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    
017    package com.google.common.base;
018    
019    import static com.google.common.base.Preconditions.checkArgument;
020    import static com.google.common.base.Preconditions.checkNotNull;
021    import static com.google.common.base.Preconditions.checkState;
022    
023    import com.google.common.annotations.Beta;
024    import com.google.common.annotations.GwtCompatible;
025    import com.google.common.annotations.GwtIncompatible;
026    
027    import java.util.Iterator;
028    import java.util.NoSuchElementException;
029    import java.util.regex.Matcher;
030    import java.util.regex.Pattern;
031    
032    /**
033     * An object that divides strings (or other instances of {@code CharSequence})
034     * into substrings, by recognizing a <i>separator</i> (a.k.a. "delimiter")
035     * which can be expressed as a single character, literal string, regular
036     * expression, {@code CharMatcher}, or by using a fixed substring length. This
037     * class provides the complementary functionality to {@link Joiner}.
038     *
039     * <p>Here is the most basic example of {@code Splitter} usage: <pre>   {@code
040     *
041     *   Splitter.on(',').split("foo,bar")}</pre>
042     *
043     * This invocation returns an {@code Iterable<String>} containing {@code "foo"}
044     * and {@code "bar"}, in that order.
045     *
046     * <p>By default {@code Splitter}'s behavior is very simplistic: <pre>   {@code
047     *
048     *   Splitter.on(',').split("foo,,bar, quux")}</pre>
049     *
050     * This returns an iterable containing {@code ["foo", "", "bar", " quux"]}.
051     * Notice that the splitter does not assume that you want empty strings removed,
052     * or that you wish to trim whitespace. If you want features like these, simply
053     * ask for them: <pre> {@code
054     *
055     *   private static final Splitter MY_SPLITTER = Splitter.on(',')
056     *       .trimResults()
057     *       .omitEmptyStrings();}</pre>
058     *
059     * Now {@code MY_SPLITTER.split("foo, ,bar, quux,")} returns an iterable
060     * containing just {@code ["foo", "bar", "quux"]}. Note that the order in which
061     * the configuration methods are called is never significant; for instance,
062     * trimming is always applied first before checking for an empty result,
063     * regardless of the order in which the {@link #trimResults()} and
064     * {@link #omitEmptyStrings()} methods were invoked.
065     *
066     * <p><b>Warning: splitter instances are always immutable</b>; a configuration
067     * method such as {@code omitEmptyStrings} has no effect on the instance it
068     * is invoked on! You must store and use the new splitter instance returned by
069     * the method. This makes splitters thread-safe, and safe to store as {@code
070     * static final} constants (as illustrated above). <pre>   {@code
071     *
072     *   // Bad! Do not do this!
073     *   Splitter splitter = Splitter.on('/');
074     *   splitter.trimResults(); // does nothing!
075     *   return splitter.split("wrong / wrong / wrong");}</pre>
076     *
077     * The separator recognized by the splitter does not have to be a single
078     * literal character as in the examples above. See the methods {@link
079     * #on(String)}, {@link #on(Pattern)} and {@link #on(CharMatcher)} for examples
080     * of other ways to specify separators.
081     *
082     * <p><b>Note:</b> this class does not mimic any of the quirky behaviors of
083     * similar JDK methods; for instance, it does not silently discard trailing
084     * separators, as does {@link String#split(String)}, nor does it have a default
085     * behavior of using five particular whitespace characters as separators, like
086     * {@link java.util.StringTokenizer}.
087     *
088     * @author Julien Silland
089     * @author Jesse Wilson
090     * @author Kevin Bourrillion
091     * @since 1
092     */
093    @GwtCompatible(emulated = true)
094    public final class Splitter {
095      private final CharMatcher trimmer;
096      private final boolean omitEmptyStrings;
097      private final Strategy strategy;
098      private final int limit;
099    
100      private Splitter(Strategy strategy) {
101        this(strategy, false, CharMatcher.NONE, Integer.MAX_VALUE);
102      }
103    
104      private Splitter(Strategy strategy, boolean omitEmptyStrings,
105          CharMatcher trimmer, int limit) {
106        this.strategy = strategy;
107        this.omitEmptyStrings = omitEmptyStrings;
108        this.trimmer = trimmer;
109        this.limit = limit;
110      }
111    
112      /**
113       * Returns a splitter that uses the given single-character separator. For
114       * example, {@code Splitter.on(',').split("foo,,bar")} returns an iterable
115       * containing {@code ["foo", "", "bar"]}.
