- Support for UTS #46.

NEWS

@@ -39,6 +39,7 @@ PHP                                                                        NEWS
   
 - Intl:
   . Fixed memory leak in several Intl locale functions. (Felipe)
+  . Added support for UTS #46. (Gustavo)
   
 - Mbstring
   . Fixed bug #60306 (Characters lost while converting from cp936 to utf8).

ext/intl/common/common_error.c

@@ -232,7 +232,6 @@ void intl_expose_icu_error_codes( INIT_FUNC_ARGS )
 	INTL_EXPOSE_CONST( U_REGEX_ERROR_LIMIT );
 
 	/* The error code in the range 0x10400-0x104ff are reserved for IDNA related error codes */
-#if defined(U_IDNA_PROHIBITED_ERROR)
 	INTL_EXPOSE_CONST( U_IDNA_PROHIBITED_ERROR );
 	INTL_EXPOSE_CONST( U_IDNA_ERROR_START );
 	INTL_EXPOSE_CONST( U_IDNA_UNASSIGNED_ERROR );
@@ -242,8 +241,8 @@ void intl_expose_icu_error_codes( INIT_FUNC_ARGS )
 	INTL_EXPOSE_CONST( U_IDNA_VERIFICATION_ERROR );
 	INTL_EXPOSE_CONST( U_IDNA_LABEL_TOO_LONG_ERROR );
 	INTL_EXPOSE_CONST( U_IDNA_ZERO_LENGTH_LABEL_ERROR );
+	INTL_EXPOSE_CONST( U_IDNA_DOMAIN_NAME_TOO_LONG_ERROR );
 	INTL_EXPOSE_CONST( U_IDNA_ERROR_LIMIT );
-#endif
 
 	/* Aliases for StringPrep */
 	INTL_EXPOSE_CONST( U_STRINGPREP_PROHIBITED_ERROR );

ext/intl/idn/idn.c

@@ -29,14 +29,25 @@
 #include "ext/standard/php_string.h"
 
 #include "intl_error.h"
- #include "intl_convert.h"
+#include "intl_convert.h"
 /* }}} */
 
+#ifdef UIDNA_INFO_INITIALIZER
+#define HAVE_46_API 1 /* has UTS#46 API (introduced in ICU 4.6) */
+#endif
+
+enum {
+	INTL_IDN_VARIANT_2003 = 0,
+	INTL_IDN_VARIANT_UTS46
+};
+
 /* {{{ grapheme_register_constants
  * Register API constants
  */
 void idn_register_constants( INIT_FUNC_ARGS )
 {
+	/* OPTIONS */
+
 	/* Option to prohibit processing of unassigned codepoints in the input and
 	   do not check if the input conforms to STD-3 ASCII rules. */
 	REGISTER_LONG_CONSTANT("IDNA_DEFAULT", UIDNA_DEFAULT, CONST_CS | CONST_PERSISTENT);
@@ -46,6 +57,50 @@ void idn_register_constants( INIT_FUNC_ARGS )
 
