Combine uhc1_ucs_table and uhc2_ucs_table for UHC/EUC-KR/ISO-2022-KR conversion

These two tables cover contiguous ranges of the KSX 1001/KSC 5601
charset. There seems to be no reason to divide them into two tables
instead of one.

Author: alexdowad

ext/mbstring/libmbfl/filters/mbfilter_euc_kr.c

@@ -118,9 +118,9 @@ int mbfl_filt_conv_euckr_wchar(int c, mbfl_convert_filter *filter)
 		}
 		if (flag > 0 && c >= 0xa1 && c <= 0xfe) {
 			if (flag == 1) { /* 1st: 0xa1..0xc6, 2nd: 0x41..0x7a, 0x81..0xfe */
-				w = (c1 - 0xa1)*190 + c - 0x41;
-				ZEND_ASSERT(w < uhc2_ucs_table_size);
-				w = uhc2_ucs_table[w];
+				w = (c1 - 0x81)*190 + c - 0x41;
+				ZEND_ASSERT(w < uhc1_ucs_table_size);
+				w = uhc1_ucs_table[w];
 			} else { /* 1st: 0xc7..0xc8,0xca..0xfe, 2nd: 0xa1..0xfe */
 				w = (c1 - 0xc7)*94 + c - 0xa1;
 				ZEND_ASSERT(w < uhc3_ucs_table_size);
@@ -216,23 +216,25 @@ static size_t mb_euckr_to_wchar(unsigned char **in, size_t *in_len, uint32_t *bu
 			*out++ = c;
 		} else if (((c >= 0xA1 && c <= 0xAC) || (c >= 0xB0 && c <= 0xFD)) && c != 0xC9 && p < e) {
 			unsigned char c2 = *p++;
+			if (c2 < 0xA1 || c2 == 0xFF) {
+				*out++ = MBFL_BAD_INPUT;
+				continue;
+			}
 
-			if (c >= 0xA1 && c <= 0xC6 && c2 >= 0xA1 && c2 <= 0xFE) {
-				unsigned int w = (c - 0xA1)*190 + c2 - 0x41;
-				ZEND_ASSERT(w < uhc2_ucs_table_size);
-				w = uhc2_ucs_table[w];
+			if (c <= 0xC6) {
+				unsigned int w = (c - 0x81)*190 + c2 - 0x41;
+				ZEND_ASSERT(w < uhc1_ucs_table_size);
+				w = uhc1_ucs_table[w];
 				if (!w)
 					w = MBFL_BAD_INPUT;
 				*out++ = w;
-			} else if (c >= 0xC7 && c <= 0xFE && c != 0xC9 && c2 >= 0xA1 && c2 <= 0xFE) {
+			} else {
 				unsigned int w = (c - 0xC7)*94 + c2 - 0xA1;
 				ZEND_ASSERT(w < uhc3_ucs_table_size);
 				w = uhc3_ucs_table[w];
 				if (!w)
 					w = MBFL_BAD_INPUT;
 				*out++ = w;
-			} else {
-				*out++ = MBFL_BAD_INPUT;
 			}
 		} else {
 			*out++ = MBFL_BAD_INPUT;

ext/mbstring/libmbfl/filters/mbfilter_iso2022_kr.c

@@ -119,9 +119,9 @@ int mbfl_filt_conv_2022kr_wchar(int c, mbfl_convert_filter *filter)
 		if (flag > 0 && c > 0x20 && c < 0x7f) {
 			if (flag == 1) {
 				if (c1 != 0x22 || c <= 0x65) {
-					w = (c1 - 0x21)*190 + (c - 0x41) + 0x80;
-					ZEND_ASSERT(w < uhc2_ucs_table_size);
-					w = uhc2_ucs_table[w];
+					w = (c1 - 1)*190 + (c - 0x41) + 0x80;
+					ZEND_ASSERT(w < uhc1_ucs_table_size);
+					w = uhc1_ucs_table[w];
 				}
 			} else {
 				w = (c1 - 0x47)*94 + c - 0x21;
@@ -329,9 +329,9 @@ static size_t mb_iso2022kr_to_wchar(unsigned char **in, size_t *in_len, uint32_t
 
 			if (c < 0x47) {
 				if (c != 0x22 || c2 <= 0x65) {
-					w = (c - 0x21)*190 + (c2 - 0x41) + 0x80;
-					ZEND_ASSERT(w < uhc2_ucs_table_size);
-					w = uhc2_ucs_table[w];
+					w = (c - 1)*190 + c2 - 0x41 + 0x80;
+					ZEND_ASSERT(w < uhc1_ucs_table_size);
+					w = uhc1_ucs_table[w];
 				}
 			} else if (c != 0x49 && c <= 0x7D) {
 				w = (c - 0x47)*94 + c2 - 0x21;

ext/mbstring/libmbfl/filters/mbfilter_uhc.c

@@ -114,16 +114,11 @@ int mbfl_filt_conv_uhc_wchar(int c, mbfl_convert_filter *filter)
 		filter->status = 0;
 		int c1 = filter->cache, w = 0;
 
