From: Alex Dowad <alexinbeijing@gmail.com>
Date: Wed, 7 Oct 2020 20:30:34 +0000 (+0200)
Subject: Don't pass invalid JIS X 0212, JIS X 0213, and Windows-CP932 characters through
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=5ffcf563bd184316c988fe3fa09a261e96999edd;p=php

Don't pass invalid JIS X 0212, JIS X 0213, and Windows-CP932 characters through

Similarly to JIS X 0208, mbstring would pass kuten codes which are not mapped
in the JIS X 0212, JIS X 0213, or CP932 character sets through silently when
converting to another Japanese encoding.
---

diff --git a/ext/mbstring/libmbfl/filters/mbfilter_cp5022x.c b/ext/mbstring/libmbfl/filters/mbfilter_cp5022x.c
index b630d3fcb0..3f4fb7132c 100644
--- a/ext/mbstring/libmbfl/filters/mbfilter_cp5022x.c
+++ b/ext/mbstring/libmbfl/filters/mbfilter_cp5022x.c
@@ -383,9 +383,8 @@ retry:
 int
 mbfl_filt_conv_wchar_jis_ms(int c, mbfl_convert_filter *filter)
 {
-	int c1, s;
+	int s = 0;
 
-	s = 0;
 	if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
 		s = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
 	} else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
@@ -408,11 +407,7 @@ mbfl_filt_conv_wchar_jis_ms(int c, mbfl_convert_filter *filter)
 
 	/* do some transliteration */
 	if (s <= 0) {
-		c1 = c & ~MBFL_WCSPLANE_MASK;
-		if (c1 == MBFL_WCSPLANE_JIS0212) {
-			s = c & MBFL_WCSPLANE_MASK;
-			s |= 0x8080;
-		} else if (c == 0xa5) {		/* YEN SIGN */
+		if (c == 0xa5) {		/* YEN SIGN */
 			s = 0x1005c;
 		} else if (c == 0x203e) {	/* OVER LINE */
 			s = 0x1007e;
diff --git a/ext/mbstring/libmbfl/filters/mbfilter_cp51932.c b/ext/mbstring/libmbfl/filters/mbfilter_cp51932.c
index 25944a10dd..6e0746b866 100644
--- a/ext/mbstring/libmbfl/filters/mbfilter_cp51932.c
+++ b/ext/mbstring/libmbfl/filters/mbfilter_cp51932.c
@@ -210,13 +210,7 @@ mbfl_filt_conv_wchar_cp51932(int c, mbfl_convert_filter *filter)
 	}
 	if (s1 >= 0x8080) s1 = -1; /* we don't support JIS X0213 */
 	if (s1 <= 0) {
-		c1 = c & ~MBFL_WCSPLANE_MASK;
-		if (c1 == MBFL_WCSPLANE_WINCP932) {
-			s1 = c & MBFL_WCSPLANE_MASK;
-			if (s1 >= ((85 + 0x20) << 8)) {	/* 85ku - 120ku */
-				s1 = -1;
-			}
-		} else if (c == 0xa5) {		/* YEN SIGN */
+		if (c == 0xa5) {		/* YEN SIGN */
 			s1 = 0x005c;			/* YEN SIGN */
 		} else if (c == 0x203e) {	/* OVER LINE */
 			s1 = 0x007e;			/* FULLWIDTH MACRON */
diff --git a/ext/mbstring/libmbfl/filters/mbfilter_cp932.c b/ext/mbstring/libmbfl/filters/mbfilter_cp932.c
index 64ccd57361..6246600de8 100644
--- a/ext/mbstring/libmbfl/filters/mbfilter_cp932.c
+++ b/ext/mbstring/libmbfl/filters/mbfilter_cp932.c
@@ -251,14 +251,7 @@ mbfl_filt_conv_wchar_cp932(int c, mbfl_convert_filter *filter)
 		s2 = 1;
 	}
 	if (s1 <= 0) {
-		c1 = c & ~MBFL_WCSPLANE_MASK;
-		if (c1 == MBFL_WCSPLANE_WINCP932) {
-			s1 = c & MBFL_WCSPLANE_MASK;
-			s2 = 1;
-		} else if (c1 == MBFL_WCSPLANE_JIS0212) {
-			s1 = c & MBFL_WCSPLANE_MASK;
-			s1 |= 0x8080;
-		} else if (c == 0xa5) {		/* YEN SIGN */
+		if (c == 0xa5) {		/* YEN SIGN */
 			s1 = 0x005c;	/* YEN SIGN */
 		} else if (c == 0x203e) {	/* OVER LINE */
 			s1 = 0x007e;	/* FULLWIDTH MACRON */
diff --git a/ext/mbstring/libmbfl/filters/mbfilter_euc_jp.c b/ext/mbstring/libmbfl/filters/mbfilter_euc_jp.c
index 1589ae7966..507a26a5b1 100644
--- a/ext/mbstring/libmbfl/filters/mbfilter_euc_jp.c
+++ b/ext/mbstring/libmbfl/filters/mbfilter_euc_jp.c
@@ -192,9 +192,8 @@ static int mbfl_filt_conv_eucjp_wchar_flush(mbfl_convert_filter *filter)
 int
 mbfl_filt_conv_wchar_eucjp(int c, mbfl_convert_filter *filter)
 {
-	int s;
+	int s = 0;
 
