001/*
002 * Copyright (C) 2009 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.base;
016
017import static com.google.common.base.Preconditions.checkArgument;
018import static com.google.common.base.Preconditions.checkNotNull;
019
020import com.google.common.annotations.Beta;
021import com.google.common.annotations.GwtCompatible;
022import com.google.common.annotations.GwtIncompatible;
023import java.util.ArrayList;
024import java.util.Collections;
025import java.util.Iterator;
026import java.util.LinkedHashMap;
027import java.util.List;
028import java.util.Map;
029import java.util.regex.Pattern;
030import java.util.stream.Stream;
031import java.util.stream.StreamSupport;
032import javax.annotation.CheckForNull;
033
034/**
035 * Extracts non-overlapping substrings from an input string, typically by recognizing appearances of
036 * a <i>separator</i> sequence. This separator can be specified as a single {@linkplain #on(char)
037 * character}, fixed {@linkplain #on(String) string}, {@linkplain #onPattern regular expression} or
038 * {@link #on(CharMatcher) CharMatcher} instance. Or, instead of using a separator at all, a
039 * splitter can extract adjacent substrings of a given {@linkplain #fixedLength fixed length}.
040 *
041 * <p>For example, this expression:
042 *
043 * <pre>{@code
044 * Splitter.on(',').split("foo,bar,qux")
045 * }</pre>
046 *
047 * ... produces an {@code Iterable} containing {@code "foo"}, {@code "bar"} and {@code "qux"}, in
048 * that order.
049 *
050 * <p>By default, {@code Splitter}'s behavior is simplistic and unassuming. The following
051 * expression:
052 *
053 * <pre>{@code
054 * Splitter.on(',').split(" foo,,,  bar ,")
055 * }</pre>
056 *
057 * ... yields the substrings {@code [" foo", "", "", " bar ", ""]}. If this is not the desired
058 * behavior, use configuration methods to obtain a <i>new</i> splitter instance with modified
059 * behavior:
060 *
061 * <pre>{@code
062 * private static final Splitter MY_SPLITTER = Splitter.on(',')
063 *     .trimResults()
064 *     .omitEmptyStrings();
065 * }</pre>
066 *
067 * <p>Now {@code MY_SPLITTER.split("foo,,, bar ,")} returns just {@code ["foo", "bar"]}. Note that
068 * the order in which these configuration methods are called is never significant.
069 *
070 * <p><b>Warning:</b> Splitter instances are immutable. Invoking a configuration method has no
071 * effect on the receiving instance; you must store and use the new splitter instance it returns
072 * instead.
073 *
074 * <pre>{@code
075 * // Do NOT do this
076 * Splitter splitter = Splitter.on('/');
077 * splitter.trimResults(); // does nothing!
078 * return splitter.split("wrong / wrong / wrong");
079 * }</pre>
080 *
081 * <p>For separator-based splitters that do not use {@code omitEmptyStrings}, an input string
082 * containing {@code n} occurrences of the separator naturally yields an iterable of size {@code n +
083 * 1}. So if the separator does not occur anywhere in the input, a single substring is returned
084 * containing the entire input. Consequently, all splitters split the empty string to {@code [""]}
085 * (note: even fixed-length splitters).
086 *
087 * <p>Splitter instances are thread-safe immutable, and are therefore safe to store as {@code static
088 * final} constants.
089 *
090 * <p>The {@link Joiner} class provides the inverse operation to splitting, but note that a
091 * round-trip between the two should be assumed to be lossy.
092 *
093 * <p>See the Guava User Guide article on <a
094 * href="https://github.com/google/guava/wiki/StringsExplained#splitter">{@code Splitter}</a>.
