001/*
002 * Copyright (C) 2008 The Guava Authors
003 *
004 * Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
005 * in compliance with the License. You may obtain a copy of the License at
006 *
007 * http://www.apache.org/licenses/LICENSE-2.0
008 *
009 * Unless required by applicable law or agreed to in writing, software distributed under the License
010 * is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
011 * or implied. See the License for the specific language governing permissions and limitations under
012 * the License.
013 */
014
015package com.google.common.net;
016
017import static com.google.common.base.Preconditions.checkNotNull;
018import static java.lang.Math.max;
019
020import com.google.common.annotations.GwtCompatible;
021import com.google.common.escape.UnicodeEscaper;
022import javax.annotation.CheckForNull;
023
024/**
025 * A {@code UnicodeEscaper} that escapes some set of Java characters using a UTF-8 based percent
026 * encoding scheme. The set of safe characters (those which remain unescaped) can be specified on
027 * construction.
028 *
029 * <p>This class is primarily used for creating URI escapers in {@link UrlEscapers} but can be used
030 * directly if required. While URI escapers impose specific semantics on which characters are
031 * considered 'safe', this class has a minimal set of restrictions.
032 *
033 * <p>When escaping a String, the following rules apply:
034 *
035 * <ul>
036 *   <li>All specified safe characters remain unchanged.
037 *   <li>If {@code plusForSpace} was specified, the space character " " is converted into a plus
038 *       sign {@code "+"}.
039 *   <li>All other characters are converted into one or more bytes using UTF-8 encoding and each
040 *       byte is then represented by the 3-character string "%XX", where "XX" is the two-digit,
041 *       uppercase, hexadecimal representation of the byte value.
042 * </ul>
043 *
044 * <p>For performance reasons the only currently supported character encoding of this class is
045 * UTF-8.
046 *
047 * <p><b>Note:</b> This escaper produces <a
048 * href="https://url.spec.whatwg.org/#percent-encode">uppercase</a> hexadecimal sequences.
049 *
050 * @author David Beaumont
051 * @since 15.0
052 */
053@GwtCompatible
054@ElementTypesAreNonnullByDefault
055public final class PercentEscaper extends UnicodeEscaper {
056
057  // In some escapers spaces are escaped to '+'
058  private static final char[] PLUS_SIGN = {'+'};
059
060  // Percent escapers output upper case hex digits (uri escapers require this).
061  private static final char[] UPPER_HEX_DIGITS = "0123456789ABCDEF".toCharArray();
062
063  /** If true we should convert space to the {@code +} character. */
064  private final boolean plusForSpace;
065
066  /**
067   * An array of flags where for any {@code char c} if {@code safeOctets[c]} is true then {@code c}
068   * should remain unmodified in the output. If {@code c >= safeOctets.length} then it should be
069   * escaped.
070   */
071  private final boolean[] safeOctets;
072
073  /**
074   * Constructs a percent escaper with the specified safe characters and optional handling of the
075   * space character.
076   *
077   * <p>Not that it is allowed, but not necessarily desirable to specify {@code %} as a safe
078   * character. This has the effect of creating an escaper which has no well-defined inverse but it
079   * can be useful when escaping additional characters.
080   *
081   * @param safeChars a non-null string specifying additional safe characters for this escaper (the
082   *     ranges 0..9, a..z and A..Z are always safe and should not be specified here)
083   * @param plusForSpace true if ASCII space should be escaped to {@code +} rather than {@code %20}
084   * @throws IllegalArgumentException if any of the parameters were invalid
085   */
086  public PercentEscaper(String safeChars, boolean plusForSpace) {
087    // TODO(dbeaumont): Switch to static factory methods for creation now that class is final.
088    // TODO(dbeaumont): Support escapers where alphanumeric chars are not safe.
089    checkNotNull(safeChars); // eager for GWT.
090    // Avoid any misunderstandings about the behavior of this escaper
091    if (safeChars.matches(".*[0-9A-Za-z].*")) {
092      throw new IllegalArgumentException(
093          "Alphanumeric characters are always 'safe' and should not be explicitly specified");
094    }
095    safeChars += "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789";
096    // Avoid ambiguous parameters. Safe characters are never modified so if
097    // space is a safe character then setting plusForSpace is meaningless.
098    if (plusForSpace && safeChars.contains(" ")) {
099      throw new IllegalArgumentException(
100          "plusForSpace cannot be specified when space is a 'safe' character");
101    }
102    this.plusForSpace = plusForSpace;
103    this.safeOctets = createSafeOctets(safeChars);
104  }
105
106  /**
107   * Creates a boolean array with entries corresponding to the character values specified in
108   * safeChars set to true. The array is as small as is required to hold the given character
109   * information.
110   */
111  private static boolean[] createSafeOctets(String safeChars) {
112    int maxChar = -1;
113    char[] safeCharArray = safeChars.toCharArray();
114    for (char c : safeCharArray) {
115      maxChar = max(c, maxChar);
116    }
117    boolean[] octets = new boolean[maxChar + 1];
118    for (char c : safeCharArray) {
119      octets[c] = true;
120    }
121    return octets;
122  }
123
124  /*
125   * Overridden for performance. For unescaped strings this improved the performance of the uri
126   * escaper from ~760ns to ~400ns as measured by {@link CharEscapersBenchmark}.
