url.c

Sofia SIP is an open-source SIP User-Agent library, compliant with the IETF RFC3261 specification.

  case 't': case 'T': 
    test_scheme(tel); break;
  case 'w': case 'W': 
    test_scheme(wv); break;


  default: break;
  }

#undef test_scheme

  if (len != span_unreserved(scheme))
    return url_invalid;
  else
    return url_unknown;
}

/**
 * Decode a URL.
 *
 * This function decodes a (SIP) URL string to a url_t structure.
 *
 * @param url structure to store the parsing result
 * @param s   NUL-terminated string to be parsed
 *
 * @note The parsed string @a s will be modified when parsing it.
 *
 * @retval 0 if successful, 
 * @retval -1 otherwise.
 */
static
int _url_d(url_t *url, char *s)
{
  size_t n, p;
  char *s0, rest_c, *host, *user;
  int have_authority = 1;

  memset(url, 0, sizeof(*url));
  
  if (strcmp(s, "*") == 0) {
    url->url_type = url_any;
    url->url_scheme = "*";
    return 0;
  }

  s0 = s;

  n = strcspn(s, ":/?#");

  if (n && s[n] == ':') {
    char *scheme;
    url->url_scheme = scheme = s; s[n] = '\0'; s = s + n + 1;

    if (!(scheme = url_canonize(scheme, scheme, SIZE_MAX, 0, "+")))
      return -1;

    n = scheme - url->url_scheme;

    url->url_type = url_get_type(url->url_scheme, n);

    have_authority = !url_type_is_opaque(url->url_type);
  }
  else {
    url->url_type = url_unknown;
  }

  user = NULL, host = s;

  if (url->url_type == url_sip || url->url_type == url_sips) {
    /* SIP URL may have /;? in user part but no path */
    /* user-unreserved  =  "&" / "=" / "+" / "$" / "," / ";" / "?" / "/" */
    /* Some #*@#* phones include unescaped # there, too */
    n = strcspn(s, "@/;?#");
    p = strcspn(s + n, "@");
    if (s[n + p] == '@') {
      n += p;
      user = s;
      host = s + n + 1;
    }

    n += strcspn(s + n, "/;?#");
  }
  else if (have_authority) {
    if (url->url_type == url_wv) {
      /* WV URL may have / in user part */
      n = strcspn(s, "@#?;");
      if (s[n] == '@') {
	user = s;
	host = s + n + 1;
	n += strcspn(s + n, ";?#");
      }
    }
    else if (host[0] == '/' && host[1] != '/') {
      /* foo:/bar or /bar - no authority, just path */
      url->url_root = '/';	/* Absolute path */
      host = NULL, n = 0;
    }
    else {
      if (host[0] == '/' && host[1] == '/') {
	/* We have authority, / / foo or foo */
	host += 2; s += 2, url->url_root = '/';
	n = strcspn(s, "/?#@[]");
      }
      else 
	n = strcspn(s, "@;/?#");

      if (s[n] == '@')
	user = host, host = user + n + 1;

      n += strcspn(s + n, ";/?#");	/* Find path, query and/or fragment */
    }
  }
  else /* !have_authority */ {
    user = host, host = NULL;
    if (url->url_type != url_invalid)
      n = strcspn(s, "/;?#");	/* Find params, query and/or fragment */
    else
      n = strcspn(s, "#");
  }

  rest_c = s[n]; s[n] = 0; s = rest_c ? s + n + 1 : NULL;

  if (user) {
    if (host) host[-1] = '\0';
    url->url_user = user;
    if (url->url_type != url_unknown) {
      n = strcspn(user, ":");
      if (user[n]) {
	user[n] = '\0';
	url->url_password = user + n + 1;
      }
    }
  }

  if (host) {
    url->url_host = host;
    /* IPv6 (and in some cases, IPv4) addresses are quoted with [] */
    if (host[0] == '[') {
      n = strcspn(host, "]");
      if (host[n] && (host[n + 1] == '\0' || host[n + 1] == ':'))
	n++;
      else
	n = 0;
    }
    else {
      n = strcspn(host, ":");
    }

