andrei Wed Jun 21 20:17:21 2006 UTC
Modified files:
/ZendEngine2 zend_constants.c zend_unicode.c zend_unicode.h
/php-src/ext/unicode unicode.c
/php-src/main/streams filter.c streams.c
Log:
Implement user conversion error handler support. Works as normal error
handler, in that it can return false to make the default one take over.
Handler signature is:
user_handler($direction, $encoding, $char_byte, $offset, $message)
Also removed support for using exceptions in default error handler.
http://cvs.php.net/viewvc.cgi/ZendEngine2/zend_constants.c?r1=1.86&r2=1.87&diff_format=u
Index: ZendEngine2/zend_constants.c
diff -u ZendEngine2/zend_constants.c:1.86 ZendEngine2/zend_constants.c:1.87
--- ZendEngine2/zend_constants.c:1.86 Sun Mar 26 01:48:33 2006
+++ ZendEngine2/zend_constants.c Wed Jun 21 20:17:21 2006
@@ -17,7 +17,7 @@
+----------------------------------------------------------------------+
*/
-/* $Id: zend_constants.c,v 1.86 2006/03/26 01:48:33 andrei Exp $ */
+/* $Id: zend_constants.c,v 1.87 2006/06/21 20:17:21 andrei Exp $ */
#include "zend.h"
#include "zend_constants.h"
@@ -124,7 +124,6 @@
REGISTER_MAIN_LONG_CONSTANT("U_CONV_ERROR_ESCAPE_JAVA",
ZEND_CONV_ERROR_ESCAPE_JAVA, CONST_PERSISTENT | CONST_CS);
REGISTER_MAIN_LONG_CONSTANT("U_CONV_ERROR_ESCAPE_XML_DEC",
ZEND_CONV_ERROR_ESCAPE_XML_DEC, CONST_PERSISTENT | CONST_CS);
REGISTER_MAIN_LONG_CONSTANT("U_CONV_ERROR_ESCAPE_XML_HEX",
ZEND_CONV_ERROR_ESCAPE_XML_HEX, CONST_PERSISTENT | CONST_CS);
- REGISTER_MAIN_LONG_CONSTANT("U_CONV_ERROR_EXCEPTION",
ZEND_CONV_ERROR_EXCEPTION, CONST_PERSISTENT | CONST_CS);
REGISTER_MAIN_LONG_CONSTANT("FROM_UNICODE", ZEND_FROM_UNICODE,
CONST_PERSISTENT | CONST_CS);
REGISTER_MAIN_LONG_CONSTANT("TO_UNICODE", ZEND_TO_UNICODE,
CONST_PERSISTENT | CONST_CS);
http://cvs.php.net/viewvc.cgi/ZendEngine2/zend_unicode.c?r1=1.22&r2=1.23&diff_format=u
Index: ZendEngine2/zend_unicode.c
diff -u ZendEngine2/zend_unicode.c:1.22 ZendEngine2/zend_unicode.c:1.23
--- ZendEngine2/zend_unicode.c:1.22 Thu May 4 21:22:17 2006
+++ ZendEngine2/zend_unicode.c Wed Jun 21 20:17:21 2006
@@ -390,8 +390,8 @@
}
/* }}} */
-/* {{{ zend_raise_conversion_error_ex */
-ZEND_API void zend_raise_conversion_error_ex(char *message, UConverter *conv,
zend_conv_direction dir, int error_char_offset, int use_exception TSRMLS_DC)
+/* {{{ zend_default_conversion_error_handler */
+static void zend_default_conversion_error_handler(char *message, UConverter
*conv, zend_conv_direction dir, int error_char_offset TSRMLS_DC)
{
const char *conv_name;
UErrorCode status = U_ZERO_ERROR;
@@ -399,15 +399,6 @@
if (!message)
return;
- if (!conv) {
- if (use_exception) {
- zend_throw_exception_ex(unicodeConversionException, 0
TSRMLS_CC, "%s", message);
- } else {
- zend_error(E_WARNING, "%s", message);
- }
- return;
- }
-
conv_name = ucnv_getName(conv, &status);
/*
* UTODO
@@ -426,11 +417,7 @@
ucnv_getInvalidUChars(conv, err_char, &err_char_len, &status);
codepoint = (err_char_len < 2) ? err_char[0] :
U16_GET_SUPPLEMENTARY(err_char[0], err_char[1]);
