added missing areas on Unicode->CP936 conversion table. added an alias CP932 on sjis-win.

ext/mbstring/mbfilter.c

@@ -653,7 +653,7 @@ static mbfl_encoding mbfl_encoding_eucjp_win = {
 	MBFL_ENCTYPE_MBCS
 };
 
-static const char *mbfl_encoding_sjis_win_aliases[] = {"SJIS-open", "MS_Kanji", "Windows-31J", NULL};
+static const char *mbfl_encoding_sjis_win_aliases[] = {"SJIS-open", "MS_Kanji", "Windows-31J", "CP932", NULL};
 
 static mbfl_encoding mbfl_encoding_sjis_win = {
 	mbfl_no_encoding_sjis_win,

ext/mbstring/mbfilter_cn.c

@@ -111,8 +111,8 @@ mbfl_filt_conv_wchar_euccn(int c, mbfl_convert_filter *filter TSRMLS_DC)
 		s = ucs_a3_cp936_table[c - ucs_a3_cp936_table_min];
 	} else if (c >= ucs_i_cp936_table_min && c < ucs_i_cp936_table_max) {
 		s = ucs_i_cp936_table[c - ucs_i_cp936_table_min];
-	} else if (c >= ucs_r_cp936_table_min && c < ucs_r_cp936_table_max) {
-		s = ucs_r_cp936_table[c - ucs_r_cp936_table_min];
+	} else if (c >= ucs_hff_cp936_table_min && c < ucs_hff_cp936_table_max) {
+		s = ucs_hff_cp936_table[c - ucs_hff_cp936_table_min];
 	}
 	c1 = (s >> 8) & 0xff;
 	c2 = s & 0xff;
@@ -223,8 +223,14 @@ mbfl_filt_conv_wchar_cp936(int c, mbfl_convert_filter *filter TSRMLS_DC)
 		s = ucs_a3_cp936_table[c - ucs_a3_cp936_table_min];
 	} else if (c >= ucs_i_cp936_table_min && c < ucs_i_cp936_table_max) {
 		s = ucs_i_cp936_table[c - ucs_i_cp936_table_min];
-	} else if (c >= ucs_r_cp936_table_min && c < ucs_r_cp936_table_max) {
-		s = ucs_r_cp936_table[c - ucs_r_cp936_table_min];
+	} else if (c >= ucs_ci_cp936_table_min && c < ucs_ci_cp936_table_max) {
+		s = ucs_ci_cp936_table[c - ucs_ci_cp936_table_min];
+	} else if (c >= ucs_cf_cp936_table_min && c < ucs_cf_cp936_table_max) {
+		s = ucs_cf_cp936_table[c - ucs_cf_cp936_table_min];
+	} else if (c >= ucs_sfv_cp936_table_min && c < ucs_sfv_cp936_table_max) {
+		s = ucs_sfv_cp936_table[c - ucs_sfv_cp936_table_min];
+	} else if (c >= ucs_hff_cp936_table_min && c < ucs_hff_cp936_table_max) {
+		s = ucs_hff_cp936_table[c - ucs_hff_cp936_table_min];
 	}
 	if (s <= 0) {
 		c1 = c & ~MBFL_WCSPLANE_MASK;

ext/mbstring/unicode_table_cn.h

@@ -3279,32 +3279,12 @@ static const unsigned short ucs_a1_cp936_table[] = {
 0xa7da,0xa7db,0xa7dc,0xa7dd,0xa7de,0xa7df,0xa7e0,0xa7e1,
 0xa7e2,0xa7e3,0xa7e4,0xa7e5,0xa7e6,0xa7e7,0xa7e8,0xa7e9,
 0xa7ea,0xa7eb,0xa7ec,0xa7ed,0xa7ee,0xa7ef,0xa7f0,0xa7f1,
-0x0000,0xa7d7,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
-0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000
-};
+0x0000,0xa7d7};
+
 static const int ucs_a1_cp936_table_min = 0x0000;
 static const int ucs_a1_cp936_table_max = 0x0000 + (sizeof (ucs_a1_cp936_table) / sizeof (unsigned short));
 
+/* 0x2000 */
 static const unsigned short ucs_a2_cp936_table[] = {
 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
@@ -3470,6 +3450,7 @@ static const unsigned short ucs_a2_cp936_table[] = {
 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+/* 0x2500 */
 0xa9a4,0xa9a5,0xa9a6,0xa9a7,0xa9a8,0xa9a9,0xa9aa,0xa9ab,
 0xa9ac,0xa9ad,0xa9ae,0xa9af,0xa9b0,0xa9b1,0xa9b2,0xa9b3,
 0xa9b4,0xa9b5,0xa9b6,0xa9b7,0xa9b8,0xa9b9,0xa9ba,0xa9bb,
@@ -3709,7 +3690,7 @@ static const unsigned short ucs_a3_cp936_table[] = {
 static const int ucs_a3_cp936_table_min = 0x2f00;
 static const int ucs_a3_cp936_table_max = 0x2f00 + (sizeof (ucs_a3_cp936_table) / sizeof (unsigned short));
 