095 *
096 * @author Julien Silland
097 * @author Jesse Wilson
098 * @author Kevin Bourrillion
099 * @author Louis Wasserman
100 * @since 1.0
101 */
102@GwtCompatible(emulated = true)
103@ElementTypesAreNonnullByDefault
104public final class Splitter {
105  private final CharMatcher trimmer;
106  private final boolean omitEmptyStrings;
107  private final Strategy strategy;
108  private final int limit;
109
110  private Splitter(Strategy strategy) {
111    this(strategy, false, CharMatcher.none(), Integer.MAX_VALUE);
112  }
113
114  private Splitter(Strategy strategy, boolean omitEmptyStrings, CharMatcher trimmer, int limit) {
115    this.strategy = strategy;
116    this.omitEmptyStrings = omitEmptyStrings;
117    this.trimmer = trimmer;
118    this.limit = limit;
119  }
120
121  /**
122   * Returns a splitter that uses the given single-character separator. For example, {@code
123   * Splitter.on(',').split("foo,,bar")} returns an iterable containing {@code ["foo", "", "bar"]}.
124   *
125   * @param separator the character to recognize as a separator
126   * @return a splitter, with default settings, that recognizes that separator
127   */
128  public static Splitter on(char separator) {
129    return on(CharMatcher.is(separator));
130  }
131
132  /**
133   * Returns a splitter that considers any single character matched by the given {@code CharMatcher}
134   * to be a separator. For example, {@code
135   * Splitter.on(CharMatcher.anyOf(";,")).split("foo,;bar,quux")} returns an iterable containing
136   * {@code ["foo", "", "bar", "quux"]}.
137   *
138   * @param separatorMatcher a {@link CharMatcher} that determines whether a character is a
139   *     separator
140   * @return a splitter, with default settings, that uses this matcher
141   */
142  public static Splitter on(final CharMatcher separatorMatcher) {
143    checkNotNull(separatorMatcher);
144
145    return new Splitter(
146        new Strategy() {
147          @Override
148          public SplittingIterator iterator(Splitter splitter, final CharSequence toSplit) {
149            return new SplittingIterator(splitter, toSplit) {
150              @Override
151              int separatorStart(int start) {
152                return separatorMatcher.indexIn(toSplit, start);
153              }
154
155              @Override
156              int separatorEnd(int separatorPosition) {
157                return separatorPosition + 1;
158              }
159            };
160          }
161        });
162  }
163
164  /**
165   * Returns a splitter that uses the given fixed string as a separator. For example, {@code
166   * Splitter.on(", ").split("foo, bar,baz")} returns an iterable containing {@code ["foo",
167   * "bar,baz"]}.
168   *
169   * @param separator the literal, nonempty string to recognize as a separator
170   * @return a splitter, with default settings, that recognizes that separator
171   */
172  public static Splitter on(final String separator) {
173    checkArgument(separator.length() != 0, "The separator may not be the empty string.");
174    if (separator.length() == 1) {
175      return Splitter.on(separator.charAt(0));
176    }
177    return new Splitter(
178        new Strategy() {
179          @Override
180          public SplittingIterator iterator(Splitter splitter, CharSequence toSplit) {
181            return new SplittingIterator(splitter, toSplit) {
182              @Override
183              public int separatorStart(int start) {
184                int separatorLength = separator.length();
185
186                positions:
187                for (int p = start, last = toSplit.length() - separatorLength; p <= last; p++) {
188                  for (int i = 0; i < separatorLength; i++) {
189                    if (toSplit.charAt(i + p) != separator.charAt(i)) {
190                      continue positions;
191                    }
192                  }
193                  return p;
194                }
195                return -1;
196              }
197
198              @Override
199              public int separatorEnd(int separatorPosition) {
200                return separatorPosition + separator.length();
201              }
202            };
203          }
204        });
205  }
206
207  /**
208   * Returns a splitter that considers any subsequence matching {@code pattern} to be a separator.
209   * For example, {@code Splitter.on(Pattern.compile("\r?\n")).split(entireFile)} splits a string
210   * into lines whether it uses DOS-style or UNIX-style line terminators.
211   *
212   * @param separatorPattern the pattern that determines whether a subsequence is a separator. This
213   *     pattern may not match the empty string.
