Merge branch 'PHP-7.2'

5 files changed, 29 insertions, 38 deletions

diff --git a/ext/mbstring/libmbfl/mbfl/mbfl_convert.c b/ext/mbstring/libmbfl/mbfl/mbfl_convert.c
index 2f98b58178..69af1476c5 100644
--- a/ext/mbstring/libmbfl/mbfl/mbfl_convert.c
+++ b/ext/mbstring/libmbfl/mbfl/mbfl_convert.c
@@ -466,14 +466,26 @@ int mbfl_convert_filter_strcat(mbfl_convert_filter *filter, const unsigned char
 int
 mbfl_filt_conv_illegal_output(int c, mbfl_convert_filter *filter)
 {
-	int mode_backup, ret, n, m, r;
+	int mode_backup, substchar_backup, ret, n, m, r;
 
 	ret = 0;
+
 	mode_backup = filter->illegal_mode;
-	filter->illegal_mode = MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE;
+	substchar_backup = filter->illegal_substchar;
+
+	/* The used substitution character may not be supported by the target character encoding.
+	 * If that happens, first try to use "?" instead and if that also fails, silently drop the
+	 * character. */
+	if (filter->illegal_mode == MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR
+			&& filter->illegal_substchar != 0x3f) {
+		filter->illegal_substchar = 0x3f;
+	} else {
+		filter->illegal_mode = MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE;
+	}
+
 	switch (mode_backup) {
 	case MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR:
-		ret = (*filter->filter_function)(filter->illegal_substchar, filter);
+		ret = (*filter->filter_function)(substchar_backup, filter);
 		break;
 	case MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG:
 		if (c >= 0) {
@@ -559,14 +571,16 @@ mbfl_filt_conv_illegal_output(int c, mbfl_convert_filter *filter)
 				}
 				ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)";");
 			} else {
-				ret = (*filter->filter_function)(filter->illegal_substchar, filter);
+				ret = (*filter->filter_function)(substchar_backup, filter);
 			}
 		}
 		break;
 	default:
 		break;
 	}
+
 	filter->illegal_mode = mode_backup;
+	filter->illegal_substchar = substchar_backup;
 	filter->num_illegalchar++;
 
 	return ret;
diff --git a/ext/mbstring/mbstring.c b/ext/mbstring/mbstring.c
index 584523ecaf..a5430b4442 100644
--- a/ext/mbstring/mbstring.c
+++ b/ext/mbstring/mbstring.c
@@ -970,6 +970,7 @@ static size_t php_mb_zend_encoding_converter(unsigned char **to, size_t *to_leng
 	if (convd == NULL) {
 		return (size_t) -1;
 	}
+
 	mbfl_buffer_converter_illegal_mode(convd, MBSTRG(current_filter_illegal_mode));
 	mbfl_buffer_converter_illegal_substchar(convd, MBSTRG(current_filter_illegal_substchar));
 
@@ -3053,27 +3054,7 @@ MBSTRING_API char *php_mb_convert_encoding_ex(const char *input, size_t length,
 	}
 
 	mbfl_buffer_converter_illegal_mode(convd, MBSTRG(current_filter_illegal_mode));
-	if (string.encoding == MBSTRG(current_internal_encoding)) {
-		mbfl_buffer_converter_illegal_substchar(convd, MBSTRG(current_filter_illegal_substchar));
-	} else if (php_mb_is_no_encoding_unicode(string.encoding->no_encoding)
-			&& php_mb_is_no_encoding_unicode(MBSTRG(current_internal_encoding)->no_encoding)) {
-
-		if (php_mb_is_no_encoding_utf8(string.encoding->no_encoding)) {
-			if (MBSTRG(current_filter_illegal_substchar) > 0xd7ff &&
-				0xe000 > MBSTRG(current_filter_illegal_substchar)
-			) {
-				mbfl_buffer_converter_illegal_substchar(convd, 0x3f);
-			} else {
-				mbfl_buffer_converter_illegal_substchar(convd,
-					MBSTRG(current_filter_illegal_substchar));
-			}
-		} else {
-			mbfl_buffer_converter_illegal_substchar(convd,
-				MBSTRG(current_filter_illegal_substchar));
-		}
-	} else {
-		mbfl_buffer_converter_illegal_substchar(convd, 0x3f);
-	}
+	mbfl_buffer_converter_illegal_substchar(convd, MBSTRG(current_filter_illegal_substchar));
 
