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    
022    import com.google.common.annotations.Beta;
023    import com.google.common.annotations.GwtCompatible;
024    import com.google.common.annotations.GwtIncompatible;
025    
026    import java.util.Collections;
027    import java.util.Iterator;
028    import java.util.LinkedHashMap;
029    import java.util.Map;
030    import java.util.regex.Matcher;
031    import java.util.regex.Pattern;
032    
033    import javax.annotation.CheckReturnValue;
034    
035    /**
036     * An object that divides strings (or other instances of {@code CharSequence})
037     * into substrings, by recognizing a <i>separator</i> (a.k.a. "delimiter")
038     * which can be expressed as a single character, literal string, regular
039     * expression, {@code CharMatcher}, or by using a fixed substring length. This
040     * class provides the complementary functionality to {@link Joiner}.
041     *
042     * <p>Here is the most basic example of {@code Splitter} usage: <pre>   {@code
043     *
044     *   Splitter.on(',').split("foo,bar")}</pre>
045     *
046     * This invocation returns an {@code Iterable<String>} containing {@code "foo"}
047     * and {@code "bar"}, in that order.
048     *
049     * <p>By default {@code Splitter}'s behavior is very simplistic: <pre>   {@code
050     *
051     *   Splitter.on(',').split("foo,,bar, quux")}</pre>
052     *
053     * This returns an iterable containing {@code ["foo", "", "bar", " quux"]}.
054     * Notice that the splitter does not assume that you want empty strings removed,
055     * or that you wish to trim whitespace. If you want features like these, simply
056     * ask for them: <pre> {@code
057     *
058     *   private static final Splitter MY_SPLITTER = Splitter.on(',')
059     *       .trimResults()
060     *       .omitEmptyStrings();}</pre>
061     *
062     * Now {@code MY_SPLITTER.split("foo, ,bar, quux,")} returns an iterable
063     * containing just {@code ["foo", "bar", "quux"]}. Note that the order in which
064     * the configuration methods are called is never significant; for instance,
065     * trimming is always applied first before checking for an empty result,
066     * regardless of the order in which the {@link #trimResults()} and
067     * {@link #omitEmptyStrings()} methods were invoked.
068     *
069     * <p><b>Warning: splitter instances are always immutable</b>; a configuration
070     * method such as {@code omitEmptyStrings} has no effect on the instance it
071     * is invoked on! You must store and use the new splitter instance returned by
072     * the method. This makes splitters thread-safe, and safe to store as {@code
073     * static final} constants (as illustrated above). <pre>   {@code
074     *
075     *   // Bad! Do not do this!
076     *   Splitter splitter = Splitter.on('/');
077     *   splitter.trimResults(); // does nothing!
078     *   return splitter.split("wrong / wrong / wrong");}</pre>
079     *
080     * The separator recognized by the splitter does not have to be a single
081     * literal character as in the examples above. See the methods {@link
082     * #on(String)}, {@link #on(Pattern)} and {@link #on(CharMatcher)} for examples
083     * of other ways to specify separators.
084     *
085     * <p><b>Note:</b> this class does not mimic any of the quirky behaviors of
086     * similar JDK methods; for instance, it does not silently discard trailing
087     * separators, as does {@link String#split(String)}, nor does it have a default
088     * behavior of using five particular whitespace characters as separators, like
089     * {@link java.util.StringTokenizer}.
090     *
091     * @author Julien Silland
092     * @author Jesse Wilson
093     * @author Kevin Bourrillion
094     * @author Louis Wasserman
095     * @since 1.0
096     */
097    @GwtCompatible(emulated = true)
098    public final class Splitter {
099      private final CharMatcher trimmer;
100      private final boolean omitEmptyStrings;
101      private final Strategy strategy;
102      private final int limit;
103    
104      private Splitter(Strategy strategy) {
105        this(strategy, false, CharMatcher.NONE, Integer.MAX_VALUE);
106      }
107    
108      private Splitter(Strategy strategy, boolean omitEmptyStrings,
109          CharMatcher trimmer, int limit) {
110        this.strategy = strategy;
111        this.omitEmptyStrings = omitEmptyStrings;
112        this.trimmer = trimmer;
113        this.limit = limit;
114      }
115    
116      /**
117       * Returns a splitter that uses the given single-character separator. For
118       * example, {@code Splitter.on(',').split("foo,,bar")} returns an iterable
119       * containing {@code ["foo", "", "bar"]}.
