001/*
002 * Copyright (C) 2012 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.io;
016
017import static com.google.common.base.Preconditions.checkArgument;
018import static com.google.common.base.Preconditions.checkNotNull;
019import static com.google.common.base.Preconditions.checkPositionIndexes;
020import static com.google.common.base.Preconditions.checkState;
021import static com.google.common.io.GwtWorkarounds.asCharInput;
022import static com.google.common.io.GwtWorkarounds.asCharOutput;
023import static com.google.common.io.GwtWorkarounds.asInputStream;
024import static com.google.common.io.GwtWorkarounds.asOutputStream;
025import static com.google.common.io.GwtWorkarounds.stringBuilderOutput;
026import static com.google.common.math.IntMath.divide;
027import static com.google.common.math.IntMath.log2;
028import static java.math.RoundingMode.CEILING;
029import static java.math.RoundingMode.FLOOR;
030import static java.math.RoundingMode.UNNECESSARY;
031
032import com.google.common.annotations.Beta;
033import com.google.common.annotations.GwtCompatible;
034import com.google.common.annotations.GwtIncompatible;
035import com.google.common.base.Ascii;
036import com.google.common.base.CharMatcher;
037import com.google.common.io.GwtWorkarounds.ByteInput;
038import com.google.common.io.GwtWorkarounds.ByteOutput;
039import com.google.common.io.GwtWorkarounds.CharInput;
040import com.google.common.io.GwtWorkarounds.CharOutput;
041
042import java.io.IOException;
043import java.io.InputStream;
044import java.io.OutputStream;
045import java.io.Reader;
046import java.io.Writer;
047import java.util.Arrays;
048
049import javax.annotation.CheckReturnValue;
050import javax.annotation.Nullable;
051
052/**
053 * A binary encoding scheme for reversibly translating between byte sequences and printable ASCII
054 * strings. This class includes several constants for encoding schemes specified by <a
055 * href="http://tools.ietf.org/html/rfc4648">RFC 4648</a>. For example, the expression:
056 * <pre>   {@code
057 *
058 *   BaseEncoding.base32().encode("foo".getBytes(Charsets.US_ASCII))
059 * }</pre>
060 * returns the string {@code "MZXW6==="}, and <pre>   {@code
061 *
062 *  byte[] decoded = BaseEncoding.base32().decode("MZXW6===");
063 * }</pre>
064 *
065 * ...returns the ASCII bytes of the string {@code "foo"}.
066 *
067 * <p>By default, {@code BaseEncoding}'s behavior is relatively strict and in accordance with
068 * RFC 4648.  Decoding rejects characters in the wrong case, though padding is optional.
069 * To modify encoding and decoding behavior, use configuration methods to obtain a new encoding
070 * with modified behavior: <pre>   {@code
071 *
072 *  BaseEncoding.base16().lowerCase().decode("deadbeef");
073 * }</pre>
074 *
075 * <p>Warning: BaseEncoding instances are immutable.  Invoking a configuration method has no effect
076 * on the receiving instance; you must store and use the new encoding instance it returns, instead.
077 * <pre>   {@code
078 *
079 *   // Do NOT do this
080 *   BaseEncoding hex = BaseEncoding.base16();
081 *   hex.lowerCase(); // does nothing!
082 *   return hex.decode("deadbeef"); // throws an IllegalArgumentException
083 * }</pre>
084 *
085 * <p>It is guaranteed that {@code encoding.decode(encoding.encode(x))} is always equal to
086 * {@code x}, but the reverse does not necessarily hold.
087 *
088 * <p>
089 * <table>
090 * <tr>
091 * <th>Encoding
092 * <th>Alphabet
093 * <th>{@code char:byte} ratio
094 * <th>Default padding
095 * <th>Comments
096 * <tr>
097 * <td>{@link #base16()}
098 * <td>0-9 A-F
099 * <td>2.00
100 * <td>N/A
101 * <td>Traditional hexadecimal.  Defaults to upper case.