116       *
117       * @param separator the character to recognize as a separator
118       * @return a splitter, with default settings, that recognizes that separator
119       */
120      public static Splitter on(char separator) {
121        return on(CharMatcher.is(separator));
122      }
123    
124      /**
125       * Returns a splitter that considers any single character matched by the
126       * given {@code CharMatcher} to be a separator. For example, {@code
127       * Splitter.on(CharMatcher.anyOf(";,")).split("foo,;bar,quux")} returns an
128       * iterable containing {@code ["foo", "", "bar", "quux"]}.
129       *
130       * @param separatorMatcher a {@link CharMatcher} that determines whether a
131       *     character is a separator
132       * @return a splitter, with default settings, that uses this matcher
133       */
134      public static Splitter on(final CharMatcher separatorMatcher) {
135        checkNotNull(separatorMatcher);
136    
137        return new Splitter(new Strategy() {
138          @Override public SplittingIterator iterator(
139              Splitter splitter, final CharSequence toSplit) {
140            return new SplittingIterator(splitter, toSplit) {
141              @Override int separatorStart(int start) {
142                return separatorMatcher.indexIn(toSplit, start);
143              }
144    
145              @Override int separatorEnd(int separatorPosition) {
146                return separatorPosition + 1;
147              }
148            };
149          }
150        });
151      }
152    
153      /**
154       * Returns a splitter that uses the given fixed string as a separator. For
155       * example, {@code Splitter.on(", ").split("foo, bar, baz,qux")} returns an
156       * iterable containing {@code ["foo", "bar", "baz,qux"]}.
157       *
158       * @param separator the literal, nonempty string to recognize as a separator
159       * @return a splitter, with default settings, that recognizes that separator
160       */
161      public static Splitter on(final String separator) {
162        checkArgument(separator.length() != 0,
163            "The separator may not be the empty string.");
164    
165        return new Splitter(new Strategy() {
166          @Override public SplittingIterator iterator(
167              Splitter splitter, CharSequence toSplit) {
168            return new SplittingIterator(splitter, toSplit) {
169              @Override public int separatorStart(int start) {
170                int delimeterLength = separator.length();
171    
172                positions:
173                for (int p = start, last = toSplit.length() - delimeterLength;
174                    p <= last; p++) {
175                  for (int i = 0; i < delimeterLength; i++) {
176                    if (toSplit.charAt(i + p) != separator.charAt(i)) {
177                      continue positions;
178                    }
179                  }
180                  return p;
181                }
182                return -1;
183              }
184    
185              @Override public int separatorEnd(int separatorPosition) {
186                return separatorPosition + separator.length();
187              }
188            };
189          }
190        });
191      }
192    
193      /**
194       * Returns a splitter that considers any subsequence matching {@code
195       * pattern} to be a separator. For example, {@code
196       * Splitter.on(Pattern.compile("\r?\n")).split(entireFile)} splits a string
197       * into lines whether it uses DOS-style or UNIX-style line terminators.
198       *
199       * @param separatorPattern the pattern that determines whether a subsequence
200       *     is a separator. This pattern may not match the empty string.
201       * @return a splitter, with default settings, that uses this pattern
202       * @throws IllegalArgumentException if {@code separatorPattern} matches the
203       *     empty string
204       */
205      @GwtIncompatible("java.util.regex")
206      public static Splitter on(final Pattern separatorPattern) {
207        checkNotNull(separatorPattern);
208        checkArgument(!separatorPattern.matcher("").matches(),
209            "The pattern may not match the empty string: %s", separatorPattern);
210    
211        return new Splitter(new Strategy() {
212          @Override public SplittingIterator iterator(
213              final Splitter splitter, CharSequence toSplit) {
214            final Matcher matcher = separatorPattern.matcher(toSplit);
215            return new SplittingIterator(splitter, toSplit) {
216              @Override public int separatorStart(int start) {
217                return matcher.find(start) ? matcher.start() : -1;
218              }
219    
220              @Override public int separatorEnd(int separatorPosition) {
221                return matcher.end();
222              }
223            };
224          }
225        });
226      }
227    
228      /**
229       * Returns a splitter that considers any subsequence matching a given
230       * pattern (regular expression) to be a separator. For example, {@code
231       * Splitter.onPattern("\r?\n").split(entireFile)} splits a string into lines
232       * whether it uses DOS-style or UNIX-style line terminators. This is
233       * equivalent to {@code Splitter.on(Pattern.compile(pattern))}.
234       *
235       * @param separatorPattern the pattern that determines whether a subsequence
236       *     is a separator. This pattern may not match the empty string.