120       *
121       * @param separator the character to recognize as a separator
122       * @return a splitter, with default settings, that recognizes that separator
123       */
124      public static Splitter on(char separator) {
125        return on(CharMatcher.is(separator));
126      }
127    
128      /**
129       * Returns a splitter that considers any single character matched by the
130       * given {@code CharMatcher} to be a separator. For example, {@code
131       * Splitter.on(CharMatcher.anyOf(";,")).split("foo,;bar,quux")} returns an
132       * iterable containing {@code ["foo", "", "bar", "quux"]}.
133       *
134       * @param separatorMatcher a {@link CharMatcher} that determines whether a
135       *     character is a separator
136       * @return a splitter, with default settings, that uses this matcher
137       */
138      public static Splitter on(final CharMatcher separatorMatcher) {
139        checkNotNull(separatorMatcher);
140    
141        return new Splitter(new Strategy() {
142          @Override public SplittingIterator iterator(
143              Splitter splitter, final CharSequence toSplit) {
144            return new SplittingIterator(splitter, toSplit) {
145              @Override int separatorStart(int start) {
146                return separatorMatcher.indexIn(toSplit, start);
147              }
148    
149              @Override int separatorEnd(int separatorPosition) {
150                return separatorPosition + 1;
151              }
152            };
153          }
154        });
155      }
156    
157      /**
158       * Returns a splitter that uses the given fixed string as a separator. For
159       * example, {@code Splitter.on(", ").split("foo, bar, baz,qux")} returns an
160       * iterable containing {@code ["foo", "bar", "baz,qux"]}.
161       *
162       * @param separator the literal, nonempty string to recognize as a separator
163       * @return a splitter, with default settings, that recognizes that separator
164       */
165      public static Splitter on(final String separator) {
166        checkArgument(separator.length() != 0,
167            "The separator may not be the empty string.");
168    
169        return new Splitter(new Strategy() {
170          @Override public SplittingIterator iterator(
171              Splitter splitter, CharSequence toSplit) {
172            return new SplittingIterator(splitter, toSplit) {
173              @Override public int separatorStart(int start) {
174                int delimeterLength = separator.length();
175    
176                positions:
177                for (int p = start, last = toSplit.length() - delimeterLength;
178                    p <= last; p++) {
179                  for (int i = 0; i < delimeterLength; i++) {
180                    if (toSplit.charAt(i + p) != separator.charAt(i)) {
181                      continue positions;
182                    }
183                  }
184                  return p;
185                }
186                return -1;
187              }
188    
189              @Override public int separatorEnd(int separatorPosition) {
190                return separatorPosition + separator.length();
191              }
192            };
193          }
194        });
195      }
196    
197      /**
198       * Returns a splitter that considers any subsequence matching {@code
199       * pattern} to be a separator. For example, {@code
200       * Splitter.on(Pattern.compile("\r?\n")).split(entireFile)} splits a string
201       * into lines whether it uses DOS-style or UNIX-style line terminators.
202       *
203       * @param separatorPattern the pattern that determines whether a subsequence
204       *     is a separator. This pattern may not match the empty string.
205       * @return a splitter, with default settings, that uses this pattern
206       * @throws IllegalArgumentException if {@code separatorPattern} matches the
207       *     empty string
208       */
209      @GwtIncompatible("java.util.regex")
210      public static Splitter on(final Pattern separatorPattern) {
211        checkNotNull(separatorPattern);
212        checkArgument(!separatorPattern.matcher("").matches(),
213            "The pattern may not match the empty string: %s", separatorPattern);
214    
215        return new Splitter(new Strategy() {
216          @Override public SplittingIterator iterator(
217              final Splitter splitter, CharSequence toSplit) {
218            final Matcher matcher = separatorPattern.matcher(toSplit);
219            return new SplittingIterator(splitter, toSplit) {
220              @Override public int separatorStart(int start) {
221                return matcher.find(start) ? matcher.start() : -1;
222              }
223    
224              @Override public int separatorEnd(int separatorPosition) {
225                return matcher.end();
226              }
227            };
228          }
229        });
230      }
231    
232      /**
233       * Returns a splitter that considers any subsequence matching a given
234       * pattern (regular expression) to be a separator. For example, {@code
235       * Splitter.onPattern("\r?\n").split(entireFile)} splits a string into lines
236       * whether it uses DOS-style or UNIX-style line terminators. This is
237       * equivalent to {@code Splitter.on(Pattern.compile(pattern))}.