102 * <tr>
103 * <td>{@link #base32()}
104 * <td>A-Z 2-7
105 * <td>1.60
106 * <td>=
107 * <td>Human-readable; no possibility of mixing up 0/O or 1/I.  Defaults to upper case.
108 * <tr>
109 * <td>{@link #base32Hex()}
110 * <td>0-9 A-V
111 * <td>1.60
112 * <td>=
113 * <td>"Numerical" base 32; extended from the traditional hex alphabet.  Defaults to upper case.
114 * <tr>
115 * <td>{@link #base64()}
116 * <td>A-Z a-z 0-9 + /
117 * <td>1.33
118 * <td>=
119 * <td>
120 * <tr>
121 * <td>{@link #base64Url()}
122 * <td>A-Z a-z 0-9 - _
123 * <td>1.33
124 * <td>=
125 * <td>Safe to use as filenames, or to pass in URLs without escaping
126 * </table>
127 *
128 * <p>
129 * All instances of this class are immutable, so they may be stored safely as static constants.
130 *
131 * @author Louis Wasserman
132 * @since 14.0
133 */
134@Beta
135@GwtCompatible(emulated = true)
136public abstract class BaseEncoding {
137  // TODO(user): consider adding encodeTo(Appendable, byte[], [int, int])
138
139  BaseEncoding() {}
140
141  /**
142   * Encodes the specified byte array, and returns the encoded {@code String}.
143   */
144  public String encode(byte[] bytes) {
145    return encode(checkNotNull(bytes), 0, bytes.length);
146  }
147
148  /**
149   * Encodes the specified range of the specified byte array, and returns the encoded
150   * {@code String}.
151   */
152  public final String encode(byte[] bytes, int off, int len) {
153    checkNotNull(bytes);
154    checkPositionIndexes(off, off + len, bytes.length);
155    CharOutput result = stringBuilderOutput(maxEncodedSize(len));
156    ByteOutput byteOutput = encodingStream(result);
157    try {
158      for (int i = 0; i < len; i++) {
159        byteOutput.write(bytes[off + i]);
160      }
161      byteOutput.close();
162    } catch (IOException impossible) {
163      throw new AssertionError("impossible");
164    }
165    return result.toString();
166  }
167
168  /**
169   * Returns an {@code OutputStream} that encodes bytes using this encoding into the specified
170   * {@code Writer}.  When the returned {@code OutputStream} is closed, so is the backing
171   * {@code Writer}.
172   */
173  @GwtIncompatible("Writer,OutputStream")
174  public final OutputStream encodingStream(Writer writer) {
175    return asOutputStream(encodingStream(asCharOutput(writer)));
176  }
177
178  /**
179   * Returns an {@code OutputSupplier} that supplies streams that encode bytes using this encoding
180   * into writers from the specified {@code OutputSupplier}.
181   */
182  @GwtIncompatible("Writer,OutputStream")
183  public final OutputSupplier<OutputStream> encodingStream(
184      final OutputSupplier<Writer> writerSupplier) {
185    checkNotNull(writerSupplier);
186    return new OutputSupplier<OutputStream>() {
187      @Override
188      public OutputStream getOutput() throws IOException {
189        return encodingStream(writerSupplier.getOutput());
190      }
191    };
192  }
193
194  // TODO(user): document the extent of leniency, probably after adding ignore(CharMatcher)
195
196  private static byte[] extract(byte[] result, int length) {
197    if (length == result.length) {
198      return result;
199    } else {
200      byte[] trunc = new byte[length];
201      System.arraycopy(result, 0, trunc, 0, length);
202      return trunc;
203    }
204  }
205
206  /**
207   * Decodes the specified character sequence, and returns the resulting {@code byte[]}.
208   * This is the inverse operation to {@link #encode(byte[])}.
209   *
210   * @throws IllegalArgumentException if the input is not a valid encoded string according to this
211   *         encoding.