214   * @return a splitter, with default settings, that uses this pattern
215   * @throws IllegalArgumentException if {@code separatorPattern} matches the empty string
216   */
217  @GwtIncompatible // java.util.regex
218  public static Splitter on(Pattern separatorPattern) {
219    return on(new JdkPattern(separatorPattern));
220  }
221
222  private static Splitter on(final CommonPattern separatorPattern) {
223    checkArgument(
224        !separatorPattern.matcher("").matches(),
225        "The pattern may not match the empty string: %s",
226        separatorPattern);
227
228    return new Splitter(
229        new Strategy() {
230          @Override
231          public SplittingIterator iterator(final Splitter splitter, CharSequence toSplit) {
232            final CommonMatcher matcher = separatorPattern.matcher(toSplit);
233            return new SplittingIterator(splitter, toSplit) {
234              @Override
235              public int separatorStart(int start) {
236                return matcher.find(start) ? matcher.start() : -1;
237              }
238
239              @Override
240              public int separatorEnd(int separatorPosition) {
241                return matcher.end();
242              }
243            };
244          }
245        });
246  }
247
248  /**
249   * Returns a splitter that considers any subsequence matching a given pattern (regular expression)
250   * to be a separator. For example, {@code Splitter.onPattern("\r?\n").split(entireFile)} splits a
251   * string into lines whether it uses DOS-style or UNIX-style line terminators. This is equivalent
252   * to {@code Splitter.on(Pattern.compile(pattern))}.
253   *
254   * @param separatorPattern the pattern that determines whether a subsequence is a separator. This
255   *     pattern may not match the empty string.
256   * @return a splitter, with default settings, that uses this pattern
257   * @throws IllegalArgumentException if {@code separatorPattern} matches the empty string or is a
258   *     malformed expression
259   */
260  @GwtIncompatible // java.util.regex
261  public static Splitter onPattern(String separatorPattern) {
262    return on(Platform.compilePattern(separatorPattern));
263  }
264
265  /**
266   * Returns a splitter that divides strings into pieces of the given length. For example, {@code
267   * Splitter.fixedLength(2).split("abcde")} returns an iterable containing {@code ["ab", "cd",
268   * "e"]}. The last piece can be smaller than {@code length} but will never be empty.
269   *
270   * <p><b>Note:</b> if {@link #fixedLength} is used in conjunction with {@link #limit}, the final
271   * split piece <i>may be longer than the specified fixed length</i>. This is because the splitter
272   * will <i>stop splitting when the limit is reached</i>, and just return the final piece as-is.
273   *
274   * <p><b>Exception:</b> for consistency with separator-based splitters, {@code split("")} does not
275   * yield an empty iterable, but an iterable containing {@code ""}. This is the only case in which
276   * {@code Iterables.size(split(input))} does not equal {@code IntMath.divide(input.length(),
277   * length, CEILING)}. To avoid this behavior, use {@code omitEmptyStrings}.
278   *
279   * @param length the desired length of pieces after splitting, a positive integer
280   * @return a splitter, with default settings, that can split into fixed sized pieces
281   * @throws IllegalArgumentException if {@code length} is zero or negative
282   */
283  public static Splitter fixedLength(final int length) {
284    checkArgument(length > 0, "The length may not be less than 1");
285
286    return new Splitter(
287        new Strategy() {
288          @Override
289          public SplittingIterator iterator(final Splitter splitter, CharSequence toSplit) {
290            return new SplittingIterator(splitter, toSplit) {
291              @Override
292              public int separatorStart(int start) {
293                int nextChunkStart = start + length;
294                return (nextChunkStart < toSplit.length() ? nextChunkStart : -1);
295              }
296
297              @Override
298              public int separatorEnd(int separatorPosition) {
299                return separatorPosition;
300              }
301            };
302          }
303        });
304  }
305
306  /**
307   * Returns a splitter that behaves equivalently to {@code this} splitter, but automatically omits
308   * empty strings from the results. For example, {@code
309   * Splitter.on(',').omitEmptyStrings().split(",a,,,b,c,,")} returns an iterable containing only
310   * {@code ["a", "b", "c"]}.
311   *
312   * <p>If either {@code trimResults} option is also specified when creating a splitter, that
313   * splitter always trims results first before checking for emptiness. So, for example, {@code
314   * Splitter.on(':').omitEmptyStrings().trimResults().split(": : : ")} returns an empty iterable.