    /* We allow empty host by default */
    if (n == 0) switch (url->url_type) {
    case url_sip:
    case url_sips:
    case url_im:
    case url_pres:
      return -1;
    default:
      break;
    }
    
    if (host[n] == ':') {
      char *port = host + n + 1;
      url->url_port = port;
      switch (url->url_type) {
      case url_any:
      case url_sip:
      case url_sips:
      case url_http:
      case url_https:
      case url_ftp:
      case url_file:
      case url_rtsp:
      case url_rtspu:
	if (!url_canonize2(port, port, SIZE_MAX, 0, RESERVED_MASK))
	  return -1;

	/* Check that port is really numeric or wildcard */
	/* Port can be *digit, empty string or "*" */
	while (*port >= '0' && *port <= '9')
	  port++;
	
	if (port != url->url_port ? port[0] != '\0' 
	    : (port[0] != '*' || port[1] != '\0'))
	  return -1;
      }
      host[n] = 0;
    }
  }

  if (rest_c == '/') {
    url->url_path = s; n = strcspn(s, "?#");
    rest_c = s[n]; s[n] = 0; s = rest_c ? s + n + 1 : NULL;
  }
  if (rest_c == ';') {
    url->url_params = s; n = strcspn(s, "?#");
    rest_c = s[n]; s[n] = 0; s = rest_c ? s + n + 1 : NULL;
  }
  if (rest_c == '?') {
    url->url_headers = s; n = strcspn(s, "#");
    rest_c = s[n]; s[n] = 0; s = rest_c ? s + n + 1 : NULL;
  }
  if (rest_c == '#') {
    url->url_fragment = s;
    rest_c = '\0';
  }
  if (rest_c)
    return -1;

  return 0;
}

/* Unreserved things */

/**
 * Decode a URL.
 *
 * This function decodes a URL string to a url_t structure.
 *
 * @param url structure to store the parsing result
 * @param s   NUL-terminated string to be parsed
 *
 * @note The parsed string @a s will be modified when parsing it.
 *
 * @retval 0 if successful, 
 * @retval -1 otherwise.
 */
int url_d(url_t *url, char *s)
{
  if (url == NULL || _url_d(url, s) < 0)
    return -1;

  /* Canonize  URL */
  /* scheme is canonized by _url_d() */
  if (url->url_type == url_sip || url->url_type == url_sips) {

#   define SIP_USER_UNRESERVED "&=+$,;?/"
    s = (char *)url->url_user;
    if (s && !url_canonize(s, s, SIZE_MAX, 0, SIP_USER_UNRESERVED))
      return -1;

    /* Having different charset in user and password does not make sense */
    /* but that is how it is defined in RFC 3261 */
#   define SIP_PASS_UNRESERVED "&=+$,"
    s = (char *)url->url_password;
    if (s && !url_canonize(s, s, SIZE_MAX, 0, SIP_PASS_UNRESERVED))
      return -1;

  }
  else {

#   define USER_UNRESERVED "&=+$,;"
    s = (char *)url->url_user;
    if (s && !url_canonize(s, s, SIZE_MAX, 0, USER_UNRESERVED))
      return -1;

#   define PASS_UNRESERVED "&=+$,;:"
    s = (char *)url->url_password;
    if (s && !url_canonize(s, s, SIZE_MAX, 0, PASS_UNRESERVED))
      return -1;
  }

  s = (char *)url->url_host;
  if (s && !url_canonize2(s, s, SIZE_MAX, 0, RESERVED_MASK))
    return -1;

