/src/w3m/libwc/charset.c

Source (jump to first uncovered line)

#include <stdlib.h>
#include <ctype.h>
#include "../alloc.h"

#include "wc.h"

#ifdef HAVE_LANGINFO_CODESET
#include <langinfo.h>
#endif

wc_locale WcLocale = 0;

static struct {
  char *lang;
  wc_ces ces;
} lang_ces_table[] = {
  { "cs", WC_CES_ISO_8859_2 },  /* cs_CZ */
  { "el", WC_CES_ISO_8859_7 },  /* el_GR */
  { "iw", WC_CES_ISO_8859_8 },  /* iw_IL */
  { "ja", WC_CES_EUC_JP },  /* ja_JP */
  { "ko", WC_CES_EUC_KR },  /* ko_KR */
  { "hu", WC_CES_ISO_8859_2 },  /* hu_HU */
  { "pl", WC_CES_ISO_8859_2 },  /* pl_PL */
  { "ro", WC_CES_ISO_8859_2 },  /* ro_RO */
  { "ru", WC_CES_ISO_8859_5 },  /* ru_SU */
  { "sk", WC_CES_ISO_8859_2 },  /* sk_SK */
  { "sl", WC_CES_ISO_8859_2 },  /* sl_CS */
  { "tr", WC_CES_ISO_8859_9 },  /* tr_TR */
  { "zh", WC_CES_EUC_CN },  /* zh_CN */
  { NULL, 0 }
};

static wc_ces
wc_codepage(int n)
{
  switch (n) {
  case 437: return WC_CES_CP437;
  case 737: return WC_CES_CP737;
  case 775: return WC_CES_CP775;
  case 850: return WC_CES_CP850;
  case 852: return WC_CES_CP852;
  case 855: return WC_CES_CP855;
  case 856: return WC_CES_CP856;
  case 857: return WC_CES_CP857;
  case 860: return WC_CES_CP860;
  case 861: return WC_CES_CP861;
  case 862: return WC_CES_CP862;
  case 863: return WC_CES_CP863;
  case 864: return WC_CES_CP864;
  case 865: return WC_CES_CP865;
  case 866: return WC_CES_CP866;
  case 869: return WC_CES_CP869;
  case 874: return WC_CES_CP874;
  case 932: return WC_CES_CP932;   /* CP932 = Shift_JIS */
  case 936: return WC_CES_CP936;   /* CP936 = GBK > EUC_CN */
  case 943: return WC_CES_CP943;   /* CP943 = Shift_JIS */
  case 949: return WC_CES_CP949;   /* CP949 = UHC > EUC_KR */
  case 950: return WC_CES_CP950;   /* CP950 = Big5 */
  case 1006: return WC_CES_CP1006;
  case 1250: return WC_CES_CP1250;
  case 1251: return WC_CES_CP1251;
  case 1252: return WC_CES_CP1252;
  case 1253: return WC_CES_CP1253;
  case 1254: return WC_CES_CP1254;
  case 1255: return WC_CES_CP1255;
  case 1256: return WC_CES_CP1256;
  case 1257: return WC_CES_CP1257;
  case 1258: return WC_CES_CP1258;
  }
  return 0;
}

wc_ces
wc_guess_charset(char *charset, wc_ces orig)
{
    wc_ces guess;

    if (charset == NULL || *charset == '\0')
  return orig;
    guess = wc_charset_to_ces(charset);
    return guess ? guess : orig;
}

wc_ces
wc_guess_charset_short(char *charset, wc_ces orig)
{
    wc_ces guess;

    if (charset == NULL || *charset == '\0')
  return orig;
    guess = wc_charset_short_to_ces(charset);
    return guess ? guess : orig;
}

wc_ces
wc_guess_locale_charset(char *locale, wc_ces orig)
{
    wc_ces guess;

    if (locale == NULL || *locale == '\0')
  return orig;
    guess = wc_locale_to_ces(locale);
    return guess ? guess : orig;
}

wc_ces
wc_charset_to_ces(char *charset)
{
    char *p = charset;
    char buf[16];
    int n;

