Update conversion error behavior and add some new modes.

# The various escape modes are what ICU calls them. We may want to come
# up with different names for UNICODE/ICU/JAVA ones..

diff --git a/Zend/zend.c b/Zend/zend.c
index e777650ac736fda52db7e45047b3b7d5b8ef059d..0ae309e0861960341b1b8b0073c7aafba62f0dc7 100644 (file)
--- a/Zend/zend.c
+++ b/Zend/zend.c
@@ -173,9 +173,15 @@ static void zend_update_converters_error_behavior(TSRMLS_D)
 static ZEND_INI_MH(OnUpdateConversionErrorMode)
 {
        if (!new_value) {
-               UG(from_u_error_mode) = ZEND_FROM_U_ERROR_SUBST;
+               UG(from_u_error_mode) = ZEND_CONV_ERROR_SUBST;
        } else {
-               UG(from_u_error_mode) = atoi(new_value);
+               uint16_t mode = atoi(new_value);
+
+               if ((mode & 0xff) > ZEND_CONV_ERROR_LAST_ENUM) {
+                       zend_error(E_WARNING, "Illegal value for conversion error mode");
+                       return FAILURE;
+               }
+               UG(from_u_error_mode) = mode;
        }
        zend_update_converters_error_behavior(TSRMLS_C);
        return SUCCESS;
@@ -194,7 +200,7 @@ static ZEND_INI_MH(OnUpdateConversionSubstChar)
                        zend_error(E_WARNING, "Substitution character string should be a hexadecimal Unicode codepoint value");
                        return FAILURE;
                }
-               if (c < 0 || c >= 0x10FFFF) {
+               if (c < 0 || c >= UCHAR_MAX_VALUE) {
                        zend_error(E_WARNING, "Substitution character value U+%06x is out of range 0-10FFFF", c);
                        return FAILURE;
                }
@@ -952,7 +958,7 @@ static void unicode_globals_ctor(zend_unicode_globals *unicode_globals TSRMLS_DC
        unicode_globals->http_input_encoding_conv = NULL;
        unicode_globals->subst_char_len = 0;
        zend_set_converter_encoding(&unicode_globals->utf8_conv, "UTF-8");
-       unicode_globals->from_u_error_mode = ZEND_FROM_U_ERROR_SUBST;
+       unicode_globals->from_u_error_mode = ZEND_CONV_ERROR_SUBST;
 
        zend_hash_init_ex(&unicode_globals->flex_compatible, 0, NULL, NULL, 1, 0);
 }

diff --git a/Zend/zend_constants.c b/Zend/zend_constants.c
index 3ed7883be021529b4332073422585d95245f2292..699b1abcbdff58baf7d40765aa4457d18699b8f7 100644 (file)
--- a/Zend/zend_constants.c
+++ b/Zend/zend_constants.c
@@ -116,10 +116,15 @@ void zend_register_standard_constants(TSRMLS_D)
 
        REGISTER_MAIN_LONG_CONSTANT("E_ALL", E_ALL, CONST_PERSISTENT | CONST_CS);
 
-       REGISTER_MAIN_LONG_CONSTANT("U_INVALID_STOP", ZEND_FROM_U_ERROR_STOP, CONST_PERSISTENT | CONST_CS);
-       REGISTER_MAIN_LONG_CONSTANT("U_INVALID_SKIP", ZEND_FROM_U_ERROR_SKIP, CONST_PERSISTENT | CONST_CS);
-       REGISTER_MAIN_LONG_CONSTANT("U_INVALID_SUBSTITUTE", ZEND_FROM_U_ERROR_SUBST, CONST_PERSISTENT | CONST_CS);
-       REGISTER_MAIN_LONG_CONSTANT("U_INVALID_ESCAPE", ZEND_FROM_U_ERROR_ESCAPE, CONST_PERSISTENT | CONST_CS);
+       REGISTER_MAIN_LONG_CONSTANT("U_CONV_ERROR_STOP", ZEND_CONV_ERROR_STOP, CONST_PERSISTENT | CONST_CS);
+       REGISTER_MAIN_LONG_CONSTANT("U_CONV_ERROR_SKIP", ZEND_CONV_ERROR_SKIP, CONST_PERSISTENT | CONST_CS);
+       REGISTER_MAIN_LONG_CONSTANT("U_CONV_ERROR_SUBST", ZEND_CONV_ERROR_SUBST, CONST_PERSISTENT | CONST_CS);
+       REGISTER_MAIN_LONG_CONSTANT("U_CONV_ERROR_ESCAPE_UNICODE", ZEND_CONV_ERROR_ESCAPE_UNICODE, CONST_PERSISTENT | CONST_CS);
+       REGISTER_MAIN_LONG_CONSTANT("U_CONV_ERROR_ESCAPE_ICU", ZEND_CONV_ERROR_ESCAPE_ICU, CONST_PERSISTENT | CONST_CS);
+       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);
 
