url.c

ELinks is an advanced and well-established feature-rich text mode web (HTTP/FTP/..) browser. ELinks

		/**d++ = s[0];*/
		if (s == dd && !s[3]) goto p;
		s += 2;
		goto r;
	}
	if (dsep(s[0]) && s[1] == '.' && s[2] == '.' && (dsep(s[3]) || !s[3])) {
		while (d > dd) {
			d--;
			if (dsep(*d)) goto b;
		}
		b:
		if (!s[3]) *d++ = *s;
		s += 3;
		goto r;
	}
	p:
	if ((*d++ = *s++)) goto r;
}

void insert_wd(unsigned char **up, unsigned char *cwd)
{
	unsigned char *url = *up;
	if (!url || !cwd || !*cwd) return;
	if (casecmp(url, "file://", 7)) return;
	if (dir_sep(url[7])) return;
#ifdef DOS_FS
	if (upcase(url[7]) >= 'A' && upcase(url[7]) <= 'Z' && url[8] == ':' && dir_sep(url[9])) return;
#endif
#ifdef SPAD
	if (_is_absolute(url + 7) != _ABS_NO) return;
#endif
	url = mem_alloc(strlen(*up) + strlen(cwd) + 2);
	memcpy(url, *up, 7);
	strcpy(url + 7, cwd);
	if (!dir_sep(cwd[strlen(cwd) - 1])) strcat(url, "/");
	strcat(url, *up + 7);
	mem_free(*up);
	*up = url;
}

/*
 * U funkce join_urls musi byt prvni url absolutni (takove, co projde funkci
 * parse_url bez chyby --- pokud neni absolutni, tak to spatne na internal) a
 * druhe url je relativni cesta vuci nemu nebo taky absolutni url. Pokud je
 * druhe url absolutni, vrati se to; pokud je relativni, tak se spoji prvni a
 * druhe url.
 */
unsigned char *join_urls(unsigned char *base, unsigned char *rel)
{
	unsigned char *p, *n, *pp, *ch;
	int l;
	int lo = !casecmp(base, "file://", 7);
	if (rel[0] == '#' || !rel[0]) {
		n = stracpy(base);
		for (p = n; *p && *p != POST_CHAR && *p != '#'; p++) ;
		*p = 0;
		add_to_strn(&n, rel);
		extend_str(&n, 1);
		translate_directories(n);
		return n;
	}
	if (rel[0] == '?' || rel[0] == '&') {
		unsigned char rj[3];
		unsigned char *d = get_url_data(base);
		if (!d) goto bad_base;
		rj[0] = rel[0];
		rj[1] = POST_CHAR;
		rj[2] = 0;
		d += strcspn(d, rj);
		n = memacpy(base, d - base);
		add_to_strn(&n, rel);
		translate_directories(n);
		return n;
	}
	if (rel[0] == '/' && rel[1] == '/') {
		unsigned char *s, *n;
		if (!(s = strstr(base, "//"))) {
			if (!(s = strchr(base, ':'))) {
				bad_base:
				internal("bad base url: %s", base);
				return NULL;
			}
			s++;
		}
		n = memacpy(base, s - base);
		add_to_strn(&n, rel);
		if (!parse_url(n, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL)) {
			extend_str(&n, 1);
			translate_directories(n);
			return n;
		}
		mem_free(n);
	}
	if (!casecmp("proxy://", rel, 8)) goto prx;
	if (!parse_url(rel, &l, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL)) {
		n = stracpy(rel);
		extend_str(&n, 1);
		translate_directories(n);
		return n;
	}
	n = stracpy(rel);
	while (n[0] && n[strlen(n) - 1] <= ' ') n[strlen(n) - 1] = 0;
	extend_str(&n, 1);
	ch = strrchr(n, '#');
	if (!ch || strchr(ch, '/')) ch = n + strlen(n);
	memmove(ch + 1, ch, strlen(ch) + 1);
	*ch = '/';
	if (!parse_url(n, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL)) {
		extend_str(&n, 1);
		translate_directories(n);
		return n;
	}
	mem_free(n);
	prx:
	if (parse_url(base, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, &p, NULL, NULL) || !p) {
		goto bad_base;
	}
	if (!dsep(*p)) p--;
	if (end_of_dir(base, rel[0])) for (; *p; p++) {
		if (end_of_dir(base, *p)) break;
	} else if (!dsep(rel[0])) for (pp = p; *pp; pp++) {
		if (end_of_dir(base, *pp)) break;
		if (dsep(*pp)) p = pp + 1;
	}
	n = mem_alloc(p - base + strlen(rel) + 2);
	memcpy(n, base, p - base);
	strcpy(n + (p - base), rel);
	translate_directories(n);
	return n;
}