    if ((*p == 'x' || *p == 'X') && *(p+1) == '-')
  p += 2;
    for (n = 0; *p && n < 15; p++) {
  if ((unsigned char)*p > 0x20 && *p != '_' && *p != '-')
      buf[n++] = *p | 32; /* tolower(*p); */
    }
    buf[n] = 0;
    p = buf;
    switch (*p) {
    case 'e':
  if (! strncmp(p, "euc", 3)) {
      p += 3;
      switch (*p) {
      case 'j': return WC_CES_EUC_JP;
      case 'c': return WC_CES_EUC_CN;
      case 't': return WC_CES_EUC_TW;
      case 'k': return WC_CES_EUC_KR;
      }
      switch (WcLocale) {
      case WC_LOCALE_JA_JP: return WC_CES_EUC_JP;
      case WC_LOCALE_ZH_CN: return WC_CES_EUC_CN;
      case WC_LOCALE_ZH_TW: return WC_CES_EUC_TW;
      case WC_LOCALE_ZH_HK: return WC_CES_EUC_CN;
      case WC_LOCALE_KO_KR: return WC_CES_EUC_KR;
      }
      return WC_CES_EUC_JP;
        }
  break;
    case 'i':
  if (! strncmp(p, "iso2022", 7)) {
      p += 7;
      switch (*p) {
      case 'j':
    if (! strncmp(p, "jp2", 3))
        return WC_CES_ISO_2022_JP_2;
    if (! strncmp(p, "jp3", 3))
        return WC_CES_ISO_2022_JP_3;
    return WC_CES_ISO_2022_JP;
      case 'c': return WC_CES_ISO_2022_CN;
      case 'k': return WC_CES_ISO_2022_KR;
      }
      return WC_CES_ISO_2022_JP;
  } else if (! strncmp(p, "iso8859", 7)) {
      n = atoi(p + 7);
      if (n >= 1 && n <= 16 && n != 12)
    return (WC_CES_E_ISO_8859 | n);
      return WC_CES_ISO_8859_1;
  } else if (! strncmp(p, "ibm", 3)) {
      p += 3;
      if (*p >= '1' && *p <= '9')
        return wc_codepage(atoi(p));
      return wc_charset_to_ces(p);
  }
  break;
    case 'j':
  if (! strncmp(p, "johab", 5))
      return WC_CES_JOHAB;
  if (! strncmp(p, "jis", 3))
      return WC_CES_ISO_2022_JP;
  break;
    case 's':
  if (! strncmp(p, "shiftjisx0213", 13) ||
      ! strncmp(p, "sjisx0213", 9))
      return WC_CES_SHIFT_JISX0213;
  if (! strncmp(p, "shiftjis", 8) ||
      ! strncmp(p, "sjis", 4))
      return WC_CES_SHIFT_JIS;
  break;
    case 'p':
  if (! strncmp(p, "pck", 3))
      return WC_CES_SHIFT_JIS;
  break;
    case 'g':
  if (! strncmp(p, "gb18030", 7) ||
      ! strncmp(p, "gbk2k", 5))
      return WC_CES_GB18030;
  if (! strncmp(p, "gbk", 3))
      return WC_CES_GBK;
  if (! strncmp(p, "gb2312", 6))
      return WC_CES_EUC_CN;
  break;
    case 'b':
  if (! strncmp(p, "big5hkscs", 9))
      return WC_CES_HKSCS;
  if (! strncmp(p, "big5", 4))
      return WC_CES_BIG5;
  break;
    case 'h':
  if (! strncmp(p, "hz", 2))
      return WC_CES_HZ_GB_2312;
  if (! strncmp(p, "hkscs", 5))
      return WC_CES_HKSCS;
  break;
    case 'k':
  if (! strncmp(p, "koi8r", 5))
      return WC_CES_KOI8_R;
  if (! strncmp(p, "koi8u", 5))
      return WC_CES_KOI8_U;
  if (! strncmp(p, "ksx1001", 7))
      return WC_CES_EUC_KR;
  if (! strncmp(p, "ksc5601", 7))
      return WC_CES_EUC_KR;
  break;
    case 't':
  if (! strncmp(p, "tis620", 6))
      return WC_CES_TIS_620;
  if (! strncmp(p, "tcvn", 4))
      return WC_CES_TCVN_5712;
  break;
    case 'n':
  if (! strncmp(p, "next", 4))
      return WC_CES_NEXTSTEP;
  break;
    case 'v':
  if (! strncmp(p, "viet", 4)) {
      p += 4;
      if (! strncmp(p, "tcvn", 4))
    return WC_CES_TCVN_5712;
  }
  if (! strncmp(p, "viscii", 6))
      return WC_CES_VISCII_11;
  if (! strncmp(p, "vps", 3))
      return WC_CES_VPS;
  break;
    case 'u':
#ifdef USE_UNICODE
  if (! strncmp(p, "utf8", 4))
      return WC_CES_UTF_8;
  if (! strncmp(p, "utf7", 4))
      return WC_CES_UTF_7;
#endif
  if (! strncmp(p, "uhc", 3))
      return WC_CES_UHC;
  if (! strncmp(p, "ujis", 4))
      return WC_CES_EUC_JP;
  if (! strncmp(p, "usascii", 7))
      return WC_CES_US_ASCII;
  break;
    case 'a':
  if (! strncmp(p, "ascii", 5))
      return WC_CES_US_ASCII;
  break;
    case 'c':
  if (! strncmp(p, "cngb", 4))
      return WC_CES_EUC_CN;
  if (*(p+1) != 'p')
      break;
  p += 2;
  if (*p >= '1' &&  *p <= '9')
      return wc_codepage(atoi(p));
  break;
    case 'w':
  if (strncmp(p, "windows", 7))
      break;
  p += 7;
  if (! strncmp(p, "31j", 3))
      return WC_CES_CP932;
  if (*p >= '1' &&  *p <= '9')
      return wc_codepage(atoi(p));
  break;
    }
    return 0;
}

