uriutil.java

Light in the box 抓取程序。 使用HttpClient

     * The path may consist of a sequence of path segments separated by a
     * single slash "/" character.  Within a path segment, the characters
     * "/", ";", "=", and "?" are reserved.
     *
     * @param unescaped an unescaped string
     * @param charset the charset
     * @return the escaped string
     * 
     * @throws URIException if the charset is not supported
     * 
     * @see #encode
     */
    public static String encodeWithinPath(String unescaped, String charset)
        throws URIException {

        return encode(unescaped, URI.allowed_within_path, charset);
    }


    /**
     * Escape and encode a string regarded as the path component of an URI with
     * the default protocol charset.
     *
     * @param unescaped an unescaped string
     * @return the escaped string
     * 
     * @throws URIException if the default protocol charset is not supported
     * 
     * @see URI#getDefaultProtocolCharset
     * @see #encode
     */
    public static String encodePath(String unescaped) throws URIException {
        return encodePath(unescaped, URI.getDefaultProtocolCharset());
    }


    /**
     * Escape and encode a string regarded as the path component of an URI with
     * a given charset.
     *
     * @param unescaped an unescaped string
     * @param charset the charset
     * @return the escaped string
     * 
     * @throws URIException if the charset is not supported
     * 
     * @see #encode
     */
    public static String encodePath(String unescaped, String charset)
        throws URIException {

        return encode(unescaped, URI.allowed_abs_path, charset);
    }


    /**
     * Escape and encode a string regarded as within the query component of an
     * URI with the default protocol charset.
     * When a query comprise the name and value pairs, it is used in order
     * to encode each name and value string.  The reserved special characters
     * within a query component are being included in encoding the query.
     *
     * @param unescaped an unescaped string
     * @return the escaped string
     * 
     * @throws URIException if the default protocol charset is not supported
     * 
     * @see URI#getDefaultProtocolCharset
     * @see #encode
     */
    public static String encodeWithinQuery(String unescaped)
        throws URIException {

        return encodeWithinQuery(unescaped, URI.getDefaultProtocolCharset());
    }


    /**
     * Escape and encode a string regarded as within the query component of an
     * URI with a given charset.
     * When a query comprise the name and value pairs, it is used in order
     * to encode each name and value string.  The reserved special characters
     * within a query component are being included in encoding the query.
     *
     * @param unescaped an unescaped string
     * @param charset the charset
     * @return the escaped string
     * 
     * @throws URIException if the charset is not supported
     * 
     * @see #encode
     */
    public static String encodeWithinQuery(String unescaped, String charset)
        throws URIException {

        return encode(unescaped, URI.allowed_within_query, charset);
    }


    /**
     * Escape and encode a string regarded as the query component of an URI with
     * the default protocol charset.
     * When a query string is not misunderstood the reserved special characters
     * ("&", "=", "+", ",", and "$") within a query component, this method
     * is recommended to use in encoding the whole query.
     *
     * @param unescaped an unescaped string
     * @return the escaped string
     * 
     * @throws URIException if the default protocol charset is not supported
     * 
     * @see URI#getDefaultProtocolCharset
     * @see #encode
     */
    public static String encodeQuery(String unescaped) throws URIException {
        return encodeQuery(unescaped, URI.getDefaultProtocolCharset());
    }


    /**
     * Escape and encode a string regarded as the query component of an URI with
     * a given charset.
     * When a query string is not misunderstood the reserved special characters
     * ("&", "=", "+", ",", and "$") within a query component, this method
     * is recommended to use in encoding the whole query.
     *
     * @param unescaped an unescaped string
     * @param charset the charset
     * @return the escaped string
     * 
     * @throws URIException if the charset is not supported
     * 
     * @see #encode
     */
    public static String encodeQuery(String unescaped, String charset)
        throws URIException {

        return encode(unescaped, URI.allowed_query, charset);
    }


    /**
     * Escape and encode a given string with allowed characters not to be
     * escaped and the default protocol charset.
     *
     * @param unescaped a string
     * @param allowed allowed characters not to be escaped
     * @return the escaped string
     * 
     * @throws URIException if the default protocol charset is not supported
     * 
     * @see URI#getDefaultProtocolCharset
     */
    public static String encode(String unescaped, BitSet allowed)
        throws URIException {

        return encode(unescaped, allowed, URI.getDefaultProtocolCharset());
    }


    /**
     * Escape and encode a given string with allowed characters not to be
     * escaped and a given charset.
     *
     * @param unescaped a string
     * @param allowed allowed characters not to be escaped
     * @param charset the charset
     * @return the escaped string
     */
    public static String encode(String unescaped, BitSet allowed,
            String charset) throws URIException {
        byte[] rawdata = URLCodec.encodeUrl(allowed, 
            EncodingUtil.getBytes(unescaped, charset));
        return EncodingUtil.getAsciiString(rawdata);
    }


