Add a global ASCII connverter, macros to return single codepoints

(UCHAR32), and zend_unicode_to_ascii() function.

diff --git a/Zend/zend.c b/Zend/zend.c
index 8e172fb521851889338a0fb99478beb6085597f1..38cb08d46974deb02bf2a1e4bc429eb7ed409723 100644 (file)
--- a/Zend/zend.c
+++ b/Zend/zend.c
@@ -42,10 +42,8 @@
 #   define GLOBAL_CONSTANTS_TABLE   EG(zend_constants)
 #endif
 
-#ifndef __GNUC__
 ZEND_API zstr null_zstr;
 ZEND_API zstr empty_zstr;
-#endif
 
 #if defined(ZEND_WIN32) && ZEND_DEBUG
 BOOL WINAPI IsDebuggerPresent(VOID);
@@ -906,6 +904,7 @@ static void unicode_globals_ctor(zend_unicode_globals *unicode_globals TSRMLS_DC
 {
        unicode_globals->unicode = 0;
        unicode_globals->utf8_conv = NULL;
+       unicode_globals->ascii_conv = NULL;
        unicode_globals->fallback_encoding_conv = NULL;
        unicode_globals->runtime_encoding_conv = NULL;
        unicode_globals->output_encoding_conv = NULL;
@@ -913,6 +912,8 @@ static void unicode_globals_ctor(zend_unicode_globals *unicode_globals TSRMLS_DC
        unicode_globals->http_input_encoding_conv = NULL;
        unicode_globals->filesystem_encoding_conv = NULL;
        zend_set_converter_encoding(&unicode_globals->utf8_conv, "UTF-8");
+       zend_set_converter_encoding(&unicode_globals->ascii_conv, "US-ASCII");
+       zend_set_converter_error_mode(unicode_globals->ascii_conv, ZEND_FROM_UNICODE, ZEND_CONV_ERROR_STOP);
        unicode_globals->from_error_mode = ZEND_CONV_ERROR_SUBST;
        memset(unicode_globals->from_subst_char, 0, 3 * sizeof(UChar));
        zend_codepoint_to_uchar(0x3f, unicode_globals->from_subst_char);
@@ -941,6 +942,9 @@ static void unicode_globals_dtor(zend_unicode_globals *unicode_globals TSRMLS_DC
        if (unicode_globals->utf8_conv) {
                ucnv_close(unicode_globals->utf8_conv);
        }
+       if (unicode_globals->ascii_conv) {
+               ucnv_close(unicode_globals->ascii_conv);
+       }
        zend_hash_destroy(&unicode_globals->flex_compatible);
 }
 

diff --git a/Zend/zend.h b/Zend/zend.h
index 6046f614a25b780fa30c2cb009c92989d994d97f..ebd5f6a1166d34782e971dcfebfba040fb04d777 100644 (file)
--- a/Zend/zend.h
+++ b/Zend/zend.h
@@ -246,11 +246,6 @@ typedef union _zstr {
        void  *v;
 } zstr;
 
-#ifdef __GNUC__
-#      define ZSTR(x)    ((zstr)(x))
-#      define NULL_ZSTR  ZSTR((void*)NULL)
-#      define EMPTY_ZSTR ZSTR("\0\0")
-#else
 extern ZEND_API zstr null_zstr;
 extern ZEND_API zstr empty_zstr;
 
@@ -263,7 +258,6 @@ static inline zstr _to_zstr(void *v) {
 #      define ZSTR(x)    _to_zstr(x)
 #      define NULL_ZSTR  null_zstr
 #      define EMPTY_ZSTR empty_zstr
-#endif
 
 #define EMPTY_STR ((UChar*)"\0\0")
 

diff --git a/Zend/zend_API.h b/Zend/zend_API.h
index 6c71fa290e35cedec47be1fac7193765793874e1..89c49a11794a6366e14f679200a73edf435c2145 100644 (file)
--- a/Zend/zend_API.h
+++ b/Zend/zend_API.h
@@ -807,6 +807,12 @@ END_EXTERN_C()
                Z_TYPE_P(z) = IS_UNICODE;                   \
 }
 