315   *
316   * <p>Note that it is ordinarily not possible for {@link #split(CharSequence)} to return an empty
317   * iterable, but when using this option, it can (if the input sequence consists of nothing but
318   * separators).
319   *
320   * @return a splitter with the desired configuration
321   */
322  public Splitter omitEmptyStrings() {
323    return new Splitter(strategy, true, trimmer, limit);
324  }
325
326  /**
327   * Returns a splitter that behaves equivalently to {@code this} splitter but stops splitting after
328   * it reaches the limit. The limit defines the maximum number of items returned by the iterator,
329   * or the maximum size of the list returned by {@link #splitToList}.
330   *
331   * <p>For example, {@code Splitter.on(',').limit(3).split("a,b,c,d")} returns an iterable
332   * containing {@code ["a", "b", "c,d"]}. When omitting empty strings, the omitted strings do not
333   * count. Hence, {@code Splitter.on(',').limit(3).omitEmptyStrings().split("a,,,b,,,c,d")} returns
334   * an iterable containing {@code ["a", "b", "c,d"}. When trim is requested, all entries are
335   * trimmed, including the last. Hence {@code Splitter.on(',').limit(3).trimResults().split(" a , b
336   * , c , d ")} results in {@code ["a", "b", "c , d"]}.
337   *
338   * @param maxItems the maximum number of items returned
339   * @return a splitter with the desired configuration
340   * @since 9.0
341   */
342  public Splitter limit(int maxItems) {
343    checkArgument(maxItems > 0, "must be greater than zero: %s", maxItems);
344    return new Splitter(strategy, omitEmptyStrings, trimmer, maxItems);
345  }
346
347  /**
348   * Returns a splitter that behaves equivalently to {@code this} splitter, but automatically
349   * removes leading and trailing {@linkplain CharMatcher#whitespace whitespace} from each returned
350   * substring; equivalent to {@code trimResults(CharMatcher.whitespace())}. For example, {@code
351   * Splitter.on(',').trimResults().split(" a, b ,c ")} returns an iterable containing {@code ["a",
352   * "b", "c"]}.
353   *
354   * @return a splitter with the desired configuration
355   */
356  public Splitter trimResults() {
357    return trimResults(CharMatcher.whitespace());
358  }
359
360  /**
361   * Returns a splitter that behaves equivalently to {@code this} splitter, but removes all leading
362   * or trailing characters matching the given {@code CharMatcher} from each returned substring. For
363   * example, {@code Splitter.on(',').trimResults(CharMatcher.is('_')).split("_a ,_b_ ,c__")}
364   * returns an iterable containing {@code ["a ", "b_ ", "c"]}.
365   *
366   * @param trimmer a {@link CharMatcher} that determines whether a character should be removed from
367   *     the beginning/end of a subsequence
368   * @return a splitter with the desired configuration
369   */
370  // TODO(kevinb): throw if a trimmer was already specified!
371  public Splitter trimResults(CharMatcher trimmer) {
372    checkNotNull(trimmer);
373    return new Splitter(strategy, omitEmptyStrings, trimmer, limit);
374  }
375
376  /**
377   * Splits {@code sequence} into string components and makes them available through an {@link
378   * Iterator}, which may be lazily evaluated. If you want an eagerly computed {@link List}, use
379   * {@link #splitToList(CharSequence)}. Java 8 users may prefer {@link #splitToStream} instead.
380   *
381   * @param sequence the sequence of characters to split
382   * @return an iteration over the segments split from the parameter
383   */
384  public Iterable<String> split(final CharSequence sequence) {
385    checkNotNull(sequence);
386
387    return new Iterable<String>() {
388      @Override
389      public Iterator<String> iterator() {
390        return splittingIterator(sequence);
391      }
392
393      @Override
394      public String toString() {
395        return Joiner.on(", ")
396            .appendTo(new StringBuilder().append('['), this)
397            .append(']')
398            .toString();
399      }
400    };
401  }
402
403  private Iterator<String> splittingIterator(CharSequence sequence) {
404    return strategy.iterator(this, sequence);
405  }
406
407  /**
408   * Splits {@code sequence} into string components and returns them as an immutable list. If you
409   * want an {@link Iterable} which may be lazily evaluated, use {@link #split(CharSequence)}.