        /* true/false constants */
        {

diff --git a/Zend/zend_globals.h b/Zend/zend_globals.h
index 8c8f67720c68b4c419b45b010807ef8af331d693..4fb31b09f41a31d5772b6f12788ea514b2b471a7 100644 (file)
--- a/Zend/zend_globals.h
+++ b/Zend/zend_globals.h
@@ -301,7 +301,7 @@ struct _zend_unicode_globals {
        UConverter *http_input_encoding_conv;/* http input encoding converter */
        UConverter *utf8_conv;                           /* all-purpose UTF-8 converter */
 
-       uint8_t from_u_error_mode;
+       uint16_t from_u_error_mode;
        UChar subst_char[3];
        uint8_t subst_char_len;
 

diff --git a/Zend/zend_unicode.c b/Zend/zend_unicode.c
index 77f2583a56ee18ea22df04c6929de99485ace448..f67b8b02629233eda9a716108d90a5de14fd140c 100644 (file)
--- a/Zend/zend_unicode.c
+++ b/Zend/zend_unicode.c
@@ -35,21 +35,36 @@ void zend_set_converter_error_mode(UConverter *conv, uint8_t error_mode)
     UErrorCode status = U_ZERO_ERROR;
 
     switch (error_mode) {
-        case ZEND_FROM_U_ERROR_STOP:
+        case ZEND_CONV_ERROR_STOP:
             ucnv_setFromUCallBack(conv, UCNV_FROM_U_CALLBACK_STOP, NULL, NULL, NULL, &status);
             break;
 
-        case ZEND_FROM_U_ERROR_SKIP:
+        case ZEND_CONV_ERROR_SKIP:
             ucnv_setFromUCallBack(conv, UCNV_FROM_U_CALLBACK_SKIP, UCNV_SKIP_STOP_ON_ILLEGAL, NULL, NULL, &status);
             break;
 
-        case ZEND_FROM_U_ERROR_ESCAPE:
-            /* UTODO replace with custom callback for various substitution patterns */
+        case ZEND_CONV_ERROR_SUBST:
+            ucnv_setFromUCallBack(conv, UCNV_FROM_U_CALLBACK_SUBSTITUTE, UCNV_SUB_STOP_ON_ILLEGAL, NULL, NULL, &status);
+            break;
+
+        case ZEND_CONV_ERROR_ESCAPE_UNICODE:
             ucnv_setFromUCallBack(conv, UCNV_FROM_U_CALLBACK_ESCAPE, UCNV_ESCAPE_UNICODE, NULL, NULL, &status);
             break;
 
-        case ZEND_FROM_U_ERROR_SUBST:
-            ucnv_setFromUCallBack(conv, UCNV_FROM_U_CALLBACK_SUBSTITUTE, UCNV_SKIP_STOP_ON_ILLEGAL, NULL, NULL, &status);
+        case ZEND_CONV_ERROR_ESCAPE_ICU:
+            ucnv_setFromUCallBack(conv, UCNV_FROM_U_CALLBACK_ESCAPE, UCNV_ESCAPE_ICU, NULL, NULL, &status);
+            break;
+
+        case ZEND_CONV_ERROR_ESCAPE_JAVA:
+            ucnv_setFromUCallBack(conv, UCNV_FROM_U_CALLBACK_ESCAPE, UCNV_ESCAPE_JAVA, NULL, NULL, &status);
+            break;
+
+        case ZEND_CONV_ERROR_ESCAPE_XML_DEC:
+            ucnv_setFromUCallBack(conv, UCNV_FROM_U_CALLBACK_ESCAPE, UCNV_ESCAPE_XML_DEC, NULL, NULL, &status);
+            break;
+
+        case ZEND_CONV_ERROR_ESCAPE_XML_HEX:
+            ucnv_setFromUCallBack(conv, UCNV_FROM_U_CALLBACK_ESCAPE, UCNV_ESCAPE_XML_HEX, NULL, NULL, &status);
             break;
 
         default:
@@ -317,50 +332,6 @@ ZEND_API int zval_unicode_to_string(zval *string, UConverter *conv TSRMLS_DC)
     char *s = NULL;
     int s_len;
 