127   */
128  @Override
129  protected int nextEscapeIndex(CharSequence csq, int index, int end) {
130    checkNotNull(csq);
131    for (; index < end; index++) {
132      char c = csq.charAt(index);
133      if (c >= safeOctets.length || !safeOctets[c]) {
134        break;
135      }
136    }
137    return index;
138  }
139
140  /*
141   * Overridden for performance. For unescaped strings this improved the performance of the uri
142   * escaper from ~400ns to ~170ns as measured by {@link CharEscapersBenchmark}.
143   */
144  @Override
145  public String escape(String s) {
146    checkNotNull(s);
147    int slen = s.length();
148    for (int index = 0; index < slen; index++) {
149      char c = s.charAt(index);
150      if (c >= safeOctets.length || !safeOctets[c]) {
151        return escapeSlow(s, index);
152      }
153    }
154    return s;
155  }
156
157  /** Escapes the given Unicode code point in UTF-8. */
158  @Override
159  @CheckForNull
160  protected char[] escape(int cp) {
161    // We should never get negative values here but if we do it will throw an
162    // IndexOutOfBoundsException, so at least it will get spotted.
163    if (cp < safeOctets.length && safeOctets[cp]) {
164      return null;
165    } else if (cp == ' ' && plusForSpace) {
166      return PLUS_SIGN;
167    } else if (cp <= 0x7F) {
168      // Single byte UTF-8 characters
169      // Start with "%--" and fill in the blanks
170      char[] dest = new char[3];
171      dest[0] = '%';
172      dest[2] = UPPER_HEX_DIGITS[cp & 0xF];
173      dest[1] = UPPER_HEX_DIGITS[cp >>> 4];
174      return dest;
175    } else if (cp <= 0x7ff) {
176      // Two byte UTF-8 characters [cp >= 0x80 && cp <= 0x7ff]
177      // Start with "%--%--" and fill in the blanks
178      char[] dest = new char[6];
179      dest[0] = '%';
180      dest[3] = '%';
181      dest[5] = UPPER_HEX_DIGITS[cp & 0xF];
182      cp >>>= 4;
183      dest[4] = UPPER_HEX_DIGITS[0x8 | (cp & 0x3)];
184      cp >>>= 2;
185      dest[2] = UPPER_HEX_DIGITS[cp & 0xF];
186      cp >>>= 4;
187      dest[1] = UPPER_HEX_DIGITS[0xC | cp];
188      return dest;
189    } else if (cp <= 0xffff) {
190      // Three byte UTF-8 characters [cp >= 0x800 && cp <= 0xffff]
191      // Start with "%E-%--%--" and fill in the blanks
192      char[] dest = new char[9];
193      dest[0] = '%';
194      dest[1] = 'E';
195      dest[3] = '%';
196      dest[6] = '%';
197      dest[8] = UPPER_HEX_DIGITS[cp & 0xF];
198      cp >>>= 4;
199      dest[7] = UPPER_HEX_DIGITS[0x8 | (cp & 0x3)];
200      cp >>>= 2;
201      dest[5] = UPPER_HEX_DIGITS[cp & 0xF];
202      cp >>>= 4;
203      dest[4] = UPPER_HEX_DIGITS[0x8 | (cp & 0x3)];
204      cp >>>= 2;
205      dest[2] = UPPER_HEX_DIGITS[cp];
206      return dest;
207    } else if (cp <= 0x10ffff) {
208      char[] dest = new char[12];
209      // Four byte UTF-8 characters [cp >= 0xffff && cp <= 0x10ffff]
210      // Start with "%F-%--%--%--" and fill in the blanks
211      dest[0] = '%';
212      dest[1] = 'F';
213      dest[3] = '%';
214      dest[6] = '%';
215      dest[9] = '%';
216      dest[11] = UPPER_HEX_DIGITS[cp & 0xF];
217      cp >>>= 4;
218      dest[10] = UPPER_HEX_DIGITS[0x8 | (cp & 0x3)];
219      cp >>>= 2;
220      dest[8] = UPPER_HEX_DIGITS[cp & 0xF];
221      cp >>>= 4;
222      dest[7] = UPPER_HEX_DIGITS[0x8 | (cp & 0x3)];
223      cp >>>= 2;
224      dest[5] = UPPER_HEX_DIGITS[cp & 0xF];
225      cp >>>= 4;
226      dest[4] = UPPER_HEX_DIGITS[0x8 | (cp & 0x3)];
227      cp >>>= 2;
228      dest[2] = UPPER_HEX_DIGITS[cp & 0x7];
229      return dest;
230    } else {
231      // If this ever happens it is due to bug in UnicodeEscaper, not bad input.
232      throw new IllegalArgumentException("Invalid unicode character value " + cp);
233    }
234  }
235}