Adjust API for scanning INI files as UTF-8.

diff --git a/Zend/zend_compile.h b/Zend/zend_compile.h
index 318cac7000ce08e947d1e537c1f2ee869479738e..852dc684af0c297c2793800590494323f911c626 100644 (file)
--- a/Zend/zend_compile.h
+++ b/Zend/zend_compile.h
@@ -342,7 +342,7 @@ ZEND_API char *zend_get_compiled_script_encoding(TSRMLS_D);
 ZEND_API zstr zend_get_compiled_variable_name(zend_op_array *op_array, zend_uint var, int* name_len);
 
 ZEND_API int zend_prepare_scanner_converters(const char *onetime_encoding, int run_time TSRMLS_DC);
-ZEND_API int zend_convert_scanner_output(UChar **target, int *target_len, const char *source, int source_len, UErrorCode *status TSRMLS_DC);
+ZEND_API int zend_convert_scanner_output(UConverter *conv, UChar **target, int *target_len, const char *source, int source_len, UErrorCode *status TSRMLS_DC);
 int zend_unicode_yyinput(zend_file_handle *file_handle, char *buf, size_t len TSRMLS_DC);
 
 #ifdef ZTS
@@ -538,6 +538,7 @@ ZEND_API void zend_file_handle_dtor(zend_file_handle *fh);
 ZEND_API int zend_cleanup_class_data(zend_class_entry **pce TSRMLS_DC);
 ZEND_API int zend_cleanup_function_data(zend_function *function TSRMLS_DC);
 ZEND_API int zend_cleanup_function_data_full(zend_function *function TSRMLS_DC);
+ZEND_API int zend_copy_scanner_string(zval *zendlval, char *str, zend_uint str_len, zend_uchar type, UConverter *conv TSRMLS_DC);
 
 ZEND_API void destroy_zend_function(zend_function *function TSRMLS_DC);
 ZEND_API void zend_function_dtor(zend_function *function);

diff --git a/Zend/zend_language_scanner.l b/Zend/zend_language_scanner.l
index 0bd6905be0a207b691c72ef28111ec2dfe8c07f5..ed32c9663cf84e7adcd21be3f687e922194098c1 100644 (file)
--- a/Zend/zend_language_scanner.l
+++ b/Zend/zend_language_scanner.l
@@ -411,13 +411,13 @@ static inline int zend_parse_charname_sequence(UChar **s, UChar *end, UChar32 *c
        return 0;
 }
 
-static inline int zend_copy_string_value(zval *zendlval, char *str, zend_uint str_len, zend_uchar type TSRMLS_DC)
+ZEND_API int zend_copy_scanner_string(zval *zendlval, char *str, zend_uint str_len, zend_uchar type, UConverter *conv TSRMLS_DC)
 {
        UErrorCode status = U_ZERO_ERROR;
        int consumed = 0;
 
        if (type == IS_UNICODE) {
-               consumed = zend_convert_scanner_output(&Z_USTRVAL_P(zendlval), &Z_USTRLEN_P(zendlval), str, str_len, &status TSRMLS_CC);
+               consumed = zend_convert_scanner_output(conv, &Z_USTRVAL_P(zendlval), &Z_USTRLEN_P(zendlval), str, str_len, &status TSRMLS_CC);
 
                if (U_FAILURE(status)) {
                        zend_error(E_COMPILE_WARNING,"Illegal or truncated character in input: offset %d, state=%d", consumed, YYSTATE);
@@ -622,16 +622,16 @@ ZEND_API int zend_prepare_scanner_converters(const char *onetime_encoding, int r
        return zend_set_converter_encoding(&SCNG(output_conv), encoding);
 }
 
-ZEND_API int zend_convert_scanner_output(UChar **target, int *target_len, const char *source, int source_len, UErrorCode *status TSRMLS_DC)
+ZEND_API int zend_convert_scanner_output(UConverter *conv, UChar **target, int *target_len, const char *source, int source_len, UErrorCode *status TSRMLS_DC)
 {
        const char *source_consumed = NULL;
 
        /* set our custom callback with context */
-       ucnv_setToUCallBack(SCNG(output_conv), zend_scanner_output_callback, &source_consumed, NULL, NULL, status);
+       ucnv_setToUCallBack(conv, zend_scanner_output_callback, &source_consumed, NULL, NULL, status);
 
        /* reset the error and perform conversion */
        *status = U_ZERO_ERROR;
-       zend_convert_to_unicode(SCNG(output_conv), target, target_len, source, source_len, status);
+       zend_convert_to_unicode(conv, target, target_len, source, source_len, status);
 
        /* figure out how many source bytes were consumed */
        if (U_SUCCESS(*status)) {
@@ -1032,7 +1032,7 @@ int zend_scan_unicode_double_string(zval *zendlval TSRMLS_DC)
 