-	s = 0;
 	if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
 		s = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
 	} else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
diff --git a/ext/mbstring/libmbfl/filters/mbfilter_euc_jp_win.c b/ext/mbstring/libmbfl/filters/mbfilter_euc_jp_win.c
index df51be0183..569dabd92d 100644
--- a/ext/mbstring/libmbfl/filters/mbfilter_euc_jp_win.c
+++ b/ext/mbstring/libmbfl/filters/mbfilter_euc_jp_win.c
@@ -273,20 +273,7 @@ mbfl_filt_conv_wchar_eucjpwin(int c, mbfl_convert_filter *filter)
 		s1 = 0x2d62;		/* NUMERO SIGN */
 	}
 	if (s1 <= 0) {
-		c1 = c & ~MBFL_WCSPLANE_MASK;
-		if (c1 == MBFL_WCSPLANE_WINCP932) {
-			s1 = c & MBFL_WCSPLANE_MASK;
-			if (s1 >= ((85 + 0x20) << 8)) {	/* 85ku - 120ku */
-				s1 = -1;
-			}
-		} else if (c1 == MBFL_WCSPLANE_JIS0212) {
-			s1 = c & MBFL_WCSPLANE_MASK;
-			if (s1 >= ((83 + 0x20) << 8)) {	/* 83ku - 94ku */
-				s1 = -1;
-			} else {
-				s1 |= 0x8080;
-			}
-		} else if (c == 0xa5) {		/* YEN SIGN */
+		if (c == 0xa5) {		/* YEN SIGN */
 			s1 = 0x216f;			/* FULLWIDTH YEN SIGN */
 		} else if (c == 0x203e) {	/* OVER LINE */
 			s1 = 0x2131;			/* FULLWIDTH MACRON */
diff --git a/ext/mbstring/libmbfl/filters/mbfilter_iso2022_jp_ms.c b/ext/mbstring/libmbfl/filters/mbfilter_iso2022_jp_ms.c
index af54a1c605..5e0a63958b 100644
--- a/ext/mbstring/libmbfl/filters/mbfilter_iso2022_jp_ms.c
+++ b/ext/mbstring/libmbfl/filters/mbfilter_iso2022_jp_ms.c
@@ -302,14 +302,7 @@ mbfl_filt_conv_wchar_2022jpms(int c, mbfl_convert_filter *filter)
 		s1 = (c1 << 8) | c2;
 	}
 	if (s1 <= 0) {
-		c1 = c & ~MBFL_WCSPLANE_MASK;
-		if (c1 == MBFL_WCSPLANE_WINCP932) {
-			s1 = c & MBFL_WCSPLANE_MASK;
-			s2 = 1;
-		} else if (c1 == MBFL_WCSPLANE_JIS0212) {
-			s1 = c & MBFL_WCSPLANE_MASK;
-			s1 |= 0x8080;
-		} else if (c == 0xa5) {		/* YEN SIGN */
+		if (c == 0xa5) {		/* YEN SIGN */
 			s1 = 0x216f;	            /* FULLWIDTH YEN SIGN */
 		} else if (c == 0x203e) {	/* OVER LINE */
 			s1 = 0x2131;	/* FULLWIDTH MACRON */
diff --git a/ext/mbstring/libmbfl/filters/mbfilter_iso2022jp_mobile.c b/ext/mbstring/libmbfl/filters/mbfilter_iso2022jp_mobile.c
index 1b4857dfe2..0099fb1ebb 100644
--- a/ext/mbstring/libmbfl/filters/mbfilter_iso2022jp_mobile.c
+++ b/ext/mbstring/libmbfl/filters/mbfilter_iso2022jp_mobile.c
@@ -330,14 +330,7 @@ mbfl_filt_conv_wchar_2022jp_mobile(int c, mbfl_convert_filter *filter)
 		s1 = (c1 << 8) | c2;
 	}
 	if (s1 <= 0) {
-		c1 = c & ~MBFL_WCSPLANE_MASK;
-		if (c1 == MBFL_WCSPLANE_WINCP932) {
-			s1 = c & MBFL_WCSPLANE_MASK;
-			s2 = 1;
-		} else if (c1 == MBFL_WCSPLANE_JIS0212) {
-			s1 = c & MBFL_WCSPLANE_MASK;
-			s1 |= 0x8080;
-		} else if (c == 0xa5) {		/* YEN SIGN */
+		if (c == 0xa5) {		/* YEN SIGN */
 			s1 = 0x216f;	            /* FULLWIDTH YEN SIGN */
 		} else if (c == 0x203e) {	/* OVER LINE */
 			s1 = 0x2131;	/* FULLWIDTH MACRON */
diff --git a/ext/mbstring/libmbfl/filters/mbfilter_jis.c b/ext/mbstring/libmbfl/filters/mbfilter_jis.c
index 8ba46a5b9a..5e0c05c8fc 100644
--- a/ext/mbstring/libmbfl/filters/mbfilter_jis.c
+++ b/ext/mbstring/libmbfl/filters/mbfilter_jis.c
@@ -274,9 +274,8 @@ retry:
 int
 mbfl_filt_conv_wchar_jis(int c, mbfl_convert_filter *filter)
 {
-	int c1, s;
+	int s = 0;
 