212   */
213  public final byte[] decode(CharSequence chars) {
214    ByteInput decodedInput = decodingStream(asCharInput(chars));
215    byte[] tmp = new byte[maxDecodedSize(chars.length())];
216    int index = 0;
217    try {
218      for (int i = decodedInput.read(); i != -1; i = decodedInput.read()) {
219        tmp[index++] = (byte) i;
220      }
221    } catch (IOException badInput) {
222      throw new IllegalArgumentException(badInput);
223    }
224    return extract(tmp, index);
225  }
226
227  /**
228   * Returns an {@code InputStream} that decodes base-encoded input from the specified
229   * {@code Reader}.
230   */
231  @GwtIncompatible("Reader,InputStream")
232  public final InputStream decodingStream(Reader reader) {
233    return asInputStream(decodingStream(asCharInput(reader)));
234  }
235
236  /**
237   * Returns an {@code InputSupplier} that supplies input streams that decode base-encoded input
238   * from readers from the specified supplier.
239   */
240  @GwtIncompatible("Reader,InputStream")
241  public InputSupplier<InputStream> decodingStream(final InputSupplier<Reader> readerSupplier) {
242    checkNotNull(readerSupplier);
243    return new InputSupplier<InputStream>() {
244      @Override
245      public InputStream getInput() throws IOException {
246        return decodingStream(readerSupplier.getInput());
247      }
248    };
249  }
250
251  // Implementations for encoding/decoding
252
253  abstract int maxEncodedSize(int bytes);
254
255  abstract ByteOutput encodingStream(CharOutput charOutput);
256
257  abstract int maxDecodedSize(int chars);
258
259  abstract ByteInput decodingStream(CharInput charInput);
260
261  // Modified encoding generators
262
263  /**
264   * Returns an encoding that behaves equivalently to this encoding, but omits any padding
265   * characters as specified by <a href="http://tools.ietf.org/html/rfc4648#section-3.2">RFC 4648
266   * section 3.2</a>, Padding of Encoded Data.
267   */
268  @CheckReturnValue
269  public abstract BaseEncoding omitPadding();
270
271  /**
272   * Returns an encoding that behaves equivalently to this encoding, but uses an alternate character
273   * for padding.
274   *
275   * @throws IllegalArgumentException if this padding character is already used in the alphabet or a
276   *         separator
277   */
278  @CheckReturnValue
279  public abstract BaseEncoding withPadChar(char padChar);
280
281  /**
282   * Returns an encoding that behaves equivalently to this encoding, but adds a separator string
283   * after every {@code n} characters. Any occurrences of any characters that occur in the separator
284   * are skipped over in decoding.
285   *
286   * @throws IllegalArgumentException if any alphabet or padding characters appear in the separator
287   *         string, or if {@code n <= 0}
288   * @throws UnsupportedOperationException if this encoding already uses a separator
289   */
290  @CheckReturnValue
291  public abstract BaseEncoding withSeparator(String separator, int n);
292
293  /**
294   * Returns an encoding that behaves equivalently to this encoding, but encodes and decodes with
295   * uppercase letters. Padding and separator characters remain in their original case.
296   *
297   * @throws IllegalStateException if the alphabet used by this encoding contains mixed upper- and
298   *         lower-case characters
299   */
300  @CheckReturnValue
301  public abstract BaseEncoding upperCase();
302
303  /**
304   * Returns an encoding that behaves equivalently to this encoding, but encodes and decodes with
305   * lowercase letters. Padding and separator characters remain in their original case.
306   *
307   * @throws IllegalStateException if the alphabet used by this encoding contains mixed upper- and
308   *         lower-case characters
309   */
310  @CheckReturnValue
311  public abstract BaseEncoding lowerCase();
312
313  private static final BaseEncoding BASE64 = new StandardBaseEncoding(
314      "base64()", "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/", '=');
315
316  /**
317   * The "base64" base encoding specified by <a
318   * href="http://tools.ietf.org/html/rfc4648#section-4">RFC 4648 section 4</a>, Base 64 Encoding.
