url.c

Internet Phone, Chat, Conferencing

/** Test if param @a tag matches to parameter string @a p. 
 */
#define URL_PARAM_MATCH(p, tag) \
 (strncasecmp(p, tag, strlen(tag)) == 0 && \
  (p[strlen(tag)] == '\0' || p[strlen(tag)] == ';' || p[strlen(tag)] == '='))

/**
 * Search for a parameter.
 *
 * This function searches for a parameter from a parameter list.
 *
 * If you want to test if there is parameter @b user=phone, 
 * call this function like 
 * @code if (url_param(url->url_param, "user=phone", NULL, 0))
 * @endcode
 *
 * @param params URL parameter string (excluding first semicolon)
 * @param tag    parameter name
 * @param value  string to which the parameter value is copied
 * @param vlen   length of string reserved for value
 *
 * @retval positive length of parameter value (including final NUL) if found
 * @retval zero     if not found.
 */
int url_param(char const *params, 
	      char const *tag, 
	      char value[], int vlen)
{
  size_t n, tlen = strlen(tag), flen;
  char *p;
  
  if (!params)
    return 0;

  for (p = (char *)params; *p; p += n + 1) {
    n = strcspn(p, ";");
    if (n < tlen) {
      if (p[n]) continue; else break;
    }
    if (strncasecmp(p, tag, tlen) == 0) {
      if (n == tlen) {
	if (vlen > 0)
	  value[0] = '\0';	
	return 1;
      }
      if (p[tlen] != '=')
	continue;
      flen = n - tlen - 1;
      if (flen >= (size_t)vlen)
	return flen + 1;
      memcpy(value, p + tlen + 1, flen);
      value[flen] = '\0';
      return flen + 1;
    }
    if (!p[n])
      break;
  }

  return 0;
}

/** Check for a parameter. 
 *
 * @deprecated 
 * Bad grammar. Use url_has_param().
 */
int url_have_param(char const *params, char const *tag)
{
  return url_param(params, tag, NULL, 0);
}

/** Check for a parameter. */
int url_has_param(url_t const *url, char const *tag)
{
  return url && url->url_params && url_param(url->url_params, tag, NULL, 0);
}

/** Add an parameter. */
int url_param_add(su_home_t *h, url_t *url, char const *param)
{
  /* XXX - should remove existing parameters? */
  int n = url->url_params ? strlen(url->url_params) + 1: 0;
  int nn = strlen(param) + 1;
  char *s = su_alloc(h, n + nn);

  if (!s)
    return -1;
  
  if (url->url_params)
    strcpy(s, url->url_params)[n - 1] = ';';
  strcpy(s + n, param);
  url->url_params = s;

  return 0;
}

int url_string_p(url_string_t const *url)
{
  return URL_STRING_P(url);
}

int url_is_string(url_string_t const *url)
{
  return URL_IS_STRING(url);
}

/** Strip transport-specific stuff. */
static
int url_strip_transport2(url_t *url, int modify)
{
  char *p, *d;
  size_t n;
  int semi;

  if (url->url_type != url_sip && url->url_type != url_sips)
    return 0;
  
  if (url->url_port != NULL) {
    if (!modify)
      return 1;
    url->url_port = NULL;
  }

  if (!url->url_params)
    return 0;

  for (d = p = (char *)url->url_params; *p; p += n + semi) {
    n = strcspn(p, ";");
    semi = (p[n] != '\0');

    if (modify && n == 0)
      continue;
    if (URL_PARAM_MATCH(p, "method"))
      continue;
    if (URL_PARAM_MATCH(p, "maddr"))
      continue;
    if (URL_PARAM_MATCH(p, "ttl"))
      continue;
    if (URL_PARAM_MATCH(p, "transport"))
      continue;

    if (p != d) {
      if (d != url->url_params)
	d++;
      if (p != d) {
	if (!modify)
	  return 1;
	memcpy(d, p, n + 1);
      }
    }
    d += n;
  }

  if (d == p)
    return 0;
  else if (d + 1 == p)		/* empty param */
    return 0;
  else if (!modify)
    return 1;

