static void php_intl_idn_to_46(INTERNAL_FUNCTION_PARAMETERS,
const zend_string *domain, uint32_t option, int mode, zval *idna_info)
{
UErrorCode status = U_ZERO_ERROR;
UIDNA *uts46;
int32_t len;
zend_string *buffer;
UIDNAInfo info = UIDNA_INFO_INITIALIZER;
uts46 = uidna_openUTS46(option, &status);
if (php_intl_idn_check_status(status, "failed to open UIDNA instance") == FAILURE) {
RETURN_FALSE;
}
if (mode == INTL_IDN_TO_ASCII) {
const int32_t buffer_capac = 255;
buffer = zend_string_alloc(buffer_capac, 0);
len = uidna_nameToASCII_UTF8(uts46, ZSTR_VAL(domain), ZSTR_LEN(domain),
ZSTR_VAL(buffer), buffer_capac, &info, &status);
if (len >= buffer_capac || php_intl_idn_check_status(status, "failed to convert name") == FAILURE) {
uidna_close(uts46);
zend_string_efree(buffer);
RETURN_FALSE;
}
} else {
const int32_t buffer_capac = 252*4;
buffer = zend_string_alloc(buffer_capac, 0);
len = uidna_nameToUnicodeUTF8(uts46, ZSTR_VAL(domain), ZSTR_LEN(domain),
ZSTR_VAL(buffer), buffer_capac, &info, &status);
if (len >= buffer_capac || php_intl_idn_check_status(status, "failed to convert name") == FAILURE) {
uidna_close(uts46);
zend_string_efree(buffer);
RETURN_FALSE;
}
}
ZSTR_VAL(buffer)[len] = '\0';
ZSTR_LEN(buffer) = len;
if (info.errors == 0) {
RETVAL_STR_COPY(buffer);
} else {
RETVAL_FALSE;
}
if (idna_info) {
add_assoc_str_ex(idna_info, "result", sizeof("result")-1, zend_string_copy(buffer));
add_assoc_bool_ex(idna_info, "isTransitionalDifferent",
sizeof("isTransitionalDifferent")-1, info.isTransitionalDifferent);
add_assoc_long_ex(idna_info, "errors", sizeof("errors")-1, (zend_long)info.errors);
}
zend_string_release(buffer);
uidna_close(uts46);
}
/*
Function:
ToUnicode
Used by System.Globalization.IdnMapping.GetUnicodeCore to convert an ASCII name
to Unicode
Return values:
0: internal error during conversion.
>0: the length of the converted string (not including the null terminator).
*/
extern "C" int32_t GlobalizationNative_ToUnicode(
int32_t flags, const UChar* lpSrc, int32_t cwSrcLength, UChar* lpDst, int32_t cwDstLength)
{
UErrorCode err = U_ZERO_ERROR;
UIDNAInfo info = UIDNA_INFO_INITIALIZER;
UIDNA* pIdna = uidna_openUTS46(GetOptions(flags), &err);
int32_t unicodeStrLen = uidna_nameToUnicode(pIdna, lpSrc, cwSrcLength, lpDst, cwDstLength, &info, &err);
uidna_close(pIdna);
return ((U_SUCCESS(err) || (err == U_BUFFER_OVERFLOW_ERROR)) && (info.errors == 0)) ? unicodeStrLen : 0;
}
static Variant php_intl_idn_to_46(const String& domain, int64_t options,
IdnVariant idn_variant, VRefParam idna_info,
int mode) {
int32_t converted_capacity;
char *converted = NULL;
int32_t converted_len;
UIDNA *uts46;
UIDNAInfo info = UIDNA_INFO_INITIALIZER;
UErrorCode status = U_ZERO_ERROR;
// Get UIDNA instance which implements UTS #46.
uts46 = uidna_openUTS46(options, &status);
SCOPE_EXIT { uidna_close(uts46); };
if (U_FAILURE(status)) return false;
// Call the appropriate IDN function
status = U_ZERO_ERROR;
converted_capacity = 255; // no domain name may exceed this
String result(converted_capacity, ReserveString); // reserves converted_capacity+1 characters.
converted = result.bufferSlice().ptr;
if (mode == INTL_IDN_TO_ASCII) {
converted_len = uidna_nameToASCII_UTF8(uts46, (char*)domain.data(), domain.size(),
converted, converted_capacity, &info, &status);
} else {
converted_len = uidna_nameToUnicodeUTF8(uts46, (char*)domain.data(), domain.size(),
converted, converted_capacity, &info, &status);
}
if (U_FAILURE(status) || converted_len > converted_capacity) return false;
if (info.errors == 0) {
result.setSize(converted_len);
} else {
result.setSize(0);
}
// Set up the array returned in idna_info.