319   * (This is the same as the base 64 encoding from <a
320   * href="http://tools.ietf.org/html/rfc3548#section-3">RFC 3548</a>.)
321   *
322   * <p>The character {@code '='} is used for padding, but can be {@linkplain #omitPadding()
323   * omitted} or {@linkplain #withPadChar(char) replaced}.
324   *
325   * <p>No line feeds are added by default, as per <a
326   * href="http://tools.ietf.org/html/rfc4648#section-3.1"> RFC 4648 section 3.1</a>, Line Feeds in
327   * Encoded Data. Line feeds may be added using {@link #withSeparator(String, int)}.
328   */
329  public static BaseEncoding base64() {
330    return BASE64;
331  }
332
333  private static final BaseEncoding BASE64_URL = new StandardBaseEncoding(
334      "base64Url()", "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_", '=');
335
336  /**
337   * The "base64url" encoding specified by <a
338   * href="http://tools.ietf.org/html/rfc4648#section-5">RFC 4648 section 5</a>, Base 64 Encoding
339   * with URL and Filename Safe Alphabet, also sometimes referred to as the "web safe Base64."
340   * (This is the same as the base 64 encoding with URL and filename safe alphabet from <a
341   * href="http://tools.ietf.org/html/rfc3548#section-4">RFC 3548</a>.)
342   *
343   * <p>The character {@code '='} is used for padding, but can be {@linkplain #omitPadding()
344   * omitted} or {@linkplain #withPadChar(char) replaced}.
345   *
346   * <p>No line feeds are added by default, as per <a
347   * href="http://tools.ietf.org/html/rfc4648#section-3.1"> RFC 4648 section 3.1</a>, Line Feeds in
348   * Encoded Data. Line feeds may be added using {@link #withSeparator(String, int)}.
349   */
350  public static BaseEncoding base64Url() {
351    return BASE64_URL;
352  }
353
354  private static final BaseEncoding BASE32 =
355      new StandardBaseEncoding("base32()", "ABCDEFGHIJKLMNOPQRSTUVWXYZ234567", '=');
356
357  /**
358   * The "base32" encoding specified by <a
359   * href="http://tools.ietf.org/html/rfc4648#section-6">RFC 4648 section 6</a>, Base 32 Encoding.
360   * (This is the same as the base 32 encoding from <a
361   * href="http://tools.ietf.org/html/rfc3548#section-5">RFC 3548</a>.)
362   *
363   * <p>The character {@code '='} is used for padding, but can be {@linkplain #omitPadding()
364   * omitted} or {@linkplain #withPadChar(char) replaced}.
365   *
366   * <p>No line feeds are added by default, as per <a
367   * href="http://tools.ietf.org/html/rfc4648#section-3.1"> RFC 4648 section 3.1</a>, Line Feeds in
368   * Encoded Data. Line feeds may be added using {@link #withSeparator(String, int)}.
369   */
370  public static BaseEncoding base32() {
371    return BASE32;
372  }
373
374  private static final BaseEncoding BASE32_HEX =
375      new StandardBaseEncoding("base32Hex()", "0123456789ABCDEFGHIJKLMNOPQRSTUV", '=');
376
377  /**
378   * The "base32hex" encoding specified by <a
379   * href="http://tools.ietf.org/html/rfc4648#section-7">RFC 4648 section 7</a>, Base 32 Encoding
380   * with Extended Hex Alphabet.  There is no corresponding encoding in RFC 3548.
381   *
382   * <p>The character {@code '='} is used for padding, but can be {@linkplain #omitPadding()
383   * omitted} or {@linkplain #withPadChar(char) replaced}.
