moriyoshi Sun Oct 6 11:38:34 2002 EDT
Modified files: (Branch: PHP_4_3)
/php4/ext/mbstring php_mbregex.c
Log:
Added support for aliases of the encoding names.
Index: php4/ext/mbstring/php_mbregex.c
diff -u php4/ext/mbstring/php_mbregex.c:1.5 php4/ext/mbstring/php_mbregex.c:1.5.2.1
--- php4/ext/mbstring/php_mbregex.c:1.5 Tue Aug 27 11:32:04 2002
+++ php4/ext/mbstring/php_mbregex.c Sun Oct 6 11:38:33 2002
@@ -16,7 +16,7 @@
+----------------------------------------------------------------------+
*/
-/* $Id: php_mbregex.c,v 1.5 2002/08/27 15:32:04 helly Exp $ */
+/* $Id: php_mbregex.c,v 1.5.2.1 2002/10/06 15:38:33 moriyoshi Exp $ */
#ifdef HAVE_CONFIG_H
@@ -90,22 +90,23 @@
mbctype = -1;
if (pname != NULL) {
- if (strcasecmp("EUC-JP", pname) == 0) {
+ if (strcasecmp("EUC-JP", pname) == 0
+ || strcasecmp("X-EUC-JP", pname) == 0
+ || strcasecmp("UJIS", pname) == 0
+ || strcasecmp("EUCJP", pname) == 0
+ || strcasecmp("EUC_JP", pname) == 0) {
mbctype = MBCTYPE_EUC;
- } else if (strcasecmp("UTF-8", pname) == 0) {
+ } else if (strcasecmp("UTF-8", pname) == 0
+ || strcasecmp("UTF8", pname) == 0) {
mbctype = MBCTYPE_UTF8;
- } else if (strcasecmp("SJIS", pname) == 0) {
+ } else if (strcasecmp("SJIS", pname) == 0
+ || strcasecmp("CP932", pname) == 0
+ || strcasecmp("MS932", pname) == 0
+ || strcasecmp("SHIFT_JIS", pname) == 0 ) {
mbctype = MBCTYPE_SJIS;
- } else if (strcasecmp("ascii", pname) == 0) {
+ } else if (strcasecmp("ASCII", pname) == 0) {
mbctype = MBCTYPE_ASCII;
- } else if (strcasecmp("euc", pname) == 0) {
mbctype = MBCTYPE_EUC;
- } else if (strcasecmp("eucJP", pname) == 0) {
- mbctype = MBCTYPE_EUC;
- } else if (strcasecmp("EUC_JP", pname) == 0) {
- mbctype = MBCTYPE_EUC;
- } else if (strcasecmp("Shift_JIS", pname) == 0) {
- mbctype = MBCTYPE_SJIS;
}
}
@@ -115,7 +116,7 @@
static const char*
php_mbregex_mbctype2name(int mbctype)
{
- const char *p;
+ const char *p = NULL;
if (mbctype == MBCTYPE_EUC) {
p = "EUC-JP";
@@ -125,8 +126,6 @@
p = "SJIS";
} else if(mbctype == MBCTYPE_ASCII) {
p = "ascii";
- } else {
- p = "unknown";
}
return p;
@@ -224,12 +223,18 @@
int mbctype;
if (ZEND_NUM_ARGS() == 0) {
-
RETVAL_STRING((char*)php_mbregex_mbctype2name(MBSTRG(current_mbctype)), 1);
- } else if (ZEND_NUM_ARGS() == 1 && zend_get_parameters_ex(1, &arg1) !=
FAILURE) {
+ const char *retval = php_mbregex_mbctype2name(MBSTRG(current_mbctype));
+ if ( retval != NULL ) {
+ RETVAL_STRING((char *)retval);
+ } else {
+ RETVAL_FALSE;
+ }
+ } else if (ZEND_NUM_ARGS() == 1 &&
+ zend_get_parameters_ex(1, &arg1) != FAILURE) {
convert_to_string_ex(arg1);
mbctype = php_mbregex_name2mbctype(Z_STRVAL_PP(arg1));
if (mbctype < 0) {
- php_error_docref(NULL TSRMLS_CC, E_WARNING, "unknown encoding
\"%s\"", Z_STRVAL_PP(arg1));
+ php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding
+\"%s\"", Z_STRVAL_PP(arg1));
RETVAL_FALSE;
} else {
MBSTRG(current_mbctype) = mbctype;
@@ -307,7 +312,9 @@
mbre_free_registers(®s);