wc_ces
wc_charset_short_to_ces(char *charset)
{
    char *p = charset;
    char buf[16];
    wc_ces ces;
    int n;

    ces = wc_charset_to_ces(charset);
    if (ces)
  return ces;

    for (n = 0; *p && n < 15; p++) {
  if ((unsigned char)*p > 0x20 && *p != '_' && *p != '-')
      buf[n++] = *p | 32; /* tolower(*p); */
    }
    buf[n] = 0;
    p = buf;
    switch (*p) {
    case 'e':
  switch (*(p+1)) {
  case 'j': return WC_CES_EUC_JP;
  case 'c': return WC_CES_EUC_CN;
  case 't': return WC_CES_EUC_TW;
  case 'k': return WC_CES_EUC_KR;
  }
  return WC_CES_EUC_JP;
    case 'j':
  p++;
  if (*p == 'o')
      return WC_CES_JOHAB;
  if (*p == 'p')
     p++;
  if (*p == '2')
     return WC_CES_ISO_2022_JP_2;
  if (*p == '3')
     return WC_CES_ISO_2022_JP_3;
  return WC_CES_ISO_2022_JP;
    case 's':
  return WC_CES_SHIFT_JIS;
    case 'g':
  return WC_CES_EUC_CN;
    case 'b':
  return WC_CES_BIG5;
    case 'h':
  if (*(p+1) == 'k')
      return WC_CES_HKSCS;
  return WC_CES_HZ_GB_2312;
    case 'k':
  if (*(p+1) == 'o')
      return WC_CES_KOI8_R;
  return WC_CES_ISO_2022_KR;
    case 'l':
  n = atoi(p + 1);
  if (n >= 1 && n <= 16 && n != 12)
      return (WC_CES_E_ISO_8859 | n);
  return WC_CES_ISO_8859_1;
    case 't':
  if (*(p+1) == 'c')
      return WC_CES_TCVN_5712;
  return WC_CES_TIS_620;
    case 'n':
  return WC_CES_NEXTSTEP;
    case 'v':
  if (*(p+1) == 'p')
      return WC_CES_VPS;
  return WC_CES_VISCII_11;
#ifdef USE_UNICODE
    case 'u':
  if (*(p+1) == '7')
      return WC_CES_UTF_7;
  return WC_CES_UTF_8;
#endif
    case 'a':
  return WC_CES_US_ASCII;
    case 'c':
  return WC_CES_ISO_2022_CN;
    case 'w':
  p++;
  if (*p >= '1' &&  *p <= '9')
      return wc_codepage(atoi(p));
  break;
    case 'r':
  return WC_CES_RAW;
    }
    return 0;
}

wc_ces
wc_locale_to_ces(char *locale)
{
    char *p = locale;
    char buf[8];
    int n;

    if (*p == 'C' && *(p+1) == '\0')
  return WC_CES_US_ASCII;
#ifdef HAVE_LANGINFO_CODESET
    {
  char *cs = nl_langinfo(CODESET);
  if (cs && strcmp(cs, "US-ASCII"))
      return wc_charset_to_ces(cs);
    }
#endif
    for (n = 0; *p && *p != '.' && n < 7; p++) {
  if ((unsigned char)*p > 0x20)
      buf[n++] = *p | 32; /* tolower(*p); */
    }
    buf[n] = 0;
    if (*p == '.') {
  p++;
  if (! strcasecmp(p, "euc")) {
      switch (buf[0]) {
      case 'j':
    WcLocale = WC_LOCALE_JA_JP;
    break;
      case 'k':
    WcLocale = WC_LOCALE_KO_KR;
    break;
      case 'z':
          if (!strcmp(buf, "zh_tw"))
        WcLocale = WC_LOCALE_ZH_TW;
          else if (!strcmp(buf, "zh_hk"))
        WcLocale = WC_LOCALE_ZH_HK;
    else
        WcLocale = WC_LOCALE_ZH_CN;
    break;
      default:
    WcLocale = 0;
    break;
      }
  }
  return wc_charset_to_ces(p);
    }