ArrayInit arr(3);
arr.set(s_result, result);
arr.set(s_isTransitionalDifferent, info.isTransitionalDifferent);
arr.set(s_errors, (long)info.errors);
// As in Zend, the previous value of idn_variant is overwritten, not modified.
idna_info = arr.create();
if (info.errors == 0) {
return result;
}
return false;
}
static void php_intl_idn_to_46(INTERNAL_FUNCTION_PARAMETERS,
const char *domain, int32_t domain_len, uint32_t option, int mode, zval *idna_info)
{
UErrorCode status = U_ZERO_ERROR;
UIDNA *uts46;
int32_t len;
int32_t buffer_capac = 255; /* no domain name may exceed this */
zend_string *buffer = zend_string_alloc(buffer_capac, 0);
UIDNAInfo info = UIDNA_INFO_INITIALIZER;
int buffer_used = 0;
uts46 = uidna_openUTS46(option, &status);
if (php_intl_idn_check_status(status, "failed to open UIDNA instance",
mode) == FAILURE) {
zend_string_free(buffer);
RETURN_FALSE;
}
if (mode == INTL_IDN_TO_ASCII) {
len = uidna_nameToASCII_UTF8(uts46, domain, domain_len,
ZSTR_VAL(buffer), buffer_capac, &info, &status);
} else {
len = uidna_nameToUnicodeUTF8(uts46, domain, domain_len,
ZSTR_VAL(buffer), buffer_capac, &info, &status);
}
if (php_intl_idn_check_status(status, "failed to convert name",
mode) == FAILURE) {
uidna_close(uts46);
zend_string_free(buffer);
RETURN_FALSE;
}
if (len >= 255) {
php_error_docref(NULL, E_ERROR, "ICU returned an unexpected length");
}
ZSTR_VAL(buffer)[len] = '\0';
ZSTR_LEN(buffer) = len;
if (info.errors == 0) {
RETVAL_STR(buffer);
buffer_used = 1;
} else {
RETVAL_FALSE;
}
if (idna_info) {
if (buffer_used) { /* used in return_value then */
zval_addref_p(return_value);
add_assoc_zval_ex(idna_info, "result", sizeof("result")-1, return_value);
} else {
zval zv;
ZVAL_NEW_STR(&zv, buffer);
buffer_used = 1;
add_assoc_zval_ex(idna_info, "result", sizeof("result")-1, &zv);
}
add_assoc_bool_ex(idna_info, "isTransitionalDifferent",
sizeof("isTransitionalDifferent")-1, info.isTransitionalDifferent);
add_assoc_long_ex(idna_info, "errors", sizeof("errors")-1, (zend_long)info.errors);
}
if (!buffer_used) {
zend_string_free(buffer);
}
uidna_close(uts46);
}
static void php_intl_idn_to(INTERNAL_FUNCTION_PARAMETERS,
const zend_string *domain, uint32_t option, int mode)
{
UChar* ustring = NULL;
int ustring_len = 0;
UErrorCode status;
zend_string *u8str;
/* convert the string to UTF-16. */
status = U_ZERO_ERROR;
intl_convert_utf8_to_utf16(&ustring, &ustring_len, ZSTR_VAL(domain), ZSTR_LEN(domain), &status);
if (U_FAILURE(status)) {
intl_error_set_code(NULL, status);
/* Set error messages. */
intl_error_set_custom_msg( NULL, "Error converting input string to UTF-16", 0 );
if (ustring) {
efree(ustring);
}
RETURN_FALSE;
} else {
UChar converted[MAXPATHLEN];
int32_t converted_ret_len;
status = U_ZERO_ERROR;
#if U_ICU_VERSION_MAJOR_NUM >= 55
UIDNAInfo info = UIDNA_INFO_INITIALIZER;
UIDNA *idna = uidna_openUTS46((int32_t)option, &status);
if (U_FAILURE(status)) {
intl_error_set( NULL, status, "idn_to_ascii: failed to create an UIDNA instance", 0 );
RETURN_FALSE;
}
if (mode == INTL_IDN_TO_ASCII) {
converted_ret_len = uidna_nameToASCII(idna, ustring, ustring_len, converted, MAXPATHLEN, &info, &status);
} else {
converted_ret_len = uidna_nameToUnicode(idna, ustring, ustring_len, converted, MAXPATHLEN, &info, &status);
