httpurl.java

Light in the box 抓取程序。 使用HttpClient

    public char[] getRawScheme() {
        return (_scheme == null) ? null : HttpURL.DEFAULT_SCHEME;
    }


    /**
     * Get the scheme.  You can get the scheme explicitly.
     *
     * @return the scheme null if empty or undefined
     */
    public String getScheme() {
        return (_scheme == null) ? null : new String(HttpURL.DEFAULT_SCHEME);
    }

    // --------------------------------------------------------------- The port

    /**
     * Get the port number.
     * @return the port number
     */
    public int getPort() {
        return (_port == -1) ? HttpURL.DEFAULT_PORT : _port;
    }

    // ----------------------------------------------------------- The userinfo

    /**
     * Set the raw-escaped user and password.
     *
     * @param escapedUser the raw-escaped user
     * @param escapedPassword the raw-escaped password; could be null
     * @throws URIException escaped user not valid or user required; escaped
     * password not valid or username missed
     */
    public void setRawUserinfo(char[] escapedUser, char[] escapedPassword)
        throws URIException {

        if (escapedUser == null || escapedUser.length == 0) {
            throw new URIException(URIException.PARSING, "user required");
        }
        if (!validate(escapedUser, within_userinfo) 
            || ((escapedPassword != null) 
            && !validate(escapedPassword, within_userinfo))) {
            throw new URIException(URIException.ESCAPING,
                    "escaped userinfo not valid");
        }
        String username = new String(escapedUser);
        String password = (escapedPassword == null) 
            ? null : new String(escapedPassword);
        String userinfo = username + ((password == null) ? "" : ":" + password);
        String hostname = new String(getRawHost());
        String hostport = (_port == -1) ? hostname : hostname + ":" + _port;
        String authority = userinfo + "@" + hostport;
        _userinfo = userinfo.toCharArray();
        _authority = authority.toCharArray();
        setURI();
    }


    /**
     * Set the raw-escaped user and password.
     *
     * @param escapedUser the escaped user
     * @param escapedPassword the escaped password; could be null
     * @throws URIException escaped user not valid or user required; escaped
     * password not valid or username missed
     * @throws NullPointerException null user
     */
    public void setEscapedUserinfo(String escapedUser, String escapedPassword)
        throws URIException, NullPointerException {

        setRawUserinfo(escapedUser.toCharArray(), (escapedPassword == null) 
            ? null : escapedPassword.toCharArray());
    }


    /**
     * Set the user and password.
     *
     * @param user the user
     * @param password the password; could be null
     * @throws URIException encoding error or username missed
     * @throws NullPointerException null user
     */
    public void setUserinfo(String user, String password) 
        throws URIException, NullPointerException {
        // set the charset to do escape encoding
        String charset = getProtocolCharset();
        setRawUserinfo(encode(user, within_userinfo, charset),
                (password == null) 
                ? null 
                : encode(password, within_userinfo, charset));
    }


    /**
     * Set the raw-escaped user.
     *
     * @param escapedUser the raw-escaped user
     * @throws URIException escaped user not valid or user required
     */
    public void setRawUser(char[] escapedUser) throws URIException {
        if (escapedUser == null || escapedUser.length == 0) {
            throw new URIException(URIException.PARSING, "user required");
        }
        if (!validate(escapedUser, within_userinfo)) {
            throw new URIException(URIException.ESCAPING,
                    "escaped user not valid");
        }
        String username = new String(escapedUser);
        char[] rawPassword = getRawPassword();
        String password = rawPassword == null ? null : new String(rawPassword);
        String userinfo = username + ((password == null) ? "" : ":" + password);
        String hostname = new String(getRawHost());
        String hostport = (_port == -1) ? hostname : hostname + ":" + _port;
        String authority = userinfo + "@" + hostport;
        _userinfo = userinfo.toCharArray();
        _authority = authority.toCharArray();
        setURI();
    }


    /**
     * Set the escaped user string.
     *
     * @param escapedUser the escaped user string
     * @throws URIException escaped user not valid
     * @throws NullPointerException null user
     */
    public void setEscapedUser(String escapedUser)
        throws URIException, NullPointerException {
        setRawUser(escapedUser.toCharArray());
    }