 	/* Option to check if input conforms to STD-3 ASCII rules */
 	REGISTER_LONG_CONSTANT("IDNA_USE_STD3_RULES", UIDNA_USE_STD3_RULES, CONST_CS | CONST_PERSISTENT);
+
+#ifdef HAVE_46_API
+
+	/* Option to check for whether the input conforms to the BiDi rules.
+	 * Ignored by the IDNA2003 implementation. (IDNA2003 always performs a BiDi check.) */
+	REGISTER_LONG_CONSTANT("IDNA_CHECK_BIDI", UIDNA_CHECK_BIDI, CONST_CS | CONST_PERSISTENT);
+
+	/* Option to check for whether the input conforms to the CONTEXTJ rules.
+	 * Ignored by the IDNA2003 implementation. (The CONTEXTJ check is new in IDNA2008.) */
+	REGISTER_LONG_CONSTANT("IDNA_CHECK_CONTEXTJ", UIDNA_CHECK_CONTEXTJ, CONST_CS | CONST_PERSISTENT);
+
+	/* Option for nontransitional processing in ToASCII().
+	 * By default, ToASCII() uses transitional processing.
+	 * Ignored by the IDNA2003 implementation. */
+	REGISTER_LONG_CONSTANT("IDNA_NONTRANSITIONAL_TO_ASCII", UIDNA_NONTRANSITIONAL_TO_ASCII, CONST_CS | CONST_PERSISTENT);
+
+	/* Option for nontransitional processing in ToUnicode().
+	 * By default, ToUnicode() uses transitional processing.
+	 * Ignored by the IDNA2003 implementation. */
+	REGISTER_LONG_CONSTANT("IDNA_NONTRANSITIONAL_TO_UNICODE", UIDNA_NONTRANSITIONAL_TO_UNICODE, CONST_CS | CONST_PERSISTENT);
+#endif
+
+	/* VARIANTS */
+	REGISTER_LONG_CONSTANT("INTL_IDNA_VARIANT_2003", INTL_IDN_VARIANT_2003, CONST_CS | CONST_PERSISTENT);
+#ifdef HAVE_46_API
+	REGISTER_LONG_CONSTANT("INTL_IDNA_VARIANT_UTS46", INTL_IDN_VARIANT_UTS46, CONST_CS | CONST_PERSISTENT);
+#endif
+
+#ifdef HAVE_46_API
+	/* PINFO ERROR CODES */
+	REGISTER_LONG_CONSTANT("IDNA_ERROR_EMPTY_LABEL", UIDNA_ERROR_EMPTY_LABEL, CONST_CS | CONST_PERSISTENT);
+	REGISTER_LONG_CONSTANT("IDNA_ERROR_LABEL_TOO_LONG", UIDNA_ERROR_LABEL_TOO_LONG, CONST_CS | CONST_PERSISTENT);
+	REGISTER_LONG_CONSTANT("IDNA_ERROR_DOMAIN_NAME_TOO_LONG", UIDNA_ERROR_DOMAIN_NAME_TOO_LONG, CONST_CS | CONST_PERSISTENT);
+	REGISTER_LONG_CONSTANT("IDNA_ERROR_LEADING_HYPHEN", UIDNA_ERROR_LEADING_HYPHEN, CONST_CS | CONST_PERSISTENT);
+	REGISTER_LONG_CONSTANT("IDNA_ERROR_TRAILING_HYPHEN", UIDNA_ERROR_TRAILING_HYPHEN, CONST_CS | CONST_PERSISTENT);
+	REGISTER_LONG_CONSTANT("IDNA_ERROR_HYPHEN_3_4", UIDNA_ERROR_HYPHEN_3_4, CONST_CS | CONST_PERSISTENT);
+	REGISTER_LONG_CONSTANT("IDNA_ERROR_LEADING_COMBINING_MARK", UIDNA_ERROR_LEADING_COMBINING_MARK, CONST_CS | CONST_PERSISTENT);
+	REGISTER_LONG_CONSTANT("IDNA_ERROR_DISALLOWED", UIDNA_ERROR_DISALLOWED, CONST_CS | CONST_PERSISTENT);
+	REGISTER_LONG_CONSTANT("IDNA_ERROR_PUNYCODE", UIDNA_ERROR_PUNYCODE, CONST_CS | CONST_PERSISTENT);
+	REGISTER_LONG_CONSTANT("IDNA_ERROR_LABEL_HAS_DOT", UIDNA_ERROR_LABEL_HAS_DOT, CONST_CS | CONST_PERSISTENT);
+	REGISTER_LONG_CONSTANT("IDNA_ERROR_INVALID_ACE_LABEL", UIDNA_ERROR_INVALID_ACE_LABEL, CONST_CS | CONST_PERSISTENT);
+	REGISTER_LONG_CONSTANT("IDNA_ERROR_BIDI", UIDNA_ERROR_BIDI, CONST_CS | CONST_PERSISTENT);
+	REGISTER_LONG_CONSTANT("IDNA_ERROR_CONTEXTJ", UIDNA_ERROR_CONTEXTJ, CONST_CS | CONST_PERSISTENT);
+#endif
 }
 /* }}} */
 
