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

  * @param p_userinfo the URI userinfo (cannot be specified if host
  *                   is null)
  * @param p_host the hostname, IPv4 address or IPv6 reference for the URI
  * @param p_port the URI port (may be -1 for "unspecified"; cannot
  *               be specified if host is null)
  * @param p_path the URI path - if the path contains '?' or '#',
  *               then the query string and/or fragment will be
  *               set from the path; however, if the query and
  *               fragment are specified both in the path and as
  *               separate parameters, an exception is thrown
  * @param p_queryString the URI query string (cannot be specified
  *                      if path is null)
  * @param p_fragment the URI fragment (cannot be specified if path
  *                   is null)
  *
  * @exception MalformedURIException if any of the parameters violates
  *                                  syntax rules or semantic rules
  */
  public URI(String p_scheme, String p_userinfo,
             String p_host, int p_port, String p_path,
             String p_queryString, String p_fragment)
         throws MalformedURIException {
    if (p_scheme == null || p_scheme.trim().length() == 0) {
      throw new MalformedURIException("Scheme is required!");
    }

    if (p_host == null) {
      if (p_userinfo != null) {
        throw new MalformedURIException(
             "Userinfo may not be specified if host is not specified!");
      }
      if (p_port != -1) {
        throw new MalformedURIException(
             "Port may not be specified if host is not specified!");
      }
    }

    if (p_path != null) {
      if (p_path.indexOf('?') != -1 && p_queryString != null) {
        throw new MalformedURIException(
          "Query string cannot be specified in path and query string!");
      }

      if (p_path.indexOf('#') != -1 && p_fragment != null) {
        throw new MalformedURIException(
          "Fragment cannot be specified in both the path and fragment!");
      }
    }

    setScheme(p_scheme);
    setHost(p_host);
    setPort(p_port);
    setUserinfo(p_userinfo);
    setPath(p_path);
    setQueryString(p_queryString);
    setFragment(p_fragment);
  }

 /**
  * Initialize all fields of this URI from another URI.
  *
  * @param p_other the URI to copy (cannot be null)
  */
  private void initialize(URI p_other) {
    m_scheme = p_other.getScheme();
    m_userinfo = p_other.getUserinfo();
    m_host = p_other.getHost();
    m_port = p_other.getPort();
    m_regAuthority = p_other.getRegBasedAuthority();
    m_path = p_other.getPath();
    m_queryString = p_other.getQueryString();
    m_fragment = p_other.getFragment();
  }

 /**
  * Initializes this URI from a base URI and a URI specification string.
  * See RFC 2396 Section 4 and Appendix B for specifications on parsing
  * the URI and Section 5 for specifications on resolving relative URIs
  * and relative paths.
  *
  * @param p_base the base URI (may be null if p_uriSpec is an absolute
  *               URI)
  * @param p_uriSpec the URI spec string which may be an absolute or
  *                  relative URI (can only be null/empty if p_base
  *                  is not null)
  *
  * @exception MalformedURIException if p_base is null and p_uriSpec
  *                                  is not an absolute URI or if
  *                                  p_uriSpec violates syntax rules
  */
  private void initialize(URI p_base, String p_uriSpec)
                         throws MalformedURIException {
	  
    String uriSpec = p_uriSpec;
    int uriSpecLen = (uriSpec != null) ? uriSpec.length() : 0;
	
    if (p_base == null && uriSpecLen == 0) {
      throw new MalformedURIException(
                  "Cannot initialize URI with empty parameters.");
    }

    // just make a copy of the base if spec is empty
    if (uriSpecLen == 0) {
      initialize(p_base);
      return;
    }

    int index = 0;

    // Check for scheme, which must be before '/', '?' or '#'. Also handle
    // names with DOS drive letters ('D:'), so 1-character schemes are not
    // allowed.
    int colonIdx = uriSpec.indexOf(':');
    if (colonIdx != -1) {
        final int searchFrom = colonIdx - 1;
        // search backwards starting from character before ':'.
        int slashIdx = uriSpec.lastIndexOf('/', searchFrom);
        int queryIdx = uriSpec.lastIndexOf('?', searchFrom);
        int fragmentIdx = uriSpec.lastIndexOf('#', searchFrom);
       
        if (colonIdx < 2 || slashIdx != -1 || 
            queryIdx != -1 || fragmentIdx != -1) {
            // A standalone base is a valid URI according to spec
            if (colonIdx == 0 || (p_base == null && fragmentIdx != 0)) {
                throw new MalformedURIException("No scheme found in URI.");
            }
        }
        else {
            initializeScheme(uriSpec);
            index = m_scheme.length()+1;
            