-		if (c1 >= 0x81 && c1 <= 0xa0 && c >= 0x41 && c <= 0xfe) {
+		if (c1 >= 0x81 && c1 <= 0xc6 && c >= 0x41 && c <= 0xfe) {
 			w = (c1 - 0x81)*190 + (c - 0x41);
 			if (w >= 0 && w < uhc1_ucs_table_size) {
 				w = uhc1_ucs_table[w];
 			}
-		} else if (c1 >= 0xa1 && c1 <= 0xc6 && c >= 0x41 && c <= 0xfe) {
-			w = (c1 - 0xa1)*190 + (c - 0x41);
-			if (w >= 0 && w < uhc2_ucs_table_size) {
-				w = uhc2_ucs_table[w];
-			}
 		} else if (c1 >= 0xc7 && c1 < 0xfe && c >= 0xa1 && c <= 0xfe) {
 			w = (c1 - 0xc7)*94 + (c - 0xa1);
 			if (w >= 0 && w < uhc3_ucs_table_size) {
@@ -214,21 +209,14 @@ static size_t mb_uhc_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf,
 			}
 			unsigned int w = 0;
 
-			if (c <= 0xA0) {
+			if (c <= 0xC6) {
 				w = (c - 0x81)*190 + c2 - 0x41;
-				if (w < uhc1_ucs_table_size) {
-					w = uhc1_ucs_table[w];
-				}
-			} else if (c <= 0xC6) {
-				w = (c - 0xA1)*190 + c2 - 0x41;
-				if (w < uhc2_ucs_table_size) {
-					w = uhc2_ucs_table[w];
-				}
+				ZEND_ASSERT(w < uhc1_ucs_table_size);
+				w = uhc1_ucs_table[w];
 			} else if (c2 >= 0xA1) {
 				w = (c - 0xC7)*94 + c2 - 0xA1;
-				if (w < uhc3_ucs_table_size) {
-					w = uhc3_ucs_table[w];
-				}
+				ZEND_ASSERT(w < uhc3_ucs_table_size);
+				w = uhc3_ucs_table[w];
 			}
 			if (!w) {
 				w = MBFL_BAD_INPUT;

ext/mbstring/libmbfl/filters/unicode_table_uhc.h

@@ -790,12 +790,8 @@ const unsigned short uhc1_ucs_table[] = {
 0xc876,0xc877,0xc879,0xc87b,0xc87c,0xc87d,0xc87e,0xc87f,
 0xc882,0xc884,0xc888,0xc889,0xc88a,0xc88e,0xc88f,0xc890,
 0xc891,0xc892,0xc893,0xc895,0xc896,0xc897,0xc898,0xc899,
-0xc89a,0xc89b,0xc89c,0xc89e,0xc8a0,0xc8a2,0xc8a3,0xc8a4
-};
-
-const int uhc1_ucs_table_size = (sizeof(uhc1_ucs_table)/sizeof(unsigned short));
+0xc89a,0xc89b,0xc89c,0xc89e,0xc8a0,0xc8a2,0xc8a3,0xc8a4,
 
-const unsigned short uhc2_ucs_table[] = {
 0xc8a5,0xc8a6,0xc8a7,0xc8a9,0xc8aa,0xc8ab,0xc8ac,0xc8ad,
 0xc8ae,0xc8af,0xc8b0,0xc8b1,0xc8b2,0xc8b3,0xc8b4,0xc8b5,
 0xc8b6,0xc8b7,0xc8b8,0xc8b9,0xc8ba,0xc8bb,0xc8be,0xc8bf,
@@ -1698,9 +1694,10 @@ const unsigned short uhc2_ucs_table[] = {
 0xd391,0xd398,0xd399,0xd39c,0xd3a0,0xd3a8,0xd3a9,0xd3ab,
 0xd3ad,0xd3b4,0xd3b8,0xd3bc,0xd3c4,0xd3c5,0xd3c8,0xd3c9,
 0xd3d0,0xd3d8,0xd3e1,0xd3e3,0xd3ec,0xd3ed,0xd3f0,0xd3f4,
-0xd3fc,0xd3fd,0xd3ff,0xd401};
+0xd3fc,0xd3fd,0xd3ff,0xd401
+};
 
-const int uhc2_ucs_table_size = (sizeof(uhc2_ucs_table)/sizeof(unsigned short));
+const int uhc1_ucs_table_size = (sizeof(uhc1_ucs_table)/sizeof(unsigned short));
 
 const unsigned short uhc3_ucs_table[] = {
 0xd408,0xd41d,0xd440,0xd444,0xd45c,0xd460,0xd464,0xd46d,