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.GwtCompatible;
021import com.google.common.annotations.GwtIncompatible;
022import com.google.common.annotations.J2ktIncompatible;
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 javax.annotation.CheckForNull;
031
032/**
033 * Extracts non-overlapping substrings from an input string, typically by recognizing appearances of
034 * a <i>separator</i> sequence. This separator can be specified as a single {@linkplain #on(char)
035 * character}, fixed {@linkplain #on(String) string}, {@linkplain #onPattern regular expression} or
036 * {@link #on(CharMatcher) CharMatcher} instance. Or, instead of using a separator at all, a
037 * splitter can extract adjacent substrings of a given {@linkplain #fixedLength fixed length}.
038 *
039 * <p>For example, this expression:
040 *
041 * <pre>{@code
042 * Splitter.on(',').split("foo,bar,qux")
043 * }</pre>
044 *
045 * ... produces an {@code Iterable} containing {@code "foo"}, {@code "bar"} and {@code "qux"}, in
046 * that order.
047 *
048 * <p>By default, {@code Splitter}'s behavior is simplistic and unassuming. The following
049 * expression:
050 *
051 * <pre>{@code
052 * Splitter.on(',').split(" foo,,,  bar ,")
053 * }</pre>
054 *
055 * ... yields the substrings {@code [" foo", "", "", " bar ", ""]}. If this is not the desired
056 * behavior, use configuration methods to obtain a <i>new</i> splitter instance with modified
057 * behavior:
058 *
059 * <pre>{@code
060 * private static final Splitter MY_SPLITTER = Splitter.on(',')
061 *     .trimResults()
062 *     .omitEmptyStrings();
063 * }</pre>
064 *
065 * <p>Now {@code MY_SPLITTER.split("foo,,, bar ,")} returns just {@code ["foo", "bar"]}. Note that
066 * the order in which these configuration methods are called is never significant.
067 *
068 * <p><b>Warning:</b> Splitter instances are immutable. Invoking a configuration method has no
069 * effect on the receiving instance; you must store and use the new splitter instance it returns
070 * instead.
071 *
072 * <pre>{@code
073 * // Do NOT do this
074 * Splitter splitter = Splitter.on('/');
075 * splitter.trimResults(); // does nothing!
076 * return splitter.split("wrong / wrong / wrong");
077 * }</pre>
078 *
079 * <p>For separator-based splitters that do not use {@code omitEmptyStrings}, an input string
080 * containing {@code n} occurrences of the separator naturally yields an iterable of size {@code n +
081 * 1}. So if the separator does not occur anywhere in the input, a single substring is returned
082 * containing the entire input. Consequently, all splitters split the empty string to {@code [""]}
083 * (note: even fixed-length splitters).
084 *
085 * <p>Splitter instances are thread-safe immutable, and are therefore safe to store as {@code static
086 * final} constants.
087 *
088 * <p>The {@link Joiner} class provides the inverse operation to splitting, but note that a
089 * round-trip between the two should be assumed to be lossy.
090 *
091 * <p>See the Guava User Guide article on <a
092 * href="https://github.com/google/guava/wiki/StringsExplained#splitter">{@code Splitter}</a>.
093 *
094 * @author Julien Silland
095 * @author Jesse Wilson
096 * @author Kevin Bourrillion
097 * @author Louis Wasserman
098 * @since 1.0
099 */
100@GwtCompatible(emulated = true)
101@ElementTypesAreNonnullByDefault
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  @J2ktIncompatible
216  @GwtIncompatible // java.util.regex
217  public static Splitter on(Pattern separatorPattern) {
218    return onPatternInternal(new JdkPattern(separatorPattern));
219  }
220
221  /** Internal utility; see {@link #on(Pattern)} instead. */
222  static Splitter onPatternInternal(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  @J2ktIncompatible
261  @GwtIncompatible // java.util.regex
262  public static Splitter onPattern(String separatorPattern) {
263    return onPatternInternal(Platform.compilePattern(separatorPattern));
264  }
265
266  /**
267   * Returns a splitter that divides strings into pieces of the given length. For example, {@code
268   * Splitter.fixedLength(2).split("abcde")} returns an iterable containing {@code ["ab", "cd",
269   * "e"]}. The last piece can be smaller than {@code length} but will never be empty.
270   *
271   * <p><b>Note:</b> if {@link #fixedLength} is used in conjunction with {@link #limit}, the final
272   * split piece <i>may be longer than the specified fixed length</i>. This is because the splitter
273   * will <i>stop splitting when the limit is reached</i>, and just return the final piece as-is.
