diff options
Diffstat (limited to 'libwc')
| -rw-r--r-- | libwc/ambwidth_map.awk | 32 | ||||
| -rw-r--r-- | libwc/charset.c | 3 | ||||
| -rw-r--r-- | libwc/gb18030.c | 1 | ||||
| -rw-r--r-- | libwc/iso2022.c | 23 | ||||
| -rw-r--r-- | libwc/johab.c | 6 | ||||
| -rw-r--r-- | libwc/map/big5_ucs.map | 4 | ||||
| -rw-r--r-- | libwc/map/cns11643_ucs.map | 8 | ||||
| -rw-r--r-- | libwc/map/gb12345_ucs.map | 4 | ||||
| -rw-r--r-- | libwc/map/gb2312_ucs.map | 4 | ||||
| -rw-r--r-- | libwc/map/gbk_ucs.map | 4 | ||||
| -rw-r--r-- | libwc/map/hkscs_ucs.map | 4 | ||||
| -rw-r--r-- | libwc/map/jisx0208x0212x0213_ucs.map | 8 | ||||
| -rw-r--r-- | libwc/map/ksx1001_ucs.map | 4 | ||||
| -rw-r--r-- | libwc/map/sjis_ext_ucs.map | 4 | ||||
| -rw-r--r-- | libwc/map/ucs_ambwidth.map | 109 | ||||
| -rw-r--r-- | libwc/map/uhc_ucs.map | 4 | ||||
| -rw-r--r-- | libwc/status.c | 3 | ||||
| -rw-r--r-- | libwc/ucs.c | 8 | ||||
| -rw-r--r-- | libwc/ucs.map | 100 | ||||
| -rw-r--r-- | libwc/wtf.c | 67 | ||||
| -rw-r--r-- | libwc/wtf.h | 3 | 
21 files changed, 305 insertions, 98 deletions
| diff --git a/libwc/ambwidth_map.awk b/libwc/ambwidth_map.awk index 8544f58..1d9d25f 100644 --- a/libwc/ambwidth_map.awk +++ b/libwc/ambwidth_map.awk @@ -3,9 +3,15 @@ BEGIN {      i = 0;  }  $2 == "A" {  -    code = sprintf("0x%s", $1); -    if (strtonum(code) < 0x10000) { -	map[i] = code +    code = code2 = strtonum(sprintf("0x%s", $1)) +    if (match($1, /[.]+[0-9A-Fa-f]+/)) { +	s = substr($1, RSTART, RLENGTH) +	sub(/[.]+/, "0x", s) +	code2 = strtonum(s) +    } +    for (; code <= code2; code++) { +	if (code >= 0x10000) { break } +	map[i] = sprintf("0x%04X", code)  	i++;      }  } @@ -15,28 +21,14 @@ END {      prev = strtonum(map[0]);      for (j = 1; j < i; j++) {  	cur = strtonum(map[j]); -	if (match(map[j], "[.]+")) { +	if (cur - prev > 1) {  	    map2[n] = sprintf("%s, %s", start, map[j - 1]);  	    n++; -	    gsub("[.]+", ", 0x", map[j]) -	    map2[n] = map[j]; -	    n++; -	    start = map[j + 1]; -	    cur = strtonum(start); -	} else { -	    if (cur - prev > 2) { -		map2[n] = sprintf("%s, %s", start, map[j - 1]); -		start = map[j]; -		n++; -	    } - -	    if (j == i - 1) { -		map2[n] = sprintf("%s, %s", start, map[j]); -		n++; -	    } +	    start = map[j];  	}  	prev = cur;      } +    if (i > 0) { map2[n] = sprintf("%s, %s", start, map[i - 1]); n++ }      printf("static wc_map ucs_ambwidth_map[] = {\n");      for (j = 0; j < n; j++) { diff --git a/libwc/charset.c b/libwc/charset.c index 3f0b74d..ea79b1c 100644 --- a/libwc/charset.c +++ b/libwc/charset.c @@ -1,8 +1,7 @@  #include <stdlib.h>  #include <ctype.h> -#include <gc.h> -#define New_N(type,n) ((type*)GC_MALLOC((n)*sizeof(type))) +#include "../alloc.h"  #include "wc.h" diff --git a/libwc/gb18030.c b/libwc/gb18030.c index c195d49..d5c9018 100644 --- a/libwc/gb18030.c +++ b/libwc/gb18030.c @@ -151,6 +151,7 @@ wc_ucs_to_gb18030(wc_uint32 ucs)  	return cc;      }      cc.ccs = WC_CCS_UNKNOWN; +    cc.code = 0;      return cc;  }  #endif diff --git a/libwc/iso2022.c b/libwc/iso2022.c index 33d9a19..a191f28 100644 --- a/libwc/iso2022.c +++ b/libwc/iso2022.c @@ -405,7 +405,8 @@ wc_push_to_iso2022(Str os, wc_wchar_t cc, wc_status *st)      case WC_CCS_A_CS94:  	if (cc.ccs == WC_CCS_US_ASCII)  	    cc.ccs = st->g0_ccs; -	g = cs94_gmap[WC_CCS_INDEX(cc.ccs) - WC_F_ISO_BASE]; +	if (WC_CCS_INDEX(cc.ccs) >= WC_F_ISO_BASE) +	    g = cs94_gmap[WC_CCS_INDEX(cc.ccs) - WC_F_ISO_BASE];  	break;      case WC_CCS_A_CS94W:  	is_wide = 1; @@ -435,35 +436,41 @@ wc_push_to_iso2022(Str os, wc_wchar_t cc, wc_status *st)  	    break;  #endif  	} -	g = cs94w_gmap[WC_CCS_INDEX(cc.ccs) - WC_F_ISO_BASE]; +	if (WC_CCS_INDEX(cc.ccs) >= WC_F_ISO_BASE) +	    g = cs94w_gmap[WC_CCS_INDEX(cc.ccs) - WC_F_ISO_BASE];  	break;      case WC_CCS_A_CS96: -	g = cs96_gmap[WC_CCS_INDEX(cc.ccs) - WC_F_ISO_BASE]; +	if (WC_CCS_INDEX(cc.ccs) >= WC_F_ISO_BASE) +	    g = cs96_gmap[WC_CCS_INDEX(cc.ccs) - WC_F_ISO_BASE];  	break;      case WC_CCS_A_CS96W:  	is_wide = 1; -	g = cs96w_gmap[WC_CCS_INDEX(cc.ccs) - WC_F_ISO_BASE]; +	if (WC_CCS_INDEX(cc.ccs) >= WC_F_ISO_BASE) +	    g = cs96w_gmap[WC_CCS_INDEX(cc.ccs) - WC_F_ISO_BASE];  	break;      case WC_CCS_A_CS942: -	g = cs942_gmap[WC_CCS_INDEX(cc.ccs) - WC_F_ISO_BASE]; +	if (WC_CCS_INDEX(cc.ccs) >= WC_F_ISO_BASE) +	    g = cs942_gmap[WC_CCS_INDEX(cc.ccs) - WC_F_ISO_BASE];  	break;      case WC_CCS_A_UNKNOWN_W:  	if (WcOption.no_replace)  	    return;  	is_wide = 1;  	cc.ccs = WC_CCS_US_ASCII; -	g = cs94_gmap[WC_CCS_INDEX(cc.ccs) - WC_F_ISO_BASE]; +	if (WC_CCS_INDEX(cc.ccs) >= WC_F_ISO_BASE) +	    g = cs94_gmap[WC_CCS_INDEX(cc.ccs) - WC_F_ISO_BASE];  	cc.code = ((wc_uint32)WC_REPLACE_W[0] << 8) | WC_REPLACE_W[1];  	break;      case WC_CCS_A_UNKNOWN:  	if (WcOption.no_replace)  	    return;  	cc.ccs = WC_CCS_US_ASCII; -	g = cs94_gmap[WC_CCS_INDEX(cc.ccs) - WC_F_ISO_BASE]; +	if (WC_CCS_INDEX(cc.ccs) >= WC_F_ISO_BASE) +	    g = cs94_gmap[WC_CCS_INDEX(cc.ccs) - WC_F_ISO_BASE];  	cc.code = (wc_uint32)WC_REPLACE[0];  	break;      default: -	if ((cc.ccs == WC_CCS_JOHAB || WC_CCS_JOHAB_1 || +	if ((cc.ccs == WC_CCS_JOHAB || cc.ccs == WC_CCS_JOHAB_1 ||  		cc.ccs == WC_CCS_JOHAB_2 || cc.ccs == WC_CCS_JOHAB_3) &&  		cs94w_gmap[WC_F_KS_X_1001 - WC_F_ISO_BASE]) {  	    wc_wchar_t cc2 = wc_johab_to_ksx1001(cc); diff --git a/libwc/johab.c b/libwc/johab.c index 8d587b8..498b1bb 100644 --- a/libwc/johab.c +++ b/libwc/johab.c @@ -160,9 +160,9 @@ wc_N_to_johab1(wc_uint32 code)  {      wc_uint32 a, b, c; -    a = N_johab1_map[0][(code / 28) / 21]; -    b = N_johab1_map[1][(code / 28) % 21]; -    c = N_johab1_map[2][ code % 28      ]; +    a = N_johab1_map[0][(code / 28) / 21 & 0x1F]; +    b = N_johab1_map[1][(code / 28) % 21 & 0x1F]; +    c = N_johab1_map[2][ code % 28       & 0x1F];      return 0x8000 | (a << 10) | (b << 5) | c;  } diff --git a/libwc/map/big5_ucs.map b/libwc/map/big5_ucs.map index 0c6fd12..ac817a9 100644 --- a/libwc/map/big5_ucs.map +++ b/libwc/map/big5_ucs.map @@ -1,6 +1,8 @@  /* Big5 (Chinese Taiwan) */ -static wc_uint16 big5_ucs_map[ 0x59 * 0x9D ] = { +#define N_big5_ucs_map (0x59 * 0x9D) + +static wc_uint16 big5_ucs_map[ N_big5_ucs_map ] = {   0x3000,	/* 0xA140 */   0xFF0C,	/* 0xA141 */   0x3001,	/* 0xA142 */ diff --git a/libwc/map/cns11643_ucs.map b/libwc/map/cns11643_ucs.map index b426dd3..fcba334 100644 --- a/libwc/map/cns11643_ucs.map +++ b/libwc/map/cns11643_ucs.map @@ -1,6 +1,8 @@  /* CNS 11643 (Chinese Taiwan) */ -static wc_uint16 cns116431_ucs_map[ 0x5E * 0x5E ] = { +#define N_cns116431_ucs_map (0x5E * 0x5E) + +static wc_uint16 cns116431_ucs_map[ N_cns116431_ucs_map ] = {   0x3000,	/* 0x2121 */   0xFF0C,	/* 0x2122 */   0x3001,	/* 0x2123 */ @@ -8839,7 +8841,9 @@ static wc_uint16 cns116431_ucs_map[ 0x5E * 0x5E ] = {   0,		/* 0x7E7E */  }; -static wc_uint16 cns116432_ucs_map[ 0x5E * 0x5E ] = { +#define N_cns116432_ucs_map (0x5E * 0x5E) + +static wc_uint16 cns116432_ucs_map[ N_cns116432_ucs_map ] = {   0x4E42,	/* 0x2121 */   0x4E5C,	/* 0x2122 */   0x51F5,	/* 0x2123 */ diff --git a/libwc/map/gb12345_ucs.map b/libwc/map/gb12345_ucs.map index 55558c7..3fb338d 100644 --- a/libwc/map/gb12345_ucs.map +++ b/libwc/map/gb12345_ucs.map @@ -1,6 +1,8 @@  /* GB 12345 (Chinese) */ -static wc_uint16 gb12345_ucs_map[ 0x5E * 0x5E ] = { +#define N_gb12345_ucs_map (0x5E * 0x5E) + +static wc_uint16 gb12345_ucs_map[ N_gb12345_ucs_map ] = {   0x3000,	/* 0x2121 */   0x3001,	/* 0x2122 */   0x3002,	/* 0x2123 */ diff --git a/libwc/map/gb2312_ucs.map b/libwc/map/gb2312_ucs.map index 38fb88f..3d37465 100644 --- a/libwc/map/gb2312_ucs.map +++ b/libwc/map/gb2312_ucs.map @@ -1,6 +1,8 @@  /* GB 2312 (Chinese) */ -static wc_uint16 gb2312_ucs_map[ 0x5E * 0x5E ] = { +#define N_gb2312_ucs_map (0x5E * 0x5E) + +static wc_uint16 gb2312_ucs_map[ N_gb2312_ucs_map ] = {   0x3000,	/* 0x2121 */   0x3001,	/* 0x2122 */   0x3002,	/* 0x2123 */ diff --git a/libwc/map/gbk_ucs.map b/libwc/map/gbk_ucs.map index 5a0d5ba..d092fd7 100644 --- a/libwc/map/gbk_ucs.map +++ b/libwc/map/gbk_ucs.map @@ -6,7 +6,9 @@ static wc_map ucs_gbk_80_map[ N_ucs_gbk_80_map ] = {    { 0x20AC, 0x0080 },  }; -static wc_uint16 gbk_ucs_map[ 0x7E * 0xBE - 0x5E * 0x5E + 0x0A + 0x16 + 0x06 ] = { +#define N_gbk_ucs_map (0x7E * 0xBE - 0x5E * 0x5E + 0x0A + 0x16 + 0x06) + +static wc_uint16 gbk_ucs_map[ N_gbk_ucs_map ] = {    0x4E02,	/* 0x8140 */    0x4E04,	/* 0x8141 */    0x4E05,	/* 0x8142 */ diff --git a/libwc/map/hkscs_ucs.map b/libwc/map/hkscs_ucs.map index 96d1566..2fbe6b4 100644 --- a/libwc/map/hkscs_ucs.map +++ b/libwc/map/hkscs_ucs.map @@ -1,6 +1,8 @@  /* HKSCS (Chinese Hong Kong) */ -static wc_uint16 hkscs_ucs_map[ 0x1E * 0x9D ] = { +#define N_hkscs_ucs_map (0x1E * 0x9D) + +static wc_uint16 hkscs_ucs_map[ N_hkscs_ucs_map ] = {   0,		/* 0x8840 */   0,		/* 0x8841 */   0,		/* 0x8842 */ diff --git a/libwc/map/jisx0208x0212x0213_ucs.map b/libwc/map/jisx0208x0212x0213_ucs.map index 1a1d706..28c2a6c 100644 --- a/libwc/map/jisx0208x0212x0213_ucs.map +++ b/libwc/map/jisx0208x0212x0213_ucs.map @@ -1,6 +1,8 @@  /* JIS X 0208, JIS X 0212, JIS X 0213 (Japanese) */ -static wc_uint16 jisx0208x02131_ucs_map[ 0x5E * 0x5E ] = { +#define N_jisx0208x02131_ucs_map (0x5E * 0x5E) + +static wc_uint16 jisx0208x02131_ucs_map[ N_jisx0208x02131_ucs_map ] = {   0x3000,	/* JIS X 0208 0x2121 */   0x3001,	/* JIS X 0208 0x2122 */   0x3002,	/* JIS X 0208 0x2123 */ @@ -8839,7 +8841,9 @@ static wc_uint16 jisx0208x02131_ucs_map[ 0x5E * 0x5E ] = {   0,		/* JIS X 0213-1 0x7E7E */  }; -static wc_uint16 jisx0212x02132_ucs_map[ 0x5E * 0x5E ] = { +#define N_jisx0212x02132_ucs_map (0x5E * 0x5E) + +static