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

    StringBuffer uriSpecString = new StringBuffer();

    if (m_scheme != null) {
      uriSpecString.append(m_scheme);
      uriSpecString.append(':');
    }
    uriSpecString.append(getSchemeSpecificPart());
    return uriSpecString.toString();
  }

 /**
  * Get the indicator as to whether this URI uses the "generic URI"
  * syntax.
  *
  * @return true if this URI uses the "generic URI" syntax, false
  *         otherwise
  */
  public boolean isGenericURI() {
    // presence of the host (whether valid or empty) means
    // double-slashes which means generic uri
    return (m_host != null);
  }

 /**
  * Determine whether a scheme conforms to the rules for a scheme name.
  * A scheme is conformant if it starts with an alphanumeric, and
  * contains only alphanumerics, '+','-' and '.'.
  *
  * @return true if the scheme is conformant, false otherwise
  */
  public static boolean isConformantSchemeName(String p_scheme) {
    if (p_scheme == null || p_scheme.trim().length() == 0) {
      return false;
    }

    if (!isAlpha(p_scheme.charAt(0))) {
      return false;
    }

    char testChar;
    int schemeLength = p_scheme.length();
    for (int i = 1; i < schemeLength; ++i) {
      testChar = p_scheme.charAt(i);
      if (!isSchemeCharacter(testChar)) {
        return false;
      }
    }

    return true;
  }

 /**
  * Determine whether a string is syntactically capable of representing
  * a valid IPv4 address, IPv6 reference or the domain name of a network host. 
  * A valid IPv4 address consists of four decimal digit groups separated by a
  * '.'. Each group must consist of one to three digits. See RFC 2732 Section 3,
  * and RFC 2373 Section 2.2, for the definition of IPv6 references. A hostname 
  * consists of domain labels (each of which must begin and end with an alphanumeric 
  * but may contain '-') separated & by a '.'. See RFC 2396 Section 3.2.2.
  *
  * @return true if the string is a syntactically valid IPv4 address, 
  * IPv6 reference or hostname
  */
  public static boolean isWellFormedAddress(String address) {
    if (address == null) {
      return false;
    }

    int addrLength = address.length();
    if (addrLength == 0) {
      return false;
    }
    
    // Check if the host is a valid IPv6reference.
    if (address.startsWith("[")) {
      return isWellFormedIPv6Reference(address);
    }

    // Cannot start with a '.', '-', or end with a '-'.
    if (address.startsWith(".") || 
        address.startsWith("-") || 
        address.endsWith("-")) {
      return false;
    }

    // rightmost domain label starting with digit indicates IP address
    // since top level domain label can only start with an alpha
    // see RFC 2396 Section 3.2.2
    int index = address.lastIndexOf('.');
    if (address.endsWith(".")) {
      index = address.substring(0, index).lastIndexOf('.');
    }

    if (index+1 < addrLength && isDigit(address.charAt(index+1))) {
      return isWellFormedIPv4Address(address);
    }
    else {
      // hostname      = *( domainlabel "." ) toplabel [ "." ]
      // domainlabel   = alphanum | alphanum *( alphanum | "-" ) alphanum
      // toplabel      = alpha | alpha *( alphanum | "-" ) alphanum
      
      // RFC 2396 states that hostnames take the form described in 
      // RFC 1034 (Section 3) and RFC 1123 (Section 2.1). According
      // to RFC 1034, hostnames are limited to 255 characters.
      if (addrLength > 255) {
      	return false;
      }
      
      // domain labels can contain alphanumerics and '-"
      // but must start and end with an alphanumeric
      char testChar;
      int labelCharCount = 0;

      for (int i = 0; i < addrLength; i++) {
        testChar = address.charAt(i);
        if (testChar == '.') {
          if (!isAlphanum(address.charAt(i-1))) {
            return false;
          }
          if (i+1 < addrLength && !isAlphanum(address.charAt(i+1))) {
            return false;
          }
          labelCharCount = 0;
        }
        else if (!isAlphanum(testChar) && testChar != '-') {
          return false;
        }
        // RFC 1034: Labels must be 63 characters or less.
        else if (++labelCharCount > 63) {
          return false;
        }
      }
    }
    return true;
  }
  