-#if 0
-    /* UTODO Putting it here for now, until we figure out the framework */
-    switch (UG(from_u_error_mode)) {
-        case ZEND_FROM_U_ERROR_STOP:
-            ucnv_setFromUCallBack(UG(runtime_encoding_conv), UCNV_FROM_U_CALLBACK_STOP, NULL, NULL, NULL, &status);
-            break;
-
-        case ZEND_FROM_U_ERROR_SKIP:
-            ucnv_setFromUCallBack(UG(runtime_encoding_conv), UCNV_FROM_U_CALLBACK_SKIP, NULL, NULL, NULL, &status);
-            break;
-
-        case ZEND_FROM_U_ERROR_ESCAPE:
-            ucnv_setFromUCallBack(UG(runtime_encoding_conv), UCNV_FROM_U_CALLBACK_ESCAPE, UCNV_ESCAPE_UNICODE, NULL, NULL, &status);
-            break;
-
-        case ZEND_FROM_U_ERROR_SUBST:
-            ucnv_setFromUCallBack(UG(runtime_encoding_conv), UCNV_FROM_U_CALLBACK_SUBSTITUTE, NULL, NULL, NULL, &status);
-            break;
-
-        default:
-            assert(0);
-            break;
-    }
-
-    if (UG(subst_chars)) {
-        char subchar[16];
-        int8_t char_len = 16;
-        status = U_ZERO_ERROR;
-        ucnv_getSubstChars(UG(runtime_encoding_conv), subchar, &char_len, &status);
-        if (U_FAILURE(status)) {
-            zend_error(E_WARNING, "Could not get substitution characters");
-            return FAILURE;
-        }
-        status = U_ZERO_ERROR;
-        ucnv_setSubstChars(UG(runtime_encoding_conv), UG(subst_chars), MIN(char_len, UG(subst_chars_len)), &status);
-        if (U_FAILURE(status)) {
-            zend_error(E_WARNING, "Could not set substitution characters");
-            return FAILURE;
-        }
-    }
-
-    status = U_ZERO_ERROR;
-#endif
-
     UChar *u = Z_USTRVAL_P(string);
     int u_len = Z_USTRLEN_P(string);
 

diff --git a/Zend/zend_unicode.h b/Zend/zend_unicode.h
index c99891cfb82730367cf65bf5cec0fbfd64d8e30a..13c3d13a80977bea44934ce9c690b4445b884d3f 100644 (file)
--- a/Zend/zend_unicode.h
+++ b/Zend/zend_unicode.h
@@ -28,10 +28,20 @@
 #include <unicode/uloc.h>
 #include <unicode/ucol.h>
 
-#define ZEND_FROM_U_ERROR_STOP   0
-#define ZEND_FROM_U_ERROR_SKIP   1
-#define ZEND_FROM_U_ERROR_SUBST  2
-#define ZEND_FROM_U_ERROR_ESCAPE 3
+enum {
+  ZEND_CONV_ERROR_STOP,
+  ZEND_CONV_ERROR_SKIP,
+  ZEND_CONV_ERROR_SUBST,
+  ZEND_CONV_ERROR_ESCAPE_UNICODE,
+  ZEND_CONV_ERROR_ESCAPE_ICU,
+  ZEND_CONV_ERROR_ESCAPE_JAVA,
+  ZEND_CONV_ERROR_ESCAPE_XML_DEC,
+  ZEND_CONV_ERROR_ESCAPE_XML_HEX,
+  ZEND_CONV_ERROR_LAST_ENUM,
+
+  ZEND_CONV_ERROR_EXCEPTION    = 0x100
+};
+
 
 /* internal functions */