wc_uint16 jisx0212x02132_ucs_map[ N_jisx0212x02132_ucs_map ] = {   0,		/* JIS X 0213-2 0x2121 */   0x4E02,	/* JIS X 0213-2 0x2122 */   0x4E0F,	/* JIS X 0213-2 0x2123 */ diff --git a/libwc/map/ksx1001_ucs.map b/libwc/map/ksx1001_ucs.map index 9a17d61..cb62f98 100644 --- a/libwc/map/ksx1001_ucs.map +++ b/libwc/map/ksx1001_ucs.map @@ -1,6 +1,8 @@  /* KS X 1001 (Korean) */ -static wc_uint16 ksx1001_ucs_map[ 0x5E * 0x5E ] = { +#define N_ksx1001_ucs_map (0x5E * 0x5E) + +static wc_uint16 ksx1001_ucs_map[ N_ksx1001_ucs_map ] = {   0x3000,	/* 0x2121 */   0x3001,	/* 0x2122 */   0x3002,	/* 0x2123 */ diff --git a/libwc/map/sjis_ext_ucs.map b/libwc/map/sjis_ext_ucs.map index a82995c..cc748ba 100644 --- a/libwc/map/sjis_ext_ucs.map +++ b/libwc/map/sjis_ext_ucs.map @@ -1,6 +1,8 @@  /* Shift_JIS/CP932 (Japanese) */ -static wc_uint16 sjis_ext_ucs_map[ 0x5E * 10 ] = { +#define N_sjis_ext_ucs_map (0x5E * 10) + +static wc_uint16 sjis_ext_ucs_map[ N_sjis_ext_ucs_map ] = {    0x2460,	/* 0x8740 */    0x2461,	/* 0x8741 */    0x2462,	/* 0x8742 */ diff --git a/libwc/map/ucs_ambwidth.map b/libwc/map/ucs_ambwidth.map index 6f03ba8..35ceedb 100644 --- a/libwc/map/ucs_ambwidth.map +++ b/libwc/map/ucs_ambwidth.map @@ -1,50 +1,82 @@  static wc_map ucs_ambwidth_map[] = {     { 0x00A1, 0x00A1 },     { 0x00A4, 0x00A4 }, -   { 0x00A7, 0x00AA }, -   { 0x00AD, 0x00BF }, +   { 0x00A7, 0x00A8 }, +   { 0x00AA, 0x00AA }, +   { 0x00AD, 0x00AE }, +   { 0x00B0, 0x00B4 }, +   { 0x00B6, 0x00BA }, +   { 0x00BC, 0x00BF },     { 0x00C6, 0x00C6 },     { 0x00D0, 0x00D0 },     { 0x00D7, 0x00D8 },     { 0x00DE, 0x00E1 }, -   { 0x00E6, 0x00ED }, -   { 0x00F0, 0x00F3 }, -   { 0x00F7, 0x00FE }, +   { 0x00E6, 0x00E6 }, +   { 0x00E8, 0x00EA }, +   { 0x00EC, 0x00ED }, +   { 0x00F0, 0x00F0 }, +   { 0x00F2, 0x00F3 }, +   { 0x00F7, 0x00FA }, +   { 0x00FC, 0x00FC }, +   { 0x00FE, 0x00FE },     { 0x0101, 0x0101 }, -   { 0x0111, 0x0113 }, +   { 0x0111, 0x0111 }, +   { 0x0113, 0x0113 },     { 0x011B, 0x011B },     { 0x0126, 0x0127 },     { 0x012B, 0x012B },     { 0x0131, 0x0133 },     { 0x0138, 0x0138 }, -   { 0x013F, 0x0144 }, -   { 0x0148, 0x014D }, +   { 0x013F, 0x0142 }, +   { 0x0144, 0x0144 }, +   { 0x0148, 0x014B }, +   { 0x014D, 0x014D },     { 0x0152, 0x0153 },     { 0x0166, 0x0167 },     { 0x016B, 0x016B }, -   { 0x01CE, 0x01DC }, +   { 0x01CE, 0x01CE }, +   { 0x01D0, 0x01D0 }, +   { 0x01D2, 0x01D2 }, +   { 0x01D4, 0x01D4 }, +   { 0x01D6, 0x01D6 }, +   { 0x01D8, 0x01D8 }, +   { 0x01DA, 0x01DA }, +   { 0x01DC, 0x01DC },     { 0x0251, 0x0251 },     { 0x0261, 0x0261 },     { 0x02C4, 0x02C4 }, -   { 0x02C7, 0x02CD }, +   { 0x02C7, 0x02C7 }, +   { 0x02C9, 0x02CB }, +   { 0x02CD, 0x02CD },     { 0x02D0, 0x02D0 }, -   { 0x02D8, 0x02DF }, +   { 0x02D8, 0x02DB }, +   { 0x02DD, 0x02DD }, +   { 0x02DF, 0x02DF },     { 0x0300, 0x036F }, -   { 0x0391, 0x03A9 }, -   { 0x03B1, 0x03C9 }, +   { 0x0391, 0x03A1 }, +   { 0x03A3, 0x03A9 }, +   { 0x03B1, 0x03C1 }, +   { 0x03C3, 0x03C9 },     { 0x0401, 0x0401 }, -   { 0x0410, 0x0451 }, +   { 0x0410, 0x044F }, +   { 0x0451, 0x0451 },     { 0x2010, 0x2010 }, -   { 0x2013, 0x2019 }, +   { 0x2013, 0x2016 }, +   { 0x2018, 0x2019 },     { 0x201C, 0x201D }, -   { 0x2020, 