  /**
   * <p>Determines whether a string is an IPv4 address as defined by 
   * RFC 2373, and under the further constraint that it must be a 32-bit
   * address. Though not expressed in the grammar, in order to satisfy 
   * the 32-bit address constraint, each segment of the address cannot 
   * be greater than 255 (8 bits of information).</p>
   *
   * <p><code>IPv4address = 1*3DIGIT "." 1*3DIGIT "." 1*3DIGIT "." 1*3DIGIT</code></p>
   *
   * @return true if the string is a syntactically valid IPv4 address
   */
  public static boolean isWellFormedIPv4Address(String address) {
      
      int addrLength = address.length();
      char testChar;
      int numDots = 0;
      int numDigits = 0;

      // make sure that 1) we see only digits and dot separators, 2) that
      // any dot separator is preceded and followed by a digit and
      // 3) that we find 3 dots
      //
      // RFC 2732 amended RFC 2396 by replacing the definition 
      // of IPv4address with the one defined by RFC 2373. - mrglavas
      //
      // IPv4address = 1*3DIGIT "." 1*3DIGIT "." 1*3DIGIT "." 1*3DIGIT
      //
      // One to three digits must be in each segment.
      for (int i = 0; i < addrLength; i++) {
        testChar = address.charAt(i);
        if (testChar == '.') {
          if ((i > 0 && !isDigit(address.charAt(i-1))) || 
              (i+1 < addrLength && !isDigit(address.charAt(i+1)))) {
            return false;
          }
          numDigits = 0;
          if (++numDots > 3) {
            return false;
          }
        }
        else if (!isDigit(testChar)) {
          return false;
        }
        // Check that that there are no more than three digits
        // in this segment.
        else if (++numDigits > 3) {
          return false;
        }
        // Check that this segment is not greater than 255.
        else if (numDigits == 3) {
          char first = address.charAt(i-2);
          char second = address.charAt(i-1);
          if (!(first < '2' || 
               (first == '2' && 
               (second < '5' || 
               (second == '5' && testChar <= '5'))))) {
            return false;
          }
        }
      }
      return (numDots == 3);
  }
  
  /**
   * <p>Determines whether a string is an IPv6 reference as defined
   * by RFC 2732, where IPv6address is defined in RFC 2373. The 
   * IPv6 address is parsed according to Section 2.2 of RFC 2373,
   * with the additional constraint that the address be composed of
   * 128 bits of information.</p>
   *
   * <p><code>IPv6reference = "[" IPv6address "]"</code></p>
   *
   * <p>Note: The BNF expressed in RFC 2373 Appendix B does not 
   * accurately describe section 2.2, and was in fact removed from
   * RFC 3513, the successor of RFC 2373.</p>
   *
   * @return true if the string is a syntactically valid IPv6 reference
   */
  public static boolean isWellFormedIPv6Reference(String address) {

      int addrLength = address.length();
      int index = 1;
      int end = addrLength-1;
      
      // Check if string is a potential match for IPv6reference.
      if (!(addrLength > 2 && address.charAt(0) == '[' 
          && address.charAt(end) == ']')) {
          return false;
      }
      
      // Counter for the number of 16-bit sections read in the address.
      int [] counter = new int[1];
      
      // Scan hex sequence before possible '::' or IPv4 address.
      index = scanHexSequence(address, index, end, counter);
      if (index == -1) {
          return false;
      }
      // Address must contain 128-bits of information.
      else if (index == end) {
          return (counter[0] == 8);
      }
      
      if (index+1 < end && address.charAt(index) == ':') {
          if (address.charAt(index+1) == ':') {
              // '::' represents at least one 16-bit group of zeros.
              if (++counter[0] > 8) {
                  return false;
              }
              index += 2;
              // Trailing zeros will fill out the rest of the address.
              if (index == end) {
                 return true;
              }
          }
          // If the second character wasn't ':', in order to be valid,
          // the remainder of the string must match IPv4Address, 
          // and we must have read exactly 6 16-bit groups.
          else {
              return (counter[0] == 6) && 
                  isWellFormedIPv4Address(address.substring(index+1, end));
          }
      }
      else {
          return false;
      }
      