410   *
411   * @param sequence the sequence of characters to split
412   * @return an immutable list of the segments split from the parameter
413   * @since 15.0
414   */
415  public List<String> splitToList(CharSequence sequence) {
416    checkNotNull(sequence);
417
418    Iterator<String> iterator = splittingIterator(sequence);
419    List<String> result = new ArrayList<>();
420
421    while (iterator.hasNext()) {
422      result.add(iterator.next());
423    }
424
425    return Collections.unmodifiableList(result);
426  }
427
428  /**
429   * Splits {@code sequence} into string components and makes them available through an {@link
430   * Stream}, which may be lazily evaluated. If you want an eagerly computed {@link List}, use
431   * {@link #splitToList(CharSequence)}.
432   *
433   * @param sequence the sequence of characters to split
434   * @return a stream over the segments split from the parameter
435   * @since 28.2
436   */
437  @Beta
438  public Stream<String> splitToStream(CharSequence sequence) {
439    // Can't use Streams.stream() from base
440    return StreamSupport.stream(split(sequence).spliterator(), false);
441  }
442
443  /**
444   * Returns a {@code MapSplitter} which splits entries based on this splitter, and splits entries
445   * into keys and values using the specified separator.
446   *
447   * @since 10.0
448   */
449  @Beta
450  public MapSplitter withKeyValueSeparator(String separator) {
451    return withKeyValueSeparator(on(separator));
452  }
453
454  /**
455   * Returns a {@code MapSplitter} which splits entries based on this splitter, and splits entries
456   * into keys and values using the specified separator.
457   *
458   * @since 14.0
459   */
460  @Beta
461  public MapSplitter withKeyValueSeparator(char separator) {
462    return withKeyValueSeparator(on(separator));
463  }
464
465  /**
466   * Returns a {@code MapSplitter} which splits entries based on this splitter, and splits entries
467   * into keys and values using the specified key-value splitter.
468   *
469   * <p>Note: Any configuration option configured on this splitter, such as {@link #trimResults},
470   * does not change the behavior of the {@code keyValueSplitter}.
471   *
472   * <p>Example:
473   *
474   * <pre>{@code
475   * String toSplit = " x -> y, z-> a ";
476   * Splitter outerSplitter = Splitter.on(',').trimResults();
477   * MapSplitter mapSplitter = outerSplitter.withKeyValueSeparator(Splitter.on("->"));
478   * Map<String, String> result = mapSplitter.split(toSplit);
479   * assertThat(result).isEqualTo(ImmutableMap.of("x ", " y", "z", " a"));
480   * }</pre>
481   *
482   * @since 10.0
483   */
484  @Beta
485  public MapSplitter withKeyValueSeparator(Splitter keyValueSplitter) {
486    return new MapSplitter(this, keyValueSplitter);
487  }
488
489  /**
490   * An object that splits strings into maps as {@code Splitter} splits iterables and lists. Like
491   * {@code Splitter}, it is thread-safe and immutable. The common way to build instances is by
492   * providing an additional {@linkplain Splitter#withKeyValueSeparator key-value separator} to
493   * {@link Splitter}.
494   *
495   * @since 10.0
496   */
497  @Beta
498  public static final class MapSplitter {
499    private static final String INVALID_ENTRY_MESSAGE = "Chunk [%s] is not a valid entry";
500    private final Splitter outerSplitter;
501    private final Splitter entrySplitter;
502
503    private MapSplitter(Splitter outerSplitter, Splitter entrySplitter) {
504      this.outerSplitter = outerSplitter; // only "this" is passed
505      this.entrySplitter = checkNotNull(entrySplitter);
506    }
507
508    /**
509     * Splits {@code sequence} into substrings, splits each substring into an entry, and returns an
510     * unmodifiable map with each of the entries. For example, {@code
511     * Splitter.on(';').trimResults().withKeyValueSeparator("=>").split("a=>b ; c=>b")} will return
512     * a mapping from {@code "a"} to {@code "b"} and {@code "c"} to {@code "b"}.