0x2027 }, -   { 0x2030, 0x2035 }, +   { 0x2020, 0x2022 }, +   { 0x2024, 0x2027 }, +   { 0x2030, 0x2030 }, +   { 0x2032, 0x2033 }, +   { 0x2035, 0x2035 },     { 0x203B, 0x203B },     { 0x203E, 0x203E },     { 0x2074, 0x2074 }, -   { 0x207F, 0x2084 }, +   { 0x207F, 0x207F }, +   { 0x2081, 0x2084 },     { 0x20AC, 0x20AC }, -   { 0x2103, 0x2105 }, +   { 0x2103, 0x2103 }, +   { 0x2105, 0x2105 },     { 0x2109, 0x2109 },     { 0x2113, 0x2113 },     { 0x2116, 0x2116 }, @@ -52,21 +84,28 @@ static wc_map ucs_ambwidth_map[] = {     { 0x2126, 0x2126 },     { 0x212B, 0x212B },     { 0x2153, 0x2154 }, -   { 0x215B, 0x216B }, +   { 0x215B, 0x215E }, +   { 0x2160, 0x216B },     { 0x2170, 0x2179 },     { 0x2189, 0x2189 },     { 0x2190, 0x2199 },     { 0x21B8, 0x21B9 }, -   { 0x21D2, 0x21D4 }, +   { 0x21D2, 0x21D2 }, +   { 0x21D4, 0x21D4 },     { 0x21E7, 0x21E7 }, -   { 0x2200, 0x2203 }, +   { 0x2200, 0x2200 }, +   { 0x2202, 0x2203 },     { 0x2207, 0x2208 },     { 0x220B, 0x220B }, -   { 0x220F, 0x2211 }, +   { 0x220F, 0x220F }, +   { 0x2211, 0x2211 },     { 0x2215, 0x2215 },     { 0x221A, 0x221A },     { 0x221D, 0x2220 }, -   { 0x2223, 0x222E }, +   { 0x2223, 0x2223 }, +   { 0x2225, 0x2225 }, +   { 0x2227, 0x222C }, +   { 0x222E, 0x222E },     { 0x2234, 0x2237 },     { 0x223C, 0x223D },     { 0x2248, 0x2248 }, @@ -83,11 +122,13 @@ static wc_map ucs_ambwidth_map[] = {     { 0x22A5, 0x22A5 },     { 0x22BF, 0x22BF },     { 0x2312, 0x2312 }, -   { 0x2460, 0x254B }, +   { 0x2460, 0x24E9 }, +   { 0x24EB, 0x254B },     { 0x2550, 0x2573 },     { 0x2580, 0x258F },     { 0x2592, 0x2595 }, -   { 0x25A0, 0x25A9 }, +   { 0x25A0, 0x25A1 }, +   { 0x25A3, 0x25A9 },     { 0x25B2, 0x25B3 },     { 0x25B6, 0x25B7 },     { 0x25BC, 0x25BD }, @@ -101,12 +142,20 @@ static wc_map ucs_ambwidth_map[] = {     { 0x2609, 0x2609 },     { 0x260E, 0x260F },     { 0x2614, 0x2615 }, -   { 0x261C, 0x261E }, -   { 0x2640, 0x2642 }, -   { 0x2660, 0x266F }, +   { 0x261C, 0x261C }, +   { 0x261E, 0x261E }, +   { 0x2640, 0x2640 }, +   { 0x2642, 0x2642 }, +   { 0x2660, 0x2661 }, +   { 0x2663, 0x2665 }, +   { 0x2667, 0x266A }, +   { 0x266C, 0x266D }, +   { 0x266F, 0x266F },     { 0x269E, 0x269F },     { 0x26BE, 0x26BF }, -   { 0x26C4, 0x26E3 }, +   { 0x26C4, 0x26CD }, +   { 0x26CF, 0x26E1 }, +   { 0x26E3, 0x26E3 },     { 0x26E8, 0x26FF },     { 0x273D, 0x273D },     { 0x2757, 0x2757 }, diff --git a/libwc/map/uhc_ucs.map b/libwc/map/uhc_ucs.map index b6b43ca..55efc09 100644 --- a/libwc/map/uhc_ucs.map +++ b/libwc/map/uhc_ucs.map @@ -1,6 +1,8 @@  /* UHC/CP949 (Korean) */ -static wc_uint16 uhc_ucs_map[ 0x20 * 0xB2 + 0x27 * 0x54 + 2 ] = { +#define N_uhc_ucs_map (0x20 * 0xB2 + 0x27 * 0x54 + 2) + +static wc_uint16 uhc_ucs_map[ N_uhc_ucs_map ] = {    0xAC02,	/* 0x8141 */    0xAC03,	/* 0x8142 */    0xAC05,	/* 0x8143 */ diff --git a/libwc/status.c b/libwc/status.c index d25c924..4a2ebf8 100644 --- a/libwc/status.c +++ b/libwc/status.c @@ -1,7 +1,6 @@  #include <string.h> -#include <gc.h> -#define New_N(type,n) ((type*)GC_MALLOC((n)*sizeof(type))) +#include "../alloc.h"  #include "wc.h"  #ifdef USE_UNICODE diff --git a/libwc/ucs.c b/libwc/ucs.c index d7b6948..18c3a67 100644 --- a/libwc/ucs.c +++ b/libwc/ucs.c @@ -100,6 +100,7 @@ wc_ucs_to_any(wc_uint32 ucs, wc_table *t)  	    return t->conv(t->ccs, map->code2);      }      cc.ccs = WC_CCS_UNKNOWN; +    cc.code = 0;      return cc;  } @@ -108,6 +109,7 @@ wc_any_to_ucs(wc_wchar_t cc)  {      int f;      wc_uint16 *map = NULL; +    wc_uint32 map_size = 0x80;      wc_map *map2;      f = WC_CCS_INDEX(cc.ccs); @@ -138,6 +140,7 @@ wc_any_to_ucs(wc_wchar_t cc)  	if (f < WC_F_ISO_BASE || f > WC_F_CS94W_END)  	    return 0;  	map = cs94w_ucs_map[f - WC_F_ISO_BASE]; +	map_size = cs94w_ucs_map_size[f - WC_F_ISO_BASE];  	cc.code = WC_CS94W_N(cc.code);  	break;      case WC_CCS_A_CS96: @@ -150,6 +153,7 @@ wc_any_to_ucs(wc_wchar_t cc)  	if (f < WC_F_ISO_BASE || f > WC_F_CS96W_END)  	    return WC_C_UCS4_ERROR;  	map = cs96w_ucs_map[f - WC_F_ISO_BASE]; +	map_size = cs96w_ucs_map_size[f - WC_F_ISO_BASE];  	cc.code = WC_CS96W_N(cc.code);  	break;      case WC_CCS_A_CS942: @@ -174,12 +178,14 @@ wc_any_to_ucs(wc_wchar_t cc)  	    return WC_C_UCS2_EURO;  	}  	map = pcs_ucs_map[f - WC_F_PCS_BASE]; +	map_size = pcs_ucs_map_size[f - WC_F_PCS_BASE];  	cc.code &= 0x7f;  	break;      case WC_CCS_A_PCSW:  	if (f < WC_F_PCS_BASE || f > WC_F_PCSW_END)  	    return WC_C_UCS4_ERROR;  	map = pcsw_ucs_map[f - WC_F_PCS_BASE]; +	map_size = pcsw_ucs_map_size[f - WC_F_PCS_BASE];  	switch (cc.ccs) {  	case WC_CCS_BIG5:  	    cc.code = WC_BIG5_N(cc.code); @@ -271,6 +277,8 @@ wc_any_to_ucs(wc_wchar_t cc)      }      if (map == NULL)  	return WC_C_UCS4_ERROR; +    if (map_size == 0 || cc.code > map_size - 1) +	return WC_C_UCS4_ERROR;      cc.code = map[cc.code];      return cc.code ? cc.code : WC_C_UCS4_ERROR;  } diff --git a/libwc/ucs.map b/libwc/ucs.map index dfac6d9..bed5dff 100644 --- a/libwc/ucs.map +++ b/libwc/ucs.map @@ -175,6 +175,49 @@ static wc_uint16 *pcs_ucs_map[] = {    NULL,			/* (Raw) */  }; +static wc_uint32 pcs_ucs_map_size[] = { +  0x80,			/* cp437_ucs_map */ +  0x80,			/* cp737_ucs_map */ +  0x80,			/* cp775_ucs_map */ +  0x80,			/* cp850_ucs_map */ +  0x80,			/* cp852_ucs_map */ +  0x80,			/* cp855_ucs_map */ +  0x80,			/* cp856_ucs_map */ +  0x80,			/* cp857_ucs_map */ +  0x80,			/* cp860_ucs_map */ +  0x80,			/* cp861_ucs_map */ +  0x80,			/* cp862_ucs_map */ +  0x80,			/* cp863_ucs_map */ +  0x80,			/* cp864_ucs_map */ +  0x80,			/* cp865_ucs_map */ +  0x80,			/* cp866_ucs_map */ +  0x80,			/* cp869_ucs_map */ +  0x80,			/* cp874_ucs_map */ +  0x80,			/* cp1006_ucs_map */ +  0x80,			/* cp1250_ucs_map */ +  0x80,			/* cp1251_ucs_map */ +  0x80,			/* cp1252_ucs_map */ +  0x80,			/* cp1253_ucs_map */ +  0x80,			/* cp1254_ucs_map */ +  0x80,			/* cp1255_ucs_map */ +  0x80,			/* cp1256_ucs_map */ +  0x80,			/* cp1257_ucs_map */ +  0x80,			/* cp1258_ucs_map */ +  0,			/* NULL */ +  0x80,			/* tcvn57121_ucs_map */ +  0x20,			/* tcvn57122_ucs_map */ +  0,			/* NULL */ +  0x80,			/* viscii111_ucs_map */ +  0x20,			/* viscii112_ucs_map */ +  0x80,			/* vps1_ucs_map */ +  0x20,			/* vps2_ucs_map */ +  0x80,			/* koi8r_ucs_map */ +  0x80,			/* koi8u_ucs_map */ +  0x80,			/* nextstep_ucs_map */ +  0,			/* NULL */ +  0,			/* NULL (Raw) */ +}; +  static wc_uint16 *cs94w_ucs_map[] = {    jisx0208x02131_ucs_map,	/* 40 (JIS C 6226) */    gb2312_ucs_map,		/* 41 (GB 2312) */ @@ -195,7 +238,28 @@ static wc_uint16 *cs94w_ucs_map[] = {    jisx0212x02132_ucs_map,	/* 50 (JIS X 0213-2) */  }; +static wc_uint32 cs94w_ucs_map_size[] = { +  N_jisx0208x02131_ucs_map,	/* 40 (JIS C 6226) */ +  N_gb2312_ucs_map,		/* 41 (GB 2312) */ +  N_jisx0208x02131_ucs_map,	/* 42 (JIS X 0208) */ +  N_ksx1001_ucs_map,		/* 43 (KS X 1001) */ +  N_jisx0212x02132_ucs_map,	/* 44 (JIS X 0212) */ +  0,				/* 45 (ISO IR 165) */ +  0,				/* 46 */ +  N_cns116431_ucs_map,		/* 47 (CNS 11643-1) */ +  N_cns116432_ucs_map,		/* 48 (CNS 11643-2) */ +  0,				/* 49 (CNS 11643-3) */ +  0,				/* 4A (CNS 11643-4) */ +  0,				/* 4B (CNS 11643-5) */ +  0,				/* 4C (CNS 11643-6) */ +  0,				/* 4D (CNS 11643-7) */ +  0,				/* 4E (KSP 9566) */ +  N_jisx0208x02131_ucs_map,	/* 4F (JIS X 0213-1) */ +  N_jisx0212x02132_ucs_map,	/* 50 (JIS X 0213-2) */ +}; +  static wc_uint16 **cs96w_ucs_map; +static wc_uint32 *cs96w_ucs_map_size;  static wc_uint16 *pcsw_ucs_map[] = {    big5_ucs_map,		/* Big5 */ @@ -233,6 +297,42 @@ static wc_uint16 *pcsw_ucs_map[] = {    hkscs_ucs_map,	/* HKSCS-2 */  }; +static wc_uint32 pcsw_ucs_map_size[] = { +  N_big5_ucs_map,	/* Big5 */ +  N_big5_ucs_map,	/* Big5-1 */ +  N_big5_ucs_map,	/* Big5-2 */ +  0,			/* CNS 11643-8 */ +  0,			/* CNS 11643-9 */ +  0,			/* CNS 11643-10 */ +  0,			/* CNS 11643-11 */ +  0,			/* CNS 11643-12 */ +  0,			/* CNS 11643-13 */ +  0,			/* CNS 11643-14 */ +  0,			/* CNS 11643-15 */ +  0,			/* CNS 11643-16 */ +  0,			/* CNS 11643-X */ +  N_gb12345_ucs_map,	/* GB 12345 */ +  0,			/* Johab (special conversion) */ +  0,			/* Johab-1 (special conversion) */ +  0,			/* Johab-2 (special conversion) */ +  N_ksx1001_ucs_map,	/* Johab-3 */ +  N_sjis_ext_ucs_map,	/* Shift_JIS(CP932) ext */ +  N_sjis_ext_ucs_map,	/* Shift_JIS(CP932) ext-1 */ +  N_sjis_ext_ucs_map,	/* Shift_JIS(CP932) ext-2 */ +  N_gbk_ucs_map,	/* GBK(CP936) */ +  N_gbk_ucs_map,	/* GBK(CP936)-1 */ +  N_gbk_ucs_map,	/* GBK(CP936)-2 */ +  0,			/* GB18030 GBK-ext (special conversion) */ +  0,			/* GB18030 GBK-ext-1 (special conversion) */ +  0,			/* GB18030 GBK-ext-2 (special conversion) */ +  N_uhc_ucs_map,	/* UHC(CP949) */ +  N_uhc_ucs_map,	/* UHC(CP949)-1 */ +  N_uhc_ucs_map,	/* UHC(CP949)-2 */ +  N_hkscs_ucs_map,	/* HKSCS */ +  N_hkscs_ucs_map,	/* HKSCS-1 */ +  N_hkscs_ucs_map,	/* HKSCS-2 */ +}; +  static wc_wchar_t  ucs_cs94_conv(wc_ccs ccs, wc_uint16 c)  { diff --git a/libwc/wtf.c b/libwc/wtf.c index b8cfdc7..94d95c1 100644 --- a/libwc/wtf.c +++ b/libwc/wtf.c @@ -120,29 +120,36 @@ int  wtf_strwidth(wc_uchar *p)  {      int w = 0; +    wc_uchar *q = p + strlen(p); -    while (*p) { +    while (p < q) {  	w += wtf_width(p);  	p += WTF_LEN_MAP[*p];      }      return w;  } -/*  size_t  wtf_len1(wc_uchar *p)  { -    return (size_t)WTF_LEN_MAP[*p]; +    size_t len, len_max = WTF_LEN_MAP[*p]; + +    for (len = 0; *(p + len); len++) +	if (len == len_max) +	    break; +    if (len == 0) +	len = 1; +    return len;  } -*/  size_t  wtf_len(wc_uchar *p)  {      wc_uchar *q = p; +    wc_uchar *strz = p + strlen(p);      q += WTF_LEN_MAP[*q]; -    while (*q && ! WTF_WIDTH_MAP[*q]) +    while (q < strz && ! WTF_WIDTH_MAP[*q])  	q += WTF_LEN_MAP[*q];      return q - p;  } @@ -166,15 +173,17 @@ wtf_type(wc_uchar *p)      ((p)[3] = (((c) >>  7) & 0x7f) | 0x80), \      ((p)[4] = ( (c)        & 0x7f) | 0x80)  #define wtf_to_wcs16(p) \ +    ((p)[0] == 0 || (p)[1] == 0 || (p)[2] == 0 ? 