url.c

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

/* url.c
 * (c) 2002 Mikulas Patocka
 * This file is a part of the Links program, released under GPL.
 */

#include "links.h"

struct {
	unsigned char *prot;
	int port;
	void (*func)(struct connection *);
	void (*nc_func)(struct session *, unsigned char *);
	int free_syntax;
	int need_slashes;
	int need_slash_after_host;
	int allow_post;
	int bypasses_socks;
} protocols[]= {
		{"file", 0, file_func, NULL,		1, 1, 0, 0, 1},
		{"https", 443, https_func, NULL,	0, 1, 1, 1, 0},
		{"http", 80, http_func, NULL,		0, 1, 1, 1, 0},
		{"proxy", 3128, proxy_func, NULL,	0, 1, 1, 1, 0},
		{"ftp", 21, ftp_func, NULL,		0, 1, 1, 0, 0},
		{"finger", 79, finger_func, NULL,	0, 1, 1, 0, 0},
#ifndef DISABLE_SMB
		{"smb", 139, smb_func, NULL,		0, 1, 1, 0, 1},
#endif
		{"mailto", 0, NULL, mailto_func,	0, 0, 0, 0, 0},
		{"telnet", 0, NULL, telnet_func,	0, 0, 0, 0, 1},
		{"tn3270", 0, NULL, tn3270_func,	0, 0, 0, 0, 1},
		{"mms", 0, NULL, mms_func,		1, 0, 1, 0, 1},
#ifdef JS
		{"javascript", 0, NULL, javascript_func,1, 0, 0, 0, 0},
#endif
		{NULL, 0, NULL, NULL,			0, 0, 0, 0, 0}
};

/* prototypes */
int check_protocol(unsigned char *, int);
int get_prot_info(unsigned char *, int *, void (**)(struct connection *), void (**)(struct session *ses, unsigned char *), int *, int *);
void translate_directories(unsigned char *);
void insert_wd(unsigned char **, unsigned char *);


int check_protocol(unsigned char *p, int l)
{
	int i;
	for (i = 0; protocols[i].prot; i++)
		if (!casecmp(protocols[i].prot, p, l) && strlen(protocols[i].prot) == (size_t)l) {
			return i;
		}
	return -1;
}

int get_prot_info(unsigned char *prot, int *port, void (**func)(struct connection *), void (**nc_func)(struct session *ses, unsigned char *), int *allow_post, int *bypasses_socks)
{
	int i;
	for (i = 0; protocols[i].prot; i++)
		if (!strcasecmp(protocols[i].prot, prot)) {
			if (port) *port = protocols[i].port;
			if (func) *func = protocols[i].func;
			if (nc_func) *nc_func = protocols[i].nc_func;
			if (allow_post) *allow_post = protocols[i].allow_post;
			if (bypasses_socks) *bypasses_socks = protocols[i].bypasses_socks;
			return 0;
		}
	return -1;
}

int parse_url(unsigned char *url, int *prlen, unsigned char **user, int *uslen, unsigned char **pass, int *palen, unsigned char **host, int *holen, unsigned char **port, int *polen, unsigned char **data, int *dalen, unsigned char **post)
{
	unsigned char *p, *q;
	unsigned char p_c[2];
	int a;
	if (prlen) *prlen = 0;
	if (user) *user = NULL;
	if (uslen) *uslen = 0;
	if (pass) *pass = NULL;
	if (palen) *palen = 0;
	if (host) *host = NULL;
	if (holen) *holen = 0;
	if (port) *port = NULL;
	if (polen) *polen = 0;
	if (data) *data = NULL;
	if (dalen) *dalen = 0;
	if (post) *post = NULL;
	if (!url || !(p = strchr(url, ':'))) return -1;
	if (prlen) *prlen = p - url;
	if ((a = check_protocol(url, p - url)) == -1) return -1;
	if (p[1] != '/' || p[2] != '/') {
		if (protocols[a].need_slashes) return -1;
		p -= 2;
	}
	if (protocols[a].free_syntax) {
		if (data) *data = p + 3;
		if (dalen) *dalen = strlen(p + 3);
		return 0;
	}
	p += 3;
	q = p + strcspn(p, "@/?");
	if (!*q && protocols[a].need_slash_after_host) return -1;
	if (*q == '@') {
		unsigned char *pp;
		while (strcspn(q + 1, "@") < strcspn(q + 1, "/?"))
			q = q + 1 + strcspn(q + 1, "@");
		pp = strchr(p, ':');
		if (!pp || pp > q) {
			if (user) *user = p;
			if (uslen) *uslen = q - p;
		} else {
			if (user) *user = p;
			if (uslen) *uslen = pp - p;
			if (pass) *pass = pp + 1;
			if (palen) *palen = q - pp - 1;
		}
		p = q + 1;
	} 
	q = p + strcspn(p, ":/?");
	if (!*q && protocols[a].need_slash_after_host) return -1;
	if (host) *host = p;
	if (holen) *holen = q - p;
	if (*q == ':') {
		unsigned char *pp = q + strcspn(q, "/");
		int cc;
		if (*pp != '/' && protocols[a].need_slash_after_host) return -1;
		if (port) *port = q + 1;
		if (polen) *polen = pp - q - 1;
		for (cc = 0; cc < pp - q - 1; cc++) if (q[cc+1] < '0' || q[cc+1] > '9') return -1;
		q = pp;
	}
	if (*q && *q != '?') q++;
	p = q;
	p_c[0] = POST_CHAR;
	p_c[1] = 0;
	q = p + strcspn(p, p_c);
	if (data) *data = p;
	if (dalen) *dalen = q - p;
	if (post) *post = *q ? q + 1 : NULL;
	return 0;
}