@@ -54,11 +109,100 @@ enum {
 	INTL_IDN_TO_UTF8
 };
 
-static void php_intl_idn_to(INTERNAL_FUNCTION_PARAMETERS, int mode)
+/* like INTL_CHECK_STATUS, but as a function and varying the name of the func */
+static int php_intl_idn_check_status(UErrorCode err, const char *msg, int mode TSRMLS_DC)
+{
+	intl_error_set_code(NULL, err TSRMLS_CC);
+	if (U_FAILURE(err)) {
+		char *buff;
+		spprintf(&buff, 0, "%s: %s",
+			mode == INTL_IDN_TO_ASCII ? "idn_to_ascii" : "idn_to_utf8",
+			msg);
+		intl_error_set_custom_msg(NULL, buff, 1 TSRMLS_CC);
+		efree(buff);
+		return FAILURE;
+	}
+
+	return SUCCESS;
+}
+
+static inline void php_intl_bad_args(const char *msg, int mode TSRMLS_DC)
+{
+	php_intl_idn_check_status(U_ILLEGAL_ARGUMENT_ERROR, msg, mode TSRMLS_CC);
+}
+
+#ifdef HAVE_46_API
+static void php_intl_idn_to_46(INTERNAL_FUNCTION_PARAMETERS,
+		const char *domain, int 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 */
+	char		  *buffer = emalloc(buffer_capac);
+	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 TSRMLS_CC) == FAILURE) {
+		efree(buffer);
+		RETURN_FALSE;
+	}
+
+	if (mode == INTL_IDN_TO_ASCII) {
+		len = uidna_nameToASCII_UTF8(uts46, domain, (int32_t)domain_len,
+				buffer, buffer_capac, &info, &status);
+	} else {
+		len = uidna_nameToUnicodeUTF8(uts46, domain, (int32_t)domain_len,
+				buffer, buffer_capac, &info, &status);
+	}
+	if (php_intl_idn_check_status(status, "failed to convert name",
+			mode TSRMLS_CC) == FAILURE) {
+		uidna_close(uts46);
+		efree(buffer);
+		RETURN_FALSE;
+	}
+	if (len >= 255) {
+		php_error_docref(NULL TSRMLS_CC, E_ERROR, "ICU returned an unexpected length");
+	}
+
+	buffer[len] = '\0';
+
+	if (info.errors == 0) {
+		RETVAL_STRINGL(buffer, len, 0);
+		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"), return_value);
+		} else {
+			zval *zv;
+			ALLOC_INIT_ZVAL(zv);
+			ZVAL_STRINGL(zv, buffer, len, 0);
+			buffer_used = 1;
+			add_assoc_zval_ex(idna_info, "result", sizeof("result"), zv);
+		}
+		add_assoc_bool_ex(idna_info, "isTransitionalDifferent",
+				sizeof("isTransitionalDifferent"), info.isTransitionalDifferent);
+		add_assoc_long_ex(idna_info, "errors", sizeof("errors"), (long)info.errors);
+	}
+
+	if (!buffer_used) {
+		efree(buffer);
+	}
+
+	uidna_close(uts46);
+}
+#endif
+
+static void php_intl_idn_to(INTERNAL_FUNCTION_PARAMETERS,
+		const char *domain, int domain_len, uint32_t option, int mode)
 {
-	unsigned char* domain;
-	int domain_len;
-	long option = 0;
 	UChar* ustring = NULL;
 	int ustring_len = 0;
 	UErrorCode status;
@@ -67,18 +211,9 @@ static void php_intl_idn_to(INTERNAL_FUNCTION_PARAMETERS, int mode)
 	UChar     converted[MAXPATHLEN];
 	int32_t   converted_ret_len;
 
