gencin.c

来自「linux 下的 oxim 输入法,简单易用.」· C语言 代码 · 共 681 行 · 第 1/2 页

    if (cmp_val == 0)
    {
	    if (aa->order < bb->order)
	    {
		return 1;
	    }
	    else if (aa->order > bb->order)
	    {
		return -1;
	    }
	    else
	    {
		return 0;
	    }
    }
    else if (cmp_val > 0)
	return 1;
    else
	return -1;
}

static void
cin_chardef(char *arg, cintab_t *cintab)
{
    char cmd1[64], arg1[1024] = {'\0'}, arg2[32]= {'\0'};
    cin_char_t *cch=NULL;
    int argc;
    unsigned int i, key_len;
    unsigned int len;
    unsigned int ucs4[256];

    while ((argc=cmd_arg(True, cmd1, 64, arg1, 1024, arg2, 32, NULL)))
    {

	if ((strcasecmp("%chardef", cmd1) == 0 && strcasecmp("end", arg1) == 0) || strcasecmp("END_TABLE", cmd1) == 0)
	{
	    break;
	}

	if (argc < 2 || argc > 3)
	{
	    oxim_perr(OXIMMSG_WARNING, N_("%s(%d): arguement expected.\n"),
		cintab->fname_cin, cintab->lineno);
	    continue;
	}

	key_len = strlen(cmd1);

	/* 
	 * 看看字根中，是否有不合法的字元
	 * 所謂不合法，就是在 selkey、keyname、endkey 沒有定義的字元 
	 */
	int allkey_valid = True;
	for (i=0 ; i < key_len ; i++)
	{
	    unsigned int kk = cmd1[i];
	    if (!th.keep_key_case && kk >= 'A' && kk <= 'Z')
	    {
		cmd1[i] = tolower(cmd1[i]);
		kk = cmd1[i];
	    }

	    if (!validkey[kk])
	    {
#if 0
		if (i || cmd1[i] != '#')
		{
		    oxim_perr(OXIMMSG_WARNING, N_("%s(%d): illegal key \"%c\" specified.\n"), cintab->fname_cin, cintab->lineno, cmd1[i]);
		}
		allkey_valid = False;
		break;
#else
		if (i || cmd1[i] != '#')
		{
		    th.keyname[kk].s[0] = kk;
		    th.keyname[kk].s[1] = '\0';
		    validkey[kk] = '\1';
		    th.n_key_name++;
		}
#endif
	    }
	}

	if (!allkey_valid)
	{
	    continue;
	}

	th.n_input_content ++;
	if (th.n_input_content == 1)
	{
	    cchar = oxim_malloc(sizeof(cin_char_t), True);
	}
	else
	{
	    cchar = oxim_realloc(cchar, (th.n_input_content+1) * sizeof(cin_char_t));
	}

	cch = (cchar + th.n_input_content) - 1;
 	bzero(cch, sizeof(cin_char_t));
	cch->key_len = key_len;
	cch->keystroke = strdup(cmd1);

	len = strlen(arg1);
	int nbytes;
	char *p = arg1;
	unsigned int n_char = 0;
	while (len && (nbytes = oxim_utf8_to_ucs4(p, &ucs4[n_char], len)) > 0)
	{
	    p += nbytes;
	    len -= nbytes;
	    n_char ++;
	}
	cch->word_len = n_char;
	unsigned int size = sizeof(unsigned int) * n_char;
	cch->word = oxim_malloc(size, False);
	memcpy(cch->word, ucs4, size);

	/* 累進單一字元數目 */
	if (n_char == 1)
	{
	    th.n_char ++;
	}

	if (arg2[0] != '\0')
	{
	    cch->order = atoi(arg2);
	}

	if (th.n_max_keystroke < key_len)
	    th.n_max_keystroke = key_len;

	arg2[0] = '\0';
    }
}


/*----------------------------------------------------------------------------

	Main Functions.

----------------------------------------------------------------------------*/

struct genfunc_s {
    char *name;
    void (*func) (char *, cintab_t *);
};

static struct genfunc_s genfunc[] = {
    {"%keep_key_case", cin_keepcase},

    {"%setting", cin_setting},

    {"%ename", cin_ename},
/*    {"NAME", cin_cname},*/

    {"%cname", cin_cname},
    {"%prompt", cin_cname},

    {"%keyname", cin_keyname},
    {"BEGIN_CHAR_PROMPTS_DEFINITION", cin_keyname},

    {"%selkey", cin_selkey},
    {"SELECT_KEYS", scim_selkey},

    {"%endkey", cin_endkey},
    {"KEY_END_CHARS", cin_endkey},

    {"%chardef", cin_chardef},
    {"BEGIN_TABLE", cin_chardef},
    {NULL, NULL}
};


void
gencin(cintab_t *cintab)
{
    unsigned int i, argc;
    char cmd[64], arg1[64], arg2[64];
    size_t ret;

    bzero(validkey, 128);
    bzero(&th, sizeof(cintab_head_v1_t));