  if (d != url->url_params)
    *d = '\0';
  else 
    url->url_params = NULL;

  return 1;
}

/** Strip transport-specific stuff. 
 *
 * The function url_strip_transport() removes transport-specific parameters
 * from a SIP or SIPS URI.  These parameters include:
 * - the port number
 * - "maddr=" parameter
 * - "transport=" parameter
 * - "ttl=" parameter
 * - "method=" parameter
 *
 * @note
 * The @a url must be a pointer to a URL structure.
 *
 * @note
 * If the parameter string contains empty parameters, they are stripped, too.
 *
 * @return 
 * The function url_strip_transport() returns @e true, if the URL was
 * modified, @e false otherwise.
 */
int url_strip_transport(url_t *url)
{
  return url_strip_transport2(url, 1);
}

/** Check for transport-specific stuff. 
 *
 * The function url_have_transport() tests if there are transport-specific
 * parameters in a SIP or SIPS URI. These parameters include:
 * - the port number
 * - "maddr=" parameters
 * - "transport=" parameters
 * 
 * @note
 * The @a url must be a pointer to a URL structure.
 *
 * @return The function url_have_transport() returns @e true, if the URL
 * contains transport parameters, @e false otherwise.
 */
int url_have_transport(url_t const *url)
{
  return url_strip_transport2((url_t *)url, 0);
}

/** Compare two URLs.
 *
 * @note
 * The @a a and @a b must be pointers to URL structures.
 *
 * @note Currently, the url parameters are not compared. This is because the
 * url_cmp() is used to sort URLs: taking parameters into account makes that
 * impossible.
 */
int url_cmp(url_t const *a, url_t const *b)
{
  int rv;

  if ((a && a->url_type == url_any) || (b && b->url_type == url_any))
    return 0;

  if (!a || !b)
    return (a != NULL) - (b != NULL);

  if ((rv = a->url_type - b->url_type)) {
    /* presence and instant messaging URLs match magically with SIP */
    enum url_type_e a_type = a->url_type;
    enum url_type_e b_type = b->url_type;
    
    if (a_type == url_im || a_type == url_pres)
      a_type = url_sip;

    if (b_type == url_im || b_type == url_pres)
      b_type = url_sip;

    if (a_type != b_type)
      return rv;
  }

  if (a->url_type <= url_unknown && 
      ((rv = !a->url_scheme - !b->url_scheme) ||
       (a->url_scheme && b->url_scheme &&
	(rv = strcasecmp(a->url_scheme, b->url_scheme)))))
    return rv;

  if (a->url_host != b->url_host && 
      ((rv = !a->url_host - !b->url_host) ||
       (rv = strcasecmp(a->url_host, b->url_host))))
    return rv;

  if (a->url_port != b->url_port) {
    char const *a_port;
    char const *b_port;

    if (a->url_type != url_sip && a->url_type != url_sips)
      a_port = b_port = url_port_default(a->url_type);
    else if (host_is_ip_address(a->url_host))
      a_port = b_port = url_port_default(a->url_type);
    else
      a_port = b_port = "";
    
    if (a->url_port) a_port = a->url_port;
    if (b->url_port) b_port = b->url_port;

    if ((rv = strcmp(a_port, b_port)))
      return rv;
  }

  if (a->url_user != b->url_user) {
    if (a->url_user == NULL) return -1;
    if (b->url_user == NULL) return +1;
    if ((rv = strcmp(a->url_user, b->url_user)))
      return rv;
  }

