URLEncoder.java
/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.catalina.util;
import java.io.ByteArrayOutputStream;
import java.io.IOException;
import java.io.OutputStreamWriter;
import java.nio.charset.Charset;
import java.util.BitSet;
/**
* This class is very similar to the java.net.URLEncoder class. Unfortunately, with java.net.URLEncoder there is no way
* to specify to the java.net.URLEncoder which characters should NOT be encoded. This code was moved from
* DefaultServlet.java
*
* @author Craig R. McClanahan
* @author Remy Maucherat
*/
public final class URLEncoder implements Cloneable {
private static final char[] hexadecimal =
{ '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F' };
public static final URLEncoder DEFAULT = new URLEncoder();
public static final URLEncoder QUERY = new URLEncoder();
static {
/*
* Encoder for URI paths, so from the spec:
*
* pchar = unreserved / pct-encoded / sub-delims / ":" / "@"
*
* unreserved = ALPHA / DIGIT / "-" / "." / "_" / "~"
*
* sub-delims = "!" / "$" / "&" / "'" / "(" / ")" / "*" / "+" / "," / ";" / "="
*/
// ALPHA and DIGIT are always treated as safe characters
// Add the remaining unreserved characters
DEFAULT.addSafeCharacter('-');
DEFAULT.addSafeCharacter('.');
DEFAULT.addSafeCharacter('_');
DEFAULT.addSafeCharacter('~');
// Add the sub-delims
DEFAULT.addSafeCharacter('!');
DEFAULT.addSafeCharacter('$');
DEFAULT.addSafeCharacter('&');
DEFAULT.addSafeCharacter('\'');
DEFAULT.addSafeCharacter('(');
DEFAULT.addSafeCharacter(')');
DEFAULT.addSafeCharacter('*');
DEFAULT.addSafeCharacter('+');
DEFAULT.addSafeCharacter(',');
DEFAULT.addSafeCharacter(';');
DEFAULT.addSafeCharacter('=');
// Add the remaining literals
DEFAULT.addSafeCharacter(':');
DEFAULT.addSafeCharacter('@');
// Add '/' so it isn't encoded when we encode a path
DEFAULT.addSafeCharacter('/');
/*
* Encoder for query strings
* https://www.w3.org/TR/html5/forms.html#application/x-www-form-urlencoded-encoding-algorithm
* // @formatter:off
* 0x20 ' ' -> '+'
* 0x2A, 0x2D, 0x2E, 0x30 to 0x39, 0x41 to 0x5A, 0x5F, 0x61 to 0x7A as-is
* '*', '-', '.', '0' to '9', 'A' to 'Z', '_', 'a' to 'z'
* Also '=' and '&' are not encoded
* Everything else %nn encoded
* // @formatter:on
*/
// Special encoding for space
QUERY.setEncodeSpaceAsPlus(true);
// Alpha and digit are safe by default
// Add the other permitted characters
QUERY.addSafeCharacter('*');
QUERY.addSafeCharacter('-');
QUERY.addSafeCharacter('.');
QUERY.addSafeCharacter('_');
QUERY.addSafeCharacter('=');
QUERY.addSafeCharacter('&');
}
// Array containing the safe characters set.
private final BitSet safeCharacters;
private boolean encodeSpaceAsPlus = false;
public URLEncoder() {
this(new BitSet(256));
for (char i = 'a'; i <= 'z'; i++) {
addSafeCharacter(i);
}
for (char i = 'A'; i <= 'Z'; i++) {
addSafeCharacter(i);
}
for (char i = '0'; i <= '9'; i++) {
addSafeCharacter(i);
}
}
private URLEncoder(BitSet safeCharacters) {
this.safeCharacters = safeCharacters;
}
public void addSafeCharacter(char c) {
safeCharacters.set(c);
}
public void removeSafeCharacter(char c) {
safeCharacters.clear(c);
}
public void setEncodeSpaceAsPlus(boolean encodeSpaceAsPlus) {
this.encodeSpaceAsPlus = encodeSpaceAsPlus;
}
/**
* URL encodes the provided path using the given character set.
*
* @param path The path to encode
* @param charset The character set to use to convert the path to bytes
*
* @return The encoded path
*/
public String encode(String path, Charset charset) {
int maxBytesPerChar = 10;
StringBuilder rewrittenPath = new StringBuilder(path.length());
ByteArrayOutputStream buf = new ByteArrayOutputStream(maxBytesPerChar);
OutputStreamWriter writer = new OutputStreamWriter(buf, charset);
for (int i = 0; i < path.length(); i++) {
int c = path.charAt(i);
if (safeCharacters.get(c)) {
rewrittenPath.append((char) c);
} else if (encodeSpaceAsPlus && c == ' ') {
rewrittenPath.append('+');
} else {
// convert to external encoding before hex conversion
try {
writer.write((char) c);
writer.flush();
} catch (IOException e) {
buf.reset();
continue;
}
byte[] ba = buf.toByteArray();
for (byte toEncode : ba) {
// Converting each byte in the buffer
rewrittenPath.append('%');
int low = toEncode & 0x0f;
int high = (toEncode & 0xf0) >> 4;
rewrittenPath.append(hexadecimal[high]);
rewrittenPath.append(hexadecimal[low]);
}
buf.reset();
}
}
return rewrittenPath.toString();
}
@Override
public Object clone() {
URLEncoder result = new URLEncoder((BitSet) safeCharacters.clone());
result.setEncodeSpaceAsPlus(encodeSpaceAsPlus);
return result;
}
}