            // Neither 'scheme:' or 'scheme:#fragment' are valid URIs.
            if (colonIdx == uriSpecLen - 1 || uriSpec.charAt(colonIdx+1) == '#') {
            	throw new MalformedURIException("Scheme specific part cannot be empty.");	
            }
        }
    }
    else if (p_base == null && uriSpec.indexOf('#') != 0) {
        throw new MalformedURIException("No scheme found in URI.");    
    }

    // Two slashes means we may have authority, but definitely means we're either
    // matching net_path or abs_path. These two productions are ambiguous in that
    // every net_path (except those containing an IPv6Reference) is an abs_path. 
    // RFC 2396 resolves this ambiguity by applying a greedy left most matching rule. 
    // Try matching net_path first, and if that fails we don't have authority so 
    // then attempt to match abs_path.
    //
    // net_path = "//" authority [ abs_path ]
    // abs_path = "/"  path_segments
    if (((index+1) < uriSpecLen) &&
        (uriSpec.charAt(index) == '/' && uriSpec.charAt(index+1) == '/')) {
      index += 2;
      int startPos = index;

      // Authority will be everything up to path, query or fragment
      char testChar = '\0';
      while (index < uriSpecLen) {
        testChar = uriSpec.charAt(index);
        if (testChar == '/' || testChar == '?' || testChar == '#') {
          break;
        }
        index++;
      }

      // Attempt to parse authority. If the section is an empty string
      // this is a valid server based authority, so set the host to this
      // value.
      if (index > startPos) {
        // If we didn't find authority we need to back up. Attempt to
        // match against abs_path next.
        if (!initializeAuthority(uriSpec.substring(startPos, index))) {
          index = startPos - 2;
        }
      }
      else {
        m_host = "";
      }
    }

    initializePath(uriSpec, index);

    // Resolve relative URI to base URI - see RFC 2396 Section 5.2
    // In some cases, it might make more sense to throw an exception
    // (when scheme is specified is the string spec and the base URI
    // is also specified, for example), but we're just following the
    // RFC specifications
    if (p_base != null) {

      // check to see if this is the current doc - RFC 2396 5.2 #2
      // note that this is slightly different from the RFC spec in that
      // we don't include the check for query string being null
      // - this handles cases where the urispec is just a query
      // string or a fragment (e.g. "?y" or "#s") -
      // see <http://www.ics.uci.edu/~fielding/url/test1.html> which
      // identified this as a bug in the RFC
      if (m_path.length() == 0 && m_scheme == null &&
          m_host == null && m_regAuthority == null) {
        m_scheme = p_base.getScheme();
        m_userinfo = p_base.getUserinfo();
        m_host = p_base.getHost();
        m_port = p_base.getPort();
        m_regAuthority = p_base.getRegBasedAuthority();
        m_path = p_base.getPath();

        if (m_queryString == null) {
          m_queryString = p_base.getQueryString();
        }
        return;
      }

      // check for scheme - RFC 2396 5.2 #3
      // if we found a scheme, it means absolute URI, so we're done
      if (m_scheme == null) {
        m_scheme = p_base.getScheme();
      }
      else {
        return;
      }

      // check for authority - RFC 2396 5.2 #4
      // if we found a host, then we've got a network path, so we're done
      if (m_host == null && m_regAuthority == null) {
        m_userinfo = p_base.getUserinfo();
        m_host = p_base.getHost();
        m_port = p_base.getPort();
        m_regAuthority = p_base.getRegBasedAuthority();
      }
      else {
        return;
      }

      // check for absolute path - RFC 2396 5.2 #5
      if (m_path.length() > 0 &&
          m_path.startsWith("/")) {
        return;
      }

