rrichards Wed Nov 15 12:08:12 2006 UTC
Modified files:
/php-src/ext/xml xml.c
/php-src/ext/xml/tests bug25666.phpt bug26528.phpt bug32001.phpt
bug35447.phpt xml006.phpt xml009.phpt
Log:
add unicode support so Andrei finally stops asking me about it
update tests
http://cvs.php.net/viewvc.cgi/php-src/ext/xml/xml.c?r1=1.167&r2=1.168&diff_format=u
Index: php-src/ext/xml/xml.c
diff -u php-src/ext/xml/xml.c:1.167 php-src/ext/xml/xml.c:1.168
--- php-src/ext/xml/xml.c:1.167 Sun Oct 8 13:34:24 2006
+++ php-src/ext/xml/xml.c Wed Nov 15 12:08:12 2006
@@ -18,7 +18,7 @@
+----------------------------------------------------------------------+
*/
-/* $Id: xml.c,v 1.167 2006/10/08 13:34:24 bjori Exp $ */
+/* $Id: xml.c,v 1.168 2006/11/15 12:08:12 rrichards Exp $ */
#define IS_EXT_MODULE
@@ -285,18 +285,19 @@
static zval *_xml_string_zval(const char *str)
{
zval *ret;
- int len = strlen(str);
+ TSRMLS_FETCH();
+
MAKE_STD_ZVAL(ret);
- Z_TYPE_P(ret) = IS_STRING;
- Z_STRLEN_P(ret) = len;
- Z_STRVAL_P(ret) = estrndup(str, len);
+ ZVAL_UTF8_STRING(ret, (char *)str, ZSTR_DUPLICATE);
return ret;
}
static zval *_xml_xmlchar_zval(const XML_Char *s, int len, const XML_Char
*encoding)
{
zval *ret;
+ TSRMLS_FETCH();
+
MAKE_STD_ZVAL(ret);
if (s == NULL) {
@@ -306,8 +307,14 @@
if (len == 0) {
len = _xml_xmlcharlen(s);
}
- Z_TYPE_P(ret) = IS_STRING;
- Z_STRVAL_P(ret) = xml_utf8_decode(s, len, &Z_STRLEN_P(ret), encoding);
+
+ if (UG(unicode)) {
+ ZVAL_UTF8_STRINGL(ret, (char *)s, len, ZSTR_DUPLICATE);
+ } else {
+ Z_TYPE_P(ret) = IS_STRING;
+ Z_STRVAL_P(ret) = xml_utf8_decode(s, len, &Z_STRLEN_P(ret),
encoding);
+ }
+
return ret;
}
/* }}} */
@@ -428,12 +435,12 @@
zval **method;
zval **obj;
- if (Z_TYPE_P(handler) == IS_STRING) {
- php_error_docref(NULL TSRMLS_CC, E_WARNING,
"Unable to call handler %s()", Z_STRVAL_P(handler));
+ if (Z_TYPE_P(handler) == IS_STRING || Z_TYPE_P(handler)
== IS_UNICODE) {
+ php_error_docref(NULL TSRMLS_CC, E_WARNING,
"Unable to call handler %R()", Z_TYPE_P(handler), Z_UNIVAL_P(handler));
} else if (zend_hash_index_find(Z_ARRVAL_P(handler), 0,
(void **) &obj) == SUCCESS &&
zend_hash_index_find(Z_ARRVAL_P(handler), 1, (void **) &method) == SUCCESS &&
Z_TYPE_PP(obj) == IS_OBJECT &&
- Z_TYPE_PP(method) == IS_STRING) {
+ (Z_TYPE_PP(method) == IS_STRING ||
Z_TYPE_PP(method) == IS_UNICODE)) {
php_error_docref(NULL TSRMLS_CC, E_WARNING,
"Unable to call handler %v::%R()", Z_OBJCE_PP(obj)->name, Z_TYPE_PP(method),
Z_UNIVAL_PP(method));
} else
php_error_docref(NULL TSRMLS_CC, E_WARNING,
"Unable to call handler");
@@ -562,11 +569,15 @@
char *newbuf = emalloc(len + 1);
unsigned short c;
char (*decoder)(unsigned short) = NULL;
- xml_encoding *enc = xml_get_encoding(encoding);
+ xml_encoding *enc = NULL;
*newlen = 0;
- if (enc) {
- decoder = enc->decoding_function;
+
+ if (encoding) {
+ enc = xml_get_encoding(encoding);
+ if (enc) {
+ decoder = enc->decoding_function;
