- MFH: Fixed bug #44617 (wrong HTML entity output when substitute_character=entity)

2 files changed, 60 insertions, 56 deletions

diff --git a/ext/mbstring/libmbfl/mbfl/mbfl_convert.c b/ext/mbstring/libmbfl/mbfl/mbfl_convert.c
index c7ec3d3a87..63003d0f0a 100644
--- a/ext/mbstring/libmbfl/mbfl/mbfl_convert.c
+++ b/ext/mbstring/libmbfl/mbfl/mbfl_convert.c
@@ -363,22 +363,6 @@ int mbfl_convert_filter_strcat(mbfl_convert_filter *filter, const unsigned char
 	return 0;
 }
 
-#if 0
-static int
-mbfl_convert_filter_strncat(mbfl_convert_filter *filter, const unsigned char *p, 
-			    int n)
-{
-	while (n > 0) {
-		if ((*filter->filter_function)(*p++, filter) < 0) {
-			return -1;
-		}
-		n--;
-	}
-
-	return n;
-}
-#endif
-
 /* illegal character output function for conv-filter */
 int
 mbfl_filt_conv_illegal_output(int c, mbfl_convert_filter *filter)
@@ -393,14 +377,9 @@ mbfl_filt_conv_illegal_output(int c, mbfl_convert_filter *filter)
 		ret = (*filter->filter_function)(filter->illegal_substchar, filter);
 		break;
 	case MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG:
-	case MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY:
 		if (c >= 0) {
 			if (c < MBFL_WCSGROUP_UCS4MAX) {	/* unicode */
-			  if (mode_backup == MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG) {
-			    ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)"U+");
-			  } else { /* entity */
-			    ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)"&#");
-			  }
+				ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)"U+");
 			} else {
 				if (c < MBFL_WCSGROUP_WCHARMAX) {
 					m = c & ~MBFL_WCSPLANE_MASK;
@@ -444,9 +423,38 @@ mbfl_filt_conv_illegal_output(int c, mbfl_convert_filter *filter)
 				if (m == 0 && ret >= 0) {
 					ret = (*filter->filter_function)(mbfl_hexchar_table[0], filter);
 				}
-				if (mode_backup == MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY) {
-				  ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)";");
+			}
+		}
+		break;
+	case MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY:
+		if (c >= 0) {
+			if (c < MBFL_WCSGROUP_UCS4MAX) {	/* unicode */
+				ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)"&#x");
+				if (ret < 0)
+					break;
+
+				m = 0;
+				r = 28;
+				while (r >= 0) {
+					n = (c >> r) & 0xf;
+					if (n || m) {
+						m = 1;
+						ret = (*filter->filter_function)(mbfl_hexchar_table[n], filter);
+						if (ret < 0) {
+							break;
+						}
+					}
+					r -= 4;
+				}
+				if (ret < 0) {
+					break;
+				}
+				if (m == 0) {
+					ret = (*filter->filter_function)(mbfl_hexchar_table[0], filter);
 				}
+				ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)";");
+			} else {
+				ret = (*filter->filter_function)(filter->illegal_substchar, filter);
 			}
 		}
 		break;
@@ -468,8 +476,8 @@ const struct mbfl_convert_vtbl * mbfl_convert_filter_get_vtbl(enum mbfl_no_encod
 	    to == mbfl_no_encoding_7bit) {
 		from = mbfl_no_encoding_8bit;
 	} else if (from == mbfl_no_encoding_base64 ||
-	           from == mbfl_no_encoding_qprint ||
-				  from == mbfl_no_encoding_uuencode) {
+			   from == mbfl_no_encoding_qprint ||
+			   from == mbfl_no_encoding_uuencode) {
 		to = mbfl_no_encoding_8bit;
 	}
 
diff --git a/ext/mbstring/tests/mb_substitute_character.phpt b/ext/mbstring/tests/mb_substitute_character.phpt
index 2adfddb10d..db1198822f 100644
--- a/ext/mbstring/tests/mb_substitute_character.phpt
+++ b/ext/mbstring/tests/mb_substitute_character.phpt
@@ -10,40 +10,36 @@ include_once('common.inc');
 
 // Note: It does not return TRUE/FALSE for setting char
 
-// Use Unicode val
-$r = mb_substitute_character(0x3013);
-//$r = mb_substitute_character('U+3013');
-($r === TRUE) ?    print "OK_UTF\n" :  print("NG_UTF: ".gettype($r)." $r\n");
-print mb_substitute_character() . "\n";
+var_dump(mb_substitute_character(0x3044));
+var_dump(mb_substitute_character());
+var_dump(bin2hex(mb_convert_encoding("\xe2\x99\xa0\xe3\x81\x82", "CP932", "UTF-8")));
 
+var_dump(mb_substitute_character('long'));
+var_dump(mb_substitute_character());
+var_dump(bin2hex(mb_convert_encoding("\xe2\x99\xa0\xe3\x81\x82", "CP932", "UTF-8")));
 
-// Use "long"
-$r = mb_substitute_character('long');
-($r === TRUE) ? print "OK_LONG\n" : print("NG_LONG: ".gettype($r)." $r\n");
-print mb_substitute_character() . "\n";
+var_dump(mb_substitute_character('none'));
+var_dump(mb_substitute_character());
+var_dump(bin2hex(mb_convert_encoding("\xe2\x99\xa0\xe3\x81\x82", "CP932", "UTF-8")));
 
+var_dump(mb_substitute_character('entity'));
+var_dump(mb_substitute_character());
+var_dump(bin2hex(mb_convert_encoding("\xe2\x99\xa0\xe3\x81\x82", "CP932", "UTF-8")));
 
-// Use "none"
-$r = mb_substitute_character('none');
-($r === TRUE) ? print "OK_NONE\n" : print("NG_NONE: ".gettype($r)." $r\n");
-print mb_substitute_character() . "\n";
-
-
-// Set invalid string. Should fail.
-print "== INVALID PARAMETER ==\n";
-$r = mb_substitute_character('BAD_NAME');
-($r === FALSE) ? print "OK_BAD_NAME\n" : print("NG_BAD_NAME: ".gettype($r)." $r\n");
-
+var_dump(mb_substitute_character('BAD_NAME'));
 ?>
-
 --EXPECT--
-OK_UTF
-12307
-OK_LONG
-long
-OK_NONE
-none
-== INVALID PARAMETER ==
+bool(true)
+int(12356)
+string(8) "82a282a0"
+bool(true)
+string(4) "long"
+string(16) "552b3236363082a0"
+bool(true)
+string(4) "none"
+string(4) "82a0"
+bool(true)
+string(6) "entity"
+string(20) "262378323636303b82a0"
 ERR: Warning
-OK_BAD_NAME
-
+bool(false)