    if (!strcmp(buf, "japanese"))
  return WC_CES_SHIFT_JIS;
    if (!strcmp(buf, "zh_tw") ||
  !strcmp(buf, "zh_hk"))
  return WC_CES_BIG5;
    for (n = 0; lang_ces_table[n].lang; n++) {
  if (!strncmp(buf, lang_ces_table[n].lang, 2))
      return lang_ces_table[n].ces;
    }
    return WC_CES_ISO_8859_1;
}

char *
wc_ces_to_charset(wc_ces ces)
{
    if (ces == WC_CES_WTF)
  return "WTF";
    return WcCesInfo[WC_CES_INDEX(ces)].name;
}

char *
wc_ces_to_charset_desc(wc_ces ces)
{
    if (ces == WC_CES_WTF)
  return "W3M Transfer Format";
    return WcCesInfo[WC_CES_INDEX(ces)].desc;
}

wc_ces
wc_guess_8bit_charset(wc_ces orig)
{
    switch (orig) {
    case WC_CES_ISO_2022_JP:
    case WC_CES_ISO_2022_JP_2:
    case WC_CES_ISO_2022_JP_3:
  return WC_CES_EUC_JP;
    case WC_CES_ISO_2022_KR:
  return WC_CES_EUC_KR;
    case WC_CES_ISO_2022_CN:
    case WC_CES_HZ_GB_2312:
  return WC_CES_EUC_CN;
    case WC_CES_US_ASCII:
  return WC_CES_ISO_8859_1;
    }
    return orig;
}

wc_bool
wc_check_ces(wc_ces ces)
{
    size_t i = WC_CES_INDEX(ces);

    return (i <= WC_CES_END && WcCesInfo[i].id == ces);
}

static int
wc_ces_list_cmp(const void *a, const void *b)
{
    return strcasecmp(((wc_ces_list *)a)->desc, ((wc_ces_list *)b)->desc);
}

static wc_ces_list *list = NULL;

wc_ces_list *
wc_get_ces_list(void)
{
    wc_ces_info *info;
    size_t n;

    if (list)
  return list;
    for (info = WcCesInfo, n = 0; info->id; info++) {
  if (info->name != NULL)
      n++;
    }
    list = New_N(wc_ces_list, n + 1);
    for (info = WcCesInfo, n = 0; info->id; info++) {
  if (info->name != NULL) {
      list[n].id = info->id;
      list[n].name = info->name;
      list[n].desc = info->desc;
      n++;
  }
    }
    list[n].id = 0;
    list[n].name = NULL;
    list[n].desc = NULL;
    qsort(list, n, sizeof(wc_ces_list), wc_ces_list_cmp);
    return list;
}