      // 3. Scan hex sequence after '::'.
      int prevCount = counter[0];
      index = scanHexSequence(address, index, end, counter);

      // We've either reached the end of the string, the address ends in
      // an IPv4 address, or it is invalid. scanHexSequence has already 
      // made sure that we have the right number of bits. 
      return (index == end) || 
          (index != -1 && isWellFormedIPv4Address(
          address.substring((counter[0] > prevCount) ? index+1 : index, end)));
  }
  
  /**
   * Helper method for isWellFormedIPv6Reference which scans the 
   * hex sequences of an IPv6 address. It returns the index of the 
   * next character to scan in the address, or -1 if the string 
   * cannot match a valid IPv6 address. 
   *
   * @param address the string to be scanned
   * @param index the beginning index (inclusive)
   * @param end the ending index (exclusive)
   * @param counter a counter for the number of 16-bit sections read
   * in the address
   *
   * @return the index of the next character to scan, or -1 if the
   * string cannot match a valid IPv6 address
   */
  private static int scanHexSequence (String address, int index, int end, int [] counter) {
  	
      char testChar;
      int numDigits = 0;
      int start = index;
      
      // Trying to match the following productions:
      // hexseq = hex4 *( ":" hex4)
      // hex4   = 1*4HEXDIG
      for (; index < end; ++index) {
      	testChar = address.charAt(index);
      	if (testChar == ':') {
      	    // IPv6 addresses are 128-bit, so there can be at most eight sections.
      	    if (numDigits > 0 && ++counter[0] > 8) {
      	        return -1;
      	    }
      	    // This could be '::'.
      	    if (numDigits == 0 || ((index+1 < end) && address.charAt(index+1) == ':')) {
      	        return index;
      	    }
      	    numDigits = 0;
        }
        // This might be invalid or an IPv4address. If it's potentially an IPv4address,
        // backup to just after the last valid character that matches hexseq.
        else if (!isHex(testChar)) {
            if (testChar == '.' && numDigits < 4 && numDigits > 0 && counter[0] <= 6) {
                int back = index - numDigits - 1;
                return (back >= start) ? back : (back+1);
            }
            return -1;
        }
        // There can be at most 4 hex digits per group.
        else if (++numDigits > 4) {
            return -1;
        }
      }
      return (numDigits > 0 && ++counter[0] <= 8) ? end : -1;
  } 


 /**
  * Determine whether a char is a digit.
  *
  * @return true if the char is betweeen '0' and '9', false otherwise
  */
  private static boolean isDigit(char p_char) {
    return p_char >= '0' && p_char <= '9';
  }

 /**
  * Determine whether a character is a hexadecimal character.
  *
  * @return true if the char is betweeen '0' and '9', 'a' and 'f'
  *         or 'A' and 'F', false otherwise
  */
  private static boolean isHex(char p_char) {
    return (p_char <= 'f' && (fgLookupTable[p_char] & ASCII_HEX_CHARACTERS) != 0);
  }

 /**
  * Determine whether a char is an alphabetic character: a-z or A-Z
  *
  * @return true if the char is alphabetic, false otherwise
  */
  private static boolean isAlpha(char p_char) {
      return ((p_char >= 'a' && p_char <= 'z') || (p_char >= 'A' && p_char <= 'Z' ));
  }

 /**
  * Determine whether a char is an alphanumeric: 0-9, a-z or A-Z
  *
  * @return true if the char is alphanumeric, false otherwise
  */
  private static boolean isAlphanum(char p_char) {
     return (p_char <= 'z' && (fgLookupTable[p_char] & MASK_ALPHA_NUMERIC) != 0);
  }

 /**
  * Determine whether a character is a reserved character:
  * ';', '/', '?', ':', '@', '&', '=', '+', '$', ',', '[', or ']'
  *
  * @return true if the string contains any reserved characters
  */
  private static boolean isReservedCharacter(char p_char) {
     return (p_char <= ']' && (fgLookupTable[p_char] & RESERVED_CHARACTERS) != 0);
  }

 /**
  * Determine whether a char is an unreserved character.
  *
  * @return true if the char is unreserved, false otherwise
  */
  private static boolean isUnreservedCharacter(char p_char) {
     return (p_char <= '~' && (fgLo