unsigned char *translate_url(unsigned char *url, unsigned char *cwd)
{
	unsigned char *ch;
	unsigned char *nu, *da;
	unsigned char *prefix;
	int sl;
	while (*url == ' ') url++;
	if (*url && url[strlen(url) - 1] == ' ') {
		nu = stracpy(url);
		while (*nu && nu[strlen(nu) - 1] == ' ') nu[strlen(nu) - 1] = 0;
		ch = translate_url(nu, cwd);
		mem_free(nu);
		return ch;
	}
	if (!casecmp("proxy://", url, 8)) return NULL;
	if (!parse_url(url, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, &da, NULL, NULL)) {
		nu = stracpy(url);
		insert_wd(&nu, cwd);
		extend_str(&nu, 1);
		translate_directories(nu);
		return nu;
	}
	if (strchr(url, POST_CHAR)) return NULL;
	if (strstr(url, "://")) {
		nu = stracpy(url);
		extend_str(&nu, 1);
		ch = strrchr(nu, '#');
		if (!ch || strchr(ch, '/')) ch = nu + strlen(nu);
		memmove(ch + 1, ch, strlen(ch) + 1);
		*ch = '/';
		if (!parse_url(nu, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL)) {
			insert_wd(&nu, cwd);
			extend_str(&nu, 1);
			translate_directories(nu);
			return nu;
		}
		mem_free(nu);
	}
	ch = url + strcspn(url, ".:/@");
	prefix = "file://";
	sl = 0;
#ifdef SPAD
	if (strchr(url, ':') && _is_local(url)) goto set_prefix;
#endif
	if (*ch != ':' || *(url + strcspn(url, "/@")) == '@') {
		if (*url != '.' && *ch == '.') {
			unsigned char *f, *e;
			int i;
			for (e = ch + 1; *(f = e + strcspn(e, ".:/")) == '.'; e = f + 1) ;
			for (i = 0; i < f - e; i++) if (e[i] < '0' || e[i] > '9') goto noip;
			goto http;
			noip:
			if (f - e == 2 && casecmp(e, "gz", 2)) {
				http:
				prefix = "http://", sl = 1;
			} else {
				unsigned char *tld[] = { "com", "edu", "net", "org", "gov", "mil", "int", "arpa", "aero", "biz", "coop", "info", "museum", "name", "pro", "cat", "jobs", "mobi", "travel", "tel", NULL };
				for (i = 0; tld[i]; i++) if ((size_t)(f - e) == strlen(tld[i]) && !casecmp(tld[i], e, f - e)) goto http;
			}
		}
		if (*ch == '@' || *ch == ':' || !cmpbeg(url, "ftp.")) prefix = "ftp://", sl = 1;
		goto set_prefix;
		set_prefix:
		nu = stracpy(prefix);
		add_to_strn(&nu, url);
		if (sl && !strchr(url, '/')) add_to_strn(&nu, "/");
		if (parse_url(nu, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL)) mem_free(nu), nu = NULL;
		else {
			insert_wd(&nu, cwd);
			extend_str(&nu, 1);
			translate_directories(nu);
		}
		return nu;
	}
	sl = 1;
#ifdef DOS_FS
	if (ch == url + 1) goto set_prefix;
#endif
	if (!(nu = memacpy(url, ch - url + 1))) return NULL;
	add_to_strn(&nu, "//");
	add_to_strn(&nu, ch + 1);
	if (!parse_url(nu, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL)) {
		insert_wd(&nu, cwd);
		extend_str(&nu, 1);
		translate_directories(nu);
		return nu;
	}
	add_to_strn(&nu, "/");
	if (!parse_url(nu, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL)) {
		insert_wd(&nu, cwd);
		extend_str(&nu, 1);
		translate_directories(nu);
		return nu;
	}
	mem_free(nu);
	return NULL;
}

unsigned char *extract_position(unsigned char *url)
{
	unsigned char *u, *uu, *r;
	if ((u = get_url_data(url))) url = u;
	if (!(u = strchr(url, POST_CHAR))) u = url + strlen(url);
	if (!(uu = memchr(url, '#', u - url))) return NULL;
	r = mem_alloc(u - uu);
	memcpy(r, uu + 1, u - uu - 1);
	r[u - uu - 1] = 0;
	memmove(uu, u, strlen(u) + 1);
	return r;
}

#define accept_char(x)	((x) != '"' && (x) != '\'' && (x) != '&' && (x) != '<' && (x) != '>')
#define special_char(x)	((x) == '%' || (x) == '#')

void add_conv_str(unsigned char **s, int *l, unsigned char *b, int ll, int encode_special)
{
	for (; ll; ll--, b++) {
		if ((unsigned char)*b < ' ') continue;
		if (special_char(*b) && encode_special == 1) {
			unsigned char h[4];
			sprintf(h, "%%%02X", (unsigned)*b & 0xff);
			add_to_str(s, l, h);
		} else if (*b == '%' && encode_special <= -1 && ll > 2 && ((b[1] >= '0' && b[1] <= '9') || (b[1] >= 'A' && b[1] <= 'F') || (b[1] >= 'a' && b[1] <= 'f'))) {
			unsigned char h = 0;
			int i;
			for (i = 1; i < 3; i++) {
				if (b[i] >= '0' && b[i] <= '9') h = h * 16 + b[i] - '0';
				if (b[i] >= 'A' && b[i] <= 'F') h = h * 16 + b[i] - 'A' + 10;
				if (b[i] >= 'a' && b[i] <= 'f') h = h * 16 + b[i] - 'a' + 10;
			}
			if (h >= ' ') add_chr_to_str(s, l, h);
			ll -= 2;
			b += 2;
		} else if (accept_char(*b) || encode_special == -2) {
			add_chr_to_str(s, l, *b);
		} else {
			add_to_str(s, l, "&#");
			add_num_to_str(s, l, (int)*b);
			add_chr_to_str(s, l, ';');
		}
	}
}