Coverage Report

Created: 2024-10-12 16:05

Line	Count	Source (jump to first uncovered line)
1
2		#include <stdlib.h>
3		#include <ctype.h>
4		#include "../alloc.h"
5
6		#include "wc.h"
7
8		#ifdef HAVE_LANGINFO_CODESET
9		#include <langinfo.h>
10		#endif
11
12		wc_locale WcLocale = 0;
13
14		static struct {
15		char *lang;
16		wc_ces ces;
17		} lang_ces_table[] = {
18		{ "cs", WC_CES_ISO_8859_2 }, /* cs_CZ */
19		{ "el", WC_CES_ISO_8859_7 }, /* el_GR */
20		{ "iw", WC_CES_ISO_8859_8 }, /* iw_IL */
21		{ "ja", WC_CES_EUC_JP }, /* ja_JP */
22		{ "ko", WC_CES_EUC_KR }, /* ko_KR */
23		{ "hu", WC_CES_ISO_8859_2 }, /* hu_HU */
24		{ "pl", WC_CES_ISO_8859_2 }, /* pl_PL */
25		{ "ro", WC_CES_ISO_8859_2 }, /* ro_RO */
26		{ "ru", WC_CES_ISO_8859_5 }, /* ru_SU */
27		{ "sk", WC_CES_ISO_8859_2 }, /* sk_SK */
28		{ "sl", WC_CES_ISO_8859_2 }, /* sl_CS */
29		{ "tr", WC_CES_ISO_8859_9 }, /* tr_TR */
30		{ "zh", WC_CES_EUC_CN }, /* zh_CN */
31		{ NULL, 0 }
32		};
33
34		static wc_ces
35		wc_codepage(int n)
36	372	{
37	372	switch (n) {
38	5	case 437: return WC_CES_CP437;
39	3	case 737: return WC_CES_CP737;
40	4	case 775: return WC_CES_CP775;
41	3	case 850: return WC_CES_CP850;
42	4	case 852: return WC_CES_CP852;
43	3	case 855: return WC_CES_CP855;
44	4	case 856: return WC_CES_CP856;
45	3	case 857: return WC_CES_CP857;
46	4	case 860: return WC_CES_CP860;
47	3	case 861: return WC_CES_CP861;
48	5	case 862: return WC_CES_CP862;
49	4	case 863: return WC_CES_CP863;
50	12	case 864: return WC_CES_CP864;
51	4	case 865: return WC_CES_CP865;
52	4	case 866: return WC_CES_CP866;
53	3	case 869: return WC_CES_CP869;
54	17	case 874: return WC_CES_CP874;
55	3	case 932: return WC_CES_CP932; /* CP932 = Shift_JIS */
56	3	case 936: return WC_CES_CP936; /* CP936 = GBK > EUC_CN */
57	4	case 943: return WC_CES_CP943; /* CP943 = Shift_JIS */
58	3	case 949: return WC_CES_CP949; /* CP949 = UHC > EUC_KR */
59	3	case 950: return WC_CES_CP950; /* CP950 = Big5 */
60	3	case 1006: return WC_CES_CP1006;
61	3	case 1250: return WC_CES_CP1250;
62	5	case 1251: return WC_CES_CP1251;
63	6	case 1252: return WC_CES_CP1252;
64	4	case 1253: return WC_CES_CP1253;
65	4	case 1254: return WC_CES_CP1254;
66	14	case 1255: return WC_CES_CP1255;
67	14	case 1256: return WC_CES_CP1256;
68	4	case 1257: return WC_CES_CP1257;
69	81	case 1258: return WC_CES_CP1258;
70	372	}
71	133	return 0;
72	372	}
73
74		wc_ces
75		wc_guess_charset(char *charset, wc_ces orig)
76	0	{
77	0	wc_ces guess;
78
79	0	if (charset == NULL \|\| *charset == '\0')
80	0	return orig;
81	0	guess = wc_charset_to_ces(charset);
82	0	return guess ? guess : orig;
83	0	}
84
85		wc_ces
86		wc_guess_charset_short(char *charset, wc_ces orig)
87	21.3k	{
88	21.3k	wc_ces guess;
89
90	21.3k	if (charset == NULL \|\| *charset == '\0')
91	5.86k	return orig;
92	15.4k	guess = wc_charset_short_to_ces(charset);
93	15.4k	return guess ? guess : orig;
94	21.3k	}
95
96		wc_ces
97		wc_guess_locale_charset(char *locale, wc_ces orig)
98	0	{
99	0	wc_ces guess;
100
101	0	if (locale == NULL \|\| *locale == '\0')
102	0	return orig;
103	0	guess = wc_locale_to_ces(locale);
104	0	return guess ? guess : orig;
105	0	}
106
107		wc_ces
108		wc_charset_to_ces(char *charset)
109	15.4k	{
110	15.4k	char *p = charset;
111	15.4k	char buf[16];
112	15.4k	int n;
113
114	15.4k	if ((p == 'x' \|\| p == 'X') && *(p+1) == '-')
115	4	p += 2;
116	67.1k	for (n = 0; *p && n < 15; p++) {
