uri.java

来自「JAVA的一些源码 JAVA2 STANDARD EDITION DEVELO」· Java 代码 · 共 1,924 行 · 第 1/5 页

      	index = (colonIndex > start) ? colonIndex : end;
      	hasPort = (index != end);
      }
    }
    host = p_uriSpec.substring(start, index);
    int port = -1;
    if (host.length() > 0) {
      // port
      if (hasPort) {
        index++;
        start = index;
        while (index < end) {
          index++;
        }
        String portStr = p_uriSpec.substring(start, index);
        if (portStr.length() > 0) {
          // REVISIT: Remove this code.
          /** for (int i = 0; i < portStr.length(); i++) {
            if (!isDigit(portStr.charAt(i))) {
              throw new MalformedURIException(
                   portStr +
                   " is invalid. Port should only contain digits!");
            }
          }**/
          // REVISIT: Remove this code.
          // Store port value as string instead of integer.
          try {
            port = Integer.parseInt(portStr);
            if (port == -1) --port;
          }
          catch (NumberFormatException nfe) {
            port = -2;
          }
        }
      }
    }
    
    if (isValidServerBasedAuthority(host, port, userinfo)) {
      m_host = host;
      m_port = port;
      m_userinfo = userinfo;
      return true;
    }
    // Note: Registry based authority is being removed from a
    // new spec for URI which would obsolete RFC 2396. If the
    // spec is added to XML errata, processing of reg_name
    // needs to be removed. - mrglavas.
    else if (isValidRegistryBasedAuthority(p_uriSpec)) {
      m_regAuthority = p_uriSpec;
      return true;
    }
    return false;
  }
  
  /**
   * Determines whether the components host, port, and user info
   * are valid as a server authority.
   * 
   * @param host the host component of authority
   * @param port the port number component of authority
   * @param userinfo the user info component of authority
   * 
   * @return true if the given host, port, and userinfo compose
   * a valid server authority
   */
  private boolean isValidServerBasedAuthority(String host, int port, String userinfo) {
    
    // Check if the host is well formed.
    if (!isWellFormedAddress(host)) {
      return false;
    }
    
    // Check that port is well formed if it exists.
    // REVISIT: There's no restriction on port value ranges, but
    // perform the same check as in setPort to be consistent. Pass
    // in a string to this method instead of an integer.
    if (port < -1 || port > 65535) {
      return false;
    }
    
    // Check that userinfo is well formed if it exists.
    if (userinfo != null) {
      // Userinfo can contain alphanumerics, mark characters, escaped
      // and ';',':','&','=','+','$',','
      int index = 0;
      int end = userinfo.length();
      char testChar = '\0';
      while (index < end) {
        testChar = userinfo.charAt(index);
        if (testChar == '%') {
          if (index+2 >= end ||
            !isHex(userinfo.charAt(index+1)) ||
            !isHex(userinfo.charAt(index+2))) {
            return false;
          }
          index += 2;
        }
        else if (!isUserinfoCharacter(testChar)) {
          return false;
        }
        ++index;
      }
    }
    return true;
  }
  
  /**
   * Determines whether the given string is a registry based authority.
   * 
   * @param authority the authority component of a URI
   * 
   * @return true if the given string is a registry based authority
   */
  private boolean isValidRegistryBasedAuthority(String authority) {
    int index = 0;
    int end = authority.length();
    char testChar;
  	
    while (index < end) {
      testChar = authority.charAt(index);
      
      // check for valid escape sequence
      if (testChar == '%') {
        if (index+2 >= end ||
            !isHex(authority.charAt(index+1)) ||
            !isHex(authority.charAt(index+2))) {
            return false;
        }
        index += 2;
      }
      // can check against path characters because the set
      // is the same except for '/' which we've already excluded.
      else if (!isPathCharacter(testChar)) {
        return false;
      }
      ++index;
    }
    return true;
  }
  	
 /**
  * Initialize the path for this URI from a URI string spec.
  *
  * @param p_uriSpec the URI specification (cannot be null)
  * @param p_nStartIndex the index to begin scanning from
  *
  * @exception MalformedURIException if p_uriSpec violates syntax rules
  */
  private void initializePath(String p_uriSpec, int p_nStartIndex)
                 throws MalformedURIException {
    if (p_uriSpec == null) {
      throw new MalformedURIException(
                "Cannot initialize path from null string!");
    }

    int index = p_nStartIndex;
    int start = p_nStartIndex;
    int end = p_uriSpec.length();
    char testChar = '\0';

    // path - everything up to query string or fragment
    if (start < end) {
    	// RFC 2732 only allows '[' and ']' to appear in the opaque part.
    	if (getScheme() == null || p_uriSpec.charAt(start) == '/') {
    	
            // Scan path.
            // abs_path = "/"  path_segments
            // rel_path = rel_segment [ abs_path ]
            while (index < end) {
                testChar = p_uriSpec.charAt(index);
            