        HANDLE_NEWLINES(yytext, yyleng);
 
-       if (!zend_copy_string_value(zendlval, yytext+1, yyleng-2, IS_UNICODE TSRMLS_CC)) {
+       if (!zend_copy_scanner_string(zendlval, yytext+1, yyleng-2, IS_UNICODE, SCNG(output_conv) TSRMLS_CC)) {
                return 0;
        }
 
@@ -1174,7 +1174,7 @@ int zend_scan_unicode_single_string(zval *zendlval TSRMLS_DC)
 
        HANDLE_NEWLINES(yytext, yyleng);
 
-       if (!zend_copy_string_value(zendlval, yytext+1, yyleng-2, IS_UNICODE TSRMLS_CC)) {
+       if (!zend_copy_scanner_string(zendlval, yytext+1, yyleng-2, IS_UNICODE, SCNG(output_conv) TSRMLS_CC)) {
                return 0;
        }
 
@@ -1558,7 +1558,7 @@ NEWLINE ("\r"|"\n"|"\r\n")
 
 <ST_LOOKING_FOR_PROPERTY>{LABEL} {
        yy_pop_state(TSRMLS_C);
-       if (!zend_copy_string_value(zendlval, yytext, yyleng, UG(unicode)?IS_UNICODE:IS_STRING TSRMLS_CC)) {
+       if (!zend_copy_scanner_string(zendlval, yytext, yyleng, UG(unicode)?IS_UNICODE:IS_STRING, SCNG(output_conv) TSRMLS_CC)) {
                return 0;
        }
        if (UG(unicode) && !zend_check_and_normalize_identifier(zendlval)) {
@@ -1836,7 +1836,7 @@ NEWLINE ("\r"|"\n"|"\r\n")
 
 
 <ST_LOOKING_FOR_VARNAME>{LABEL} {
-       if (!zend_copy_string_value(zendlval, yytext, yyleng, UG(unicode)?IS_UNICODE:IS_STRING TSRMLS_CC)) {
+       if (!zend_copy_scanner_string(zendlval, yytext, yyleng, UG(unicode)?IS_UNICODE:IS_STRING, SCNG(output_conv) TSRMLS_CC)) {
                return 0;
        }
        if (UG(unicode) && !zend_check_and_normalize_identifier(zendlval)) {
@@ -1889,7 +1889,7 @@ NEWLINE ("\r"|"\n"|"\r\n")
 }
 
 <ST_DOUBLE_QUOTES,ST_BACKQUOTE,ST_HEREDOC>{LNUM}|{HNUM} { /* treat numbers (almost) as strings inside encapsulated strings */
-       if (!zend_copy_string_value(zendlval, yytext, yyleng, CG(literal_type) TSRMLS_CC)) {
+       if (!zend_copy_scanner_string(zendlval, yytext, yyleng, CG(literal_type), SCNG(output_conv) TSRMLS_CC)) {
                return 0;
        }
        return T_NUM_STRING;
@@ -2063,7 +2063,7 @@ NEWLINE ("\r"|"\n"|"\r\n")
 }
 
 <ST_IN_SCRIPTING,ST_DOUBLE_QUOTES,ST_HEREDOC,ST_BACKQUOTE>"$"{LABEL} {
-       if (!zend_copy_string_value(zendlval, (yytext+1), (yyleng-1), UG(unicode)?IS_UNICODE:IS_STRING TSRMLS_CC)) {
+       if (!zend_copy_scanner_string(zendlval, (yytext+1), (yyleng-1), UG(unicode)?IS_UNICODE:IS_STRING, SCNG(output_conv) TSRMLS_CC)) {
                return 0;
        }
        if (UG(unicode) && !zend_check_and_normalize_identifier(zendlval)) {
@@ -2073,7 +2073,7 @@ NEWLINE ("\r"|"\n"|"\r\n")
 }
 
 <ST_IN_SCRIPTING>{LABEL} {
-       if (!zend_copy_string_value(zendlval, yytext, yyleng, UG(unicode)?IS_UNICODE:IS_STRING TSRMLS_CC)) {
+       if (!zend_copy_scanner_string(zendlval, yytext, yyleng, UG(unicode)?IS_UNICODE:IS_STRING, SCNG(output_conv) TSRMLS_CC)) {
                return 0;
        }
        if (UG(unicode) && !zend_check_and_normalize_identifier(zendlval)) {
@@ -2083,7 +2083,7 @@ NEWLINE ("\r"|"\n"|"\r\n")
 }
 
