001/*
002 * Copyright (C) 2009 The Guava Authors
003 *
004 * Licensed under the Apache License, Version 2.0 (the "License");
005 * you may not use this file except in compliance with the License.
006 * You may obtain a copy of the License at
007 *
008 * http://www.apache.org/licenses/LICENSE-2.0
009 *
010 * Unless required by applicable law or agreed to in writing, software
011 * distributed under the License is distributed on an "AS IS" BASIS,
012 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
013 * See the License for the specific language governing permissions and
014 * limitations under the License.
015 */
016
017package com.google.common.base;
018
019import static com.google.common.base.Preconditions.checkArgument;
020import static com.google.common.base.Preconditions.checkNotNull;
021
022import com.google.common.annotations.Beta;
023import com.google.common.annotations.GwtCompatible;
024import com.google.common.annotations.GwtIncompatible;
025
026import java.util.ArrayList;
027import java.util.Collections;
028import java.util.Iterator;
029import java.util.LinkedHashMap;
030import java.util.List;
031import java.util.Map;
032import java.util.regex.Matcher;
033import java.util.regex.Pattern;
034
035import javax.annotation.CheckReturnValue;
036
037/**
038 * Extracts non-overlapping substrings from an input string, typically by
039 * recognizing appearances of a <i>separator</i> sequence. This separator can be
040 * specified as a single {@linkplain #on(char) character}, fixed {@linkplain
041 * #on(String) string}, {@linkplain #onPattern regular expression} or {@link
042 * #on(CharMatcher) CharMatcher} instance. Or, instead of using a separator at
043 * all, a splitter can extract adjacent substrings of a given {@linkplain
044 * #fixedLength fixed length}.
045 *
046 * <p>For example, this expression: <pre>   {@code
047 *
048 *   Splitter.on(',').split("foo,bar,qux")}</pre>
049 *
050 * ... produces an {@code Iterable} containing {@code "foo"}, {@code "bar"} and
051 * {@code "qux"}, in that order.
052 *
053 * <p>By default, {@code Splitter}'s behavior is simplistic and unassuming. The
054 * following expression: <pre>   {@code
055 *
056 *   Splitter.on(',').split(" foo,,,  bar ,")}</pre>
057 *
058 * ... yields the substrings {@code [" foo", "", "", "  bar ", ""]}. If this
059 * is not the desired behavior, use configuration methods to obtain a <i>new</i>
060 * splitter instance with modified behavior: <pre>   {@code
061 *
062 *   private static final Splitter MY_SPLITTER = Splitter.on(',')
063 *       .trimResults()
064 *       .omitEmptyStrings();}</pre>
065 *
066 * <p>Now {@code MY_SPLITTER.split("foo,,,  bar ,")} returns just {@code ["foo",
067 * "bar"]}. Note that the order in which these configuration methods are called
068 * is never significant.
069 *
070 * <p><b>Warning:</b> Splitter instances are immutable. Invoking a configuration
071 * method has no effect on the receiving instance; you must store and use the
072 * new splitter instance it returns instead. <pre>   {@code
073 *
074 *   // Do NOT do this
075 *   Splitter splitter = Splitter.on('/');
076 *   splitter.trimResults(); // does nothing!
077 *   return splitter.split("wrong / wrong / wrong");}</pre>
078 *
079 * <p>For separator-based splitters that do not use {@code omitEmptyStrings}, an
080 * input string containing {@code n} occurrences of the separator naturally
081 * yields an iterable of size {@code n + 1}. So if the separator does not occur
082 * anywhere in the input, a single substring is returned containing the entire
083 * 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
087 * store as {@code static final} constants.
088 *
089 * <p>The {@link Joiner} class provides the inverse operation to splitting, but
090 * note that a round-trip between the two should be assumed to be lossy.
091 *
092 * <p>See the Guava User Guide article on <a href=
093 * "http://code.google.com/p/guava-libraries/wiki/StringsExplained#Splitter">
094 * {@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)
103public final class Splitter {
104  private final CharMatcher trimmer;
105  private final boolean omitEmptyStrings;
106  private final Strategy strategy;
107  private final int limit;
108
109  private Splitter(Strategy strategy) {
110    this(strategy, false, CharMatcher.NONE, Integer.MAX_VALUE);
111  }
112
113  private Splitter(Strategy strategy, boolean omitEmptyStrings,
114      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
123   * example, {@code Splitter.on(',').split("foo,,bar")} returns an iterable
124   * containing {@code ["foo", "", "bar"]}.