- if (use_exception) {
- zend_throw_exception_ex(unicodeConversionException, 0
TSRMLS_CC, message_fmt, message, conv_name?conv_name:"<unknown>", codepoint,
error_char_offset-1);
- } else {
- zend_error(E_WARNING, message_fmt, message,
conv_name?conv_name:"", codepoint, error_char_offset-1);
- }
+ zend_error(E_WARNING, message_fmt, message,
conv_name?conv_name:"", codepoint, error_char_offset-1);
} else {
char err_char[8]; /* UTF-8 uses up to 8 bytes */
char buf[32]; /* 4x number of error bytes */
@@ -446,11 +433,106 @@
p += 5;
}
- if (use_exception) {
- zend_throw_exception_ex(unicodeConversionException, 0
TSRMLS_CC, message_fmt, message, conv_name?conv_name:"<unknown>", buf,
error_char_offset-err_char_len);
- } else {
- zend_error(E_WARNING, message_fmt, message,
conv_name?conv_name:"", buf, error_char_offset-err_char_len);
+ zend_error(E_WARNING, message_fmt, message,
conv_name?conv_name:"", buf, error_char_offset-err_char_len);
+ }
+}
+/* }}} */
+
+/* {{{ zend_call_conversion_error_handler */
+static void zend_call_conversion_error_handler(char *message, UConverter
*conv, zend_conv_direction dir, int error_char_offset TSRMLS_DC)
+{
+ zval *z_message, *z_dir, *z_encoding, *z_char, *z_offset;
+ zval ***params;
+ zval *retval;
+ zval *orig_user_error_handler;
+ const char *conv_name;
+ UErrorCode status = U_ZERO_ERROR;
+
+ ALLOC_INIT_ZVAL(z_message);
+ ALLOC_INIT_ZVAL(z_dir);
+ ALLOC_INIT_ZVAL(z_encoding);
+ ALLOC_INIT_ZVAL(z_char);
+ ALLOC_INIT_ZVAL(z_offset);
+
+ if (message) {
+ ZVAL_STRING(z_message, message, 1);
+ } else {
+ ZVAL_NULL(z_message);
+ }
+
+ ZVAL_LONG(z_dir, dir);
+
+ conv_name = ucnv_getName(conv, &status);
+ /*
+ * UTODO
+ * use some other standard than MIME? or fallback onto IANA? or use
+ * internal converter name? ponder
+ * maybe pass Converter object, when it's implemented?
+ */
+ conv_name = ucnv_getStandardName(conv_name, "MIME", &status);
+ ZVAL_STRING(z_encoding, (char *) conv_name, 1);
+
+ if (dir == ZEND_FROM_UNICODE) {
+ UChar err_char[U16_MAX_LENGTH];
+ int8_t err_char_len = sizeof(err_char);
+
+ ucnv_getInvalidUChars(conv, err_char, &err_char_len, &status);
+ ZVAL_UNICODEL(z_char, err_char, err_char_len, 1);
+ ZVAL_LONG(z_offset, error_char_offset-1);
+ } else {
+ char err_char[8]; /* UTF-8 uses up to 8 bytes */
+ int8_t err_char_len = sizeof(err_char);
+
+ ucnv_getInvalidChars(conv, err_char, &err_char_len, &status);
+ ZVAL_STRINGL(z_char, err_char, err_char_len, 1);
+ ZVAL_LONG(z_offset, error_char_offset-err_char_len);
+ }
+
+ params = (zval ***) emalloc(sizeof(zval **) * 6);
+ params[0] = &z_dir;
+ params[1] = &z_encoding;
+ params[2] = &z_char;
+ params[3] = &z_offset;
+ params[4] = &z_message;
+
+ orig_user_error_handler = UG(conv_error_handler);
+ UG(conv_error_handler) = NULL;
+
+ if (call_user_function_ex(EG(function_table), NULL,
orig_user_error_handler, &retval, 5, params, 1, NULL TSRMLS_CC)==SUCCESS) {
+ if (retval) {
+ /* user error handler returned 'false', use built-in
error handler */