                // check for valid escape sequence
                if (testChar == '%') {
                    if (index+2 >= end ||
                    !isHex(p_uriSpec.charAt(index+1)) ||
                    !isHex(p_uriSpec.charAt(index+2))) {
                        throw new MalformedURIException(
                            "Path contains invalid escape sequence!");
                    }
                    index += 2;
                }
                // Path segments cannot contain '[' or ']' since pchar
                // production was not changed by RFC 2732.
                else if (!isPathCharacter(testChar)) {
      	            if (testChar == '?' || testChar == '#') {
      	                break;
      	            }
                    throw new MalformedURIException(
                        "Path contains invalid character: " + testChar);
                }
                ++index;
            }
        }
        else {
            
            // Scan opaque part.
            // opaque_part = uric_no_slash *uric
            while (index < end) {
                testChar = p_uriSpec.charAt(index);
            
                if (testChar == '?' || testChar == '#') {
                    break;
      	        }
                
                // check for valid escape sequence
                if (testChar == '%') {
                    if (index+2 >= end ||
                    !isHex(p_uriSpec.charAt(index+1)) ||
                    !isHex(p_uriSpec.charAt(index+2))) {
                        throw new MalformedURIException(
                            "Opaque part contains invalid escape sequence!");
                    }
                    index += 2;
                }
                // If the scheme specific part is opaque, it can contain '['
                // and ']'. uric_no_slash wasn't modified by RFC 2732, which
                // I've interpreted as an error in the spec, since the 
                // production should be equivalent to (uric - '/'), and uric
                // contains '[' and ']'. - mrglavas
                else if (!isURICharacter(testChar)) {
                    throw new MalformedURIException(
                        "Opaque part contains invalid character: " + testChar);
                }
                ++index;
            }
        }
    }
    m_path = p_uriSpec.substring(start, index);

    // query - starts with ? and up to fragment or end
    if (testChar == '?') {
      index++;
      start = index;
      while (index < end) {
        testChar = p_uriSpec.charAt(index);
        if (testChar == '#') {
          break;
        }
        if (testChar == '%') {
           if (index+2 >= end ||
              !isHex(p_uriSpec.charAt(index+1)) ||
              !isHex(p_uriSpec.charAt(index+2))) {
            throw new MalformedURIException(
                    "Query string contains invalid escape sequence!");
           }
           index += 2;
        }
        else if (!isURICharacter(testChar)) {
          throw new MalformedURIException(
                "Query string contains invalid character: " + testChar);
        }
        index++;
      }
      m_queryString = p_uriSpec.substring(start, index);
    }

    // fragment - starts with #
    if (testChar == '#') {
      index++;
      start = index;
      while (index < end) {
        testChar = p_uriSpec.charAt(index);

        if (testChar == '%') {
           if (index+2 >= end ||
              !isHex(p_uriSpec.charAt(index+1)) ||
              !isHex(p_uriSpec.charAt(index+2))) {
            throw new MalformedURIException(
                    "Fragment contains invalid escape sequence!");
           }
           index += 2;
        }
        else if (!isURICharacter(testChar)) {
          throw new MalformedURIException(
                "Fragment contains invalid character: "+testChar);
        }
        index++;
      }
      m_fragment = p_uriSpec.substring(start, index);
    }
  }

 /**
  * Get the scheme for this URI.
  *
  * @return the scheme for this URI
  */
  public String getScheme() {
    return m_scheme;
  }

 /**
  * Get the scheme-specific part for this URI (everything following the
  * scheme and the first colon). See RFC 2396 Section 5.2 for spec.
  *
  * @return the scheme-specific part for this URI
  */
  public String getSchemeSpecificPart() {
    StringBuffer schemespec = new StringBuffer();

    if (m_host != null || m_regAuthority != null) {
      schemespec.append("//");
    
      // Server based authority.
      if (m_host != null) {

        if (m_userinfo != null) {
          schemespec.append(m_userinfo);
          schemespec.append('@');
        }
        
        schemespec.append(m_host);
        
        if (m_port != -1) {
          schemespec.append(':');
          schemespec.append(m_port);
        }
      }
      // Registry based authority.
      else {
      	schemespec.append(m_regAuthority);
      }
    }

    if (m_path != null) {
      schemespec.append((m_path));
    }

    if (m_queryString != null) {
      schemespec.append('?');
      schemespec.append(m_queryString);
    }

    if (m_fragment != null) {
      schemespec.append('#');
      schemespec.append(m_fragment);
    }

    return schemespec.toString();
  }

 /**
  * Get the userinfo for this URI.
  *
  * @return the userinfo for this URI (null if not specified).
  */
  public String getUserinfo() {
    return m_userinfo;
  }

  /**
  * Get the host for this URI.
  *
  * @return the host for this URI (null if not specified).
  */
  public String getHost() {
    return m_host;
  }

 /**
  * Get the port for this URI.
  *
  * @return the port for this URI (-1 if not specified).
  */
  public int getPort() {
    return m_port;
  }
  
  /**
   * Get the registry based authority for this URI.
   * 
   * @return the registry based authority (null if not specified).
   */
  public String getRegBasedAuthority() {
    return m_regAuthority;
  }

 /**
  * Get the path for this URI (optionally with the query string and
  * fragment).
  *
  * @param p_includeQueryString if true (and query string is not null),
  *                             then a "?" followed by the query string
  *                             will be appended
  * @param p_includeFragment if true (and fragment is not null),