RETURN_FALSE;
}
-
+ if (regs.beg[0] == regs.end[0]) {
+ php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty regular
+expression");
+ }
match_len = 1;
str = Z_STRVAL_PP(arg_string);
if (array) {
@@ -423,6 +430,10 @@
pos = 0;
while (err >= 0) {
err = mbre_search(&re, string, string_len, pos, string_len - pos,
®s);
+ if ( regs.beg[0] == regs.end[0] ) {
+ php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty regular
+expression");
+ break;
+ }
if (err <= -2) {
php_error_docref(NULL TSRMLS_CC, E_WARNING, "mbregex search
failure in php_mbereg_replace_exec()");
break;
@@ -565,20 +576,19 @@
/* churn through str, generating array entries as we go */
while ((count != 0) &&
(err = mbre_search(&re, string, string_len, pos, string_len - pos,
®s)) >= 0) {
+ if ( regs.beg[0] == regs.end[0] ) {
+ php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty regular
+expression");
+ break;
+ }
+
n = regs.beg[0];
- if (n == pos) {
- /* match is at start of string, return empty string */
- add_next_index_stringl(return_value, empty_string, 0, 1);
+ /* add it to the array */
+ if (n < string_len && n <= pos) {
+ n -= pos;
+ add_next_index_stringl(return_value, &string[pos], n, 1);
} else {
- /* On a real match */
- /* add it to the array */
- if (n < string_len) {
- n -= pos;
- add_next_index_stringl(return_value, &string[pos], n,
1);
- } else {
- err = -2;
- break;
- }
+ err = -2;
+ break;
}
/* point at our new starting point */
n = regs.end[0];
@@ -721,11 +731,11 @@
}
if (!MBSTRG(search_re)) {
- php_error_docref(NULL TSRMLS_CC, E_WARNING, "no regex for search");
+ php_error_docref(NULL TSRMLS_CC, E_WARNING, "No regex given");
RETURN_FALSE;
}
if (!str) {
- php_error_docref(NULL TSRMLS_CC, E_WARNING, "no string for search");
+ php_error_docref(NULL TSRMLS_CC, E_WARNING, "No string given");
RETURN_FALSE;
}
if (MBSTRG(search_regs)) {
@@ -736,7 +746,6 @@
}
err = mbre_search(MBSTRG(search_re), str, len, pos, len - pos,
MBSTRG(search_regs));
-
if (err <= -2) {
php_error_docref(NULL TSRMLS_CC, E_WARNING, "mbregex search failure in
mbregex_search()");
RETVAL_FALSE;
@@ -744,6 +753,9 @@
MBSTRG(search_pos) = len;
RETVAL_FALSE;
} else {
+ if (MBSTRG(search_regs)->beg[0] == MBSTRG(search_regs)->end[0]) {
+ php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty regular
+expression");
+ }
switch (mode) {
case 1:
if (array_init(return_value) != FAILURE) {
@@ -955,8 +967,11 @@
}
convert_to_long_ex(arg_pos);
n = Z_LVAL_PP(arg_pos);
- if (n < 0) {
- php_error_docref(NULL TSRMLS_CC, E_WARNING, "position is minus value");
+ if (n < 0
+ || ( MBSTRG(search_str) != NULL && *MBSTRG(search_str) != NULL &&
+ Z_TYPE_PP(MBSTRG(search_str)) == IS_STRING &&
+ n >= Z_STRLEN_PP(MBSTRG(search_str)) ) ) {
+ php_error_docref(NULL TSRMLS_CC, E_WARNING, "Position is out of
+range");
MBSTRG(search_pos) = 0;
RETVAL_FALSE;
} else {
--
PHP CVS Mailing List (http://www.php.net/)
To unsubscribe, visit: http://www.php.net/unsub.php