-	if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|l", (char **)&domain, &domain_len, &option) == FAILURE) {
-		return;
-	}
-
-	if (domain_len < 1) {
-		intl_error_set( NULL, U_ILLEGAL_ARGUMENT_ERROR, "idn_to_ascii: empty domain name", 0 TSRMLS_CC );
-		RETURN_FALSE;
-	}
-
 	/* convert the string to UTF-16. */
 	status = U_ZERO_ERROR;
-	intl_convert_utf8_to_utf16(&ustring, &ustring_len, (char*) domain, domain_len, &status );
+	intl_convert_utf8_to_utf16(&ustring, &ustring_len, domain, domain_len, &status);
 
 	if (U_FAILURE(status)) {
 		intl_error_set_code(NULL, status TSRMLS_CC);
@@ -123,11 +258,75 @@ static void php_intl_idn_to(INTERNAL_FUNCTION_PARAMETERS, int mode)
 	RETURN_STRINGL(((char *)converted_utf8), converted_utf8_len, 0);
 }
 
+static void php_intl_idn_handoff(INTERNAL_FUNCTION_PARAMETERS, int mode)
+{
+	char *domain;
+	int domain_len;
+	long option = 0,
+		 variant = INTL_IDN_VARIANT_2003;
+	zval *idna_info = NULL;
+
+	intl_error_reset(NULL TSRMLS_CC);
+
+	if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|llz",
+			&domain, &domain_len, &option, &variant, &idna_info) == FAILURE) {
+		php_intl_bad_args("bad arguments", mode TSRMLS_CC);
+		RETURN_NULL(); /* don't set FALSE because that's not the way it was before... */
+	}
+
+#ifdef HAVE_46_API
+	if (variant != INTL_IDN_VARIANT_2003 && variant != INTL_IDN_VARIANT_UTS46) {
+		php_intl_bad_args("invalid variant, must be one of {"
+			"INTL_IDNA_VARIANT_2003, INTL_IDNA_VARIANT_UTS46}", mode TSRMLS_CC);
+		RETURN_FALSE;
+	}
+#else
+	if (variant != INTL_IDN_VARIANT_2003) {
+		php_intl_bad_args("invalid variant, PHP was compiled against "
+			"an old version of ICU and only supports INTL_IDN_VARIANT_2003",
+			mode TSRMLS_CC);
+		RETURN_FALSE;
+	}
+#endif
+
+	if (domain_len < 1) {
+		php_intl_bad_args("empty domain name", mode TSRMLS_CC);
+		RETURN_FALSE;
+	}
+	if (domain_len > INT32_MAX - 1) {
+		php_intl_bad_args("domain name too large", mode TSRMLS_CC);
+		RETURN_FALSE;
+	}
+	/* don't check options; it wasn't checked before */
+
+	if (idna_info != NULL) {
+		if (variant == INTL_IDN_VARIANT_2003) {
+			php_error_docref0(NULL TSRMLS_CC, E_NOTICE,
+				"4 arguments were provided, but INTL_IDNA_VARIANT_2003 only "
+				"takes 3 - extra argument ignored");
+		} else {
+			zval_dtor(idna_info);
+			array_init(idna_info);
+		}
+	}
+	
+	if (variant == INTL_IDN_VARIANT_2003) {
+		php_intl_idn_to(INTERNAL_FUNCTION_PARAM_PASSTHRU,
+				domain, domain_len, (uint32_t)option, mode);
+	}
+#ifdef HAVE_46_API
+	else {
+		php_intl_idn_to_46(INTERNAL_FUNCTION_PARAM_PASSTHRU, domain, domain_len,
+				(uint32_t)option, mode, idna_info);
+	}
+#endif
+}
+
 /* {{{ proto int idn_to_ascii(string domain[, int options])
    Converts an Unicode domain to ASCII representation, as defined in the IDNA RFC */
 PHP_FUNCTION(idn_to_ascii)
 {
-	php_intl_idn_to(INTERNAL_FUNCTION_PARAM_PASSTHRU, INTL_IDN_TO_ASCII);
+	php_intl_idn_handoff(INTERNAL_FUNCTION_PARAM_PASSTHRU, INTL_IDN_TO_ASCII);
 }
 /* }}} */
 