 <ST_DOUBLE_QUOTES,ST_BACKQUOTE,ST_HEREDOC>{LABEL} {
-       if (!zend_copy_string_value(zendlval, yytext, yyleng, CG(literal_type) TSRMLS_CC)) {
+       if (!zend_copy_scanner_string(zendlval, yytext, yyleng, CG(literal_type), SCNG(output_conv) TSRMLS_CC)) {
                return 0;
        }
        return T_STRING;
@@ -2309,7 +2309,7 @@ NEWLINE ("\r"|"\n"|"\r\n")
                return T_END_HEREDOC;
        } else {
                CG(zend_lineno)++;
-               if (!zend_copy_string_value(zendlval, yytext, yyleng, CG(literal_type) TSRMLS_CC)) {
+               if (!zend_copy_scanner_string(zendlval, yytext, yyleng, CG(literal_type), SCNG(output_conv) TSRMLS_CC)) {
                        return 0;
                }
                return T_STRING;
@@ -2319,14 +2319,14 @@ NEWLINE ("\r"|"\n"|"\r\n")
 
 <ST_DOUBLE_QUOTES,ST_BACKQUOTE,ST_HEREDOC>{ESCAPED_AND_WHITESPACE} {
        HANDLE_NEWLINES(yytext, yyleng);
-       if (!zend_copy_string_value(zendlval, yytext, yyleng, CG(literal_type) TSRMLS_CC)) {
+       if (!zend_copy_scanner_string(zendlval, yytext, yyleng, CG(literal_type), SCNG(output_conv) TSRMLS_CC)) {
                return 0;
        }
        return T_ENCAPSED_AND_WHITESPACE;
 }
 
 <ST_DOUBLE_QUOTES>[`]+ {
-       if (!zend_copy_string_value(zendlval, yytext, yyleng, CG(literal_type) TSRMLS_CC)) {
+       if (!zend_copy_scanner_string(zendlval, yytext, yyleng, CG(literal_type), SCNG(output_conv) TSRMLS_CC)) {
                return 0;
        }
        return T_ENCAPSED_AND_WHITESPACE;
@@ -2334,7 +2334,7 @@ NEWLINE ("\r"|"\n"|"\r\n")
 
 
 <ST_BACKQUOTE>["]+ {
-       if (!zend_copy_string_value(zendlval, yytext, yyleng, CG(literal_type) TSRMLS_CC)) {
+       if (!zend_copy_scanner_string(zendlval, yytext, yyleng, CG(literal_type), SCNG(output_conv) TSRMLS_CC)) {
                return 0;
        }
        return T_ENCAPSED_AND_WHITESPACE;
@@ -2406,7 +2406,7 @@ NEWLINE ("\r"|"\n"|"\r\n")
                        return 0;
                }
        } else {
-               zend_copy_string_value(zendlval, yytext, yyleng, CG(literal_type) TSRMLS_CC);
+               zend_copy_scanner_string(zendlval, yytext, yyleng, CG(literal_type), SCNG(output_conv) TSRMLS_CC);
                return T_STRING;
        }
 }
@@ -2432,13 +2432,13 @@ NEWLINE ("\r"|"\n"|"\r\n")
                        return 0;
                }
        } else {
-               zend_copy_string_value(zendlval, yytext, yyleng, CG(literal_type) TSRMLS_CC);
+               zend_copy_scanner_string(zendlval, yytext, yyleng, CG(literal_type), SCNG(output_conv) TSRMLS_CC);
                return T_STRING;
        }
 }
 
 <ST_DOUBLE_QUOTES,ST_BACKQUOTE,ST_HEREDOC>"\\{" {
-       if (!zend_copy_string_value(zendlval, yytext, yyleng, CG(literal_type) TSRMLS_CC)) {
+       if (!zend_copy_scanner_string(zendlval, yytext, yyleng, CG(literal_type), SCNG(output_conv) TSRMLS_CC)) {
                return 0;
        }
        return T_STRING;
@@ -2462,7 +2462,7 @@ NEWLINE ("\r"|"\n"|"\r\n")
                        Z_LVAL_P(zendlval) = (long) yytext[1];
                        break;
                default:
-                       if (!zend_copy_string_value(zendlval, yytext, yyleng, CG(literal_type) TSRMLS_CC)) {
+                       if (!zend_copy_scanner_string(zendlval, yytext, yyleng, CG(literal_type), SCNG(output_conv) TSRMLS_CC)) {
                                return 0;
                        }
                        return T_BAD_CHARACTER;
@@ -2473,7 +2473,7 @@ NEWLINE ("\r"|"\n"|"\r\n")
 
 
 <ST_HEREDOC>["'`]+ {
-       if (!zend_copy_string_value(zendlval, yytext, yyleng, CG(literal_type) TSRMLS_CC)) {
+       if (!zend_copy_scanner_string(zendlval, yytext, yyleng, CG(literal_type), SCNG(output_conv) TSRMLS_CC)) {
                return 0;
        }
        return T_ENCAPSED_AND_WHITESPACE;