 	/* do it */
 	ret = mbfl_buffer_converter_feed_result(convd, &string, &result);
@@ -5017,17 +4998,7 @@ static inline char* php_mb_chr(zend_long cp, const char* enc_name, size_t *outpu
 	if (php_mb_is_no_encoding_utf8(no_enc)) {
 
 		if (0 > cp || cp > 0x10ffff || (cp > 0xd7ff && 0xe000 > cp)) {
-			if (php_mb_is_no_encoding_utf8(MBSTRG(current_internal_encoding)->no_encoding)) {
-				cp = MBSTRG(current_filter_illegal_substchar);
-			} else if (php_mb_is_no_encoding_unicode(MBSTRG(current_internal_encoding)->no_encoding)) {
-				if (0xd800 > MBSTRG(current_filter_illegal_substchar) || MBSTRG(current_filter_illegal_substchar) > 0xdfff) {
-					cp = MBSTRG(current_filter_illegal_substchar);
-				} else {
-					cp = 0x3f;
-				}
-			} else {
-				cp = 0x3f;
-			}
+			cp = MBSTRG(current_filter_illegal_substchar);
 		}
 
 		if (cp < 0x80) {
diff --git a/ext/mbstring/tests/bug69086.phpt b/ext/mbstring/tests/bug69086.phpt
index 921d61cca4..9566e10968 100644
--- a/ext/mbstring/tests/bug69086.phpt
+++ b/ext/mbstring/tests/bug69086.phpt
@@ -8,7 +8,13 @@ mb_substitute_character(0xfffd);
 var_dump("?" === mb_convert_encoding("\x80", "Shift_JIS", "EUC-JP"));
 mb_internal_encoding("UCS-4BE");
 var_dump("\x00\x00\xff\xfd" === mb_convert_encoding("\x80", "UCS-4BE", "UTF-8"));
+
+mb_internal_encoding("UTF-8");
+mb_substitute_character(0xfffd);
+var_dump("\u{fffd}" === mb_convert_encoding("\x80", "UTF-8", "EUC-JP-2004"));
+
 ?>
 --EXPECT--
 bool(true)
 bool(true)
+bool(true)
diff --git a/ext/mbstring/tests/mb_chr.phpt b/ext/mbstring/tests/mb_chr.phpt
index 19e1a704ec..8ec35920c3 100644
--- a/ext/mbstring/tests/mb_chr.phpt
+++ b/ext/mbstring/tests/mb_chr.phpt
@@ -22,7 +22,7 @@ var_dump(
 mb_internal_encoding("EUC-JP");
 mb_substitute_character(0xa4a2);
 var_dump(
-    "?" === mb_chr(0xd800, "UTF-8")
+    "\u{a4a2}" === mb_chr(0xd800, "UTF-8")
 );
 
 // Invalid
diff --git a/ext/mbstring/tests/mb_substitute_character_variation2.phpt b/ext/mbstring/tests/mb_substitute_character_variation2.phpt
index 202561afc7..6248174aa6 100644
--- a/ext/mbstring/tests/mb_substitute_character_variation2.phpt
+++ b/ext/mbstring/tests/mb_substitute_character_variation2.phpt
@@ -35,5 +35,5 @@ var_dump(bin2hex(mb_convert_encoding($string_mb, "ISO-8859-1", "UTF-8")));
 string(14) "3f3f3f3f3f3f3f"
 string(14) "42424242424242"
 string(0) ""
-string(0) ""
+string(14) "3f3f3f3f3f3f3f"
 ===DONE===