125   *
126   * @param separator the character to recognize as a separator
127   * @return a splitter, with default settings, that recognizes that separator
128   */
129  public static Splitter on(char separator) {
130    return on(CharMatcher.is(separator));
131  }
132
133  /**
134   * Returns a splitter that considers any single character matched by the
135   * given {@code CharMatcher} to be a separator. For example, {@code
136   * Splitter.on(CharMatcher.anyOf(";,")).split("foo,;bar,quux")} returns an
137   * iterable containing {@code ["foo", "", "bar", "quux"]}.
138   *
139   * @param separatorMatcher a {@link CharMatcher} that determines whether a
140   *     character is a separator
141   * @return a splitter, with default settings, that uses this matcher
142   */
143  public static Splitter on(final CharMatcher separatorMatcher) {
144    checkNotNull(separatorMatcher);
145
146    return new Splitter(new Strategy() {
147      @Override public SplittingIterator iterator(
148          Splitter splitter, final CharSequence toSplit) {
149        return new SplittingIterator(splitter, toSplit) {
150          @Override int separatorStart(int start) {
151            return separatorMatcher.indexIn(toSplit, start);
152          }
153
154          @Override 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
164   * example, {@code Splitter.on(", ").split("foo, bar,baz")} returns an
165   * iterable containing {@code ["foo", "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,
172        "The separator may not be the empty string.");
173
174    return new Splitter(new Strategy() {
175      @Override public SplittingIterator iterator(
176          Splitter splitter, CharSequence toSplit) {
177        return new SplittingIterator(splitter, toSplit) {
178          @Override public int separatorStart(int start) {
179            int separatorLength = separator.length();
180
181            positions:
182            for (int p = start, last = toSplit.length() - separatorLength;
183                p <= last; p++) {
184              for (int i = 0; i < separatorLength; i++) {
185                if (toSplit.charAt(i + p) != separator.charAt(i)) {
186                  continue positions;
187                }
188              }
189              return p;
190            }
191            return -1;
192          }
193
194          @Override public int separatorEnd(int separatorPosition) {
195            return separatorPosition + separator.length();
196          }
197        };
198      }
199    });
200  }
201
202  /**
203   * Returns a splitter that considers any subsequence matching {@code
204   * pattern} to be a separator. For example, {@code
205   * Splitter.on(Pattern.compile("\r?\n")).split(entireFile)} splits a string
206   * into lines whether it uses DOS-style or UNIX-style line terminators.
207   *
208   * @param separatorPattern the pattern that determines whether a subsequence
209   *     is a separator. This pattern may not match the empty string.
210   * @return a splitter, with default settings, that uses this pattern
211   * @throws IllegalArgumentException if {@code separatorPattern} matches the
212   *     empty string
213   */
214  @GwtIncompatible("java.util.regex")
215  public static Splitter on(final Pattern separatorPattern) {
216    checkNotNull(separatorPattern);
217    checkArgument(!separatorPattern.matcher("").matches(),
218        "The pattern may not match the empty string: %s", separatorPattern);
219
220    return new Splitter(new Strategy() {
221      @Override public SplittingIterator iterator(
222          final Splitter splitter, CharSequence toSplit) {
223        final Matcher matcher = separatorPattern.matcher(toSplit);
224        return new SplittingIterator(splitter, toSplit) {
225          @Override public int separatorStart(int start) {
226            return matcher.find(start) ? matcher.start() : -1;
227          }
228
229          @Override public int separatorEnd(int separatorPosition) {
230            return matcher.end();
231          }
232        };
233      }
234    });
235  }
236
237  /**
238   * Returns a splitter that considers any subsequence matching a given
239   * pattern (regular expression) to be a separator. For example, {@code
240   * Splitter.onPattern("\r?\n").split(entireFile)} splits a string into lines
241   * whether it uses DOS-style or UNIX-style line terminators. This is
242   * equivalent to {@code Splitter.on(Pattern.compile(pattern))}.