-	s = 0;
 	if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
 		s = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
 	} else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
@@ -287,11 +286,7 @@ mbfl_filt_conv_wchar_jis(int c, mbfl_convert_filter *filter)
 		s = ucs_r_jis_table[c - ucs_r_jis_table_min];
 	}
 	if (s <= 0) {
-		c1 = c & ~MBFL_WCSPLANE_MASK;
-		if (c1 == MBFL_WCSPLANE_JIS0212) {
-			s = c & MBFL_WCSPLANE_MASK;
-			s |= 0x8080;
-		} else if (c == 0xa5) {		/* YEN SIGN */
+		if (c == 0xa5) {		/* YEN SIGN */
 			s = 0x1005c;
 		} else if (c == 0x203e) {	/* OVER LINE */
 			s = 0x1007e;
diff --git a/ext/mbstring/libmbfl/filters/mbfilter_sjis_2004.c b/ext/mbstring/libmbfl/filters/mbfilter_sjis_2004.c
index 02bbdccdc8..1e549fdebb 100644
--- a/ext/mbstring/libmbfl/filters/mbfilter_sjis_2004.c
+++ b/ext/mbstring/libmbfl/filters/mbfilter_sjis_2004.c
@@ -590,10 +590,6 @@ retry:
 	}
 
 	if (s1 <= 0) {
-		c1 = c & ~MBFL_WCSPLANE_MASK;
-		if (c1 == MBFL_WCSPLANE_JIS0213) {
-			s1 = c & MBFL_WCSPLANE_MASK;
-		}
 		if (c == 0) {
 			s1 = 0;
 		} else if (s1 <= 0) {
diff --git a/ext/mbstring/libmbfl/filters/mbfilter_sjis_mac.c b/ext/mbstring/libmbfl/filters/mbfilter_sjis_mac.c
index 6361195606..4dead25a78 100644
--- a/ext/mbstring/libmbfl/filters/mbfilter_sjis_mac.c
+++ b/ext/mbstring/libmbfl/filters/mbfilter_sjis_mac.c
@@ -416,14 +416,7 @@ mbfl_filt_conv_wchar_sjis_mac(int c, mbfl_convert_filter *filter)
 		}
 
 		if (s1 <= 0) {
-			c1 = c & ~MBFL_WCSPLANE_MASK;
-			if (c1 == MBFL_WCSPLANE_WINCP932) {
-				s1 = c & MBFL_WCSPLANE_MASK;
-				s2 = 1;
-			} else if (c1 == MBFL_WCSPLANE_JIS0212) {
-				s1 = c & MBFL_WCSPLANE_MASK;
-				s1 |= 0x8080;
-			} else if (c == 0xa0) {
+			if (c == 0xa0) {
