charmap.c

举世闻名的joe记事本源程序

	map->from_uni = from_uni;
	map->from_size = 0;
	map->to_map = builtin->to_uni;
	for (x=0; x!=256; ++x) {
		if (map->to_map[x]!=-1) {
			map->from_map[map->from_size].first = map->to_map[x];
			map->from_map[map->from_size].last = x;
			++map->from_size;
		}
	}
		
	qsort(map->from_map,map->from_size,sizeof(struct pair), (int (*)(const void *, const void *))pair_cmp);

	/* Fill in print, alpha and alnum bit maps */

	for (x=0;x!=32;++x) {
		map->print_map[x] = 0;
		map->alpha__map[x] = 0;
		map->alnum__map[x] = 0;
	}

	for (x=0; x!=256; ++x)
		if (map->to_map[x] != -1) {
			if (joe_iswprint(NULL,map->to_map[x]))
				set_bit(map->print_map,x);
			if (joe_iswalpha(NULL,map->to_map[x])) {
				set_bit(map->alpha__map,x);
				set_bit(map->alnum__map,x);
			}
		}

	/* Set underbar <U+005F> */

	x = from_uni(map,0x5F);
	if (x != -1) {
		set_bit(map->alpha__map,x);
		set_bit(map->alnum__map,x);
	}

	/* Put digits in alnum map */

	for (x=0x30; x!=0x3A; ++x) {
		int y = from_uni(map,x);
		if (y != -1)
			set_bit(map->alnum__map,y);
	}

	/* Build case conversion tables */

	for (x=0; x!=256; ++x) {
		map->lower_map[x] = x;
		if (map->to_map[x] != -1) {
			int y = joe_towlower(NULL,map->to_map[x]);
			int z = from_uni(map,y);
			if (z != -1)
				map->lower_map[x] = z;
		}
	}

	for (x=0; x!=256; ++x) {
		map->upper_map[x] = x;
		if (map->to_map[x] != -1) {
			int y = joe_towupper(NULL,map->to_map[x]);
			int z = from_uni(map,y);
			if (z != -1)
				map->upper_map[x] = z;
		}
	}

	map->next = charmaps;
	charmaps = map;

	return map;
}

static void load_builtins(void)
{
	struct charmap *map;

	/* install UTF-8 map (ties into i18n module) */
	map = joe_malloc(sizeof(struct charmap));
	map->name = USTR "utf-8";
	map->type = 1;
	map->to_uni = rtn_arg;
	map->from_uni = rtn_arg;
	map->is_punct = joe_iswpunct;
	map->is_print = joe_iswprint;
	map->is_space = joe_iswspace;
	map->is_alpha_ = joe_iswalpha_;
	map->is_alnum_ = joe_iswalnum_;
	map->to_lower = joe_towlower;
	map->to_upper = joe_towupper;
	map->next = charmaps;
	charmaps = map;

	/* Load all built-in byte maps */
	/*
	for (y=0; y!=sizeof(builtin_charmaps)/sizeof(struct builtin_charmap); ++y)
		process_builtin(builtin_charmaps + y);
	*/
}

/* Parse character map file */

struct builtin_charmap *parse_charmap(unsigned char *name,FILE *f)
{
	unsigned char buf[1024];
	unsigned char bf1[1024];
	unsigned comment_char = '#';
	int in_map = 0;
	int x;

	struct builtin_charmap *b;

	if (!f)
		return 0;

	b = joe_malloc(sizeof(struct builtin_charmap));

	b->name = zdup(name);

	for (x=0; x!=256; ++x)
		b->to_uni[x]= -1;

	/* This is a _really_bad_ parser.  The file has to be perfect. */
	while (fgets((char *)buf,1023,f)) {
		unsigned char *p = buf;
		parse_ws(&p, comment_char);
		parse_tows(&p, bf1);
		if (!zcmp(bf1,USTR "<comment_char>")) {
			parse_ws(&p, comment_char);
			parse_tows(&p, bf1);
			comment_char = bf1[0];
		} else if (!zcmp(bf1,USTR "<escape_char>")) {
			parse_ws(&p, comment_char);
			parse_tows(&p, bf1);
		} else if (!zcmp(bf1,USTR "CHARMAP")) {
			in_map = 1;
		} else if (!zcmp(bf1,USTR "END")) {
			in_map = 0;
		} else if (in_map && bf1[0]=='<' && bf1[1]=='U') {
			int uni;
			int byt;
			sscanf((char *)bf1+2,"%x",(unsigned *)&uni);
			parse_ws(&p, comment_char);
			parse_tows(&p, bf1);
			sscanf((char *)bf1+2,"%x",(unsigned *)&byt);
			b->to_uni[byt]=uni;
		}
	}

	/* For generating builtin maps from /usr/share/i18n/charmaps/ * */
/*
	printf("	{ \"%s\"\n",name);
	for (y=0;y!=256;y+=8) {
		printf("\t");
		for(x=0;x!=8;++x) {
			if (b->to_uni[y+x]==-1)
				printf("    -1, ");
			else
				printf("0x%4.4x, ",b->to_uni[y+x]);
		}
		printf("\n");
	}
*/
	fclose(f);
	return b;
}

/* Byte wide character map to unicode conversion */

/* Compare character map names.  Ignores '-'s and terminates string on '.' */
/* Chicken and egg problem here.. */

static int map_up(int c)
{
	if (c>='a' && c<='z')
		return c+'A'-'a';
	else
		return c;
}

int map_name_cmp(unsigned char *a,unsigned char *b)
{
	while (*a=='-') ++a;
	while (*b=='-') ++b;
	while (*a && *b && map_up(*a)==map_up(*b)) {
		++a;
		++b;
		while (*a=='-') ++a;
		while (*b=='-') ++b;
	}
	if (!*a && (*b=='.' || !*b))
		return 0;
	else
		return 1;
}

/* Find a character map */

struct charmap *find_charmap(unsigned char *name)
{
	unsigned char buf[1024];
	unsigned char *p;
	struct charmap *m;
	struct builtin_charmap *b;
	FILE *f;
	int y;

	if (!name)
		return 0;

	/* Install some initial character maps */
	if (!charmaps)
		load_builtins();

	/* Alias? */
	for (y=0; alias_table[y].alias; ++y)
		if (!map_name_cmp(alias_table[y].alias,name)) {
			name = alias_table[y].builtin;
			break;
		}

	/* Already loaded? */
	for (m=charmaps; m; m=m->next)
		if (!map_name_cmp(m->name,name))
			return m;

	/* Check ~/.joe/charmaps */
	p = (unsigned char *)getenv("HOME");
	f = 0;
	if (p) {
		joe_snprintf_2(buf,sizeof(buf),"%s/.joe/charmaps/%s",p,name);
		f = fopen((char *)buf,"r");
	}

	/* Check JOERCcharmaps */
	if (!f) {
		joe_snprintf_2(buf,sizeof(buf),"%scharmaps/%s",JOERC,name);
		f = fopen((char *)buf,"r");
	}

	/* Parse and install character map from file */
	if (f && (b = parse_charmap(name,f)))
		return process_builtin(b);

	/* Check builtin sets */
	for (y=0; y!=sizeof(builtin_charmaps)/sizeof(struct builtin_charmap); ++y)
		if (!map_name_cmp(builtin_charmaps[y].name,name))
			return process_builtin(builtin_charmaps + y);

	return NULL;
}

/* Test */

#if 0
main(int argc,char *argv[])
{
	struct charmap *map=find_charma