384   *
385   * <p>No line feeds are added by default, as per <a
386   * href="http://tools.ietf.org/html/rfc4648#section-3.1"> RFC 4648 section 3.1</a>, Line Feeds in
387   * Encoded Data. Line feeds may be added using {@link #withSeparator(String, int)}.
388   */
389  public static BaseEncoding base32Hex() {
390    return BASE32_HEX;
391  }
392
393  private static final BaseEncoding BASE16 =
394      new StandardBaseEncoding("base16()", "0123456789ABCDEF", null);
395
396  /**
397   * The "base16" encoding specified by <a
398   * href="http://tools.ietf.org/html/rfc4648#section-8">RFC 4648 section 8</a>, Base 16 Encoding.
399   * (This is the same as the base 16 encoding from <a
400   * href="http://tools.ietf.org/html/rfc3548#section-6">RFC 3548</a>.) This is commonly known as
401   * "hexadecimal" format.
402   *
403   * <p>No padding is necessary in base 16, so {@link #withPadChar(char)} and
404   * {@link #omitPadding()} have no effect.
405   *
406   * <p>No line feeds are added by default, as per <a
407   * href="http://tools.ietf.org/html/rfc4648#section-3.1"> RFC 4648 section 3.1</a>, Line Feeds in
408   * Encoded Data. Line feeds may be added using {@link #withSeparator(String, int)}.
409   */
410  public static BaseEncoding base16() {
411    return BASE16;
412  }
413
414  private static final class Alphabet extends CharMatcher {
415    private final String name;
416    // this is meant to be immutable -- don't modify it!
417    private final char[] chars;
418    final int mask;
419    final int bitsPerChar;
420    final int charsPerChunk;
421    final int bytesPerChunk;
422    private final byte[] decodabet;
423    private final boolean[] validPadding;
424
425    Alphabet(String name, char[] chars) {
426      this.name = checkNotNull(name);
427      this.chars = checkNotNull(chars);
428      try {
429        this.bitsPerChar = log2(chars.length, UNNECESSARY);
430      } catch (ArithmeticException e) {
431        throw new IllegalArgumentException("Illegal alphabet length " + chars.length, e);
432      }
433
434      /*
435       * e.g. for base64, bitsPerChar == 6, charsPerChunk == 4, and bytesPerChunk == 3. This makes
436       * for the smallest chunk size that still has charsPerChunk * bitsPerChar be a multiple of 8.
437       */
438      int gcd = Math.min(8, Integer.lowestOneBit(bitsPerChar));
439      this.charsPerChunk = 8 / gcd;
440      this.bytesPerChunk = bitsPerChar / gcd;
441
442      this.mask = chars.length - 1;
443
444      byte[] decodabet = new byte[Ascii.MAX + 1];
445      Arrays.fill(decodabet, (byte) -1);
446      for (int i = 0; i < chars.length; i++) {
447        char c = chars[i];
448        checkArgument(CharMatcher.ASCII.matches(c), "Non-ASCII character: %s", c);
449        checkArgument(decodabet[c] == -1, "Duplicate character: %s", c);
450        decodabet[c] = (byte) i;
451      }
452      this.decodabet = decodabet;
453
454      boolean[] validPadding = new boolean[charsPerChunk];
455      for (int i = 0; i < bytesPerChunk; i++) {
456        validPadding[divide(i * 8, bitsPerChar, CEILING)] = true;
457      }
458      this.validPadding = validPadding;
459    }
460
461    char encode(int bits) {
462      return chars[bits];
463    }
464
465    boolean isValidPaddingStartPosition(int index) {
466      return validPadding[index % charsPerChunk];
467    }
468
469    int decode(char ch) throws IOException {
470      if (ch > Ascii.MAX || decodabet[ch] == -1) {
471        throw new IOException("Unrecognized character: " + ch);
472      }
473      return decodabet[ch];
474    }
475
476    private boolean hasLowerCase() {
477      for (char c : chars) {
478        if (Ascii.isLowerCase(c)) {
479          return true;
480        }
481      }
482      return false;
483    }
484
485    private boolean hasUpperCase() {
486      for (char c : chars) {
487        if (Ascii.isUpperCase(c)) {
488          return true;
489        }
490      }
491      return false;
492    }