    /**
     * Set the user string.
     *
     * @param user the user string
     * @throws URIException user encoding error
     * @throws NullPointerException null user
     */
    public void setUser(String user) throws URIException, NullPointerException {
        setRawUser(encode(user, allowed_within_userinfo, getProtocolCharset()));
    }


    /**
     * Get the raw-escaped user.
     *
     * @return the raw-escaped user
     */
    public char[] getRawUser() {
        if (_userinfo == null || _userinfo.length == 0) {
            return null;
        }
        int to = indexFirstOf(_userinfo, ':');
        // String.indexOf(':', 0, _userinfo.length, _userinfo, 0, 1, 0);
        if (to == -1) {
            return _userinfo; // only user.
        }
        char[] result = new char[to];
        System.arraycopy(_userinfo, 0, result, 0, to);
        return result;
    }


    /**
     * Get the escaped user
     *
     * @return the escaped user
     */
    public String getEscapedUser() {
        char[] user = getRawUser();
        return (user == null) ? null : new String(user);
    }


    /**
     * Get the user.
     *
     * @return the user name
     * @throws URIException If {@link #decode} fails
     */
    public String getUser() throws URIException {
        char[] user = getRawUser();
        return (user == null) ? null : decode(user, getProtocolCharset());
    }


    /**
     * Set the raw-escaped password.
     *
     * @param escapedPassword the raw-escaped password; could be null
     * @throws URIException escaped password not valid or username missed
     */
    public void setRawPassword(char[] escapedPassword) throws URIException {
        if (escapedPassword != null 
            && !validate(escapedPassword, within_userinfo)) {
            throw new URIException(URIException.ESCAPING,
               "escaped password not valid");
        }
        if (getRawUser() == null || getRawUser().length == 0) {
            throw new URIException(URIException.PARSING, "username required");
        }
        String username = new String(getRawUser());
        String password = escapedPassword == null ? null : new String(escapedPassword);
        // an emtpy string is allowed as a password
        String userinfo = username + ((password == null) ? "" : ":" + password);
        String hostname = new String(getRawHost());
        String hostport = (_port == -1) ? hostname : hostname + ":" + _port;
        String authority = userinfo + "@" + hostport;
        _userinfo = userinfo.toCharArray();
        _authority = authority.toCharArray();
        setURI();
    }


    /**
     * Set the escaped password string.
     *
     * @param escapedPassword the escaped password string; could be null
     * @throws URIException escaped password not valid or username missed
     */
    public void setEscapedPassword(String escapedPassword) throws URIException {
        setRawPassword((escapedPassword == null) ? null 
            : escapedPassword.toCharArray());
    }


    /**
     * Set the password string.
     *
     * @param password the password string; could be null
     * @throws URIException encoding error or username missed
     */
    public void setPassword(String password) throws URIException {
        setRawPassword((password == null) ? null : encode(password,
                    allowed_within_userinfo, getProtocolCharset()));
    }


    /**
     * Get the raw-escaped password.
     *
     * @return the raw-escaped password
     */
    public char[] getRawPassword() {
        int from = indexFirstOf(_userinfo, ':');
        if (from == -1) {
            return null; // null or only user.
        }
        int len = _userinfo.length - from - 1;
        char[] result = new char[len];
        System.arraycopy(_userinfo, from + 1, result, 0, len);
        return result;
    }


    /**
     * Get the escaped password.
     *
     * @return the escaped password
     */
    public String getEscapedPassword() {
        char[] password = getRawPassword();
        return (password == null) ? null : new String(password);
    }


    /**
     * Get the password.
     *
     * @return the password
     * @throws URIException If {@link #decode(char[],String)} fails.
     */
    public String getPassword() throws URIException {
        char[] password = getRawPassword();
        return (password == null) ? null : decode(password,
                getProtocolCharset());
    }

    // --------------------------------------------------------------- The path

    /**
     * Get the raw-escaped current hierarchy level.
     *
     * @return the raw-escaped current hierarchy level
     * @throws URIException If {@link #getRawCurrentHierPath(char[])} fails.
     */
    public char[] getRawCurrentHierPath() throws URIException {
        return (_path == null || _path.length == 0) ? rootPath 
            : super.getRawCurrentHierPath(_path);
    }