+ if (Z_TYPE_P(retval) == IS_BOOL && Z_LVAL_P(retval) ==
0) {
+ zend_default_conversion_error_handler(message,
conv, dir, error_char_offset TSRMLS_CC);
+ }
+ zval_ptr_dtor(&retval);
}
+ } else if (!EG(exception)) {
+ /* The user error handler failed, use built-in error handler */
+ zend_default_conversion_error_handler(message, conv, dir,
error_char_offset TSRMLS_CC);
+ }
+
+ if (!UG(conv_error_handler)) {
+ UG(conv_error_handler) = orig_user_error_handler;
+ } else {
+ zval_ptr_dtor(&orig_user_error_handler);
+ }
+
+ efree(params);
+ zval_ptr_dtor(&z_dir);
+ zval_ptr_dtor(&z_encoding);
+ zval_ptr_dtor(&z_char);
+ zval_ptr_dtor(&z_offset);
+ zval_ptr_dtor(&z_message);
+}
+/* }}} */
+
+/* {{{ zend_raise_conversion_error_ex */
+ZEND_API void zend_raise_conversion_error_ex(char *message, UConverter *conv,
zend_conv_direction dir, int error_char_offset TSRMLS_DC)
+{
+ if (UG(conv_error_handler)) {
+ zend_call_conversion_error_handler(message, conv, dir,
error_char_offset TSRMLS_CC);
+ } else {
+ zend_default_conversion_error_handler(message, conv, dir,
error_char_offset TSRMLS_CC);
}
}
/* }}} */
@@ -471,7 +553,7 @@
if (U_FAILURE(status)) {
int32_t offset = u_countChar32(u, num_conv);
- zend_raise_conversion_error_ex("Could not convert Unicode
string to binary string", conv, ZEND_FROM_UNICODE, offset, (UG(from_error_mode)
& ZEND_CONV_ERROR_EXCEPTION) TSRMLS_CC);
+ zend_raise_conversion_error_ex("Could not convert Unicode
string to binary string", conv, ZEND_FROM_UNICODE, offset TSRMLS_CC);
if (s) {
efree(s);
}
@@ -500,7 +582,7 @@
num_conv = zend_convert_to_unicode(conv, &u, &u_len, s, s_len, &status);
if (U_FAILURE(status)) {
- zend_raise_conversion_error_ex("Could not convert binary string
to Unicode string", conv, ZEND_TO_UNICODE, num_conv, (UG(to_error_mode) &
ZEND_CONV_ERROR_EXCEPTION) TSRMLS_CC);
+ zend_raise_conversion_error_ex("Could not convert binary string
to Unicode string", conv, ZEND_TO_UNICODE, num_conv TSRMLS_CC);
if (u) {
efree(u);
}
http://cvs.php.net/viewvc.cgi/ZendEngine2/zend_unicode.h?r1=1.16&r2=1.17&diff_format=u
Index: ZendEngine2/zend_unicode.h
diff -u ZendEngine2/zend_unicode.h:1.16 ZendEngine2/zend_unicode.h:1.17
--- ZendEngine2/zend_unicode.h:1.16 Thu May 4 21:22:17 2006
+++ ZendEngine2/zend_unicode.h Wed Jun 21 20:17:21 2006
@@ -87,9 +87,7 @@
ZEND_API int zend_is_valid_identifier(UChar *ident, int ident_len);
ZEND_API int zend_normalize_identifier(UChar **dest, int *dest_len, UChar
*ident, int ident_len, zend_bool fold_case);
-#define zend_raise_conversion_error(message, exception) \
- zend_raise_conversion_error_ex(message, NULL, 0, 0, exception TSRMLS_CC)
-ZEND_API void zend_raise_conversion_error_ex(char *message, UConverter *conv,
zend_conv_direction dir, int error_char_offset, int use_exception TSRMLS_DC);
+ZEND_API void zend_raise_conversion_error_ex(char *message, UConverter *conv,
zend_conv_direction dir, int error_char_offset TSRMLS_DC);
/*
* Function to get a codepoint at position n. Iterates over codepoints
starting from the
