- Implement zend_codepoint_to_uchar().

- Rename and fix zend_get_unified_string_type() so that it does not
  allow mixing of binary and Unicode types.

diff --git a/Zend/zend_API.c b/Zend/zend_API.c
index 9422bb865b7212b0f619cea5abbb2b4d89de3a35..248441b73239abed8e0bc27ae25a4ef6fc724a3f 100644 (file)
--- a/Zend/zend_API.c
+++ b/Zend/zend_API.c
@@ -2928,30 +2928,40 @@ ZEND_API zval *zend_read_property(zend_class_entry *scope, zval *object, char *n
 }
 
 /*
- * Return the most precise string type out of the list.
- * If none of the types are string types, IS_STRING is returned.
- * Binary > Unicode > string.
+ * Return the string type that all the passed in types should be converted to.
+ * If none of the types are string types, IS_UNICODE or IS_STRING is returned,
+ * depending on the Unicode semantics switch.
+ * Binary and Unicode types cannot be mixed, so we return -1 (or 255 since it's
+ * uchar)
  */
-ZEND_API zend_uchar zend_get_best_string_type(int num_args, ...)
+ZEND_API zend_uchar zend_get_unified_string_type(int num_args TSRMLS_DC, ...)
 {
        va_list ap;
-       int best_type = IS_STRING;
+       int best_type = UG(unicode) ? IS_UNICODE : IS_STRING;
+       zend_bool seen_unicode = 0;
        int type;
 
-       if (num_args <= 0) return -1;
+       if (num_args <= 0) return (zend_uchar)-1;
 
        va_start(ap, num_args);
        while (num_args--) {
                type = va_arg(ap, int);
                if (type == IS_BINARY) {
                        best_type = IS_BINARY;
-                       break;
-               } else if (type == IS_UNICODE && best_type == IS_STRING) {
-                       best_type = IS_UNICODE;
+               } else if (type == IS_UNICODE) {
+                       seen_unicode = 1;
+                       if (best_type == IS_STRING) {
+                               best_type = IS_UNICODE;
+                       }
                }
        }
        va_end(ap);
 
+       /* We do not allow mixing binary and Unicode types */
+       if (best_type == IS_BINARY && seen_unicode) {
+               return (zend_uchar)-1;
+       }
+
        return best_type;
 }
 

diff --git a/Zend/zend_API.h b/Zend/zend_API.h
index c5c54ddeb519f303b531966994b5a813946d3004..6a5ee07893a9dfc9a8125f09934973da73412943 100644 (file)
--- a/Zend/zend_API.h
+++ b/Zend/zend_API.h
@@ -235,7 +235,7 @@ ZEND_API zval *zend_read_property(zend_class_entry *scope, zval *object, char *n
 
 ZEND_API zend_class_entry *zend_get_class_entry(zval *zobject TSRMLS_DC);
 ZEND_API int zend_get_object_classname(zval *object, char **class_name, zend_uint *class_name_len TSRMLS_DC);
-ZEND_API zend_uchar zend_get_best_string_type(int num_args, ...);
+ZEND_API zend_uchar zend_get_unified_string_type(int num_args TSRMLS_DC, ...);
 
 #define getThis() (this_ptr)
 

diff --git a/Zend/zend_operators.h b/Zend/zend_operators.h
index d5fa434e89c776e7cf5d91e38972e2b26897e5d1..1cdea5aac631f1cdfd96e8f15e8bf9e29db10b01 100644 (file)
--- a/Zend/zend_operators.h
+++ b/Zend/zend_operators.h
@@ -68,7 +68,7 @@ ZEND_API long zend_u_strtol(const UChar *nptr, UChar **endptr, int base);
 ZEND_API double zend_u_strtod(const UChar *nptr, UChar **endptr);
 END_EXTERN_C()
 
-static inline zend_bool is_numeric_string(char *str, int length, long *lval, double *dval, zend_bool allow_errors)
+static inline zend_uchar is_numeric_string(char *str, int length, long *lval, double *dval, zend_bool allow_errors)
 {
        long local_lval;
        double local_dval;
@@ -131,7 +131,7 @@ static inline zend_bool is_numeric_string(char *str, int length, long *lval, dou
        return 0;
 }
 
-static inline zend_bool is_numeric_unicode(UChar *str, int32_t length, long *lval, double *dval, zend_bool allow_errors)
+static inline zend_uchar is_numeric_unicode(UChar *str, int32_t length, long *lval, double *dval, zend_bool allow_errors)
 {
        int32_t local_lval;
        double local_dval;
@@ -320,7 +320,7 @@ END_EXTERN_C()
                                convert_to_double(pzv);                 \
                                break;                                                  \
                        case IS_BOOL:                                           \
-                               convert_to_bool(pzv);                   \
+                               convert_to_boolean(pzv);                \
                                break;                                                  \
                        case IS_ARRAY:                                          \
                                convert_to_array(pzv);                  \

diff --git a/Zend/zend_unicode.h b/Zend/zend_unicode.h
index 950dc7c1fd9e935b88e30ee4b8024b6c932d5bb5..6de553893628df1358efcf9692bd0bdc59d28218 100644 (file)
--- a/Zend/zend_unicode.h
+++ b/Zend/zend_unicode.h
@@ -71,6 +71,24 @@ static inline UChar32 zend_get_codepoint_at(UChar *str, int32_t length, int32_t
        return c;
 }
 
+/*
+ * Convert a single codepoint to UChar sequence (1 or 2).
+ * The UChar buffer is assumed to be large enough.
+ */
+static inline int zend_codepoint_to_uchar(UChar32 codepoint, UChar *buf)
+{
+       if (U_IS_BMP(codepoint)) {
+               *buf++ = (UChar) codepoint;
+               return 1;
+       } else if (codepoint <= UCHAR_MAX_VALUE) {
+               *buf++ = (UChar) U16_LEAD(codepoint);
+               *buf++ = (UChar) U16_TRAIL(codepoint);
+               return 2;
+       } else {
+               return 0;
+       }
+}
+
 #define ZEND_U_CONVERTER(c) ((c)?(c):UG(fallback_encoding_conv))
 
 #define USTR_FREE(ustr) do { if (ustr) { efree(ustr); } } while (0);