493
494    Alphabet upperCase() {
495      if (!hasLowerCase()) {
496        return this;
497      } else {
498        checkState(!hasUpperCase(), "Cannot call upperCase() on a mixed-case alphabet");
499        char[] upperCased = new char[chars.length];
500        for (int i = 0; i < chars.length; i++) {
501          upperCased[i] = Ascii.toUpperCase(chars[i]);
502        }
503        return new Alphabet(name + ".upperCase()", upperCased);
504      }
505    }
506
507    Alphabet lowerCase() {
508      if (!hasUpperCase()) {
509        return this;
510      } else {
511        checkState(!hasLowerCase(), "Cannot call lowerCase() on a mixed-case alphabet");
512        char[] lowerCased = new char[chars.length];
513        for (int i = 0; i < chars.length; i++) {
514          lowerCased[i] = Ascii.toLowerCase(chars[i]);
515        }
516        return new Alphabet(name + ".lowerCase()", lowerCased);
517      }
518    }
519
520    @Override
521    public boolean matches(char c) {
522      return CharMatcher.ASCII.matches(c) && decodabet[c] != -1;
523    }
524
525    @Override
526    public String toString() {
527      return name;
528    }
529  }
530
531  static final class StandardBaseEncoding extends BaseEncoding {
532    // TODO(user): provide a useful toString
533    private final Alphabet alphabet;
534
535    @Nullable
536    private final Character paddingChar;
537
538    StandardBaseEncoding(String name, String alphabetChars, @Nullable Character paddingChar) {
539      this(new Alphabet(name, alphabetChars.toCharArray()), paddingChar);
540    }
541
542    StandardBaseEncoding(Alphabet alphabet, Character paddingChar) {
543      this.alphabet = checkNotNull(alphabet);
544      checkArgument(paddingChar == null || !alphabet.matches(paddingChar),
545          "Padding character %s was already in alphabet", paddingChar);
546      this.paddingChar = paddingChar;
547    }
548
549    private CharMatcher paddingMatcher() {
550      return (paddingChar == null) ? CharMatcher.NONE : CharMatcher.is(paddingChar.charValue());
551    }
552
553    @Override
554    int maxEncodedSize(int bytes) {
555      return alphabet.charsPerChunk * divide(bytes, alphabet.bytesPerChunk, CEILING);
556    }
557
558    @Override
559    ByteOutput encodingStream(final CharOutput out) {
560      checkNotNull(out);
561      return new ByteOutput() {
562        int bitBuffer = 0;
563        int bitBufferLength = 0;
564        int writtenChars = 0;
565
566        @Override
567        public void write(byte b) throws IOException {
568          bitBuffer <<= 8;
569          bitBuffer |= b & 0xFF;
570          bitBufferLength += 8;
571          while (bitBufferLength >= alphabet.bitsPerChar) {
572            int charIndex = (bitBuffer >> (bitBufferLength - alphabet.bitsPerChar))
573                & alphabet.mask;
574            out.write(alphabet.encode(charIndex));
575            writtenChars++;
576            bitBufferLength -= alphabet.bitsPerChar;
577          }
578        }
579
580        @Override
581        public void flush() throws IOException {
582          out.flush();
583        }
584
585        @Override
586        public void close() throws IOException {
587          if (bitBufferLength > 0) {
588            int charIndex = (bitBuffer << (alphabet.bitsPerChar - bitBufferLength))
589                & alphabet.mask;
590            out.write(alphabet.encode(charIndex));
591            writtenChars++;
592            if (paddingChar != null) {
593              while (writtenChars % alphabet.charsPerChunk != 0) {
594                out.write(paddingChar.charValue());
595                writtenChars++;
596              }
597            }
598          }
599          out.close();
600        }
601      };
602    }
603
604    @Override
605    int maxDecodedSize(int chars) {
606      return alphabet.bytesPerChunk * divide(chars, alphabet.charsPerChunk, CEILING);
607    }
608
609    @Override
610    ByteInput decodingStream(final CharInput reader) {