unsigned char *get_protocol_name(unsigned char *url)
{
	int l;
	if (parse_url(url, &l, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL)) return NULL;
	return memacpy(url, l);
}

unsigned char *get_host_and_pass(unsigned char *url)
{
	unsigned char *u, *h, *p, *z, *k;
	int hl, pl;
	if (parse_url(url, NULL, &u, NULL, NULL, NULL, &h, &hl, &p, &pl, NULL, NULL, NULL)) return NULL;
	z = u ? u : h;
	k = p ? p + pl : h + hl;
	return memacpy(z, k - z);
}

unsigned char *get_host_name(unsigned char *url)
{
	unsigned char *h;
	int hl;
	if (parse_url(url, NULL, NULL, NULL, NULL, NULL, &h, &hl, NULL, NULL, NULL, NULL, NULL)) return stracpy("");
	return memacpy(h, hl);
}

unsigned char *get_user_name(unsigned char *url)
{
	unsigned char *h;
	int hl;
	if (parse_url(url, NULL, &h, &hl, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL)) return NULL;
	return memacpy(h, hl);
}

unsigned char *get_pass(unsigned char *url)
{
	unsigned char *h;
	int hl;
	if (parse_url(url, NULL,NULL,  NULL, &h, &hl, NULL, NULL, NULL, NULL, NULL, NULL, NULL)) return NULL;
	return memacpy(h, hl);
}

unsigned char *get_port_str(unsigned char *url)
{
	unsigned char *h;
	int hl;
	if (parse_url(url, NULL, NULL, NULL, NULL, NULL, NULL, NULL, &h, &hl, NULL, NULL, NULL)) return NULL;
	return hl ? memacpy(h, hl) : NULL;
}

int get_port(unsigned char *url)
{
	unsigned char *h;
	int hl;
	long n = -1;
	if (parse_url(url, NULL, NULL, NULL, NULL, NULL, NULL, NULL, &h, &hl, NULL, NULL, NULL)) return -1;
	if (h) {
		n = strtol(h, NULL, 10);
		if (n && n < MAXINT) return n;
	}
	if ((h = get_protocol_name(url))) {
		int nn;
		get_prot_info(h, &nn, NULL, NULL, NULL, NULL);
		mem_free(h);
		n = nn;
	}
	return n;
}

void (*get_protocol_handle(unsigned char *url))(struct connection *)
{
	unsigned char *p;
	void (*f)(struct connection *) = NULL;
	int post = 0;
	if (!(p = get_protocol_name(url))) return NULL;
	get_prot_info(p, NULL, &f, NULL, &post, NULL);
	mem_free(p);
	if (!post && strchr(url, POST_CHAR)) return NULL;
	return f;
}

void (*get_external_protocol_function(unsigned char *url))(struct session *, unsigned char *)
{
	unsigned char *p;
	void (*f)(struct session *, unsigned char *) = NULL;
	int post = 0;
	if (!(p = get_protocol_name(url))) return NULL;
	get_prot_info(p, NULL, NULL, &f, &post, NULL);
	mem_free(p);
	if (!post && strchr(url, POST_CHAR)) return NULL;
	return f;
}

int url_bypasses_socks(unsigned char *url)
{
	int ret;
	unsigned char *p;
	if (!(p = get_protocol_name(url))) return 1;
	get_prot_info(p, NULL, NULL, NULL, NULL, &ret);
	mem_free(p);
	return ret;
}

unsigned char *get_url_data(unsigned char *url)
{
	unsigned char *d;
	if (parse_url(url, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, &d, NULL, NULL)) return NULL;
	return d;
}

#define dsep(x) (lo ? dir_sep(x) : (x) == '/')

void translate_directories(unsigned char *url)
{
	unsigned char *dd = get_url_data(url);
	unsigned char *s, *d;
	int lo = !casecmp(url, "file://", 7);
	if (!casecmp(url, "javascript:", 11)) return;
	if (!dd || dd == url/* || *--dd != '/'*/) return;
	if (!dsep(*dd)) {
		dd--;
		if (!dsep(*dd)) {
			dd++;
			memmove(dd + 1, dd, strlen(dd) + 1);
			*dd = '/';
		}
	}
	s = dd;
	d = dd;
	r:
	if (end_of_dir(url, s[0])) {
		memmove(d, s, strlen(s) + 1);
		return;
	}
	if (dsep(s[0]) && s[1] == '.' && dsep(s[2])) {