  /* port is canonized by _url_d() */
  s = (char *)url->url_path;
  if (s && !url_canonize(s, s, SIZE_MAX, 
			 /* Allow all URI characters but ? */
			 /* Allow unescaped /;?@, - but do not convert */
			 SYN33('/') | SYN33(';') | SYN33('=') | SYN33('@') |
			 SYN33(','),
			 /* Convert escaped :&+$ to unescaped */
			 ":&+$"))
    return -1;

  s = (char *)url->url_params;
  if (s && !url_canonize(s, s, SIZE_MAX,
			 /* Allow all URI characters but ? */
			 /* Allow unescaped ;=@, - but do not convert */
			 SYN33(';') | SYN33('=') | SYN33('@') | SYN33(','),
			 /* Convert escaped /:&+$ to unescaped */
			 "/:&+$"))
    return -1;
  
  /* Unhex alphanumeric and unreserved URI characters */
  s = (char *)url->url_headers;
  if (s && !url_canonize3(s, s, SIZE_MAX, RESERVED_MASK))
    return -1;

  /* Allow all URI characters (including reserved ones) */
  s = (char *)url->url_fragment;
  if (s && !url_canonize2(s, s, SIZE_MAX, 0, URIC_MASK))
    return -1;

  return 0;
}

/** Encode an URL. 
 * 
 * The function url_e() combines a URL from substrings in url_t structure
 * according the @ref url_syntax "URL syntax" presented above.  The encoded
 * @a url is stored in a @a buffer of @a n bytes.
 *
 * @param buffer memory area to store the encoded @a url.
 * @param n      size of @a buffer.
 * @param url    URL to be encoded.
 *
 * @return 
 * Return the number of bytes in the encoding.  
 *
 * @note The function follows the convention set by C99 snprintf().  Even if
 * the result does not fit into the @a buffer and it is truncated, the
 * function returns the number of bytes in an untruncated encoding.
 */
issize_t url_e(char buffer[], isize_t n, url_t const *url)
{
  size_t i;
  char *b = buffer;
  size_t m = n;
  int do_copy = n > 0;

  if (url == NULL)
    return -1;

  if (URL_STRING_P(url)) {
    char const *u = (char *)url;
    i = strlen(u);
    if (!buffer)
      return i;

    if (i >= n) {
      memcpy(buffer, u, n - 2);
      buffer[n - 1] = '\0';
    } else {
      memcpy(buffer, u, i + 1);
    }

    return i;
  }


  if (url->url_type == url_any) {
    if (b && m > 0) {
      if (m > 1) strcpy(b, "*"); else b[0] = '\0';
    }
    return 1;
  }

  if (url->url_scheme && url->url_scheme[0]) {
    i = strlen(url->url_scheme) + 1;
    if (do_copy && (do_copy = i <= n)) {
      memcpy(b, url->url_scheme, i - 1);
      b[i - 1] = ':';
    }
    b += i; n -= i;
  }

  if (url->url_root && (url->url_host || url->url_user)) {
    if (do_copy && (do_copy = 2 <= n))
      memcpy(b, "//", 2);
    b += 2; n -= 2;
  }

  if (url->url_user) {
    i = strlen(url->url_user);
    if (do_copy && (do_copy = i <= n))
      memcpy(b, url->url_user, i);
    b += i; n -= i;

    if (url->url_password) {
      if (do_copy && (do_copy = 1 <= n))
	*b = ':'; 
      b++; n--;
      i = strlen(url->url_password);
      if (do_copy && (do_copy = i <= n))
	memcpy(b, url->url_password, i);
      b += i; n -= i;
    }

    if (url->url_host) {
      if (do_copy && (do_copy = 1 <= n))
	*b = '@'; 
      b++; n--;
    }
  }

  if (url->url_host) {
    i = strlen(url->url_host);
    if (do_copy && (do_copy = i <= n))
      memcpy(b, url->url_host, i);
    b += i; n -= i;

    if (url->url_port) {
      i = strlen(url->url_port) + 1;
      if (do_copy && (do_copy = i <= n)) {
	b[0] = ':';
	memcpy(b + 1, url->url_port, i - 1);
      } 
      b += i; n -= i;
    }
  }

  if (url->url_path) {
    if (url->url_root) {
      if (do_copy && (do_copy = 1 <= n))
	b[0] = '/';
      b++, n--;
    }
    i = strlen(url->url_path);
    if (do_copy && (do_copy = i < n))
      memcpy(b, url->url_path, i);
    b += i; n -= i;
  }
  