    /**
     * Unescape and decode a given string regarded as an escaped string with the
     * default protocol charset.
     *
     * @param escaped a string
     * @return the unescaped string
     * 
     * @throws URIException if the string cannot be decoded (invalid)
     * 
     * @see URI#getDefaultProtocolCharset
     */
    public static String decode(String escaped) throws URIException {
        try {
            byte[] rawdata = URLCodec.decodeUrl(EncodingUtil.getAsciiBytes(escaped));
            return EncodingUtil.getString(rawdata, URI.getDefaultProtocolCharset());
        } catch (DecoderException e) {
            throw new URIException(e.getMessage());
        }
    }

    /**
     * Unescape and decode a given string regarded as an escaped string.
     *
     * @param escaped a string
     * @param charset the charset
     * @return the unescaped string
     * 
     * @throws URIException if the charset is not supported
     * 
     * @see Coder#decode
     */
    public static String decode(String escaped, String charset)
        throws URIException {

        return Coder.decode(escaped.toCharArray(), charset);
    }

    // ---------------------------------------------------------- Inner classes

    /**
     * The basic and internal utility for URI escape and character encoding and
     * decoding.
     * 
     * @deprecated use org.apache.commons.codec.net.URLCodec
     */
    protected static class Coder extends URI {

        /**
         * Escape and encode a given string with allowed characters not to be
         * escaped.
         *
         * @param unescapedComponent an unescaped component
         * @param allowed allowed characters not to be escaped
         * @param charset the charset to encode
         * @return the escaped and encoded string
         * 
         * @throws URIException if the charset is not supported
         * 
         * @deprecated use org.apache.commons.codec.net.URLCodec
         */
        public static char[] encode(String unescapedComponent, BitSet allowed, String charset) 
            throws URIException {

            return URI.encode(unescapedComponent, allowed, charset);
        }


        /**
         * Unescape and decode a given string.
         *
         * @param escapedComponent an being-unescaped component
         * @param charset the charset to decode
         * @return the escaped and encoded string
         * 
         * @throws URIException if the charset is not supported
         * 
         * @deprecated use org.apache.commons.codec.net.URLCodec
         */
        public static String decode(char[] escapedComponent, String charset)
            throws URIException {

            return URI.decode(escapedComponent, charset);
        }


        /**
         * Verify whether a given string is escaped or not
         *
         * @param original given characters
         * @return true if the given character array is 7 bit ASCII-compatible.
         */
        public static boolean verifyEscaped(char[] original) {
            for (int i = 0; i < original.length; i++) {
                int c = original[i];
                if (c > 128) {
                    return false;
                } else if (c == '%') {
                    if (Character.digit(original[++i], 16) == -1 
                        || Character.digit(original[++i], 16) == -1) {
                        return false;
                    }
                }
            }
            return true;
        }


        /**
         * Replace from a given character to given character in an array order
         * for a given string.
         *
         * @param original a given string
         * @param from a replacing character array
         * @param to a replaced character array
         * @return the replaced string
         */
        public static String replace(String original, char[] from, char[] to) {
            for (int i = from.length; i > 0; --i) {
                original = replace(original, from[i], to[i]);
            }
            return original;
        }


        /**
         * Replace from a given character to given character for a given string.
         *
         * @param original a given string
         * @param from a replacing character array
         * @param to a replaced character array
         * @return the replaced string
         */
        public static String replace(String original, char from, char to) {
            StringBuffer result = new StringBuffer(original.length());
            int at, saved = 0;
            do {
                at = original.indexOf(from);
                if (at >= 0) {
                    result.append(original.substring(0, at));
                    result.append(to);
                } else {
                    result.append(original.substring(saved));
                }
                saved = at;
            } while (at >= 0);
            return result.toString();
        }
    }

}