0 : \        ((wc_uint32)((p)[0] & 0x03) << 14) \      | ((wc_uint32)((p)[1] & 0x7f) <<  7) \ -    | ((wc_uint32)((p)[2] & 0x7f)      ) +    | ((wc_uint32)((p)[2] & 0x7f)      ))  #define wtf_to_wcs32(p) \ +    ((p)[0] == 0 || (p)[1] == 0 || (p)[2] == 0 || (p)[3] == 0 || (p)[4] == 0 ? 0 : \        ((wc_uint32)((p)[0] & 0x0f) << 28) \      | ((wc_uint32)((p)[1] & 0x7f) << 21) \      | ((wc_uint32)((p)[2] & 0x7f) << 14) \      | ((wc_uint32)((p)[3] & 0x7f) <<  7) \ -    | ((wc_uint32)((p)[4] & 0x7f)      ) +    | ((wc_uint32)((p)[4] & 0x7f)      ))  void  wtf_push(Str os, wc_ccs ccs, wc_uint32 code) @@ -388,7 +397,7 @@ wtf_parse1(wc_uchar **p)  	cc.code = *(q++);      } else if (*q > 0xa0) {  	cc.ccs = wtf_gr_ccs; -	if (WC_CCS_IS_WIDE(cc.ccs)) { +	if (WC_CCS_IS_WIDE(cc.ccs) && *(q+1)) {  	    cc.code = ((wc_uint32)*q << 8) | *(q+1);  	    q += 2;  	} else @@ -401,27 +410,47 @@ wtf_parse1(wc_uchar **p)  	case WC_CCS_A_CS942:  	case WC_CCS_A_PCS:  	case WC_CCS_A_UNKNOWN: -	    cc.ccs |= *(q++) & 0x7f; -	    cc.code = *(q++); +	    if (*q && *(q+1)) { +		cc.ccs |= *(q++) & 0x7f; +		cc.code = *(q++); +	    } else { +		cc.ccs = WC_CCS_US_ASCII; +		cc.code = (wc_uint32)' '; +	    }  	    break;  	case WC_CCS_A_CS94W:  	case WC_CCS_A_CS96W:  	case WC_CCS_A_PCSW: -	    cc.ccs |= *(q++) & 0x7f; -	    cc.code = ((wc_uint32)*q << 8) | *(q+1); -	    q += 2; +	    if (*q && *(q+1) && *(q+2)) { +		cc.ccs |= *(q++) & 0x7f; +		cc.code = ((wc_uint32)*q << 8) | *(q+1); +		q += 2; +	    } else { +		cc.ccs = WC_CCS_US_ASCII; +		cc.code = (wc_uint32)' '; +	    }  	    break;  	case WC_CCS_A_WCS16:  	case WC_CCS_A_WCS16W: -	    cc.ccs |= (*q & 0x7c) >> 2; -	    cc.code = wtf_to_wcs16(q); -	    q += 3; +	    if (*q && *(q+1) && *(q+2)) { +		cc.ccs |= (*q & 0x7c) >> 2; +		cc.code = wtf_to_wcs16(q); +		q += 3; +	    } else { +		cc.ccs = WC_CCS_US_ASCII; +		cc.code = (wc_uint32)' '; +	    }  	    break;  	case WC_CCS_A_WCS32:  	case WC_CCS_A_WCS32W: -	    cc.ccs |= (*q & 0x70) >> 4; -	    cc.code = wtf_to_wcs32(q); -	    q += 5; +	    if (*q && *(q+1) && *(q+2) && *(q+3) && *(q+4)) { +		cc.ccs |= (*q & 0x70) >> 4; +		cc.code = wtf_to_wcs32(q); +		q += 5; +	    } else { +		cc.ccs = WC_CCS_US_ASCII; +		cc.code = (wc_uint32)' '; +	    }  	    break;  	default:  	/* case 0: */ diff --git a/libwc/wtf.h b/libwc/wtf.h index ad47973..435526f 100644 --- a/libwc/wtf.h +++ b/libwc/wtf.h @@ -59,8 +59,7 @@ extern void       wtf_init(wc_ces ces1, wc_ces ces2);  #define wtf_width(p) (WcOption.use_wide ? (int)WTF_WIDTH_MAP[(wc_uchar)*(p)] \  		      : ((int)WTF_WIDTH_MAP[(wc_uchar)*(p)] ? 1 : 0))  extern int        wtf_strwidth(wc_uchar *p); -/* extern size_t  wtf_len1(wc_uchar *p); */ -#define wtf_len1(p) ((int)WTF_LEN_MAP[(wc_uchar)*(p)]) +extern size_t     wtf_len1(wc_uchar *p);  extern size_t     wtf_len(wc_uchar *p);  /* extern int     wtf_type(wc_uchar *p); */  #define wtf_type(p) WTF_TYPE_MAP[(wc_uchar)*(p)] | 