  {
    static char const sep[] = ";?#";
    char const *pp[3];
    size_t j;

    pp[0] = url->url_params;
    pp[1] = url->url_headers;
    pp[2] = url->url_fragment;

    for (j = 0; j < 3; j++) {
      char const *p = pp[j];
      if (!p) continue;
      i = strlen(p) + 1;
      if (do_copy && (do_copy = i <= n)) {
	*b = sep[j];
	memcpy(b + 1, p, i - 1);
      }
      b += i; n -= i;
    }
  }

  if (do_copy && (do_copy = 1 <= n))
    *b = '\0';
  else if (buffer && m > 0)
    buffer[m - 1] = '\0';

  assert((size_t)(b - buffer) == (size_t)(m - n));
  
  /* This follows the snprintf(C99) return value, 
   * Number of characters written (excluding NUL)
   */
  return b - buffer;
}


/** Calculate the length of URL when encoded. 
 *
 */
isize_t url_len(url_t const * url)
{
  size_t rv = 0;

  if (url->url_scheme) rv += strlen(url->url_scheme) + 1; /* plus ':' */
  if (url->url_user) {
    rv += strlen(url->url_user);
    if (url->url_password) 
      rv += strlen(url->url_password) + 1;   /* plus ':' */
    rv += url->url_host != NULL;  /* plus '@' */
  }
  if (url->url_host) rv += strlen(url->url_host);
  if (url->url_port) rv += strlen(url->url_port) + 1;	        /* plus ':' */
  if (url->url_path) rv += strlen(url->url_path) + 1;     /* plus initial / */
  if (url->url_params) rv += strlen(url->url_params) + 1; /* plus initial ; */
  if (url->url_headers) rv += strlen(url->url_headers) + 1;	/* plus '?' */
  if (url->url_fragment) rv += strlen(url->url_fragment) + 1;   /* plus '#' */

  return rv;
}

/**@def URL_E(buf, end, url)
 * Encode an URL: use @a buf up to @a end. 
 * @hideinitializer
 */

/** 
 * Calculate the size of strings associated with a #url_t sructure.
 *
 * @param url pointer to a #url_t structure or string
 * @return Number of bytes for URL
 */
isize_t url_xtra(url_t const *url)
{
  size_t xtra;

  if (URL_STRING_P(url)) {
    xtra = strlen((char const *)url) + 1;
  }
  else {
    size_t len_scheme, len_user, len_password,
      len_host, len_port, len_path, len_params, 
      len_headers, len_fragment;

    len_scheme = (url->url_type <= url_unknown && url->url_scheme) ?
      strlen(url->url_scheme) + 1 : 0;
    len_user = url->url_user ? strlen(url->url_user) + 1 : 0;
    len_password = url->url_password ? strlen(url->url_password) + 1 : 0;
    len_host = url->url_host ? strlen(url->url_host) + 1 : 0;
    len_port = url->url_port ? strlen(url->url_port) + 1 : 0;
    len_path = url->url_path ? strlen(url->url_path) + 1 : 0;
    len_params = url->url_params ? strlen(url->url_params) + 1 : 0;
    len_headers = url->url_headers ? strlen(url->url_headers) + 1 : 0;
    len_fragment = url->url_fragment ? strlen(url->url_fragment) + 1 : 0;

    xtra =
      len_scheme + len_user + len_password + len_host + len_port +
      len_path + len_params + len_headers + len_fragment;
  }

  return xtra;
}

su_inline
char *copy(char *buf, char *end, char const *src)
{
#if HAVE_MEMCCPY
  char *b = memccpy(buf, src, '\0', end - buf);
  if (b)
    return b;
  else
    return end + strlen(src + (end - buf)) + 1;
#else
  for (; buf < end && (*buf = *src); buf++, src++)
    ;

  if (buf >= end) 
    while (*src++)
      buf++;

  return buf + 1;
#endif
}