274   *
275   * <p><b>Exception:</b> for consistency with separator-based splitters, {@code split("")} does not
276   * yield an empty iterable, but an iterable containing {@code ""}. This is the only case in which
277   * {@code Iterables.size(split(input))} does not equal {@code IntMath.divide(input.length(),
278   * length, CEILING)}. To avoid this behavior, use {@code omitEmptyStrings}.
279   *
280   * @param length the desired length of pieces after splitting, a positive integer
281   * @return a splitter, with default settings, that can split into fixed sized pieces
282   * @throws IllegalArgumentException if {@code length} is zero or negative
283   */
284  public static Splitter fixedLength(final int length) {
285    checkArgument(length > 0, "The length may not be less than 1");
286
287    return new Splitter(
288        new Strategy() {
289          @Override
290          public SplittingIterator iterator(final Splitter splitter, CharSequence toSplit) {
291            return new SplittingIterator(splitter, toSplit) {
292              @Override
293              public int separatorStart(int start) {
294                int nextChunkStart = start + length;
295                return (nextChunkStart < toSplit.length() ? nextChunkStart : -1);
296              }
297
298              @Override
299              public int separatorEnd(int separatorPosition) {
300                return separatorPosition;
301              }
302            };
303          }
304        });
305  }
306
307  /**
308   * Returns a splitter that behaves equivalently to {@code this} splitter, but automatically omits
309   * empty strings from the results. For example, {@code
310   * Splitter.on(',').omitEmptyStrings().split(",a,,,b,c,,")} returns an iterable containing only
311   * {@code ["a", "b", "c"]}.
312   *
313   * <p>If either {@code trimResults} option is also specified when creating a splitter, that
314   * splitter always trims results first before checking for emptiness. So, for example, {@code
315   * Splitter.on(':').omitEmptyStrings().trimResults().split(": : : ")} returns an empty iterable.
316   *
317   * <p>Note that it is ordinarily not possible for {@link #split(CharSequence)} to return an empty
318   * iterable, but when using this option, it can (if the input sequence consists of nothing but
319   * separators).
320   *
321   * @return a splitter with the desired configuration
322   */
323  public Splitter omitEmptyStrings() {
324    return new Splitter(strategy, true, trimmer, limit);
325  }
326
327  /**
328   * Returns a splitter that behaves equivalently to {@code this} splitter but stops splitting after
329   * it reaches the limit. The limit defines the maximum number of items returned by the iterator,
330   * or the maximum size of the list returned by {@link #splitToList}.
331   *
332   * <p>For example, {@code Splitter.on(',').limit(3).split("a,b,c,d")} returns an iterable
333   * containing {@code ["a", "b", "c,d"]}. When omitting empty strings, the omitted strings do not
334   * count. Hence, {@code Splitter.on(',').limit(3).omitEmptyStrings().split("a,,,b,,,c,d")} returns
335   * an iterable containing {@code ["a", "b", "c,d"}. When trim is requested, all entries are
336   * trimmed, including the last. Hence {@code Splitter.on(',').limit(3).trimResults().split(" a , b
337   * , c , d ")} results in {@code ["a", "b", "c , d"]}.
338   *
339   * @param maxItems the maximum number of items returned
340   * @return a splitter with the desired configuration
341   * @since 9.0
342   */
343  public Splitter limit(int maxItems) {
344    checkArgument(maxItems > 0, "must be greater than zero: %s", maxItems);
345    return new Splitter(strategy, omitEmptyStrings, trimmer, maxItems);
346  }
347
348  /**
349   * Returns a splitter that behaves equivalently to {@code this} splitter, but automatically
350   * removes leading and trailing {@linkplain CharMatcher#whitespace whitespace} from each returned
351   * substring; equivalent to {@code trimResults(CharMatcher.whitespace())}. For example, {@code
352   * Splitter.on(',').trimResults().split(" a, b ,c ")} returns an iterable containing {@code ["a",
353   * "b", "c"]}.
354   *
355   * @return a splitter with the desired configuration
356   */
357  public Splitter trimResults() {
358    return trimResults(CharMatcher.whitespace());
359  }
360
361  /**
362   * Returns a splitter that behaves equivalently to {@code this} splitter, but removes all leading
363   * or trailing characters matching the given {@code CharMatcher} from each returned substring. For
364   * example, {@code Splitter.on(',').trimResults(CharMatcher.is('_')).split("_a ,_b_ ,c__")}
365   * returns an iterable containing {@code ["a ", "b_ ", "c"]}.
