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