n",
idn__debug_ucs4xstring(from, 50), (int)tolen));
if (ctx->nschemes <= 0) {
if (tolen < idn_ucs4_strlen(from) + 1) {
r = idn_buffer_overflow;
goto ret;
}
idn_ucs4_strcpy(to, from);
r = idn_success;
goto ret;
}
/*
* Normalize.
*/
src = (void *)from;
dstlen = idn_ucs4_strlen(from) + 1;
i = 0;
while (i < ctx->nschemes) {
TRACE(("idn_normalizer_normalize(): normalize %s\n",
ctx->schemes[i]->name));
/*
* Choose destination area to restore the result of a mapping.
*/
if (i + 1 == ctx->nschemes) {
dst = to;
dstlen = tolen;
} else {
if (src == buffers[0])
idx = 1;
else
idx = 0;
if (buflen[idx] < dstlen) {
void *newbuf;
newbuf = realloc(buffers[idx],
sizeof(long) * dstlen);
if (newbuf == NULL) {
r = idn_nomemory;
goto ret;
}
buffers[idx] = (unsigned long *)newbuf;
buflen[idx] = dstlen;
}
dst = buffers[idx];
dstlen = buflen[idx];
}
/*
* Perform i-th normalization scheme.
* If buffer size is not enough, we double it and try again.
*/
r = (ctx->schemes[i]->proc)(src, dst, dstlen);
if (r == idn_buffer_overflow && dst != to) {
dstlen *= 2;
continue;
}
if (r != idn_success)
goto ret;
src = dst;
i++;
}
r = idn_success;
ret:
free(buffers[0]);
free(buffers[1]);
if (r == idn_success) {
TRACE(("idn_normalizer_normalize(): success (to=\"%s\")\n",
idn__debug_ucs4xstring(to, 50)));
} else {
TRACE(("idn_normalizer_normalize(): %s\n",
idn_result_tostring(r)));
}
return (r);
}
static idn_result_t
roundtrip_check(idn_converter_t ctx, const unsigned long *from, const char *to)
{
/*
* One problem with iconv() convertion is that
* iconv() doesn't signal an error if the input
* string contains characters which are valid but
* do not have mapping to the output codeset.
* (the behavior of iconv() for that case is defined as
* `implementation dependent')
* One way to check this case is to perform round-trip
* conversion and see if it is same as the original string.
*/
idn_result_t r;
unsigned long *back;
unsigned long backbuf[256];
size_t fromlen;
size_t backlen;
TRACE(("idn_converter_convert: round-trip checking (from=\"%s\")\n",
idn__debug_ucs4xstring(from, 50)));
/* Allocate enough buffer. */
fromlen = idn_ucs4_strlen(from) + 1;
if (fromlen * sizeof(*back) <= sizeof(backbuf)) {
backlen = sizeof(backbuf);
back = backbuf;
} else {
backlen = fromlen;
back = (unsigned long *)malloc(backlen * sizeof(*back));
if (back == NULL)
return (idn_nomemory);
}
/*
* Perform backward conversion.
*/
r = idn_converter_convtoucs4(ctx, to, back, backlen);
switch (r) {
case idn_success:
if (memcmp(back, from, sizeof(*from) * fromlen) != 0)
r = idn_nomapping;
break;
case idn_invalid_encoding:
case idn_buffer_overflow:
r = idn_nomapping;
break;
default:
break;
}
if (back != backbuf)
free(back);
if (r != idn_success) {
TRACE(("round-trip check failed: %s\n",
idn_result_tostring(r)));
}
return (r);
}
static idn_result_t
label_idnencode_ace(idn_resconf_t ctx, labellist_t label) {
idn_converter_t idn_converter = NULL;
const unsigned long *from;
char *ascii_to = NULL;
unsigned long *to = NULL;
size_t to_length;
idn_result_t r;
from = labellist_getname(label);
TRACE(("res ucs4toidn(label=\"%s\")\n",
idn__debug_ucs4xstring(from, 50)));
idn_converter = idn_resconf_getidnconverter(ctx);
if (idn_converter == NULL) {
r = idn_success;
goto ret;
}
ascii_to = NULL;
to_length = idn_ucs4_strlen(from) * 4 + 16; /* add mergin */
for (;;) {
char *new_buffer;
new_buffer = (char *) realloc(ascii_to, to_length);
if (new_buffer == NULL) {
r = idn_nomemory;
goto ret;
}
ascii_to = new_buffer;
r = idn_converter_convfromucs4(idn_converter, from, ascii_to,
to_length);
if (r == idn_success)
break;
else if (r != idn_buffer_overflow)
goto ret;
to_length *= 2;
}
for (;;) {
unsigned long *new_buffer;
new_buffer = (unsigned long *)