}
uidna_close(idna);
#else
UParseError parse_error;
if (mode == INTL_IDN_TO_ASCII) {
converted_ret_len = uidna_IDNToASCII(ustring, ustring_len, converted, MAXPATHLEN, (int32_t)option, &parse_error, &status);
} else {
converted_ret_len = uidna_IDNToUnicode(ustring, ustring_len, converted, MAXPATHLEN, (int32_t)option, &parse_error, &status);
}
#endif
efree(ustring);
if (U_FAILURE(status)) {
intl_error_set( NULL, status, "idn_to_ascii: cannot convert to ASCII", 0 );
RETURN_FALSE;
}
status = U_ZERO_ERROR;
u8str = intl_convert_utf16_to_utf8(converted, converted_ret_len, &status);
if (!u8str) {
/* Set global error code. */
intl_error_set_code(NULL, status);
/* Set error messages. */
intl_error_set_custom_msg( NULL, "Error converting output string to UTF-8", 0 );
RETURN_FALSE;
}
}
/* return the allocated string, not a duplicate */
RETVAL_NEW_STR(u8str);
}
/**
* psl_load_fp:
* @fp: FILE pointer
*
* This function loads the public suffixes from a FILE pointer.
* To free the allocated resources, call psl_free().
*
* The suffixes are expected to be lowercase UTF-8 encoded if they are international.
*
* Returns: Pointer to a PSL context or %NULL on failure.
*
* Since: 0.1
*/
psl_ctx_t *psl_load_fp(FILE *fp)
{
psl_ctx_t *psl;
_psl_entry_t suffix, *suffixp;
char buf[256], *linep, *p;
#ifdef WITH_LIBICU
UIDNA *idna;
UErrorCode status = 0;
#endif
if (!fp)
return NULL;
if (!(psl = calloc(1, sizeof(psl_ctx_t))))
return NULL;
#ifdef WITH_LIBICU
idna = uidna_openUTS46(UIDNA_USE_STD3_RULES, &status);
#endif
/*
* as of 02.11.2012, the list at http://publicsuffix.org/list/ contains ~6000 rules and 40 exceptions.
* as of 19.02.2014, the list at http://publicsuffix.org/list/ contains ~6500 rules and 19 exceptions.
*/
psl->suffixes = _vector_alloc(8*1024, _suffix_compare);
psl->suffix_exceptions = _vector_alloc(64, _suffix_compare);
while ((linep = fgets(buf, sizeof(buf), fp))) {
while (isspace(*linep)) linep++; /* ignore leading whitespace */
if (!*linep) continue; /* skip empty lines */
if (*linep == '/' && linep[1] == '/')
continue; /* skip comments */
/* parse suffix rule */
for (p = linep; *linep && !isspace(*linep);) linep++;
*linep = 0;
if (*p == '!') {
/* add to exceptions */
if (_suffix_init(&suffix, p + 1, linep - p - 1) == 0) {
suffixp = _vector_get(psl->suffix_exceptions, _vector_add(psl->suffix_exceptions, &suffix));
suffixp->label = suffixp->label_buf; /* set label to changed address */
#ifdef WITH_LIBICU
_add_punycode_if_needed(idna, psl->suffix_exceptions, suffixp);
#elif defined(WITH_LIBIDN2) || defined(WITH_LIBIDN)
_add_punycode_if_needed(psl->suffix_exceptions, suffixp);
#endif
}
} else {
/* add to suffixes */
if (_suffix_init(&suffix, p, linep - p) == 0) {
suffixp = _vector_get(psl->suffixes, _vector_add(psl->suffixes, &suffix));
suffixp->label = suffixp->label_buf; /* set label to changed address */
#ifdef WITH_LIBICU
_add_punycode_if_needed(idna, psl->suffixes, suffixp);
#elif defined(WITH_LIBIDN2) || defined(WITH_LIBIDN)
_add_punycode_if_needed(psl->suffixes, suffixp);
#endif
}
}
}
_vector_sort(psl->suffix_exceptions);
_vector_sort(psl->suffixes);
#ifdef WITH_LIBICU
if (idna)
uidna_close(idna);
#endif
return psl;
}