 				s1 = 0x00a0;
 			} else if (c == 0xa5) { /* YEN SIGN */
 				/* Unicode has codepoint 0xFFE5 for a fullwidth Yen sign;
diff --git a/ext/mbstring/libmbfl/filters/mbfilter_sjis_mobile.c b/ext/mbstring/libmbfl/filters/mbfilter_sjis_mobile.c
index 10ac7d9ce4..a434d9bf54 100644
--- a/ext/mbstring/libmbfl/filters/mbfilter_sjis_mobile.c
+++ b/ext/mbstring/libmbfl/filters/mbfilter_sjis_mobile.c
@@ -790,14 +790,7 @@ mbfl_filt_conv_wchar_sjis_mobile(int c, mbfl_convert_filter *filter)
 		s2 = 1;
 	}
 	if (s1 <= 0) {
-		c1 = c & ~MBFL_WCSPLANE_MASK;
-		if (c1 == MBFL_WCSPLANE_WINCP932) {
-			s1 = c & MBFL_WCSPLANE_MASK;
-			s2 = 1;
-		} else if (c1 == MBFL_WCSPLANE_JIS0212) {
-			s1 = c & MBFL_WCSPLANE_MASK;
-			s1 |= 0x8080;
-		} else if (c == 0xa5) {		/* YEN SIGN */
+		if (c == 0xa5) {		/* YEN SIGN */
 			s1 = 0x216f;	/* FULLWIDTH YEN SIGN */
 		} else if (c == 0x203e) {	/* OVER LINE */
 			s1 = 0x2131;	/* FULLWIDTH MACRON */
diff --git a/ext/mbstring/libmbfl/filters/mbfilter_sjis_open.c b/ext/mbstring/libmbfl/filters/mbfilter_sjis_open.c
index ea19e6b105..d1dd5efd95 100644
--- a/ext/mbstring/libmbfl/filters/mbfilter_sjis_open.c
+++ b/ext/mbstring/libmbfl/filters/mbfilter_sjis_open.c
@@ -241,14 +241,7 @@ mbfl_filt_conv_wchar_sjis_open(int c, mbfl_convert_filter *filter)
 		s2 = 1;
 	}
 	if (s1 <= 0) {
-		c1 = c & ~MBFL_WCSPLANE_MASK;
-		if (c1 == MBFL_WCSPLANE_WINCP932) {
-			s1 = c & MBFL_WCSPLANE_MASK;
-			s2 = 1;
-		} else if (c1 == MBFL_WCSPLANE_JIS0212) {
-			s1 = c & MBFL_WCSPLANE_MASK;
-			s1 |= 0x8080;
-		} else if (c == 0xa5) {		/* YEN SIGN */
+		if (c == 0xa5) {		/* YEN SIGN */
 			s1 = 0x216f;	/* FULLWIDTH YEN SIGN */
 		} else if (c == 0x203e) {	/* OVER LINE */
 			s1 = 0x2131;	/* FULLWIDTH MACRON */