http://cvs.php.net/viewvc.cgi/php-src/ext/unicode/unicode.c?r1=1.37&r2=1.38&diff_format=u
Index: php-src/ext/unicode/unicode.c
diff -u php-src/ext/unicode/unicode.c:1.37 php-src/ext/unicode/unicode.c:1.38
--- php-src/ext/unicode/unicode.c:1.37 Tue Jun 20 23:00:02 2006
+++ php-src/ext/unicode/unicode.c Wed Jun 21 20:17:21 2006
@@ -15,7 +15,7 @@
+----------------------------------------------------------------------+
*/
-/* $Id: unicode.c,v 1.37 2006/06/20 23:00:02 andrei Exp $ */
+/* $Id: unicode.c,v 1.38 2006/06/21 20:17:21 andrei Exp $ */
#include "php_unicode.h"
#include "zend_unicode.h"
@@ -62,7 +62,7 @@
status = U_ZERO_ERROR;
num_conv = zend_convert_to_unicode(conv, &dest, &dest_len, str,
str_len, &status);
if (U_FAILURE(status)) {
- zend_raise_conversion_error_ex("could not decode binary
string", conv, ZEND_TO_UNICODE, num_conv, (flags & ZEND_CONV_ERROR_EXCEPTION)
TSRMLS_CC);
+ zend_raise_conversion_error_ex("could not decode binary
string", conv, ZEND_TO_UNICODE, num_conv TSRMLS_CC);
efree(dest);
ucnv_close(conv);
RETURN_FALSE;
@@ -114,7 +114,7 @@
num_conv = zend_convert_from_unicode(conv, &dest, &dest_len, uni,
uni_len, &status);
if (U_FAILURE(status)) {
int32_t offset = u_countChar32(uni, num_conv);
- zend_raise_conversion_error_ex("could not encode Unicode
string", conv, ZEND_FROM_UNICODE, offset, (flags & ZEND_CONV_ERROR_EXCEPTION)
TSRMLS_CC);
+ zend_raise_conversion_error_ex("could not encode Unicode
string", conv, ZEND_FROM_UNICODE, offset TSRMLS_CC);
efree(dest);
ucnv_close(conv);
RETURN_FALSE;
http://cvs.php.net/viewvc.cgi/php-src/main/streams/filter.c?r1=1.32&r2=1.33&diff_format=u
Index: php-src/main/streams/filter.c
diff -u php-src/main/streams/filter.c:1.32 php-src/main/streams/filter.c:1.33
--- php-src/main/streams/filter.c:1.32 Fri May 19 10:23:43 2006
+++ php-src/main/streams/filter.c Wed Jun 21 20:17:21 2006
@@ -16,7 +16,7 @@
+----------------------------------------------------------------------+
*/
-/* $Id: filter.c,v 1.32 2006/05/19 10:23:43 tony2001 Exp $ */
+/* $Id: filter.c,v 1.33 2006/06/21 20:17:21 andrei Exp $ */
#include "php.h"
#include "php_globals.h"
@@ -750,7 +750,7 @@
if (U_FAILURE(status)) {
int32_t offset = u_countChar32(bucket->buf.u,
num_conv);
- zend_raise_conversion_error_ex("Could not
convert Unicode string to binary string", conv, ZEND_FROM_UNICODE, offset,
(UG(from_error_mode) & ZEND_CONV_ERROR_EXCEPTION) TSRMLS_CC);
+ zend_raise_conversion_error_ex("Could not
convert Unicode string to binary string", conv, ZEND_FROM_UNICODE, offset
TSRMLS_CC);
}
if (bucket->own_buf) {
http://cvs.php.net/viewvc.cgi/php-src/main/streams/streams.c?r1=1.126&r2=1.127&diff_format=u
Index: php-src/main/streams/streams.c
diff -u php-src/main/streams/streams.c:1.126
php-src/main/streams/streams.c:1.127
--- php-src/main/streams/streams.c:1.126 Wed Jun 21 17:10:13 2006
+++ php-src/main/streams/streams.c Wed Jun 21 20:17:21 2006
@@ -19,7 +19,7 @@
+----------------------------------------------------------------------+
*/
