Fixed bug #35147 (__HALT_COMPILER() breaks with --enable-zend-multibyte)

Author: dstogov

NEWS

@@ -9,6 +9,8 @@ PHP                                                                        NEWS
 - Fixed bug #35179 (tokenizer extension needs T_HALT_COMPILER). (Greg)
 - Fixed bug #35176 (include()/require()/*_once() produce wrong error messages
   about main()). (Dmitry)
+- Fixed bug #35147 (__HALT_COMPILER() breaks with --enable-zend-multibyte).
+  (Dmitry, Moriyoshi)
 - Fixed bug #35142 (SOAP Client/Server Complex Object Support). (Dmitry)
 - Fixed bug #35135 (PDOStatment without related PDO object may crash). (Ilia)
 - Fixed bug #35091 (SoapClient leaks memory). (Dmitry)

Zend/zend.c

@@ -78,6 +78,9 @@ static ZEND_INI_MH(OnUpdateErrorReporting)
 ZEND_INI_BEGIN()
 	ZEND_INI_ENTRY("error_reporting",				NULL,		ZEND_INI_ALL,		OnUpdateErrorReporting)
 	STD_ZEND_INI_BOOLEAN("zend.ze1_compatibility_mode",	"0",	ZEND_INI_ALL,		OnUpdateBool,	ze1_compatibility_mode,	zend_executor_globals,	executor_globals)
+#ifdef ZEND_MULTIBYTE
+	STD_ZEND_INI_BOOLEAN("detect_unicode", "1", ZEND_INI_ALL, OnUpdateBool, detect_unicode, zend_compiler_globals, compiler_globals)
+#endif
 ZEND_INI_END()
 
 

Zend/zend_globals.h

@@ -135,6 +135,7 @@ struct _zend_compiler_globals {
 #ifdef ZEND_MULTIBYTE
 	zend_encoding **script_encoding_list;
 	int script_encoding_list_size;
+	zend_bool detect_unicode;
 
 	zend_encoding *internal_encoding;
 

Zend/zend_language_scanner.l

@@ -506,7 +506,9 @@ ZEND_API int zend_get_scanned_file_offset(TSRMLS_D)
 
 		return zend_stream_ftell(yyin TSRMLS_CC) - offset_from_the_end;
 	} else {
-		return -1;
+		/* The entire file is in the buffer; probably zend multibyte
+		   is enabled */
+		return (yy_c_buf_p - (YY_CURRENT_BUFFER)->yy_ch_buf);
 	}
 }
 

Zend/zend_multibyte.c

@@ -994,11 +994,13 @@ static zend_encoding* zend_multibyte_find_script_encoding(zend_encoding *onetime
 		return onetime_encoding;
 	}
 
-	/* check out bom(byte order mark) and see if containing wchars */
-	script_encoding = zend_multibyte_detect_unicode(TSRMLS_C);
-	if (script_encoding != NULL) {
-		/* bom or wchar detection is prior to 'script_encoding' option */
-		return script_encoding;
+	if (CG(detect_unicode)) {
+		/* check out bom(byte order mark) and see if containing wchars */
+		script_encoding = zend_multibyte_detect_unicode(TSRMLS_C);
+		if (script_encoding != NULL) {
+			/* bom or wchar detection is prior to 'script_encoding' option */
+			return script_encoding;
+		}
 	}
 
 	/* if no script_encoding specified, just leave alone */

pear/Makefile.frag

@@ -3,7 +3,7 @@
 peardir=$(PEAR_INSTALLDIR)
 
 # Skip all php.ini files altogether
-PEAR_INSTALL_FLAGS = -n -dshort_open_tag=0 -dsafe_mode=0 -derror_reporting=E_ALL
+PEAR_INSTALL_FLAGS = -n -dshort_open_tag=0 -dsafe_mode=0 -derror_reporting=E_ALL -ddetect_unicode=0
 
 install-pear-installer: $(top_builddir)/sapi/cli/php
 	@$(top_builddir)/sapi/cli/php $(PEAR_INSTALL_FLAGS) $(srcdir)/install-pear-nozlib.phar -d "$(peardir)" -b "$(bindir)"