611      checkNotNull(reader);
612      return new ByteInput() {
613        int bitBuffer = 0;
614        int bitBufferLength = 0;
615        int readChars = 0;
616        boolean hitPadding = false;
617        final CharMatcher paddingMatcher = paddingMatcher();
618
619        @Override
620        public int read() throws IOException {
621          while (true) {
622            int readChar = reader.read();
623            if (readChar == -1) {
624              if (!hitPadding && !alphabet.isValidPaddingStartPosition(readChars)) {
625                throw new IOException("Invalid input length " + readChars);
626              }
627              return -1;
628            }
629            readChars++;
630            char ch = (char) readChar;
631            if (paddingMatcher.matches(ch)) {
632              if (!hitPadding
633                  && (readChars == 1 || !alphabet.isValidPaddingStartPosition(readChars - 1))) {
634                throw new IOException("Padding cannot start at index " + readChars);
635              }
636              hitPadding = true;
637            } else if (hitPadding) {
638              throw new IOException(
639                  "Expected padding character but found '" + ch + "' at index " + readChars);
640            } else {
641              bitBuffer <<= alphabet.bitsPerChar;
642              bitBuffer |= alphabet.decode(ch);
643              bitBufferLength += alphabet.bitsPerChar;
644
645              if (bitBufferLength >= 8) {
646                bitBufferLength -= 8;
647                return (bitBuffer >> bitBufferLength) & 0xFF;
648              }
649            }
650          }
651        }
652
653        @Override
654        public void close() throws IOException {
655          reader.close();
656        }
657      };
658    }
659
660    @Override
661    public BaseEncoding omitPadding() {
662      return (paddingChar == null) ? this : new StandardBaseEncoding(alphabet, null);
663    }
664
665    @Override
666    public BaseEncoding withPadChar(char padChar) {
667      if (8 % alphabet.bitsPerChar == 0 ||
668          (paddingChar != null && paddingChar.charValue() == padChar)) {
669        return this;
670      } else {
671        return new StandardBaseEncoding(alphabet, padChar);
672      }
673    }
674
675    @Override
676    public BaseEncoding withSeparator(String separator, int afterEveryChars) {
677      checkNotNull(separator);
678      checkArgument(paddingMatcher().or(alphabet).matchesNoneOf(separator),
679          "Separator cannot contain alphabet or padding characters");
680      return new SeparatedBaseEncoding(this, separator, afterEveryChars);
681    }
682
683    private transient BaseEncoding upperCase;
684    private transient BaseEncoding lowerCase;
685
686    @Override
687    public BaseEncoding upperCase() {
688      BaseEncoding result = upperCase;
689      if (result == null) {
690        Alphabet upper = alphabet.upperCase();
691        result = upperCase =
692            (upper == alphabet) ? this : new StandardBaseEncoding(upper, paddingChar);
693      }
694      return result;
695    }
696
697    @Override
698    public BaseEncoding lowerCase() {
699      BaseEncoding result = lowerCase;
700      if (result == null) {
701        Alphabet lower = alphabet.lowerCase();
702        result = lowerCase =
703            (lower == alphabet) ? this : new StandardBaseEncoding(lower, paddingChar);
704      }
705      return result;
706    }
707
708    @Override
709    public String toString() {
710      StringBuilder builder = new StringBuilder("BaseEncoding.");
711      builder.append(alphabet.toString());
712      if (8 % alphabet.bitsPerChar != 0) {
713        if (paddingChar == null) {
714          builder.append(".omitPadding()");
715        } else {
716          builder.append(".withPadChar(").append(paddingChar).append(')');
717        }
718      }
719      return builder.toString();
720    }
721  }
722