513     *
514     * <p>The returned map preserves the order of the entries from {@code sequence}.
515     *
516     * @throws IllegalArgumentException if the specified sequence does not split into valid map
517     *     entries, or if there are duplicate keys
518     */
519    public Map<String, String> split(CharSequence sequence) {
520      Map<String, String> map = new LinkedHashMap<>();
521      for (String entry : outerSplitter.split(sequence)) {
522        Iterator<String> entryFields = entrySplitter.splittingIterator(entry);
523
524        checkArgument(entryFields.hasNext(), INVALID_ENTRY_MESSAGE, entry);
525        String key = entryFields.next();
526        checkArgument(!map.containsKey(key), "Duplicate key [%s] found.", key);
527
528        checkArgument(entryFields.hasNext(), INVALID_ENTRY_MESSAGE, entry);
529        String value = entryFields.next();
530        map.put(key, value);
531
532        checkArgument(!entryFields.hasNext(), INVALID_ENTRY_MESSAGE, entry);
533      }
534      return Collections.unmodifiableMap(map);
535    }
536  }
537
538  private interface Strategy {
539    Iterator<String> iterator(Splitter splitter, CharSequence toSplit);
540  }
541
542  private abstract static class SplittingIterator extends AbstractIterator<String> {
543    final CharSequence toSplit;
544    final CharMatcher trimmer;
545    final boolean omitEmptyStrings;
546
547    /**
548     * Returns the first index in {@code toSplit} at or after {@code start} that contains the
549     * separator.
550     */
551    abstract int separatorStart(int start);
552
553    /**
554     * Returns the first index in {@code toSplit} after {@code separatorPosition} that does not
555     * contain a separator. This method is only invoked after a call to {@code separatorStart}.
556     */
557    abstract int separatorEnd(int separatorPosition);
558
559    int offset = 0;
560    int limit;
561
562    protected SplittingIterator(Splitter splitter, CharSequence toSplit) {
563      this.trimmer = splitter.trimmer;
564      this.omitEmptyStrings = splitter.omitEmptyStrings;
565      this.limit = splitter.limit;
566      this.toSplit = toSplit;
567    }
568
569    @CheckForNull
570    @Override
571    protected String computeNext() {
572      /*
573       * The returned string will be from the end of the last match to the beginning of the next
574       * one. nextStart is the start position of the returned substring, while offset is the place
575       * to start looking for a separator.
576       */
577      int nextStart = offset;
578      while (offset != -1) {
579        int start = nextStart;
580        int end;
581
582        int separatorPosition = separatorStart(offset);
583        if (separatorPosition == -1) {
584          end = toSplit.length();
585          offset = -1;
586        } else {
587          end = separatorPosition;
588          offset = separatorEnd(separatorPosition);
589        }
590        if (offset == nextStart) {
591          /*
592           * This occurs when some pattern has an empty match, even if it doesn't match the empty
593           * string -- for example, if it requires lookahead or the like. The offset must be
594           * increased to look for separators beyond this point, without changing the start position
595           * of the next returned substring -- so nextStart stays the same.
596           */
597          offset++;
598          if (offset > toSplit.length()) {
599            offset = -1;
600          }
601          continue;
602        }
603
604        while (start < end && trimmer.matches(toSplit.charAt(start))) {
605          start++;
606        }
607        while (end > start && trimmer.matches(toSplit.charAt(end - 1))) {
608          end--;
609        }
610
611        if (omitEmptyStrings && start == end) {
612          // Don't include the (unused) separator in next split string.
613          nextStart = offset;
614          continue;
615        }
616
617        if (limit == 1) {
618          // The limit has been reached, return the rest of the string as the
619          // final item. This is tested after empty string removal so that
620          // empty strings do not count towards the limit.
621          end = toSplit.length();
622          offset = -1;
623          // Since we may have changed the end, we need to trim it again.
624          while (end > start && trimmer.matches(toSplit.charAt(end - 1))) {
625            end--;
626          }
627        } else {
628          limit--;
629        }
630
631        return toSplit.subSequence(start, end).toString();
632      }
633      return endOfData();
634    }
635  }
636}