243   *
244   * @param separatorPattern the pattern that determines whether a subsequence
245   *     is a separator. This pattern may not match the empty string.
246   * @return a splitter, with default settings, that uses this pattern
247   * @throws java.util.regex.PatternSyntaxException if {@code separatorPattern}
248   *     is a malformed expression
249   * @throws IllegalArgumentException if {@code separatorPattern} matches the
250   *     empty string
251   */
252  @GwtIncompatible("java.util.regex")
253  public static Splitter onPattern(String separatorPattern) {
254    return on(Pattern.compile(separatorPattern));
255  }
256
257  /**
258   * Returns a splitter that divides strings into pieces of the given length.
259   * For example, {@code Splitter.fixedLength(2).split("abcde")} returns an
260   * iterable containing {@code ["ab", "cd", "e"]}. The last piece can be
261   * smaller than {@code length} but will never be empty.
262   *
263   * <p><b>Exception:</b> for consistency with separator-based splitters, {@code
264   * split("")} does not yield an empty iterable, but an iterable containing
265   * {@code ""}. This is the only case in which {@code
266   * Iterables.size(split(input))} does not equal {@code
267   * IntMath.divide(input.length(), length, CEILING)}. To avoid this behavior,
268   * use {@code omitEmptyStrings}.
269   *
270   * @param length the desired length of pieces after splitting, a positive
271   *     integer
272   * @return a splitter, with default settings, that can split into fixed sized
273   *     pieces
274   * @throws IllegalArgumentException if {@code length} is zero or negative
275   */
276  public static Splitter fixedLength(final int length) {
277    checkArgument(length > 0, "The length may not be less than 1");
278
279    return new Splitter(new Strategy() {
280      @Override public SplittingIterator iterator(
281          final Splitter splitter, CharSequence toSplit) {
282        return new SplittingIterator(splitter, toSplit) {
283          @Override public int separatorStart(int start) {
284            int nextChunkStart = start + length;
285            return (nextChunkStart < toSplit.length() ? nextChunkStart : -1);
286          }
287
288          @Override public int separatorEnd(int separatorPosition) {
289            return separatorPosition;
290          }
291        };
292      }
293    });
294  }
295
296  /**
297   * Returns a splitter that behaves equivalently to {@code this} splitter, but
298   * automatically omits empty strings from the results. For example, {@code
299   * Splitter.on(',').omitEmptyStrings().split(",a,,,b,c,,")} returns an
300   * iterable containing only {@code ["a", "b", "c"]}.
301   *
302   * <p>If either {@code trimResults} option is also specified when creating a
303   * splitter, that splitter always trims results first before checking for
304   * emptiness. So, for example, {@code
305   * Splitter.on(':').omitEmptyStrings().trimResults().split(": : : ")} returns
306   * an empty iterable.
307   *
308   * <p>Note that it is ordinarily not possible for {@link #split(CharSequence)}
309   * to return an empty iterable, but when using this option, it can (if the
310   * input sequence consists of nothing but separators).
311   *
312   * @return a splitter with the desired configuration
313   */
314  @CheckReturnValue
315  public Splitter omitEmptyStrings() {
316    return new Splitter(strategy, true, trimmer, limit);
317  }
318
319  /**
320   * Returns a splitter that behaves equivalently to {@code this} splitter but
321   * stops splitting after it reaches the limit.
322   * The limit defines the maximum number of items returned by the iterator.
323   *
324   * <p>For example,
325   * {@code Splitter.on(',').limit(3).split("a,b,c,d")} returns an iterable
326   * containing {@code ["a", "b", "c,d"]}.  When omitting empty strings, the
327   * omitted strings do no count.  Hence,
328   * {@code Splitter.on(',').limit(3).omitEmptyStrings().split("a,,,b,,,c,d")}
329   * returns an iterable containing {@code ["a", "b", "c,d"}.
330   * When trim is requested, all entries, including the last are trimmed.  Hence
331   * {@code Splitter.on(',').limit(3).trimResults().split(" a , b , c , d ")}
332   * results in @{code ["a", "b", "c , d"]}.