237       * @return a splitter, with default settings, that uses this pattern
238       * @throws java.util.regex.PatternSyntaxException if {@code separatorPattern}
239       *     is a malformed expression
240       * @throws IllegalArgumentException if {@code separatorPattern} matches the
241       *     empty string
242       */
243      @GwtIncompatible("java.util.regex")
244      public static Splitter onPattern(String separatorPattern) {
245        return on(Pattern.compile(separatorPattern));
246      }
247    
248      /**
249       * Returns a splitter that divides strings into pieces of the given length.
250       * For example, {@code Splitter.fixedLength(2).split("abcde")} returns an
251       * iterable containing {@code ["ab", "cd", "e"]}. The last piece can be
252       * smaller than {@code length} but will never be empty.
253       *
254       * @param length the desired length of pieces after splitting
255       * @return a splitter, with default settings, that can split into fixed sized
256       *     pieces
257       */
258      public static Splitter fixedLength(final int length) {
259        checkArgument(length > 0, "The length may not be less than 1");
260    
261        return new Splitter(new Strategy() {
262          @Override public SplittingIterator iterator(
263              final Splitter splitter, CharSequence toSplit) {
264            return new SplittingIterator(splitter, toSplit) {
265              @Override public int separatorStart(int start) {
266                int nextChunkStart = start + length;
267                return (nextChunkStart < toSplit.length() ? nextChunkStart : -1);
268              }
269    
270              @Override public int separatorEnd(int separatorPosition) {
271                return separatorPosition;
272              }
273            };
274          }
275        });
276      }
277    
278      /**
279       * Returns a splitter that behaves equivalently to {@code this} splitter, but
280       * automatically omits empty strings from the results. For example, {@code
281       * Splitter.on(',').omitEmptyStrings().split(",a,,,b,c,,")} returns an
282       * iterable containing only {@code ["a", "b", "c"]}.
283       *
284       * <p>If either {@code trimResults} option is also specified when creating a
285       * splitter, that splitter always trims results first before checking for
286       * emptiness. So, for example, {@code
287       * Splitter.on(':').omitEmptyStrings().trimResults().split(": : : ")} returns
288       * an empty iterable.
289       *
290       * <p>Note that it is ordinarily not possible for {@link #split(CharSequence)}
291       * to return an empty iterable, but when using this option, it can (if the
292       * input sequence consists of nothing but separators).
293       *
294       * @return a splitter with the desired configuration
295       */
296      public Splitter omitEmptyStrings() {
297        return new Splitter(strategy, true, trimmer, limit);
298      }
299    
300      /**
301       * Returns a splitter that behaves equivalently to {@code this} splitter but
302       * stops splitting after it reaches the limit.
303       * The limit defines the maximum number of items returned by the iterator.
304       *
305       * <p>For example,
306       * {@code Splitter.on(',').limit(3).split("a,b,c,d")} returns an iterable
307       * containing {@code ["a", "b", "c,d"]}.  When omitting empty strings, the
308       * omitted strings do no count.  Hence,
309       * {@code Splitter.on(',').limit(3).omitEmptyStrings().split("a,,,b,,,c,d")}
310       * returns an iterable containing {@code ["a", "b", "c,d"}.
311       * When trim is requested, all entries, including the last are trimmed.  Hence
312       * {@code Splitter.on(',').limit(3).trimResults().split(" a , b , c , d ")}
313       * results in @{code ["a", "b", "c , d"]}.
314       *
315       * @param limit the maximum number of items returns
316       * @return a splitter with the desired configuration
317       * @since 9
318       */
319      @Beta
320      public Splitter limit(int limit) {
321        checkArgument(limit > 0, "must be greater then zero: %s", limit);
322        return new Splitter(strategy, omitEmptyStrings, trimmer, limit);
323      }
324    
325      /**
326       * Returns a splitter that behaves equivalently to {@code this} splitter, but
327       * automatically removes leading and trailing {@linkplain
328       * CharMatcher#WHITESPACE whitespace} from each returned substring; equivalent
329       * to {@code trimResults(CharMatcher.WHITESPACE)}. For example, {@code
330       * Splitter.on(',').trimResults().split(" a, b ,c ")} returns an iterable
331       * containing {@code ["a", "b", "c"]}.
332       *
333       * @return a splitter with the desired configuration
334       */
335      public Splitter trimResults() {
336        return trimResults(CharMatcher.WHITESPACE);
337      }
338    
339      /**
340       * Returns a splitter that behaves equivalently to {@code this} splitter, but
341       * removes all leading or trailing characters matching the given {@code
342       * CharMatcher} from each returned substring. For example, {@code
343       * Splitter.on(',').trimResults(CharMatcher.is('_')).split("_a ,_b_ ,c__")}
344       * returns an iterable containing {@code ["a ", "b_ ", "c"]}.