-/* $Id: streams.c,v 1.126 2006/06/21 17:10:13 andrei Exp $ */
+/* $Id: streams.c,v 1.127 2006/06/21 20:17:21 andrei Exp $ */
#define _GNU_SOURCE
#include "php.h"
@@ -1222,7 +1222,7 @@
num_conv =
zend_convert_from_unicode(ZEND_U_CONVERTER(UG(runtime_encoding_conv)), &str,
&len, buf.u, buflen, &status);
if (U_FAILURE(status)) {
zend_raise_conversion_error_ex("Unable to convert data
to be written", ZEND_U_CONVERTER(UG(runtime_encoding_conv)),
-
ZEND_FROM_UNICODE, num_conv, (UG(from_error_mode) & ZEND_CONV_ERROR_EXCEPTION)
TSRMLS_CC);
+
ZEND_FROM_UNICODE, num_conv TSRMLS_CC);
} else {
php_error_docref(NULL TSRMLS_CC, E_NOTICE, "%d
character unicode buffer downcoded for binary stream runtime_encoding", ulen);
}
@@ -1270,7 +1270,7 @@
} else {
/* Figure out how didwrite corresponds to the input
buffer */
char *tmp = emalloc(didwrite + 1), *t = tmp;
- UChar *s = buf_orig;
+ const UChar *s = buf_orig;
UErrorCode status = U_ZERO_ERROR;
ucnv_resetFromUnicode(ZEND_U_CONVERTER(UG(runtime_encoding_conv)));
@@ -1570,7 +1570,7 @@
while ((b = php_stream_read_unicode(stream, inbuf_start,
sizeof(inbuf_start))) > 0) {
char *outbuf = outbuf_start;
- UChar *inbuf = inbuf_start;
+ const UChar *inbuf = inbuf_start;
UErrorCode status = U_ZERO_ERROR;
int len;
@@ -1580,7 +1580,7 @@
/* Memory overflow isn't a problem becuase
MAX_BYTES_FOR_STRING was allocated,
anything else is a more serious problem */
zend_raise_conversion_error_ex("Unable to
convert Unicode character using output_encoding, at least one character was
lost",
- conv,
ZEND_FROM_UNICODE, len, (UG(from_error_mode) & ZEND_CONV_ERROR_EXCEPTION)
TSRMLS_CC);
+
conv, ZEND_FROM_UNICODE, len TSRMLS_CC);
}
if (outbuf > outbuf_start) {
PHPWRITE(outbuf_start, outbuf - outbuf_start);
@@ -2594,8 +2594,7 @@
num_conv = zend_convert_from_unicode(UG(utf8_conv),
&scheme, &scheme_len, path, (p - path) + delim_len, &status);
if (U_FAILURE(status)) {
if (options & REPORT_ERRORS) {
- zend_raise_conversion_error_ex("Unable
to convert filepath", UG(utf8_conv), ZEND_FROM_UNICODE,
-
num_conv, (UG(from_error_mode) & ZEND_CONV_ERROR_EXCEPTION) TSRMLS_CC);
+ zend_raise_conversion_error_ex("Unable
to convert filepath", UG(utf8_conv), ZEND_FROM_UNICODE, num_conv TSRMLS_CC);
}
*pathenc = NULL;
*pathenc_len = 0;
@@ -2634,7 +2633,7 @@
if (U_FAILURE(status)) {
if (options & REPORT_ERRORS) {
zend_raise_conversion_error_ex("Unable to convert
filepath", ZEND_U_CONVERTER(UG(filesystem_encoding_conv)),
- ZEND_FROM_UNICODE,
num_conv, (UG(from_error_mode) & ZEND_CONV_ERROR_EXCEPTION) TSRMLS_CC);
+ ZEND_FROM_UNICODE,
num_conv TSRMLS_CC);
}
*pathenc = NULL;
@@ -2676,7 +2675,7 @@
if (U_FAILURE(status)) {
if (options & REPORT_ERRORS) {
zend_raise_conversion_error_ex("Unable to convert
filepath", ZEND_U_CONVERTER(UG(filesystem_encoding_conv)),
- ZEND_TO_UNICODE,
num_conv, (UG(to_error_mode) & ZEND_CONV_ERROR_EXCEPTION) TSRMLS_CC);
+ ZEND_TO_UNICODE,
num_conv TSRMLS_CC);
}
*pathdec = NULL;
--
PHP CVS Mailing List (http://www.php.net/)
To unsubscribe, visit: http://www.php.net/unsub.php