      // if we get to this point, we need to resolve relative path
      // RFC 2396 5.2 #6
      String path = "";
      String basePath = p_base.getPath();

      // 6a - get all but the last segment of the base URI path
      if (basePath != null && basePath.length() > 0) {
        int lastSlash = basePath.lastIndexOf('/');
        if (lastSlash != -1) {
          path = basePath.substring(0, lastSlash+1);
        }
      }
      else if (m_path.length() > 0) {
      	path = "/";
      }

      // 6b - append the relative URI path
      path = path.concat(m_path);

      // 6c - remove all "./" where "." is a complete path segment
      index = -1;
      while ((index = path.indexOf("/./")) != -1) {
        path = path.substring(0, index+1).concat(path.substring(index+3));
      }

      // 6d - remove "." if path ends with "." as a complete path segment
      if (path.endsWith("/.")) {
        path = path.substring(0, path.length()-1);
      }

      // 6e - remove all "<segment>/../" where "<segment>" is a complete
      // path segment not equal to ".."
      index = 1;
      int segIndex = -1;
      String tempString = null;

      while ((index = path.indexOf("/../", index)) > 0) {
        tempString = path.substring(0, path.indexOf("/../"));
        segIndex = tempString.lastIndexOf('/');
        if (segIndex != -1) {
          if (!tempString.substring(segIndex).equals("..")) {
            path = path.substring(0, segIndex+1).concat(path.substring(index+4));
            index = segIndex;
          }
          else
            index += 4;
        }
        else
          index += 4;
      }

      // 6f - remove ending "<segment>/.." where "<segment>" is a
      // complete path segment
      if (path.endsWith("/..")) {
        tempString = path.substring(0, path.length()-3);
        segIndex = tempString.lastIndexOf('/');
        if (segIndex != -1) {
          path = path.substring(0, segIndex+1);
        }
      }
      m_path = path;
    }
  }

 /**
  * Initialize the scheme for this URI from a URI string spec.
  *
  * @param p_uriSpec the URI specification (cannot be null)
  *
  * @exception MalformedURIException if URI does not have a conformant
  *                                  scheme
  */
  private void initializeScheme(String p_uriSpec)
                 throws MalformedURIException {
    int uriSpecLen = p_uriSpec.length();
    int index = 0;
    String scheme = null;
    char testChar = '\0';

    while (index < uriSpecLen) {
      testChar = p_uriSpec.charAt(index);
      if (testChar == ':' || testChar == '/' ||
          testChar == '?' || testChar == '#') {
        break;
      }
      index++;
    }
    scheme = p_uriSpec.substring(0, index);

    if (scheme.length() == 0) {
      throw new MalformedURIException("No scheme found in URI.");
    }
    else {
      setScheme(scheme);
    }
  }

 /**
  * Initialize the authority (either server or registry based)
  * for this URI from a URI string spec.
  *
  * @param p_uriSpec the URI specification (cannot be null)
  * 
  * @return true if the given string matched server or registry
  * based authority
  */
  private boolean initializeAuthority(String p_uriSpec) {
    
    int index = 0;
    int start = 0;
    int end = p_uriSpec.length();

    char testChar = '\0';
    String userinfo = null;

    // userinfo is everything up to @
    if (p_uriSpec.indexOf('@', start) != -1) {
      while (index < end) {
        testChar = p_uriSpec.charAt(index);
        if (testChar == '@') {
          break;
        }
        index++;
      }
      userinfo = p_uriSpec.substring(start, index);
      index++;
    }

    // host is everything up to last ':', or up to 
    // and including ']' if followed by ':'.
    String host = null;
    start = index;
    boolean hasPort = false;
    if (index < end) {
      if (p_uriSpec.charAt(start) == '[') {
      	int bracketIndex = p_uriSpec.indexOf(']', start);
      	index = (bracketIndex != -1) ? bracketIndex : end;
      	if (index+1 < end && p_uriSpec.charAt(index+1) == ':') {
      	  ++index;
      	  hasPort = true;
      	}
      	else {
      	  index = end;
      	}
      }
      else {
      	int colonIndex = p_uriSpec.lastIndexOf(':', end);