+#define ZVAL_UCHAR32(z, ch) {                                                  \
+               UChar buf[3];                                                                   \
+               int buf_len = zend_codepoint_to_uchar(ch, buf); \
+               ZVAL_UNICODEL(z, buf, buf_len, 1);                              \
+}
+
 #define ZVAL_EMPTY_STRING(z) {                         \
                Z_STRLEN_P(z) = 0;                                      \
                Z_STRVAL_P(z) = STR_EMPTY_ALLOC();      \
@@ -881,6 +887,7 @@ END_EXTERN_C()
 #define RETVAL_EMPTY_STRING()                  ZVAL_EMPTY_STRING(return_value)
 #define RETVAL_UNICODE(u, duplicate)           ZVAL_UNICODE(return_value, u, duplicate)
 #define RETVAL_UNICODEL(u, l, duplicate)       ZVAL_UNICODEL(return_value, u, l, duplicate)
+#define RETVAL_UCHAR32(ch)                             ZVAL_UCHAR32(return_value, ch)
 #define RETVAL_EMPTY_UNICODE()                         ZVAL_EMPTY_UNICODE(return_value)
 #define RETVAL_ZVAL(zv, copy, dtor)            ZVAL_ZVAL(return_value, zv, copy, dtor)
 #define RETVAL_FALSE                                   ZVAL_BOOL(return_value, 0)
@@ -899,6 +906,7 @@ END_EXTERN_C()
 #define RETURN_EMPTY_STRING()                  { RETVAL_EMPTY_STRING(); return; }
 #define RETURN_UNICODE(u, duplicate)   { RETVAL_UNICODE(u, duplicate); return; }
 #define RETURN_UNICODEL(u, l, duplicate) { RETVAL_UNICODEL(u, l, duplicate); return; }
+#define RETURN_UCHAR32(ch)                             { RETVAL_UCHAR32(ch); return; }
 #define RETURN_EMPTY_UNICODE()                         { RETVAL_EMPTY_UNICODE(); return; }
 #define RETURN_ZVAL(zv, copy, dtor)            { RETVAL_ZVAL(zv, copy, dtor); return; }
 #define RETURN_FALSE                                   { RETVAL_FALSE; return; }

diff --git a/Zend/zend_globals.h b/Zend/zend_globals.h
index e4bdef54a02747adf5b2d347aacbbd35eac544c8..537ea3440829360bd69915e0c9135b031cf53ad8 100644 (file)
--- a/Zend/zend_globals.h
+++ b/Zend/zend_globals.h
@@ -301,6 +301,7 @@ struct _zend_unicode_globals {
        UConverter *http_input_encoding_conv;/* http input encoding converter */
        UConverter *filesystem_encoding_conv;/* default filesystem converter (entries, not contents) */ 
        UConverter *utf8_conv;                           /* all-purpose UTF-8 converter */
+       UConverter *ascii_conv;                          /* all-purpose ASCII converter */
 
        uint16_t from_error_mode;
        UChar from_subst_char[3];

diff --git a/Zend/zend_unicode.c b/Zend/zend_unicode.c
index ffa7ee254d74563839c532403600c0195bc7fed6..f5ef2d146849774afb805157aa82d51f65de1313 100644 (file)
--- a/Zend/zend_unicode.c
+++ b/Zend/zend_unicode.c
@@ -373,6 +373,23 @@ ZEND_API void zend_convert_encodings(UConverter *target_conv, UConverter *source
 }
 /* }}} */
 
+/* {{{ zend_unicode_to_ascii */
+ZEND_API char* zend_unicode_to_ascii(const UChar *us, int us_len TSRMLS_DC)
+{
+       char *cs;
+       int   cs_len;
+       UErrorCode status = U_ZERO_ERROR;
+
+       zend_convert_from_unicode(UG(ascii_conv), &cs, &cs_len, us, us_len, &status);
+       if (U_FAILURE(status)) {
+               efree(cs);
+               return NULL;
+       }
+
+       return cs;
+}
+/* }}} */
+
 /* {{{ 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)
 {

diff --git a/Zend/zend_unicode.h b/Zend/zend_unicode.h
index 9423dfa7d2d2b2d913afd569d17bb203a5bde2ac..46a2023a82fc76d3c74d36d3c686d7bc770e9fa6 100644 (file)
--- a/Zend/zend_unicode.h
+++ b/Zend/zend_unicode.h
@@ -73,6 +73,8 @@ void zend_collator_destroy(zend_collator *zcoll);
 ZEND_API int zend_convert_to_unicode(UConverter *conv, UChar **target, int *target_len, const char *source, int source_len, UErrorCode *status);
 ZEND_API int zend_convert_from_unicode(UConverter *conv, char **target, int *target_len, const UChar *source, int source_len, UErrorCode *status);
 ZEND_API void zend_convert_encodings(UConverter *target_conv, UConverter *source_conv, char **target, int *target_len, const char *source, int source_len, UErrorCode *status);
+ZEND_API char* zend_unicode_to_ascii(const UChar *us, int us_len TSRMLS_DC);
+
 ZEND_API int zval_string_to_unicode_ex(zval *string, UConverter *conv TSRMLS_DC);
 ZEND_API int zval_string_to_unicode(zval *string TSRMLS_DC);
 ZEND_API int zval_unicode_to_string(zval *string, UConverter *conv TSRMLS_DC);