+ }
}
if (decoder == NULL) {
/* If the target encoding was unknown, or no decoder function
@@ -661,7 +672,13 @@
char *newstr;
int out_len;
- newstr = xml_utf8_decode(tag, strlen(tag), &out_len,
parser->target_encoding);
+ TSRMLS_FETCH();
+
+ if (UG(unicode)) {
+ newstr = xml_utf8_decode(tag, strlen(tag), &out_len, NULL);
+ } else {
+ newstr = xml_utf8_decode(tag, strlen(tag), &out_len,
parser->target_encoding);
+ }
if (parser->case_folding) {
php_strtoupper(newstr, out_len);
@@ -682,6 +699,8 @@
zval *retval, *args[3];
if (parser) {
+ TSRMLS_FETCH();
+
parser->level++;
tag_name = _xml_decode_tag(parser, name);
@@ -694,9 +713,13 @@
while (attributes && *attributes) {
att = _xml_decode_tag(parser, attributes[0]);
- val = xml_utf8_decode(attributes[1],
strlen(attributes[1]), &val_len, parser->target_encoding);
+ if (UG(unicode)) {
+ val = xml_utf8_decode(attributes[1],
strlen(attributes[1]), &val_len, NULL);
+ } else {
+ val = xml_utf8_decode(attributes[1],
strlen(attributes[1]), &val_len, parser->target_encoding);
+ }
- add_assoc_stringl(args[2], att, val, val_len,
0);
+ add_utf8_assoc_utf8_stringl(args[2], att, val,
val_len, ZSTR_AUTOFREE);
attributes += 2;
@@ -720,9 +743,9 @@
_xml_add_to_info(parser,((char *) tag_name) +
parser->toffset);
- add_assoc_string(tag,"tag",((char *) tag_name) +
parser->toffset,1); /* cast to avoid gcc-warning */
- add_assoc_string(tag,"type","open",1);
- add_assoc_long(tag,"level",parser->level);
+ add_ascii_assoc_utf8_string(tag,"tag",((char *)
tag_name) + parser->toffset,1); /* cast to avoid gcc-warning */
+ add_ascii_assoc_ascii_string(tag,"type","open",1);
+ add_ascii_assoc_long(tag,"level",parser->level);
parser->ltags[parser->level-1] = estrdup(tag_name);
parser->lastwasopen = 1;
@@ -731,9 +754,13 @@
while (attributes && *attributes) {
att = _xml_decode_tag(parser, attributes[0]);
- val = xml_utf8_decode(attributes[1],
strlen(attributes[1]), &val_len, parser->target_encoding);
-
- add_assoc_stringl(atr,att,val,val_len,0);
+ if (UG(unicode)) {
+ val = xml_utf8_decode(attributes[1],
strlen(attributes[1]), &val_len, NULL);
+ } else {
+ val = xml_utf8_decode(attributes[1],
strlen(attributes[1]), &val_len, parser->target_encoding);
+ }
+
+ add_utf8_assoc_utf8_stringl(atr, att, val,
val_len, ZSTR_AUTOFREE);
atcnt++;
attributes += 2;
@@ -742,7 +769,7 @@
}
if (atcnt) {
-
zend_hash_add(Z_ARRVAL_P(tag),"attributes",sizeof("attributes"),&atr,sizeof(zval*),NULL);
+
zend_ascii_hash_add(Z_ARRVAL_P(tag),"attributes",sizeof("attributes"),&atr,sizeof(zval*),NULL);
} else {
zval_ptr_dtor(&atr);
}
@@ -764,6 +791,8 @@
if (parser) {
zval *retval, *args[2];
+ TSRMLS_FETCH();
+
tag_name = _xml_decode_tag(parser, name);
if (parser->endElementHandler) {
@@ -779,7 +808,7 @@
zval *tag;
if (parser->lastwasopen) {
-
add_assoc_string(*(parser->ctag),"type","complete",1);
+
add_ascii_assoc_ascii_string(*(parser->ctag),"type","complete",1);
} else {