realloc(to, sizeof(long) * to_length);
if (new_buffer == NULL) {
r = idn_nomemory;
goto ret;
}
to = new_buffer;
r = idn_ucs4_utf8toucs4(ascii_to, to, to_length);
if (r == idn_success)
break;
else if (r != idn_buffer_overflow)
goto ret;
to_length *= 2;
}
if (r != idn_success)
goto ret;
r = labellist_setname(label, to);
ret:
if (r == idn_success) {
TRACE(("res ucs4toidn(): success (label=\"%s\")\n",
idn__debug_ucs4xstring(labellist_getname(label),
50)));
} else {
TRACE(("res ucs4toidn(): %s\n", idn_result_tostring(r)));
}
if (idn_converter != NULL)
idn_converter_destroy(idn_converter);
free(to);
free(ascii_to);
return (r);
}
idn_result_t
idn__punycode_encode(idn_converter_t ctx, void *privdata,
const unsigned long *from, char *to, size_t tolen) {
char *to_org = to;
unsigned long cur_code, next_code, delta;
size_t prefixlen = strlen(IDN_PUNYCODE_PREFIX);
size_t fromlen;
size_t ucsdone;
size_t toidx;
int uidx, bias, first;
idn_result_t r;
assert(ctx != NULL);
TRACE(("idn__punycode_encode(from=\"%s\", tolen=%d)\n",
idn__debug_ucs4xstring(from, 50), (int)tolen));
if (*from == '\0') {
r = idn_ucs4_ucs4toutf8(from, to, tolen);
goto ret;
} else if (idn__util_ucs4haveaceprefix(from, IDN_PUNYCODE_PREFIX)) {
r = idn_prohibited;
goto ret;
}
if (tolen < prefixlen) {
r = idn_buffer_overflow;
goto ret;
}
memcpy(to, IDN_PUNYCODE_PREFIX, prefixlen);
to += prefixlen;
tolen -= prefixlen;
fromlen = idn_ucs4_strlen(from);
/*
* If the input string is too long (actually too long to be sane),
* return failure in order to prevent possible overflow.
*/
if (fromlen > PUNYCODE_MAXINPUT) {
ERROR(("idn__punycode_encode(): "
"the input string is too long to convert Punycode\n",
idn__debug_ucs4xstring(from, 50)));
r = idn_failure;
goto ret;
}
ucsdone = 0; /* number of characters processed */
toidx = 0;
/*
* First, pick up basic code points and copy them to 'to'.
*/
for (uidx = 0; uidx < fromlen; uidx++) {
if (from[uidx] < 0x80) {
if (toidx >= tolen) {
r = idn_buffer_overflow;
goto ret;
}
to[toidx++] = from[uidx];
ucsdone++;
}
}
/*
* If there are any basic code points, output a delimiter
* (hyphen-minus).
*/
if (toidx > 0) {
if (toidx >= tolen) {
r = idn_buffer_overflow;
goto ret;
}
to[toidx++] = '-';
to += toidx;
tolen -= toidx;
}
/*
* Then encode non-basic characters.
*/
first = 1;
cur_code = PUNYCODE_INITIAL_N;
bias = PUNYCODE_INITIAL_BIAS;
delta = 0;
while (ucsdone < fromlen) {
int limit = -1, rest;
/*
* Find the smallest code point equal to or greater
* than 'cur_code'. Also remember the index of the
* last occurence of the code point.
*/
for (next_code = MAX_UCS, uidx = fromlen - 1;
uidx >= 0; uidx--) {
if (from[uidx] >= cur_code && from[uidx] < next_code) {
next_code = from[uidx];
limit = uidx;
}
}
/* There must be such code point. */
assert(limit >= 0);
delta += (next_code - cur_code) * (ucsdone + 1);
cur_code = next_code;
/*
* Scan the input string again, and encode characters
* whose code point is 'cur_code'. Use 'limit' to avoid
* unnecessary scan.
*/
for (uidx = 0, rest = ucsdone; uidx <= limit; uidx++) {
if (from[uidx] < cur_code) {
delta++;
rest--;
} else if (from[uidx] == cur_code) {
int sz = punycode_putwc(to, tolen, delta, bias);
if (sz == 0) {
r = idn_buffer_overflow;
goto ret;
}
to += sz;
tolen -= sz;
ucsdone++;
bias = punycode_update_bias(delta, ucsdone,
first);
delta = 0;
first = 0;
}
}
delta += rest + 1;
cur_code++;
}
/*
* Terminate with NUL.
*/
if (tolen <= 0) {
r = idn_buffer_overflow;
goto ret;
}
*to = '\0';
r = idn_success;
ret:
if (r == idn_success) {
TRACE(("idn__punycode_encode(): succcess (to=\"%s\")\n",
idn__debug_xstring(to_org, 50)));
} else {
TRACE(("idn__punycode_encode(): %s\n", idn_result_tostring(r)));
}
return (r);
}