117	51.6k	if ((unsigned char)p > 0x20 && p != '_' && *p != '-')
118	48.9k	buf[n++] = p \| 32; / tolower(p); /
119	51.6k	}
120	15.4k	buf[n] = 0;
121	15.4k	p = buf;
122	15.4k	switch (*p) {
123	1.22k	case 'e':
124	1.22k	if (! strncmp(p, "euc", 3)) {
125	16	p += 3;
126	16	switch (*p) {
127	3	case 'j': return WC_CES_EUC_JP;
128	3	case 'c': return WC_CES_EUC_CN;
129	4	case 't': return WC_CES_EUC_TW;
130	3	case 'k': return WC_CES_EUC_KR;
131	16	}
132	3	switch (WcLocale) {
133	0	case WC_LOCALE_JA_JP: return WC_CES_EUC_JP;
134	0	case WC_LOCALE_ZH_CN: return WC_CES_EUC_CN;
135	0	case WC_LOCALE_ZH_TW: return WC_CES_EUC_TW;
136	0	case WC_LOCALE_ZH_HK: return WC_CES_EUC_CN;
137	0	case WC_LOCALE_KO_KR: return WC_CES_EUC_KR;
138	3	}
139	3	return WC_CES_EUC_JP;
140	3	}
141	1.21k	break;
142	1.21k	case 'i':
143	235	if (! strncmp(p, "iso2022", 7)) {
144	52	p += 7;
145	52	switch (*p) {
146	41	case 'j':
147	41	if (! strncmp(p, "jp2", 3))
148	3	return WC_CES_ISO_2022_JP_2;
149	38	if (! strncmp(p, "jp3", 3))
150	2	return WC_CES_ISO_2022_JP_3;
151	36	return WC_CES_ISO_2022_JP;
152	3	case 'c': return WC_CES_ISO_2022_CN;
153	3	case 'k': return WC_CES_ISO_2022_KR;
154	52	}
155	5	return WC_CES_ISO_2022_JP;
156	183	} else if (! strncmp(p, "iso8859", 7)) {
157	31	n = atoi(p + 7);
158	31	if (n >= 1 && n <= 16 && n != 12)
159	6	return (WC_CES_E_ISO_8859 \| n);
160	25	return WC_CES_ISO_8859_1;
161	152	} else if (! strncmp(p, "ibm", 3)) {
162	51	p += 3;
163	51	if (p >= '1' && p <= '9')
164	4	return wc_codepage(atoi(p));
165	47	return wc_charset_to_ces(p);
166	51	}
167	101	break;
168	1.36k	case 'j':
169	1.36k	if (! strncmp(p, "johab", 5))
170	7	return WC_CES_JOHAB;
171	1.36k	if (! strncmp(p, "jis", 3))
172	3	return WC_CES_ISO_2022_JP;
173	1.35k	break;
174	1.60k	case 's':
175	1.60k	if (! strncmp(p, "shiftjisx0213", 13) \|\|
176	1.60k	! strncmp(p, "sjisx0213", 9))
177	315	return WC_CES_SHIFT_JISX0213;
178	1.29k	if (! strncmp(p, "shiftjis", 8) \|\|
179	1.29k	! strncmp(p, "sjis", 4))
180	159	return WC_CES_SHIFT_JIS;
181	1.13k	break;
182	1.13k	case 'p':
183	28	if (! strncmp(p, "pck", 3))
184	3	return WC_CES_SHIFT_JIS;
185	25	break;
186	2.94k	case 'g':
187	2.94k	if (! strncmp(p, "gb18030", 7) \|\|
188	2.94k	! strncmp(p, "gbk2k", 5))
189	1.95k	return WC_CES_GB18030;
190	990	if (! strncmp(p, "gbk", 3))
191	523	return WC_CES_GBK;
192	467	if (! strncmp(p, "gb2312", 6))
193	7	return WC_CES_EUC_CN;
194	460	break;
195	460	case 'b':
196	457	if (! strncmp(p, "big5hkscs", 9))
197	5	return WC_CES_HKSCS;
198	452	if (! strncmp(p, "big5", 4))
199	38	return WC_CES_BIG5;
200	414	break;
201	1.17k	case 'h':
202	1.17k	if (! strncmp(p, "hz", 2))
203	6	return WC_CES_HZ_GB_2312;
204	1.16k	if (! strncmp(p, "hkscs", 5))
205	8	return WC_CES_HKSCS;
206	1.15k	break;
207	1.15k	case 'k':
208	538	if (! strncmp(p, "koi8r", 5))
209	3	return WC_CES_KOI8_R;
210	535	if (! strncmp(p, "koi8u", 5))
211	3	return WC_CES_KOI8_U;
212	532	if (! strncmp(p, "ksx1001", 7))
213	2	return WC_CES_EUC_KR;
214	530	if (! strncmp(p, "ksc5601", 7))
215	3	return WC_CES_EUC_KR;
216	527	break;
217	613	case 't':
218	613	if (! strncmp(p, "tis620", 6))
219	4	return WC_CES_TIS_620;
220	609	if (! strncmp(p, "tcvn", 4))
221	16	return WC_CES_TCVN_5712;
222	593	break;
223	593	case 'n':
224	198	if (! strncmp(p, "next", 4))
225	3	return WC_CES_NEXTSTEP;
226	195	break;
227	475	case 'v':
228	475	if (! strncmp(p, "viet", 4)) {