MAKE_STD_ZVAL(tag);
@@ -787,9 +816,9 @@
_xml_add_to_info(parser,((char *) tag_name) +
parser->toffset);
- add_assoc_string(tag,"tag",((char *) tag_name)
+ parser->toffset,1); /* cast to avoid gcc-warning */
- add_assoc_string(tag,"type","close",1);
- add_assoc_long(tag,"level",parser->level);
+ add_ascii_assoc_utf8_string(tag,"tag",((char *)
tag_name) + parser->toffset,1); /* cast to avoid gcc-warning */
+
add_ascii_assoc_ascii_string(tag,"type","close",1);
+ add_ascii_assoc_long(tag,"level",parser->level);
zend_hash_next_index_insert(Z_ARRVAL_P(parser->data),&tag,sizeof(zval*),NULL);
}
@@ -830,8 +859,14 @@
char *decoded_value;
int decoded_len;
+
+ TSRMLS_FETCH();
- decoded_value =
xml_utf8_decode(s,len,&decoded_len,parser->target_encoding);
+ if (UG(unicode)) {
+ decoded_value =
xml_utf8_decode(s,len,&decoded_len,NULL);
+ } else {
+ decoded_value =
xml_utf8_decode(s,len,&decoded_len,parser->target_encoding);
+ }
for (i = 0; i < decoded_len; i++) {
switch (decoded_value[i]) {
case ' ':
@@ -851,14 +886,28 @@
zval **myval;
/* check if the current tag already has
a value - if yes append to that! */
- if
(zend_hash_find(Z_ARRVAL_PP(parser->ctag),"value",sizeof("value"),(void **)
&myval) == SUCCESS) {
- int newlen = Z_STRLEN_PP(myval)
+ decoded_len;
- Z_STRVAL_PP(myval) =
erealloc(Z_STRVAL_PP(myval),newlen+1);
- strcpy(Z_STRVAL_PP(myval) +
Z_STRLEN_PP(myval),decoded_value);
- Z_STRLEN_PP(myval) +=
decoded_len;
+ if
(zend_ascii_hash_find(Z_ARRVAL_PP(parser->ctag),"value",sizeof("value"),(void
**) &myval) == SUCCESS) {
+ if (Z_TYPE_PP(myval) ==
IS_UNICODE) {
+ UErrorCode status =
U_ZERO_ERROR;
+ UChar *u_str;
+ int u_len, newlen;
+
+
zend_string_to_unicode_ex(UG(utf8_conv), &u_str, &u_len, decoded_value,
decoded_len, &status);
+
+ newlen =
Z_USTRLEN_PP(myval) + u_len;
+ Z_USTRVAL_PP(myval) =
eurealloc(Z_USTRVAL_PP(myval),newlen+1);
+
u_strcpy(Z_USTRVAL_PP(myval) + Z_USTRLEN_PP(myval),u_str);
+ Z_USTRLEN_PP(myval) +=
u_len;
+ efree(u_str);
+ } else {
+ int newlen =
Z_STRLEN_PP(myval) + decoded_len;
+ Z_STRVAL_PP(myval) =
erealloc(Z_STRVAL_PP(myval),newlen+1);
+
strcpy(Z_STRVAL_PP(myval) + Z_STRLEN_PP(myval),decoded_value);
+ Z_STRLEN_PP(myval) +=
decoded_len;
+ }
efree(decoded_value);
} else {
-
add_assoc_string(*(parser->ctag),"value",decoded_value,0);
+
add_ascii_assoc_utf8_string(*(parser->ctag),"value",decoded_value,ZSTR_AUTOFREE);
}
} else {
@@ -869,13 +918,28 @@
zend_hash_internal_pointer_end_ex(Z_ARRVAL_P(parser->data), &hpos);
if (hpos &&
(zend_hash_get_current_data_ex(Z_ARRVAL_P(parser->data), (void **) &curtag,
&hpos) == SUCCESS)) {
- if
(zend_hash_find(Z_ARRVAL_PP(curtag),"type",sizeof("type"),(void **) &mytype) ==
SUCCESS) {
- if
(!strcmp(Z_STRVAL_PP(mytype), "cdata")) {
- if
(zend_hash_find(Z_ARRVAL_PP(curtag),"value",sizeof("value"),(void **) &myval)
== SUCCESS) {
- int
newlen = Z_STRLEN_PP(myval) + decoded_len;
-