333   *
334   * @param limit the maximum number of items returns
335   * @return a splitter with the desired configuration
336   * @since 9.0
337   */
338  @CheckReturnValue
339  public Splitter limit(int limit) {
340    checkArgument(limit > 0, "must be greater than zero: %s", limit);
341    return new Splitter(strategy, omitEmptyStrings, trimmer, limit);
342  }
343
344  /**
345   * Returns a splitter that behaves equivalently to {@code this} splitter, but
346   * automatically removes leading and trailing {@linkplain
347   * CharMatcher#WHITESPACE whitespace} from each returned substring; equivalent
348   * to {@code trimResults(CharMatcher.WHITESPACE)}. For example, {@code
349   * Splitter.on(',').trimResults().split(" a, b ,c ")} returns an iterable
350   * containing {@code ["a", "b", "c"]}.
351   *
352   * @return a splitter with the desired configuration
353   */
354  @CheckReturnValue
355  public Splitter trimResults() {
356    return trimResults(CharMatcher.WHITESPACE);
357  }
358
359  /**
360   * Returns a splitter that behaves equivalently to {@code this} splitter, but
361   * removes all leading or trailing characters matching the given {@code
362   * CharMatcher} from each returned substring. For example, {@code
363   * 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
367   *     should be removed from 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  @CheckReturnValue
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
379   * through an {@link Iterator}, which may be lazily evaluated. If you want
380   * an eagerly computed {@link List}, use {@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 public Iterator<String> iterator() {
390        return splittingIterator(sequence);
391      }
392      @Override 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
407   * an immutable list. If you want an {@link Iterable} which may be lazily
408   * evaluated, use {@link #split(CharSequence)}.
409   *
410   * @param sequence the sequence of characters to split
411   * @return an immutable list of the segments split from the parameter
412   * @since 15.0
413   */
414  @Beta
415  public List<String> splitToList(CharSequence sequence) {
416    checkNotNull(sequence);
417
418    Iterator<String> iterator = splittingIterator(sequence);
419    List<String> result = new ArrayList<String>();
420
421    while (iterator.hasNext()) {
422      result.add(iterator.next());
423    }
424
425    return Collections.unmodifiableList(result);
426  }
427
428  /**
429   * Returns a {@code MapSplitter} which splits entries based on this splitter,
430   * and splits entries into keys and values using the specified separator.
431   *
432   * @since 10.0
433   */
434  @CheckReturnValue
435  @Beta
436  public MapSplitter withKeyValueSeparator(String separator) {
437    return withKeyValueSeparator(on(separator));
438  }
439
440  /**
441   * Returns a {@code MapSplitter} which splits entries based on this splitter,
442   * and splits entries into keys and values using the specified separator.
443   *
444   * @since 14.0
445   */
446  @CheckReturnValue
447  @Beta
448  public MapSplitter withKeyValueSeparator(char separator) {
449    return withKeyValueSeparator(on(separator));
450  }
451
452  /**
453   * Returns a {@code MapSplitter} which splits entries based on this splitter,
454   * and splits entries into keys and values using the specified key-value
455   * splitter.
456   *
457   * @since 10.0
458   */
459  @CheckReturnValue
460  @Beta
461  public MapSplitter withKeyValueSeparator(Splitter keyValueSplitter) {
462    return new MapSplitter(this, keyValueSplitter);
463  }
464
465  /**
466   * An object that splits strings into maps as {@code Splitter} splits
467   * iterables and lists. Like {@code Splitter}, it is thread-safe and
468   * immutable.
469   *
470   * @since 10.0
471   */
472  @Beta
473  public static final class MapSplitter {
474    private static final String INVALID_ENTRY_MESSAGE =
475        "Chunk [%s] is not a valid entry";
476    private final Splitter outerSplitter;
477    private final Splitter entrySplitter;
478
479    private MapSplitter(Splitter outerSplitter, Splitter entrySplitter) {
480      this.outerSplitter = outerSplitter; // only "this" is passed
481      this.entrySplitter = checkNotNull(entrySplitter);
482    }
483
484    /**
485     * Splits {@code sequence} into substrings, splits each substring into
486     * an entry, and returns an unmodifiable map with each of the entries. For
487     * example, <code>
488     * Splitter.on(';').trimResults().withKeyValueSeparator("=>")
489     * .split("a=>b ; c=>b")
490     * </code> will return a mapping from {@code "a"} to {@code "b"} and
491     * {@code "c"} to {@code b}.