229	41	p += 4;
230	41	if (! strncmp(p, "tcvn", 4))
231	4	return WC_CES_TCVN_5712;
232	41	}
233	471	if (! strncmp(p, "viscii", 6))
234	4	return WC_CES_VISCII_11;
235	467	if (! strncmp(p, "vps", 3))
236	7	return WC_CES_VPS;
237	460	break;
238	2.95k	case 'u':
239	2.95k	#ifdef USE_UNICODE
240	2.95k	if (! strncmp(p, "utf8", 4))
241	16	return WC_CES_UTF_8;
242	2.93k	if (! strncmp(p, "utf7", 4))
243	190	return WC_CES_UTF_7;
244	2.74k	#endif
245	2.74k	if (! strncmp(p, "uhc", 3))
246	411	return WC_CES_UHC;
247	2.33k	if (! strncmp(p, "ujis", 4))
248	4	return WC_CES_EUC_JP;
249	2.32k	if (! strncmp(p, "usascii", 7))
250	4	return WC_CES_US_ASCII;
251	2.32k	break;
252	2.32k	case 'a':
253	109	if (! strncmp(p, "ascii", 5))
254	3	return WC_CES_US_ASCII;
255	106	break;
256	369	case 'c':
257	369	if (! strncmp(p, "cngb", 4))
258	3	return WC_CES_EUC_CN;
259	366	if (*(p+1) != 'p')
260	344	break;
261	22	p += 2;
262	22	if (p >= '1' && p <= '9')
263	9	return wc_codepage(atoi(p));
264	13	break;
265	446	case 'w':
266	446	if (strncmp(p, "windows", 7))
267	406	break;
268	40	p += 7;
269	40	if (! strncmp(p, "31j", 3))
270	3	return WC_CES_CP932;
271	37	if (p >= '1' && p <= '9')
272	18	return wc_codepage(atoi(p));
273	19	break;
274	15.4k	}
275	11.6k	return 0;
276	15.4k	}
277
278		wc_ces
279		wc_charset_short_to_ces(char *charset)
280	15.4k	{
281	15.4k	char *p = charset;
282	15.4k	char buf[16];
283	15.4k	wc_ces ces;
284	15.4k	int n;
285
286	15.4k	ces = wc_charset_to_ces(charset);
287	15.4k	if (ces)
288	3.81k	return ces;
289
290	40.1k	for (n = 0; *p && n < 15; p++) {
291	28.4k	if ((unsigned char)p > 0x20 && p != '_' && *p != '-')
292	26.4k	buf[n++] = p \| 32; / tolower(p); /
293	28.4k	}
294	11.6k	buf[n] = 0;
295	11.6k	p = buf;
296	11.6k	switch (*p) {
297	1.21k	case 'e':
298	1.21k	switch (*(p+1)) {
299	0	case 'j': return WC_CES_EUC_JP;
300	6	case 'c': return WC_CES_EUC_CN;
301	466	case 't': return WC_CES_EUC_TW;
302	17	case 'k': return WC_CES_EUC_KR;
303	1.21k	}
304	723	return WC_CES_EUC_JP;
305	1.35k	case 'j':
306	1.35k	p++;
307	1.35k	if (*p == 'o')
308	685	return WC_CES_JOHAB;
309	672	if (*p == 'p')
310	6	p++;
311	672	if (*p == '2')
312	95	return WC_CES_ISO_2022_JP_2;
313	577	if (*p == '3')
314	7	return WC_CES_ISO_2022_JP_3;
315	570	return WC_CES_ISO_2022_JP;
316	1.13k	case 's':
317	1.13k	return WC_CES_SHIFT_JIS;
318	460	case 'g':
319	460	return WC_CES_EUC_CN;
320	413	case 'b':
321	413	return WC_CES_BIG5;
322	1.15k	case 'h':
323	1.15k	if (*(p+1) == 'k')
324	387	return WC_CES_HKSCS;
325	772	return WC_CES_HZ_GB_2312;
326	527	case 'k':
327	527	if (*(p+1) == 'o')
328	40	return WC_CES_KOI8_R;
329	487	return WC_CES_ISO_2022_KR;
330	308	case 'l':
331	308	n = atoi(p + 1);
332	308	if (n >= 1 && n <= 16 && n != 12)
333	33	return (WC_CES_E_ISO_8859 \| n);
334	275	return WC_CES_ISO_8859_1;
335	593	case 't':
336	593	if (*(p+1) == 'c')
337	130	return WC_CES_TCVN_5712;
338	463	return WC_CES_TIS_620;
339	195	case 'n':
340	195	return WC_CES_NEXTSTEP;
341	460	case 'v':
342	460	if (*(p+1) == 'p')
343	41	return WC_CES_VPS;
344	419	return WC_CES_VISCII_11;
345	0	#ifdef USE_UNICODE
346	2.32k	case 'u':
347	2.32k	if (*(p+1) == '7')
348	701	return WC_CES_UTF_7;
349	1.62k	return WC_CES_UTF_8;
350	0	#endif
351	106	case 'a':
352	106	return WC_CES_US_ASCII;
353	365	case 'c':
354	365	return WC_CES_ISO_2022_CN;
355	443	case 'w':
356	443	p++;
357	443	if (p >= '1' && p <= '9')