345       *
346       * @param trimmer a {@link CharMatcher} that determines whether a character
347       *     should be removed from the beginning/end of a subsequence
348       * @return a splitter with the desired configuration
349       */
350      // TODO(kevinb): throw if a trimmer was already specified!
351      public Splitter trimResults(CharMatcher trimmer) {
352        checkNotNull(trimmer);
353        return new Splitter(strategy, omitEmptyStrings, trimmer, limit);
354      }
355    
356      /**
357       * Splits {@code sequence} into string components and makes them available
358       * through an {@link Iterator}, which may be lazily evaluated.
359       *
360       * @param sequence the sequence of characters to split
361       * @return an iteration over the segments split from the parameter.
362       */
363      public Iterable<String> split(final CharSequence sequence) {
364        checkNotNull(sequence);
365    
366        return new Iterable<String>() {
367          @Override public Iterator<String> iterator() {
368            return strategy.iterator(Splitter.this, sequence);
369          }
370        };
371      }
372    
373      private interface Strategy {
374        Iterator<String> iterator(Splitter splitter, CharSequence toSplit);
375      }
376    
377      private abstract static class SplittingIterator
378          extends AbstractIterator<String> {
379        final CharSequence toSplit;
380        final CharMatcher trimmer;
381        final boolean omitEmptyStrings;
382    
383        /**
384         * Returns the first index in {@code toSplit} at or after {@code start}
385         * that contains the separator.
386         */
387        abstract int separatorStart(int start);
388    
389        /**
390         * Returns the first index in {@code toSplit} after {@code
391         * separatorPosition} that does not contain a separator. This method is only
392         * invoked after a call to {@code separatorStart}.
393         */
394        abstract int separatorEnd(int separatorPosition);
395    
396        int offset = 0;
397        int limit;
398    
399        protected SplittingIterator(Splitter splitter, CharSequence toSplit) {
400          this.trimmer = splitter.trimmer;
401          this.omitEmptyStrings = splitter.omitEmptyStrings;
402          this.limit = splitter.limit;
403          this.toSplit = toSplit;
404        }
405    
406        @Override protected String computeNext() {
407          while (offset != -1) {
408            int start = offset;
409            int end;
410    
411            int separatorPosition = separatorStart(offset);
412            if (separatorPosition == -1) {
413              end = toSplit.length();
414              offset = -1;
415            } else {
416              end = separatorPosition;
417              offset = separatorEnd(separatorPosition);
418            }
419    
420            while (start < end && trimmer.matches(toSplit.charAt(start))) {
421              start++;
422            }
423            while (end > start && trimmer.matches(toSplit.charAt(end - 1))) {
424              end--;
425            }
426    
427            if (omitEmptyStrings && start == end) {
428              continue;
429            }
430    
431            if (limit == 1) {
432              // The limit has been reached, return the rest of the string as the
433              // final item.  This is tested after empty string removal so that
434              // empty strings do not count towards the limit.
435              end = toSplit.length();
436              offset = -1;
437              // Since we may have changed the end, we need to trim it again.
438              while (end > start && trimmer.matches(toSplit.charAt(end - 1))) {
439                end--;
440              }
441            } else {
442              limit--;
443            }
444    
445            return toSplit.subSequence(start, end).toString();
446          }
447          return endOfData();
448        }
449      }
450    
451      /*
452       * Copied from common.collect.AbstractIterator. TODO(kevinb): un-fork if these
453       * packages are ever combined into a single library.
454       */
455      private abstract static class AbstractIterator<T> implements Iterator<T> {
456        State state = State.NOT_READY;
457    
458        enum State {
459          READY, NOT_READY, DONE, FAILED,
460        }
461    
462        T next;
463    
464        protected abstract T computeNext();
465    
466        protected final T endOfData() {
467          state = State.DONE;
468          return null;
469        }
470    
471        @Override
472        public final boolean hasNext() {
473          checkState(state != State.FAILED);
474          switch (state) {
475            case DONE:
476              return false;
477            case READY:
478              return true;
479            default:
480          }
481          return tryToComputeNext();
482        }
483    
484        boolean tryToComputeNext() {
485          state = State.FAILED; // temporary pessimism
486          next = computeNext();
487          if (state != State.DONE) {
488            state = State.READY;
489            return true;
490          }
491          return false;
492        }
493    
494        @Override
495        public final T next() {
496          if (!hasNext()) {
497            throw new NoSuchElementException();
498          }
499          state = State.NOT_READY;
500          return next;
501        }
502    
503        @Override public void remove() {
504          throw new UnsupportedOperationException();
505        }
506      }
507    }