Z_STRVAL_PP(myval) = erealloc(Z_STRVAL_PP(myval),newlen+1);
-
strcpy(Z_STRVAL_PP(myval) + Z_STRLEN_PP(myval),decoded_value);
-
Z_STRLEN_PP(myval) += decoded_len;
+ if
(zend_ascii_hash_find(Z_ARRVAL_PP(curtag),"type",sizeof("type"),(void **)
&mytype) == SUCCESS) {
+ if ((Z_TYPE_PP(mytype)
== IS_STRING && !strcmp(Z_STRVAL_PP(mytype), "cdata")) ||
+
(Z_TYPE_PP(mytype) == IS_UNICODE &&
!zend_cmp_unicode_and_literal(Z_USTRVAL_PP(mytype), Z_USTRLEN_PP(mytype),
"cdata", 5))) {
+ if
(zend_ascii_hash_find(Z_ARRVAL_PP(curtag),"value",sizeof("value"),(void **)
&myval) == SUCCESS) {
+ if
(Z_TYPE_PP(myval) == IS_UNICODE) {
+
UErrorCode status = U_ZERO_ERROR;
+
UChar *u_str;
+
int u_len, newlen;
+
+
zend_string_to_unicode_ex(UG(utf8_conv), &u_str, &u_len, decoded_value,
decoded_len, &status);
+
+
newlen = Z_USTRLEN_PP(myval) + u_len;
+
Z_USTRVAL_PP(myval) = eurealloc(Z_USTRVAL_PP(myval),newlen+1);
+
u_strcpy(Z_USTRVAL_PP(myval) + Z_USTRLEN_PP(myval),u_str);
+
Z_USTRLEN_PP(myval) += u_len;
+
efree(u_str);
+ } else {
+
int newlen = Z_STRLEN_PP(myval) + decoded_len;
+
Z_STRVAL_PP(myval) = erealloc(Z_STRVAL_PP(myval),newlen+1);
+
strcpy(Z_STRVAL_PP(myval) + Z_STRLEN_PP(myval),decoded_value);
+
Z_STRLEN_PP(myval) += decoded_len;
+ }
efree(decoded_value);
return;
}
@@ -889,10 +953,10 @@
_xml_add_to_info(parser,parser->ltags[parser->level-1] + parser->toffset);
-
add_assoc_string(tag,"tag",parser->ltags[parser->level-1] + parser->toffset,1);
-
add_assoc_string(tag,"value",decoded_value,0);
- add_assoc_string(tag,"type","cdata",1);
-
add_assoc_long(tag,"level",parser->level);
+
add_ascii_assoc_utf8_string(tag,"tag",parser->ltags[parser->level-1] +
parser->toffset,1);
+
add_ascii_assoc_utf8_string(tag,"value",decoded_value,ZSTR_AUTOFREE);
+
add_ascii_assoc_ascii_string(tag,"type","cdata",1);
+
add_ascii_assoc_long(tag,"level",parser->level);
zend_hash_next_index_insert(Z_ARRVAL_P(parser->data),&tag,sizeof(zval*),NULL);
}
@@ -1129,7 +1193,7 @@
}
/* }}} */
-/* {{{ proto int xml_set_object(resource parser, object &obj)
+/* {{{ proto int xml_set_object(resource parser, object &obj) U
Set up object which should be used for callbacks */
PHP_FUNCTION(xml_set_object)
{
@@ -1408,7 +1472,7 @@
}
/* }}} */
-/* {{{ proto int xml_get_error_code(resource parser)
+/* {{{ proto int xml_get_error_code(resource parser) U
Get XML parser error code */
PHP_FUNCTION(xml_get_error_code)
{
@@ -1424,7 +1488,7 @@
}
/* }}} */
-/* {{{ proto string xml_error_string(int code)
+/* {{{ proto string xml_error_string(int code) U
Get XML parser error string */
PHP_FUNCTION(xml_error_string)
{
@@ -1442,7 +1506,7 @@
}
/* }}} */
-/* {{{ proto int xml_get_current_line_number(resource parser)
+/* {{{ proto int xml_get_current_line_number(resource parser) U
Get current line number for an XML parser */
PHP_FUNCTION(xml_get_current_line_number)
{
@@ -1458,7 +1522,7 @@
}
/* }}} */
-/* {{{ proto int xml_get_current_column_number(resource parser)