@@ -136,7 +335,7 @@ PHP_FUNCTION(idn_to_ascii)
    Converts an ASCII representation of the domain to Unicode (UTF-8), as defined in the IDNA RFC */
 PHP_FUNCTION(idn_to_utf8)
 {
-	php_intl_idn_to(INTERNAL_FUNCTION_PARAM_PASSTHRU, INTL_IDN_TO_UTF8);
+	php_intl_idn_handoff(INTERNAL_FUNCTION_PARAM_PASSTHRU, INTL_IDN_TO_UTF8);
 }
 /* }}} */
 

ext/intl/php_intl.c

@@ -335,13 +335,15 @@ ZEND_END_ARG_INFO()
 ZEND_BEGIN_ARG_INFO_EX(arginfo_idn_to_ascii, 0, 0, 1)
 	ZEND_ARG_INFO(0, domain)
 	ZEND_ARG_INFO(0, option)
-	ZEND_ARG_INFO(0, status)
+	ZEND_ARG_INFO(0, variant)
+	ZEND_ARG_INFO(1, idn_info)
 ZEND_END_ARG_INFO()
 
 ZEND_BEGIN_ARG_INFO_EX(arginfo_idn_to_utf8, 0, 0, 1)
 	ZEND_ARG_INFO(0, domain)
 	ZEND_ARG_INFO(0, option)
-	ZEND_ARG_INFO(0, status)
+	ZEND_ARG_INFO(0, variant)
+	ZEND_ARG_INFO(1, idn_info)
 ZEND_END_ARG_INFO()
 
 ZEND_BEGIN_ARG_INFO_EX( arginfo_resourcebundle_create_proc, 0, 0, 2 )

ext/intl/tests/idn_uts46_basic.phpt

@@ -0,0 +1,53 @@
+--TEST--
+IDN UTS #46 API basic tests
+--SKIPIF--
+<?php
+	if (!extension_loaded('intl'))
+		die('skip');
+	if (!defined('INTL_IDNA_VARIANT_UTS46'))
+		die('skip no UTS #46 API');
+--FILE--
+<?php
+$utf8dn = "www.fußball.com";
+$asciiNonTrans = "www.xn--fuball-cta.com";
+
+echo "all ok, no details:", "\n";
+var_dump(idn_to_ascii($utf8dn, 
+	IDNA_NONTRANSITIONAL_TO_ASCII, INTL_IDNA_VARIANT_UTS46));
+	
+echo "all ok, no details, transitional:", "\n";
+var_dump(idn_to_ascii($utf8dn, 0, INTL_IDNA_VARIANT_UTS46));
+
+echo "all ok, with details:", "\n";
+var_dump(idn_to_ascii($utf8dn, IDNA_NONTRANSITIONAL_TO_ASCII,
+	INTL_IDNA_VARIANT_UTS46, $info));
+var_dump($info);
+
+echo "reverse, ok, with details:", "\n";
+var_dump(idn_to_utf8($asciiNonTrans, 0, INTL_IDNA_VARIANT_UTS46, $info));
+var_dump($info);
+--EXPECT--
+all ok, no details:
+string(22) "www.xn--fuball-cta.com"
+all ok, no details, transitional:
+string(16) "www.fussball.com"
+all ok, with details:
+string(22) "www.xn--fuball-cta.com"
+array(3) {
+  ["result"]=>
+  string(22) "www.xn--fuball-cta.com"
+  ["isTransitionalDifferent"]=>
+  bool(true)
+  ["errors"]=>
+  int(0)
+}
+reverse, ok, with details:
+string(16) "www.fußball.com"
+array(3) {
+  ["result"]=>
+  string(16) "www.fußball.com"
+  ["isTransitionalDifferent"]=>
+  bool(false)
+  ["errors"]=>
+  int(0)
+}