    /**
     * Get the level above the this hierarchy level.
     *
     * @return the raw above hierarchy level
     * @throws URIException If {@link #getRawCurrentHierPath(char[])} fails.
     */
    public char[] getRawAboveHierPath() throws URIException {
        char[] path = getRawCurrentHierPath();
        return (path == null || path.length == 0) ? rootPath : getRawCurrentHierPath(path);
    }


    /**
     * Get the raw escaped path.
     *
     * @return the path '/' if empty or undefined
     */
    public char[] getRawPath() {
        char[] path =  super.getRawPath();
        return (path == null || path.length == 0) ? rootPath : path;
    }

    // -------------------------------------------------------------- The query

    /**
     * Set the query as the name and value pair.
     *
     * @param queryName the query string.
     * @param queryValue the query string.
     * @throws URIException incomplete trailing escape pattern
     * Or unsupported character encoding
     * @throws NullPointerException null query
     * @see #encode
     */
    public void setQuery(String queryName, String queryValue)
        throws URIException, NullPointerException {

        StringBuffer buff = new StringBuffer();
        // set the charset to do escape encoding
        String charset = getProtocolCharset();
        buff.append(encode(queryName, allowed_within_query, charset));
        buff.append('=');
        buff.append(encode(queryValue, allowed_within_query, charset));
        _query = buff.toString().toCharArray();
        setURI();
    }


    /**
     * Set the query as the name and value pairs.
     *
     * @param queryName the array of the query string.
     * @param queryValue the array of the query string.
     * @throws URIException incomplete trailing escape pattern,
     * unsupported character encoding or wrong array size
     * @throws NullPointerException null query
     * @see #encode
     */
    public void setQuery(String[] queryName, String[] queryValue)
        throws URIException, NullPointerException {

        int length = queryName.length;
        if (length != queryValue.length) {
            throw new URIException("wrong array size of query");
        }

        StringBuffer buff = new StringBuffer();
        // set the charset to do escape encoding
        String charset = getProtocolCharset();
        for (int i = 0; i < length; i++) {
            buff.append(encode(queryName[i], allowed_within_query, charset));
            buff.append('=');
            buff.append(encode(queryValue[i], allowed_within_query, charset));
            if (i + 1 < length) { 
                buff.append('&');
            }
        }
        _query = buff.toString().toCharArray();
        setURI();
    }

    // ---------------------------------------------------------------- Utility

    /**
     * Verify the valid class use for construction.
     *
     * @throws URIException the wrong scheme use
     */
    protected void checkValid() throws URIException {
        // could be explicit protocol or undefined.
        if (!(equals(_scheme, DEFAULT_SCHEME) || _scheme == null)) {
            throw new URIException(URIException.PARSING, "wrong class use");
        }
    }

    /**
     * Once it's parsed successfully, set this URI.
     *
     * @see #getRawURI
     */
    protected void setURI() {
        // set _uri
        StringBuffer buf = new StringBuffer();
        // ^(([^:/?#]+):)?(//([^/?#]*))?([^?#]*)(\?([^#]*))?(#(.*))?
        if (_scheme != null) {
            buf.append(_scheme);
            buf.append(':');
        }
        if (_is_net_path) {
            buf.append("//");
            if (_authority != null) { // has_authority
                if (_userinfo != null) { // by default, remove userinfo part
                    if (_host != null) {
                        buf.append(_host);
                        if (_port != -1) {
                            buf.append(':');
                            buf.append(_port);
                        }
                    }
                } else {
                    buf.append(_authority);
                }
            }
        }
        if (_opaque != null && _is_opaque_part) {
            buf.append(_opaque);
        } else if (_path != null) {
            // _is_hier_part or _is_relativeURI
            if (_path.length != 0) {
                buf.append(_path);
            }
        }
        if (_query != null) { // has_query
            buf.append('?');
            buf.append(_query);
        }
        // ignore the fragment identifier
        _uri = buf.toString().toCharArray();
        hash = 0;
    }
    
}