+/* {{{ proto int xml_get_current_column_number(resource parser) U
Get current column number for an XML parser */
PHP_FUNCTION(xml_get_current_column_number)
{
@@ -1474,7 +1538,7 @@
}
/* }}} */
-/* {{{ proto int xml_get_current_byte_index(resource parser)
+/* {{{ proto int xml_get_current_byte_index(resource parser) U
Get current byte index for an XML parser */
PHP_FUNCTION(xml_get_current_byte_index)
{
@@ -1490,7 +1554,7 @@
}
/* }}} */
-/* {{{ proto int xml_parser_free(resource parser)
+/* {{{ proto int xml_parser_free(resource parser) U
Free an XML parser */
PHP_FUNCTION(xml_parser_free)
{
@@ -1563,7 +1627,7 @@
}
/* }}} */
-/* {{{ proto int xml_parser_get_option(resource parser, int option)
+/* {{{ proto int xml_parser_get_option(resource parser, int option) U
Get options from an XML parser */
PHP_FUNCTION(xml_parser_get_option)
{
@@ -1582,7 +1646,7 @@
RETURN_LONG(parser->case_folding);
break;
case PHP_XML_OPTION_TARGET_ENCODING:
- RETURN_STRING(parser->target_encoding, 1);
+ RETURN_ASCII_STRING(parser->target_encoding, 1);
break;
default:
php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown
option");
@@ -1598,18 +1662,30 @@
Encodes an ISO-8859-1 string to UTF-8 */
PHP_FUNCTION(utf8_encode)
{
- zval **arg;
XML_Char *encoded;
- int len;
+ zstr data;
+ int len, data_len;
+ zend_uchar data_type;
- if (ZEND_NUM_ARGS() != 1 || zend_get_parameters_ex(1, &arg) == FAILURE)
{
- WRONG_PARAM_COUNT;
- }
- convert_to_string_ex(arg);
- encoded = xml_utf8_encode(Z_STRVAL_PP(arg), Z_STRLEN_PP(arg), &len,
"ISO-8859-1");
- if (encoded == NULL) {
+ if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "t", &data,
&data_len, &data_type) == FAILURE) {
RETURN_FALSE;
}
+
+ if (data_type == IS_UNICODE) {
+ UErrorCode status = U_ZERO_ERROR;
+
+ zend_unicode_to_string_ex(UG(utf8_conv), &encoded, &len,
data.u, data_len, &status);
+ if (U_FAILURE(status)) {
+ efree(encoded);
+ RETURN_FALSE;
+ }
+ } else {
+ encoded = xml_utf8_encode(data.s, data_len, &len, "ISO-8859-1");
+ if (encoded == NULL) {
+ RETURN_FALSE;
+ }
+ }
+
RETVAL_STRINGL(encoded, len, 0);
}
/* }}} */
@@ -1618,19 +1694,25 @@
Converts a UTF-8 encoded string to ISO-8859-1 */
PHP_FUNCTION(utf8_decode)
{
- zval **arg;
XML_Char *decoded;
- int len;
+ char *data;
+ int len, data_len;
+ zend_uchar data_type;
- if (ZEND_NUM_ARGS() != 1 || zend_get_parameters_ex(1, &arg) == FAILURE)
{
- WRONG_PARAM_COUNT;
+ if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "t", &data,
&data_len, &data_type) == FAILURE) {
+ RETURN_FALSE;
}
- convert_to_string_ex(arg);
- decoded = xml_utf8_decode(Z_STRVAL_PP(arg), Z_STRLEN_PP(arg), &len,
"ISO-8859-1");
+
+ if (UG(unicode)) {
+ decoded = xml_utf8_decode(data, data_len, &len, NULL);
+ } else {
+ decoded = xml_utf8_decode(data, data_len, &len, "ISO-8859-1");
+ }
+
if (decoded == NULL) {
RETURN_FALSE;
}
- RETVAL_STRINGL(decoded, len, 0);
+ RETVAL_UTF8_STRINGL(decoded, len, ZSTR_AUTOFREE);