358	341	return wc_codepage(atoi(p));
359	102	break;
360	126	case 'r':
361	126	return WC_CES_RAW;
362	11.6k	}
363	563	return 0;
364	11.6k	}
365
366		wc_ces
367		wc_locale_to_ces(char *locale)
368	0	{
369	0	char *p = locale;
370	0	char buf[8];
371	0	int n;
372
373	0	if (p == 'C' && (p+1) == '\0')
374	0	return WC_CES_US_ASCII;
375	0	#ifdef HAVE_LANGINFO_CODESET
376	0	{
377	0	char *cs = nl_langinfo(CODESET);
378	0	if (cs && strcmp(cs, "US-ASCII"))
379	0	return wc_charset_to_ces(cs);
380	0	}
381	0	#endif
382	0	for (n = 0; p && p != '.' && n < 7; p++) {
383	0	if ((unsigned char)*p > 0x20)
384	0	buf[n++] = p \| 32; / tolower(p); /
385	0	}
386	0	buf[n] = 0;
387	0	if (*p == '.') {
388	0	p++;
389	0	if (! strcasecmp(p, "euc")) {
390	0	switch (buf[0]) {
391	0	case 'j':
392	0	WcLocale = WC_LOCALE_JA_JP;
393	0	break;
394	0	case 'k':
395	0	WcLocale = WC_LOCALE_KO_KR;
396	0	break;
397	0	case 'z':
398	0	if (!strcmp(buf, "zh_tw"))
399	0	WcLocale = WC_LOCALE_ZH_TW;
400	0	else if (!strcmp(buf, "zh_hk"))
401	0	WcLocale = WC_LOCALE_ZH_HK;
402	0	else
403	0	WcLocale = WC_LOCALE_ZH_CN;
404	0	break;
405	0	default:
406	0	WcLocale = 0;
407	0	break;
408	0	}
409	0	}
410	0	return wc_charset_to_ces(p);
411	0	}
412
413	0	if (!strcmp(buf, "japanese"))
414	0	return WC_CES_SHIFT_JIS;
415	0	if (!strcmp(buf, "zh_tw") \|\|
416	0	!strcmp(buf, "zh_hk"))
417	0	return WC_CES_BIG5;
418	0	for (n = 0; lang_ces_table[n].lang; n++) {
419	0	if (!strncmp(buf, lang_ces_table[n].lang, 2))
420	0	return lang_ces_table[n].ces;
421	0	}
422	0	return WC_CES_ISO_8859_1;
423	0	}
424
425		char *
426		wc_ces_to_charset(wc_ces ces)
427	0	{
428	0	if (ces == WC_CES_WTF)
429	0	return "WTF";
430	0	return WcCesInfo[WC_CES_INDEX(ces)].name;
431	0	}
432
433		char *
434		wc_ces_to_charset_desc(wc_ces ces)
435	0	{
436	0	if (ces == WC_CES_WTF)
437	0	return "W3M Transfer Format";
438	0	return WcCesInfo[WC_CES_INDEX(ces)].desc;
439	0	}
440
441		wc_ces
442		wc_guess_8bit_charset(wc_ces orig)
443	0	{
444	0	switch (orig) {
445	0	case WC_CES_ISO_2022_JP:
446	0	case WC_CES_ISO_2022_JP_2:
447	0	case WC_CES_ISO_2022_JP_3:
448	0	return WC_CES_EUC_JP;
449	0	case WC_CES_ISO_2022_KR:
450	0	return WC_CES_EUC_KR;
451	0	case WC_CES_ISO_2022_CN:
452	0	case WC_CES_HZ_GB_2312:
453	0	return WC_CES_EUC_CN;
454	0	case WC_CES_US_ASCII:
455	0	return WC_CES_ISO_8859_1;
456	0	}
457	0	return orig;
458	0	}
459
460		wc_bool
461		wc_check_ces(wc_ces ces)
462	2	{
463	2	size_t i = WC_CES_INDEX(ces);
464
465	2	return (i <= WC_CES_END && WcCesInfo[i].id == ces);
466	2	}
467
468		static int
469		wc_ces_list_cmp(const void a, const void b)
470	0	{
471	0	return strcasecmp(((wc_ces_list )a)->desc, ((wc_ces_list )b)->desc);
472	0	}
473
474		static wc_ces_list *list = NULL;
475
476		wc_ces_list *
477		wc_get_ces_list(void)
478	0	{
479	0	wc_ces_info *info;
480	0	size_t n;
481
482	0	if (list)
483	0	return list;
484	0	for (info = WcCesInfo, n = 0; info->id; info++) {
485	0	if (info->name != NULL)
486	0	n++;
487	0	}
488	0	list = New_N(wc_ces_list, n + 1);
489	0	for (info = WcCesInfo, n = 0; info->id; info++) {
490	0	if (info->name != NULL) {
491	0	list[n].id = info->id;
492	0	list[n].name = info->name;
493	0	list[n].desc = info->desc;
494	0	n++;
495	0	}
496	0	}
497	0	list[n].id = 0;
498	0	list[n].name = NULL;
499	0	list[n].desc = NULL;
500	0	qsort(list, n, sizeof(wc_ces_list), wc_ces_list_cmp);
501	0	return list;
502	0	}