492     *
493     * <p>The returned map preserves the order of the entries from
494     * {@code sequence}.
495     *
496     * @throws IllegalArgumentException if the specified sequence does not split
497     *         into valid map entries, or if there are duplicate keys
498     */
499    public Map<String, String> split(CharSequence sequence) {
500      Map<String, String> map = new LinkedHashMap<String, String>();
501      for (String entry : outerSplitter.split(sequence)) {
502        Iterator<String> entryFields = entrySplitter.splittingIterator(entry);
503
504        checkArgument(entryFields.hasNext(), INVALID_ENTRY_MESSAGE, entry);
505        String key = entryFields.next();
506        checkArgument(!map.containsKey(key), "Duplicate key [%s] found.", key);
507
508        checkArgument(entryFields.hasNext(), INVALID_ENTRY_MESSAGE, entry);
509        String value = entryFields.next();
510        map.put(key, value);
511
512        checkArgument(!entryFields.hasNext(), INVALID_ENTRY_MESSAGE, entry);
513      }
514      return Collections.unmodifiableMap(map);
515    }
516  }
517
518  private interface Strategy {
519    Iterator<String> iterator(Splitter splitter, CharSequence toSplit);
520  }
521
522  private abstract static class SplittingIterator extends AbstractIterator<String> {
523    final CharSequence toSplit;
524    final CharMatcher trimmer;
525    final boolean omitEmptyStrings;
526
527    /**
528     * Returns the first index in {@code toSplit} at or after {@code start}
529     * that contains the separator.
530     */
531    abstract int separatorStart(int start);
532
533    /**
534     * Returns the first index in {@code toSplit} after {@code
535     * separatorPosition} that does not contain a separator. This method is only
536     * invoked after a call to {@code separatorStart}.
537     */
538    abstract int separatorEnd(int separatorPosition);
539
540    int offset = 0;
541    int limit;
542
543    protected SplittingIterator(Splitter splitter, CharSequence toSplit) {
544      this.trimmer = splitter.trimmer;
545      this.omitEmptyStrings = splitter.omitEmptyStrings;
546      this.limit = splitter.limit;
547      this.toSplit = toSplit;
548    }
549
550    @Override protected String computeNext() {
551      /*
552       * The returned string will be from the end of the last match to the
553       * beginning of the next one. nextStart is the start position of the
554       * returned substring, while offset is the place to start looking for a
555       * separator.
556       */
557      int nextStart = offset;
558      while (offset != -1) {
559        int start = nextStart;
560        int end;
561
562        int separatorPosition = separatorStart(offset);
563        if (separatorPosition == -1) {
564          end = toSplit.length();
565          offset = -1;
566        } else {
567          end = separatorPosition;
568          offset = separatorEnd(separatorPosition);
569        }
570        if (offset == nextStart) {
571          /*
572           * This occurs when some pattern has an empty match, even if it
573           * doesn't match the empty string -- for example, if it requires
574           * lookahead or the like. The offset must be increased to look for
575           * separators beyond this point, without changing the start position
576           * of the next returned substring -- so nextStart stays the same.
577           */
578          offset++;
579          if (offset >= toSplit.length()) {
580            offset = -1;
581          }
582          continue;
583        }
584
585        while (start < end && trimmer.matches(toSplit.charAt(start))) {
586          start++;
587        }
588        while (end > start && trimmer.matches(toSplit.charAt(end - 1))) {
589          end--;
590        }
591
592        if (omitEmptyStrings && start == end) {
593          // Don't include the (unused) separator in next split string.
594          nextStart = offset;
595          continue;
596        }
597
598        if (limit == 1) {
599          // The limit has been reached, return the rest of the string as the
600          // final item.  This is tested after empty string removal so that
601          // empty strings do not count towards the limit.
602          end = toSplit.length();
603          offset = -1;
604          // Since we may have changed the end, we need to trim it again.
605          while (end > start && trimmer.matches(toSplit.charAt(end - 1))) {
606            end--;
607          }
608        } else {
609          limit--;
610        }
611
612        return toSplit.subSequence(start, end).toString();
613      }
614      return endOfData();
615    }
616  }
617}