}
/* }}} */
http://cvs.php.net/viewvc.cgi/php-src/ext/xml/tests/bug25666.phpt?r1=1.4&r2=1.5&diff_format=u
Index: php-src/ext/xml/tests/bug25666.phpt
diff -u php-src/ext/xml/tests/bug25666.phpt:1.4
php-src/ext/xml/tests/bug25666.phpt:1.5
--- php-src/ext/xml/tests/bug25666.phpt:1.4 Wed Dec 21 10:50:10 2005
+++ php-src/ext/xml/tests/bug25666.phpt Wed Nov 15 12:08:12 2006
@@ -33,3 +33,7 @@
string(24) "http://example.com/[EMAIL PROTECTED]"
string(24) "http://example.com/[EMAIL PROTECTED]"
string(24) "http://example.com/[EMAIL PROTECTED]"
+--UEXPECT--
+unicode(24) "http://example.com/[EMAIL PROTECTED]"
+unicode(24) "http://example.com/[EMAIL PROTECTED]"
+unicode(24) "http://example.com/[EMAIL PROTECTED]"
http://cvs.php.net/viewvc.cgi/php-src/ext/xml/tests/bug26528.phpt?r1=1.4&r2=1.5&diff_format=u
Index: php-src/ext/xml/tests/bug26528.phpt
diff -u php-src/ext/xml/tests/bug26528.phpt:1.4
php-src/ext/xml/tests/bug26528.phpt:1.5
--- php-src/ext/xml/tests/bug26528.phpt:1.4 Tue Sep 19 11:25:07 2006
+++ php-src/ext/xml/tests/bug26528.phpt Wed Nov 15 12:08:12 2006
@@ -29,3 +29,20 @@
}
}
}
+--UEXPECT--
+array(1) {
+ [0]=>
+ array(4) {
+ [u"tag"]=>
+ unicode(4) "TEST"
+ [u"type"]=>
+ unicode(8) "complete"
+ [u"level"]=>
+ int(1)
+ [u"attributes"]=>
+ array(1) {
+ [u"ATTR"]=>
+ unicode(13) "angle<bracket"
+ }
+ }
+}
http://cvs.php.net/viewvc.cgi/php-src/ext/xml/tests/bug32001.phpt?r1=1.10&r2=1.11&diff_format=u
Index: php-src/ext/xml/tests/bug32001.phpt
diff -u php-src/ext/xml/tests/bug32001.phpt:1.10
php-src/ext/xml/tests/bug32001.phpt:1.11
--- php-src/ext/xml/tests/bug32001.phpt:1.10 Fri Dec 23 20:03:31 2005
+++ php-src/ext/xml/tests/bug32001.phpt Wed Nov 15 12:08:12 2006
@@ -23,8 +23,16 @@
}
function start_element($parser, $name, $attrs) {
- $attrs = array_map('bin2hex', $attrs);
- $this->tags[] = bin2hex($name).": ".implode(', ', $attrs);
+ if (is_unicode($name)) {
+ $name = utf8_encode($name);
+ $binatts = array();
+ foreach ($attrs AS $att) {
+ $binatts[] = (binary)bin2hex(utf8_encode($att));
+ }
+ } else {
+ $binatts = array_map('bin2hex', $attrs);
+ }
+ $this->tags[] = (binary)bin2hex($name).(binary)":
".(binary)implode(', ', $binatts);
}
function end_element($parser, $name) {
@@ -35,10 +43,10 @@
if ($this->prologue) {
$canonical_name = preg_replace('/BE|LE/i', '',
$this->encoding);
- $data .= "<?xml version=\"1.0\"
encoding=\"$canonical_name\" ?>\n";
+ $data .= b"<?xml version=\"1.0\"
encoding=\"$canonical_name\" ?>\n";
}
- $data .= <<<HERE
+ $data .= b<<<HERE
<ãã¹ã:ãã¹ã1 xmlns:ãã¹ã="http://www.example.com/ãã¹ã/";
ãã¹ã="ãã¹ã">
<ãã¹ã:ãã¹ã2 ãã¹ã="ãã¹ã">
<ãã¹ã:ãã¹ã3>
@@ -54,7 +62,7 @@
switch (strtoupper($this->encoding)) {
case 'UTF-8':
case 'UTF8':
- $data = "\xef\xbb\xbf".$data;
+ $data = b"\xef\xbb\xbf".$data;
break;
case 'UTF-16':
@@ -65,14 +73,14 @@
case 'UCS2':
case 'UCS-2BE':
case 'UCS2BE':
- $data = "\xfe\xff".$data;
+ $data = b"\xfe\xff".$data;