    while (argc = cmd_arg(False, cmd, 64, arg1, 64, arg2, 64, NULL))
    {
	for (i=0; genfunc[i].name != NULL; i++)
	{
	    if (strcasecmp(genfunc[i].name, cmd) == 0)
	    {
		if (strlen(arg1) == 1 && arg1[0] == '=')
		{
		    genfunc[i].func(arg2, cintab);
		}
		else if (argc == 2)
		{
		    genfunc[i].func(arg1, cintab);
		}
		else
		{
		    genfunc[i].func(NULL, cintab);
		}
		break;
	    }
	    /* SCIM 的 Locale Name */
	    else if (strcasecmp("NAME", cmd) == 0)
	    {
	    	/* */
	        char tmp[128];
		bzero(tmp, sizeof(tmp));
		sprintf(tmp, "%s:en;", arg2);
		cin_cname_scim(tmp, cintab);
		break;
	    }
	    else if (strncasecmp("NAME.", cmd, 5) == 0)
	    {
	    	/* 以 XXX:zh_TW;YYY:zh_CN 的格式將輸入法表格名稱與locale一起加進去*/
	        char tmp[128];
		char *charset = cmd+5;
		bzero(tmp, sizeof(tmp));
		sprintf(tmp, "%s:%s;", arg2, charset);
		cin_cname_scim(tmp, cintab);
		break;
	    }
	}
	if (genfunc[i].name == NULL)
	    oxim_perr(OXIMMSG_NORMAL, N_("%s(%d): unknown command: %s, ignore.\n"), cintab->fname_cin, cintab->lineno, cmd);
    }


#if 1
    /* 對輸入內容排序 */
    stable_sort(cchar, th.n_input_content, sizeof(cin_char_t), icode_cmp);

    unsigned int prefix_size = sizeof(table_prefix_t); /* 檔案識別結構 size */
    unsigned int hdr_size = sizeof(cintab_head_v1_t); /* 表頭 size */
    /* 國家地區列表 size */
    unsigned int locale_size = sizeof(cintab_locale_table) * th.n_locale;
    /* 設定列表 size */
    unsigned int setting_size = sizeof(cintab_setting_table) * th.n_setting;
    /* 位置索引列表 size */
    unsigned int offset_size = sizeof(unsigned int) * (th.n_input_content + 1);
    unsigned int *offset_table = oxim_malloc(offset_size, False);

    /* 單字索引 size */
    unsigned int char_index_size = sizeof(cintab_char_index) * th.n_char;
    cintab_char_index *char_idx = NULL;
    if (th.n_char)
    {
	char_idx = oxim_malloc(char_index_size, False);
    }

    cin_char_t *cch;
    unsigned int idx = 0;
    unsigned int offset = prefix_size + hdr_size + locale_size + setting_size + offset_size + char_index_size;
    unsigned int rec_len = 0;
    for (i=0, cch = cchar ; i < th.n_input_content ; i++, cch++)
    {
	offset_table[i] = offset;
	/* 建立字元索引 */
	if (cch->word_len == 1)
	{
	    char_idx[idx].ucs4 = cch->word[0];
	    char_idx[idx].offset = offset;
	    idx ++;
	}
	/* 1 + 1 + 字根長度 + 字詞長度 */
	rec_len = sizeof(unsigned char) + sizeof(unsigned char) + cch->key_len + (cch->word_len * sizeof(unsigned int));
	offset += rec_len;
    }
    offset_table[i] = offset;

    /* 依據字元排序 */
    if (char_idx)
    {
	stable_sort(char_idx, th.n_char, sizeof(cintab_char_index), char_cmp);
    }

    /* locale table offset */
    th.locale_table_offset = prefix_size + hdr_size;
    /* setting table offset */
    th.setting_table_offset = th.locale_table_offset + locale_size;
    /* offset table offset */
    th.offset_table_offset = th.setting_table_offset + setting_size;
    /* char offset */
    th.char_offset = th.offset_table_offset + offset_size;
    /* input content offset */
    th.input_content_offset = th.char_offset + char_index_size;

    /* 計算 Checksum */
    th.chksum = crc32(th.chksum, (char *)&th, sizeof(cintab_head_v1_t) - sizeof(unsigned int));
    /* 寫入表頭 */
    gzwrite(cintab->fw, &th, hdr_size);
    /* 寫入國家地區列表 */
    if (th.n_locale)
    {
	gzwrite(cintab->fw, locale_table, locale_size);
    }
    /* 寫入設定列表 */
    if (th.n_setting)
    {
	gzwrite(cintab->fw, setting_table, setting_size);
    }
    /* 寫入位置索引列表 */
    gzwrite(cintab->fw, offset_table, offset_size);
    /* 寫入字元索引列表 */
    if (th.n_char)
    {
	gzwrite(cintab->fw, char_idx, char_index_size);
    }
    /* 最後寫入真正的資料 */
    for (i=0, cch = cchar ; i < th.n_input_content ; i++, cch++)
    {
	/* 字根長度 */
	gzwrite(cintab->fw, &cch->key_len, sizeof(unsigned char));
	/* 字詞長度 */
	gzwrite(cintab->fw, &cch->word_len, sizeof(unsigned char));
	/* 字根資料 */
	gzwrite(cintab->fw, cch->keystroke, cch->key_len);
	/* 字詞資料 */
	gzwrite(cintab->fw, cch->word, cch->word_len * sizeof(unsigned int));
    }

#endif
    oxim_perr(OXIMMSG_NORMAL,
	N_("number of locale: %d\n"), th.n_locale);
    oxim_perr(OXIMMSG_NORMAL,
	N_("number of settings: %d\n"), th.n_setting);
    oxim_perr(OXIMMSG_NORMAL,
	N_("number of keyname: %d\n"), th.n_key_name);
    oxim_perr(OXIMMSG_NORMAL, 
	N_("max length of keystroke: %d\n"), th.n_max_keystroke);
    oxim_perr(OXIMMSG_NORMAL, 
	N_("number of keystroke code defined: %d\n"), th.n_input_content);
    oxim_perr(OXIMMSG_NORMAL, 
	N_("number of char defined: %d\n"), th.n_char);
    oxim_perr(OXIMMSG_NORMAL, 
	N_("number of phrase define: %d\n"), th.n_input_content - th.n_char);
}