#if 0
  if (a->url_path != b->url_path) {
    if (a->url_path == NULL) return -1;
    if (b->url_path == NULL) return +1;
    if ((rv = strcmp(a->url_path, b->url_path)))
      return rv;
  }
#endif

  return 0;
}

/** Return default port number corresponding to the url type */
char const *url_port_default(enum url_type_e url_type)
{
  switch (url_type) {
  case url_sip:			/* "sip:" */
    return "5060";
  case url_sips:		/* "sips:" */
    return "5061";
  case url_http:		/* "http:" */
    return "80";
  case url_https:		/* "https:" */
    return "443";
  case url_ftp:			/* "ftp:" */
  case url_file:		/* "file:" */
    return "21";
  case url_rtsp:		/* "rtsp:" */
  case url_rtspu:		/* "rtspu:" */
    return "554";
  case url_mailto:		/* "mailto:" */
    return "25";

  case url_any:			/* "*" */
    return "*";

  case url_msrp:
    return "9999";		/* XXXX */

  case url_tel:	
  case url_fax:	
  case url_modem:
  case url_im:	
  case url_pres:
  case url_cid:
  case url_wv:

  default:			/* Unknown scheme */
    return "";
  }
}

/** Return default transport name corresponding to the url type */
char const *url_tport_default(enum url_type_e url_type)
{
  switch (url_type) {
  case url_sip:
    return "*";
  case url_sips:
    return "tls";
  case url_http:
    return "tcp";
  case url_https:
    return "tls";
  case url_ftp:
  case url_file:
    return "tcp";
  case url_rtsp:
    return "tcp";
  case url_rtspu:
    return "udp";
  case url_mailto:
    return "tcp";
  case url_msrp:
    return "tcp";

  case url_any:			/* "*" */
  case url_tel:
  case url_fax:
  case url_modem:
  case url_im:
  case url_pres:
  case url_cid:
  case url_wv:

  default:			/* Unknown scheme */
    return "*";
  }
}


/** Return the URL port string */
char const *url_port(url_t const *u)
{
  if (!u)
    return "";
  else if (u->url_port)
    return u->url_port;
  else
    return url_port_default(u->url_type);
}

/** Sanitize URL.
 *
 * The function url_sanitize() adds a scheme to an incomplete URL.  It
 * modifies its parameter structure @a url.  Currently, the function follows
 * simple heuristics:
 *
 * - URL with host name starting with @c ftp. is an FTP URL
 * - URL with host name starting with @c www. is an HTTP URL
 * - URL with host and path, e.g., @c host/foo;bar, is an HTTP URL
 * - URL with host name, no path is a SIP URL.
 *
 * @param url pointer to URL struct to be sanitized (IN/OUT)
 *
 * @return
 * The function url_sanitize() returns 0 if it considers URL to be
 * sane, and -1 otherwise.
 */
int url_sanitize(url_t *url)
{
  if (!url)
    return -1;
  else if (url->url_scheme != NULL)
    /* xyzzy */;
  else if (url->url_host == NULL)
    return -1;
  else if (strncasecmp(url->url_host, "ftp.", strlen("ftp.")) == 0)
    url->url_type = url_ftp, url->url_scheme = "ftp", url->url_root = '/';
  else if (strncasecmp(url->url_host, "www.", strlen("www.")) == 0
	   || url->url_path)
    url->url_type = url_http, url->url_scheme = "http", url->url_root = '/';
  else
    url->url_type = url_sip, url->url_scheme = "sip";

  return 0;
}

#include <sofia-sip/su_md5.h>

static
void canon_update(su_md5_t *md5, char const *s, int n, char const *allow)
{
  char const *s0 = s, *b = s;

  for (;*s && s - s0 < n; s++) {
    char c;

    if (*s == '%' && IS_HEX(s[1]) && IS_HEX(s[2]) && s - s0 + 2 < n) {
#define   UNHEX(a) (a - (a >= 'a' ? 'a' - 10 : (a >= 'A' ? 'A' - 10 : '0')))
      c = (UNHEX(s[1]) << 4) | UNHEX(s[2]);
#undef    UNHEX
      if (c != '%' && c > ' ' && c < '\177' && 
	  (!strchr(EXCLUDED, c) || strchr(allow, c))) {
	if (b != s)
	  su_md5_iupdate(md5, b, s - b);
	su_md5_iupdate(md5, &c, 1);
	b = s + 3;
      }
      s += 2;
    }
  }
  if (b != s)
    su_md5_iupdate(md5, b, s - b);
}