break;
case 'UTF-16LE':
case 'UTF16LE':
case 'UCS-2LE':
case 'UCS2LE':
- $data = "\xff\xfe".$data;
+ $data = b"\xff\xfe".$data;
break;
case 'UTF-32':
@@ -83,14 +91,14 @@
case 'UCS4':
case 'UCS-4BE':
case 'UCS4BE':
- $data = "\x00\x00\xfe\xff".$data;
+ $data = b"\x00\x00\xfe\xff".$data;
break;
case 'UTF-32LE':
case 'UTF32LE':
case 'UCS-4LE':
case 'UCS4LE':
- $data = "\xff\xfe\x00\x00".$data;
+ $data = b"\xff\xfe\x00\x00".$data;
break;
}
}
http://cvs.php.net/viewvc.cgi/php-src/ext/xml/tests/bug35447.phpt?r1=1.4&r2=1.5&diff_format=u
Index: php-src/ext/xml/tests/bug35447.phpt
diff -u php-src/ext/xml/tests/bug35447.phpt:1.4
php-src/ext/xml/tests/bug35447.phpt:1.5
--- php-src/ext/xml/tests/bug35447.phpt:1.4 Tue Sep 19 11:25:07 2006
+++ php-src/ext/xml/tests/bug35447.phpt Wed Nov 15 12:08:12 2006
@@ -7,7 +7,7 @@
?>
--FILE--
<?php
-$data = <<<END_OF_XML
+$data = b<<<END_OF_XML
\xEF\xBB\xBF<?xml version="1.0" encoding="utf-8"?\x3e
<!DOCTYPE bundle [
<!ELEMENT bundle (resource)+>
@@ -47,3 +47,24 @@
string(13) "A bient&244;t"
}
}
+--UEXPECT--
+array(1) {
+ [0]=>
+ array(5) {
+ [u"tag"]=>
+ unicode(8) "resource"
+ [u"type"]=>
+ unicode(8) "complete"
+ [u"level"]=>
+ int(1)
+ [u"attributes"]=>
+ array(2) {
+ [u"key"]=>
+ unicode(7) "rSeeYou"
+ [u"type"]=>
+ unicode(7) "literal"
+ }
+ [u"value"]=>
+ unicode(13) "A bient&244;t"
+ }
+}
http://cvs.php.net/viewvc.cgi/php-src/ext/xml/tests/xml006.phpt?r1=1.2&r2=1.3&diff_format=u
Index: php-src/ext/xml/tests/xml006.phpt
diff -u php-src/ext/xml/tests/xml006.phpt:1.2
php-src/ext/xml/tests/xml006.phpt:1.3
--- php-src/ext/xml/tests/xml006.phpt:1.2 Wed Aug 17 14:01:06 2005
+++ php-src/ext/xml/tests/xml006.phpt Wed Nov 15 12:08:12 2006
@@ -7,8 +7,8 @@
unicode.output_encoding=ISO-8859-1
--FILE--
<?php
-printf("%s -> %s\n", urlencode("æ"), urlencode(utf8_encode("æ")));
-printf("%s <- %s\n", urlencode(utf8_decode(urldecode("%C3%A6"))), "%C3%A6");
+printf("%s -> %s\n", urlencode((binary)"æ"), urlencode(utf8_encode("æ")));
+printf("%s <- %s\n",
urlencode((binary)utf8_decode(urldecode((binary)"%C3%A6"))), (binary)"%C3%A6");
?>
--EXPECT--
%E6 -> %C3%A6
http://cvs.php.net/viewvc.cgi/php-src/ext/xml/tests/xml009.phpt?r1=1.4&r2=1.5&diff_format=u
Index: php-src/ext/xml/tests/xml009.phpt
diff -u php-src/ext/xml/tests/xml009.phpt:1.4
php-src/ext/xml/tests/xml009.phpt:1.5
--- php-src/ext/xml/tests/xml009.phpt:1.4 Wed Dec 21 10:50:10 2005
+++ php-src/ext/xml/tests/xml009.phpt Wed Nov 15 12:08:12 2006
@@ -33,3 +33,7 @@
string(24) "http://example.com/[EMAIL PROTECTED]"
string(24) "http://example.com/[EMAIL PROTECTED]"
string(24) "http://example.com/[EMAIL PROTECTED]"
+--UEXPECT--
+unicode(24) "http://example.com/[EMAIL PROTECTED]"
+unicode(24) "http://example.com/[EMAIL PROTECTED]"
+unicode(24) "http://example.com/[EMAIL PROTECTED]"
--
PHP CVS Mailing List (http://www.php.net/)
To unsubscribe, visit: http://www.php.net/unsub.php