ext/intl/tests/idn_uts46_errors.phpt

@@ -0,0 +1,89 @@
+--TEST--
+IDN UTS #46 API error tests
+--SKIPIF--
+<?php
+	if (!extension_loaded('intl'))
+		die('skip');
+	if (!defined('INTL_IDNA_VARIANT_UTS46'))
+		die('skip no UTS #46 API');
+--FILE--
+<?php
+ini_set("intl.error_level", E_WARNING);
+echo "=> PHP level errors", "\n";
+
+echo "bad args:", "\n";
+var_dump(idn_to_ascii("", 0, array()));
+var_dump(idn_to_ascii("", 0, INTL_IDNA_VARIANT_UTS46, $foo, null));
+
+echo "bad variant:", "\n";
+var_dump(idn_to_ascii("", 0, INTL_IDNA_VARIANT_UTS46 + 10));
+
+echo "empty domain:", "\n";
+var_dump(idn_to_ascii("", 0, INTL_IDNA_VARIANT_UTS46));
+
+echo "fourth arg for 2003 variant (only notice raised):", "\n";
+var_dump(idn_to_ascii("foo.com", 0, INTL_IDNA_VARIANT_2003, $foo));
+
+echo "with error, but no details arg:", "\n";
+var_dump(idn_to_ascii("www.fußball.com-", 0, INTL_IDNA_VARIANT_UTS46));
+
+echo "with error, with details arg:", "\n";
+var_dump(idn_to_ascii("www.fußball.com-", IDNA_NONTRANSITIONAL_TO_ASCII,
+	INTL_IDNA_VARIANT_UTS46, $foo));
+var_dump($foo);
+
+echo "with error, with details arg, contextj:", "\n";
+var_dump(idn_to_ascii(
+		html_entity_decode("www.a&#x200D;b.com", 0, "UTF-8"),
+		IDNA_NONTRANSITIONAL_TO_ASCII | IDNA_CHECK_CONTEXTJ,
+		INTL_IDNA_VARIANT_UTS46, $foo));
+var_dump($foo);
+var_dump($foo["errors"]==IDNA_ERROR_CONTEXTJ);
+--EXPECTF--
+=> PHP level errors
+bad args:
+
+Warning: idn_to_ascii() expects parameter 3 to be long, array given in %s on line %d
+
+Warning: idn_to_ascii(): idn_to_ascii: bad arguments in %s on line %d
+NULL
+
+Warning: idn_to_ascii() expects at most 4 parameters, 5 given in %s on line %d
+
+Warning: idn_to_ascii(): idn_to_ascii: bad arguments in %s on line %d
+NULL
+bad variant:
+
+Warning: idn_to_ascii(): idn_to_ascii: invalid variant, must be one of {INTL_IDNA_VARIANT_2003, INTL_IDNA_VARIANT_UTS46} in %s on line %d
+bool(false)
+empty domain:
+
+Warning: idn_to_ascii(): idn_to_ascii: empty domain name in %s on line %d
+bool(false)
+fourth arg for 2003 variant (only notice raised):
+
+Notice: idn_to_ascii(): 4 arguments were provided, but INTL_IDNA_VARIANT_2003 only takes 3 - extra argument ignored in %s on line %d
+string(7) "foo.com"
+with error, but no details arg:
+bool(false)
+with error, with details arg:
+bool(false)
+array(3) {
+  ["result"]=>
+  string(23) "www.xn--fuball-cta.com-"
+  ["isTransitionalDifferent"]=>
+  bool(true)
+  ["errors"]=>
+  int(16)
+}
+with error, with details arg, contextj:
+bool(false)
+array(3) {
+  ["result"]=>
+  string(18) "www.xn--ab-m1t.com"
+  ["isTransitionalDifferent"]=>
+  bool(true)
+  ["errors"]=>
+  int(4096)
+}
+bool(true)