366   *
367   * @param trimmer a {@link CharMatcher} that determines whether a character should be removed from
368   *     the beginning/end of a subsequence
369   * @return a splitter with the desired configuration
370   */
371  // TODO(kevinb): throw if a trimmer was already specified!
372  public Splitter trimResults(CharMatcher trimmer) {
373    checkNotNull(trimmer);
374    return new Splitter(strategy, omitEmptyStrings, trimmer, limit);
375  }
376
377  /**
378   * Splits {@code sequence} into string components and makes them available through an {@link
379   * Iterator}, which may be lazily evaluated. If you want an eagerly computed {@link List}, use
380   * {@link #splitToList(CharSequence)}.
381   *
382   * @param sequence the sequence of characters to split
383   * @return an iteration over the segments split from the parameter
384   */
385  public Iterable<String> split(final CharSequence sequence) {
386    checkNotNull(sequence);
387
388    return new Iterable<String>() {
389      @Override
390      public Iterator<String> iterator() {
391        return splittingIterator(sequence);
392      }
393
394      @Override
395      public String toString() {
396        return Joiner.on(", ")
397            .appendTo(new StringBuilder().append('['), this)
398            .append(']')
399            .toString();
400      }
401    };
402  }
403
404  private Iterator<String> splittingIterator(CharSequence sequence) {
405    return strategy.iterator(this, sequence);
406  }
407
408  /**
409   * Splits {@code sequence} into string components and returns them as an immutable list. If you
410   * want an {@link Iterable} which may be lazily evaluated, use {@link #split(CharSequence)}.
411   *
412   * @param sequence the sequence of characters to split
413   * @return an immutable list of the segments split from the parameter
414   * @since 15.0
415   */
416  public List<String> splitToList(CharSequence sequence) {
417    checkNotNull(sequence);
418
419    Iterator<String> iterator = splittingIterator(sequence);
420    List<String> result = new ArrayList<>();
421
422    while (iterator.hasNext()) {
423      result.add(iterator.next());
424    }
425
426    return Collections.unmodifiableList(result);
427  }
428
429  /**
430   * Returns a {@code MapSplitter} which splits entries based on this splitter, and splits entries
431   * into keys and values using the specified separator.
432   *
433   * @since 10.0
434   */
435  public MapSplitter withKeyValueSeparator(String separator) {
436    return withKeyValueSeparator(on(separator));
437  }
438
439  /**
440   * Returns a {@code MapSplitter} which splits entries based on this splitter, and splits entries
441   * into keys and values using the specified separator.
442   *
443   * @since 14.0
444   */
445  public MapSplitter withKeyValueSeparator(char separator) {
446    return withKeyValueSeparator(on(separator));
447  }
448
449  /**
450   * Returns a {@code MapSplitter} which splits entries based on this splitter, and splits entries
451   * into keys and values using the specified key-value splitter.
452   *
453   * <p>Note: Any configuration option configured on this splitter, such as {@link #trimResults},
454   * does not change the behavior of the {@code keyValueSplitter}.
455   *
456   * <p>Example:
457   *
458   * <pre>{@code
459   * String toSplit = " x -> y, z-> a ";
460   * Splitter outerSplitter = Splitter.on(',').trimResults();
461   * MapSplitter mapSplitter = outerSplitter.withKeyValueSeparator(Splitter.on("->"));
462   * Map<String, String> result = mapSplitter.split(toSplit);
463   * assertThat(result).isEqualTo(ImmutableMap.of("x ", " y", "z", " a"));
464   * }</pre>
465   *
466   * @since 10.0
467   */
468  public MapSplitter withKeyValueSeparator(Splitter keyValueSplitter) {
469    return new MapSplitter(this, keyValueSplitter);
470  }
471
472  /**
473   * An object that splits strings into maps as {@code Splitter} splits iterables and lists. Like
474   * {@code Splitter}, it is thread-safe and immutable. The common way to build instances is by
475   * providing an additional {@linkplain Splitter#withKeyValueSeparator key-value separator} to
476   * {@link Splitter}.
477   *
478   * @since 10.0
479   */
480  public static final class MapSplitter {
481    private static final String INVALID_ENTRY_MESSAGE = "Chunk [%s] is not a valid entry";
482    private final Splitter outerSplitter;
483    private final Splitter entrySplitter;
484
485    private MapSplitter(Splitter outerSplitter, Splitter entrySplitter) {
486      this.outerSplitter = outerSplitter; // only "this" is passed
487      this.entrySplitter = checkNotNull(entrySplitter);
488    }
489
490    /**
491     * Splits {@code sequence} into substrings, splits each substring into an entry, and returns an
492     * unmodifiable map with each of the entries. For example, {@code
493     * Splitter.on(';').trimResults().withKeyValueSeparator("=>").split("a=>b ; c=>b")} will return
494     * a mapping from {@code "a"} to {@code "b"} and {@code "c"} to {@code "b"}.