-/* 4d00h */
+/* 4d00h CJK Unified Ideographs (+ Extension A) */
 static const unsigned short ucs_i_cp936_table[] = {
 /* 0x4d00 */
 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
@@ -6372,8 +6353,76 @@ static const unsigned short ucs_i_cp936_table[] = {
 static const int ucs_i_cp936_table_min = 0x4d00;
 static const int ucs_i_cp936_table_max = 0x4d00 + (sizeof (ucs_i_cp936_table) / sizeof (unsigned short));
 
- /* FF00h */
-static const unsigned short ucs_r_cp936_table[] = {
+
+
+/* 0xf900 CJK Compatibility Ideographs */
+static const unsigned short ucs_ci_cp936_table[] = {
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0xfd9c,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0xfd9d,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0xfd9e,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0xfd9f,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0xfda0,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+/* 0xfa00 */
+0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
+0x0000,0x0000,0x0000,0x0000,0xfe40,0xfe41,0xfe42,0xfe43,
+0x0000,0xfe44,0x0000,0xfe45,0xfe46,0x0000,0x0000,0x0000,
+0xfe47,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0xfe48,
+0xfe49,0xfe4a,0x0000,0xfe4b,0xfe4c,0x0000,0x0000,0xfe4d,
+0xfe4e,0xfe4f,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000};
+static const int ucs_ci_cp936_table_min = 0xf900;
+static const int ucs_ci_cp936_table_max = 0xf900 + (sizeof (ucs_ci_cp936_table) / sizeof (unsigned short));
+
+
+ /* FE30h CJK Compatibility Forms  */
+static const unsigned short ucs_cf_cp936_table[] = {
+0xa955,0xa6f2,0x0000,0xa6f4,0xa6f5,0xa6e0,0xa6e1,0xa6f0,
+0xa6f1,0xa6e2,0xa6e3,0xa6ee,0xa6ef,0xa6e6,0xa6e7,0xa6e4,
+0xa6e5,0xa6e8,0xa6e9,0xa6ea,0xa6eb,0x0000,0x0000,0x0000,
+0x0000,0xa968,0xa969,0xa96a,0xa96b,0xa96c,0xa96d,0xa96e};
+
+static const int ucs_cf_cp936_table_min = 0xfe30;
+static const int ucs_cf_cp936_table_max = 0xfe30 + (sizeof (ucs_cf_cp936_table) / sizeof (unsigned short));
+
+
+ /* FE50h Small Form Variants  */
+static const unsigned short ucs_sfv_cp936_table[] = {
+0xa96f,0xa970,0xa971,0x0000,0xa972,0xa973,0xa974,0xa975,
+0x0000,0xa976,0xa977,0xa978,0xa979,0xa97a,0xa97b,0xa97c,
+0xa97d,0xa97e,0xa980,0xa981,0xa982,0xa983,0xa984,0x0000,
+0xa985,0xa986,0xa987,0xa988,0x0000,0x0000,0x0000,0x0000};
+
+static const int ucs_sfv_cp936_table_min = 0xfe50;
+static const int ucs_sfv_cp936_table_max = 0xfe50 + (sizeof (ucs_sfv_cp936_table) / sizeof (unsigned short));
+
+ /* FF00h Halfwidth and Fullwidth Forms */
+static const unsigned short ucs_hff_cp936_table[] = {
 0x0000,0xa3a1,0xa3a2,0xa3a3,0xa1e7,0xa3a5,0xa3a6,0xa3a7,
 0xa3a8,0xa3a9,0xa3aa,0xa3ab,0xa3ac,0xa3ad,0xa3ae,0xa3af,
 0xa3b0,0xa3b1,0xa3b2,0xa3b3,0xa3b4,0xa3b5,0xa3b6,0xa3b7,
@@ -6407,8 +6456,8 @@ static const unsigned short ucs_r_cp936_table[] = {
 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,
 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000
 };
-static const int ucs_r_cp936_table_min = 0xff00;
-static const int ucs_r_cp936_table_max = 0xff00 + (sizeof (ucs_r_cp936_table) / sizeof (unsigned short));
+static const int ucs_hff_cp936_table_min = 0xff00;
+static const int ucs_hff_cp936_table_max = 0xff00 + (sizeof (ucs_hff_cp936_table) / sizeof (unsigned short));