723  static CharInput ignoringInput(final CharInput delegate, final CharMatcher toIgnore) {
724    checkNotNull(delegate);
725    checkNotNull(toIgnore);
726    return new CharInput() {
727      @Override
728      public int read() throws IOException {
729        int readChar;
730        do {
731          readChar = delegate.read();
732        } while (readChar != -1 && toIgnore.matches((char) readChar));
733        return readChar;
734      }
735
736      @Override
737      public void close() throws IOException {
738        delegate.close();
739      }
740    };
741  }
742
743  static CharOutput separatingOutput(
744      final CharOutput delegate, final String separator, final int afterEveryChars) {
745    checkNotNull(delegate);
746    checkNotNull(separator);
747    checkArgument(afterEveryChars > 0);
748    return new CharOutput() {
749      int charsUntilSeparator = afterEveryChars;
750
751      @Override
752      public void write(char c) throws IOException {
753        if (charsUntilSeparator == 0) {
754          for (int i = 0; i < separator.length(); i++) {
755            delegate.write(separator.charAt(i));
756          }
757          charsUntilSeparator = afterEveryChars;
758        }
759        delegate.write(c);
760        charsUntilSeparator--;
761      }
762
763      @Override
764      public void flush() throws IOException {
765        delegate.flush();
766      }
767
768      @Override
769      public void close() throws IOException {
770        delegate.close();
771      }
772    };
773  }
774
775  static final class SeparatedBaseEncoding extends BaseEncoding {
776    private final BaseEncoding delegate;
777    private final String separator;
778    private final int afterEveryChars;
779    private final CharMatcher separatorChars;
780
781    SeparatedBaseEncoding(BaseEncoding delegate, String separator, int afterEveryChars) {
782      this.delegate = checkNotNull(delegate);
783      this.separator = checkNotNull(separator);
784      this.afterEveryChars = afterEveryChars;
785      checkArgument(
786          afterEveryChars > 0, "Cannot add a separator after every %s chars", afterEveryChars);
787      this.separatorChars = CharMatcher.anyOf(separator).precomputed();
788    }
789
790    @Override
791    int maxEncodedSize(int bytes) {
792      int unseparatedSize = delegate.maxEncodedSize(bytes);
793      return unseparatedSize + separator.length()
794          * divide(Math.max(0, unseparatedSize - 1), afterEveryChars, FLOOR);
795    }
796
797    @Override
798    ByteOutput encodingStream(final CharOutput output) {
799      return delegate.encodingStream(separatingOutput(output, separator, afterEveryChars));
800    }
801
802    @Override
803    int maxDecodedSize(int chars) {
804      return delegate.maxDecodedSize(chars);
805    }
806
807    @Override
808    ByteInput decodingStream(final CharInput input) {
809      return delegate.decodingStream(ignoringInput(input, separatorChars));
810    }
811
812    @Override
813    public BaseEncoding omitPadding() {
814      return delegate.omitPadding().withSeparator(separator, afterEveryChars);
815    }
816
817    @Override
818    public BaseEncoding withPadChar(char padChar) {
819      return delegate.withPadChar(padChar).withSeparator(separator, afterEveryChars);
820    }
821
822    @Override
823    public BaseEncoding withSeparator(String separator, int afterEveryChars) {
824      throw new UnsupportedOperationException("Already have a separator");
825    }
826
827    @Override
828    public BaseEncoding upperCase() {
829      return delegate.upperCase().withSeparator(separator, afterEveryChars);
830    }
831
832    @Override
833    public BaseEncoding lowerCase() {
834      return delegate.lowerCase().withSeparator(separator, afterEveryChars);
835    }
836
837    @Override
838    public String toString() {
839      return delegate.toString() +
840          ".withSeparator(\"" + separator + "\", " + afterEveryChars + ")";
841    }
842  }
843}