/** Update MD5 sum with url-string contents */
static
void url_string_update(su_md5_t *md5, char const *s)
{
  int n;
  int hostpart = 1;
  enum url_type_e type = url_any;
  char const *at, *colon;
  char schema[48];

  if (s == NULL || strlen(s) == 0 || strcmp(s, "*") == 0) {
    su_md5_update(md5, "*\0\0*", 4);
    return;
  }

  n = strcspn(s, ":/?#");
  if (n >= sizeof schema) {
    su_md5_update(md5, ":", 1);
  } else if (n && s[n] == ':' ) {
    at = url_canonize(schema, s, n, "+");

    type = url_get_type(schema, at - schema);
    su_md5_iupdate(md5, schema, at - schema);

    hostpart = !url_type_is_opaque(type);
    s += n + 1;
  }
  else {
    su_md5_update(md5, "", 1);
  }
  
  if (type == url_sip || type == url_sips) {
    n = strcspn(s, "@#");	/* Opaque part */
    if (s[n] != '@')
      n = 0;
    n += strcspn(s + n, "/;?#");
  }
  else if (!hostpart || s[0] != '/') {
    n = strcspn(s, "/;?#");	/* Opaque part */
  }
  else if (s[1] == '/') {
    s += 2;
    n = strcspn(s, "/;?#");	/* Until path, query or fragment */
  }
  else {
    /* foo:/bar */
    su_md5_update(md5, "\0\0", 2); /* user, host */
    su_md5_striupdate(md5, url_port_default(type));
    return;
  }
  
  if (!hostpart) {
    char const *colon = memchr(s, ':', n);
    
    if (colon) n = colon - s;
    canon_update(md5, s, n, ""); /* user */
    su_md5_update(md5, "\0", 1); /* host, no port */
    su_md5_striupdate(md5, url_port_default(type));
    return;
  }

  at = memchr(s, '@', n);

  if (at) {
    char const *allow = 
      (type == url_sip || type == url_sips) 
      ? SIP_USER_UNRESERVED
      : USER_UNRESERVED;
    colon = memchr(s, ':', at - s);

    /* Updated only user part */
    if (colon)
      canon_update(md5, s, colon - s, allow);
    else
      canon_update(md5, s, at - s, allow);
    n = n - (at + 1 - s);
    s = at + 1;
  }
  else
    su_md5_iupdate(md5, "", 1);	/* user */

  colon = memchr(s, ':', n);	/* XXX - IPv6! */
  if (colon) {
    canon_update(md5, s, colon - s, ""); /* host */
    canon_update(md5, colon + 1, (s + n) - (colon + 1), "");
  }
  else {
    canon_update(md5, s, n, ""); /* host */
    su_md5_strupdate(md5, url_port_default(type));	/* port */
  }

  /* ignore parameters/path/headers.... */
}


/** Update md5 digest with contents of URL.
 *
 */
void url_update(su_md5_t *md5, url_t const *url)
{
  if (url_string_p((url_string_t *)url)) {
    url_string_update(md5, (char const *)url);
  }
  else {
    SU_MD5_STRI0UPDATE(md5, url->url_scheme);
    SU_MD5_STRI0UPDATE(md5, url->url_user);
    SU_MD5_STRI0UPDATE(md5, url->url_host);
    su_md5_striupdate(md5, URL_PORT(url));
    /* XXX - parameters/path.... */
    /* SU_MD5_STRI0UPDATE(md5, url->url_path); */
  }
}

/** Calculate a digest from URL contents. */
void url_digest(void *hash, int hsize, url_t const *url, char const *key)
{
  su_md5_t md5[1];
  uint8_t digest[SU_MD5_DIGEST_SIZE];

  su_md5_init(md5);
  if (key) su_md5_strupdate(md5, key);
  url_update(md5, url);
  su_md5_digest(md5, digest);

  if (hsize > SU_MD5_DIGEST_SIZE) {
    memset((char *)hash + SU_MD5_DIGEST_SIZE, 0, hsize - SU_MD5_DIGEST_SIZE);
    hsize = SU_MD5_DIGEST_SIZE;
  }

  memcpy(hash, digest, hsize);
}