495     *
496     * <p>The returned map preserves the order of the entries from {@code sequence}.
497     *
498     * @throws IllegalArgumentException if the specified sequence does not split into valid map
499     *     entries, or if there are duplicate keys
500     */
501    public Map<String, String> split(CharSequence sequence) {
502      Map<String, String> map = new LinkedHashMap<>();
503      for (String entry : outerSplitter.split(sequence)) {
504        Iterator<String> entryFields = entrySplitter.splittingIterator(entry);
505
506        checkArgument(entryFields.hasNext(), INVALID_ENTRY_MESSAGE, entry);
507        String key = entryFields.next();
508        checkArgument(!map.containsKey(key), "Duplicate key [%s] found.", key);
509
510        checkArgument(entryFields.hasNext(), INVALID_ENTRY_MESSAGE, entry);
511        String value = entryFields.next();
512        map.put(key, value);
513
514        checkArgument(!entryFields.hasNext(), INVALID_ENTRY_MESSAGE, entry);
515      }
516      return Collections.unmodifiableMap(map);
517    }
518  }
519
520  private interface Strategy {
521    Iterator<String> iterator(Splitter splitter, CharSequence toSplit);
522  }
523
524  private abstract static class SplittingIterator extends AbstractIterator<String> {
525    final CharSequence toSplit;
526    final CharMatcher trimmer;
527    final boolean omitEmptyStrings;
528
529    /**
530     * Returns the first index in {@code toSplit} at or after {@code start} that contains the
531     * separator.
532     */
533    abstract int separatorStart(int start);
534
535    /**
536     * Returns the first index in {@code toSplit} after {@code separatorPosition} that does not
537     * contain a separator. This method is only invoked after a call to {@code separatorStart}.
538     */
539    abstract int separatorEnd(int separatorPosition);
540
541    int offset = 0;
542    int limit;
543
544    protected SplittingIterator(Splitter splitter, CharSequence toSplit) {
545      this.trimmer = splitter.trimmer;
546      this.omitEmptyStrings = splitter.omitEmptyStrings;
547      this.limit = splitter.limit;
548      this.toSplit = toSplit;
549    }
550
551    @CheckForNull
552    @Override
553    protected String computeNext() {
554      /*
555       * The returned string will be from the end of the last match to the beginning of the next
556       * one. nextStart is the start position of the returned substring, while offset is the place
557       * to start looking for a separator.
558       */
559      int nextStart = offset;
560      while (offset != -1) {
561        int start = nextStart;
562        int end;
563
564        int separatorPosition = separatorStart(offset);
565        if (separatorPosition == -1) {
566          end = toSplit.length();
567          offset = -1;
568        } else {
569          end = separatorPosition;
570          offset = separatorEnd(separatorPosition);
571        }
572        if (offset == nextStart) {
573          /*
574           * This occurs when some pattern has an empty match, even if it doesn't match the empty
575           * string -- for example, if it requires lookahead or the like. The offset must be
576           * increased to look for separators beyond this point, without changing the start position
577           * of the next returned substring -- so nextStart stays the same.
578           */
579          offset++;
580          if (offset > toSplit.length()) {
581            offset = -1;
582          }
583          continue;
584        }
585
586        while (start < end && trimmer.matches(toSplit.charAt(start))) {
587          start++;
588        }
589        while (end > start && trimmer.matches(toSplit.charAt(end - 1))) {
590          end--;
591        }
592
593        if (omitEmptyStrings && start == end) {
594          // Don't include the (unused) separator in next split string.
595          nextStart = offset;
596          continue;
597        }
598
599        if (limit == 1) {
600          // The limit has been reached, return the rest of the string as the
601          // final item. This is tested after empty string removal so that
602          // empty strings do not count towards the limit.
603          end = toSplit.length();
604          offset = -1;
605          // Since we may have changed the end, we need to trim it again.
606          while (end > start && trimmer.matches(toSplit.charAt(end - 1))) {
607            end--;
608          }
609        } else {
610          limit--;
611        }
612
613        return toSplit.subSequence(start, end).toString();
614      }
615      return endOfData();
616    }
617  }
618}