238       *
239       * @param separatorPattern the pattern that determines whether a subsequence
240       *     is a separator. This pattern may not match the empty string.
241       * @return a splitter, with default settings, that uses this pattern
242       * @throws java.util.regex.PatternSyntaxException if {@code separatorPattern}
243       *     is a malformed expression
244       * @throws IllegalArgumentException if {@code separatorPattern} matches the
245       *     empty string
246       */
247      @GwtIncompatible("java.util.regex")
248      public static Splitter onPattern(String separatorPattern) {
249        return on(Pattern.compile(separatorPattern));
250      }
251    
252      /**
253       * Returns a splitter that divides strings into pieces of the given length.
254       * For example, {@code Splitter.fixedLength(2).split("abcde")} returns an
255       * iterable containing {@code ["ab", "cd", "e"]}. The last piece can be
256       * smaller than {@code length} but will never be empty.
257       *
258       * @param length the desired length of pieces after splitting
259       * @return a splitter, with default settings, that can split into fixed sized
260       *     pieces
261       */
262      public static Splitter fixedLength(final int length) {
263        checkArgument(length > 0, "The length may not be less than 1");
264    
265        return new Splitter(new Strategy() {
266          @Override public SplittingIterator iterator(
267              final Splitter splitter, CharSequence toSplit) {
268            return new SplittingIterator(splitter, toSplit) {
269              @Override public int separatorStart(int start) {
270                int nextChunkStart = start + length;
271                return (nextChunkStart < toSplit.length() ? nextChunkStart : -1);
272              }
273    
274              @Override public int separatorEnd(int separatorPosition) {
275                return separatorPosition;
276              }
277            };
278          }
279        });
280      }
281    
282      /**
283       * Returns a splitter that behaves equivalently to {@code this} splitter, but
284       * automatically omits empty strings from the results. For example, {@code
285       * Splitter.on(',').omitEmptyStrings().split(",a,,,b,c,,")} returns an
286       * iterable containing only {@code ["a", "b", "c"]}.
287       *
288       * <p>If either {@code trimResults} option is also specified when creating a
289       * splitter, that splitter always trims results first before checking for
290       * emptiness. So, for example, {@code
291       * Splitter.on(':').omitEmptyStrings().trimResults().split(": : : ")} returns
292       * an empty iterable.
293       *
294       * <p>Note that it is ordinarily not possible for {@link #split(CharSequence)}
295       * to return an empty iterable, but when using this option, it can (if the
296       * input sequence consists of nothing but separators).
297       *
298       * @return a splitter with the desired configuration
299       */
300      @CheckReturnValue
301      public Splitter omitEmptyStrings() {
302        return new Splitter(strategy, true, trimmer, limit);
303      }
304    
305      /**
306       * Returns a splitter that behaves equivalently to {@code this} splitter but
307       * stops splitting after it reaches the limit.
308       * The limit defines the maximum number of items returned by the iterator.
309       *
310       * <p>For example,
311       * {@code Splitter.on(',').limit(3).split("a,b,c,d")} returns an iterable
312       * containing {@code ["a", "b", "c,d"]}.  When omitting empty strings, the
313       * omitted strings do no count.  Hence,
314       * {@code Splitter.on(',').limit(3).omitEmptyStrings().split("a,,,b,,,c,d")}
315       * returns an iterable containing {@code ["a", "b", "c,d"}.
316       * When trim is requested, all entries, including the last are trimmed.  Hence
317       * {@code Splitter.on(',').limit(3).trimResults().split(" a , b , c , d ")}
318       * results in @{code ["a", "b", "c , d"]}.
319       *
320       * @param limit the maximum number of items returns
321       * @return a splitter with the desired configuration
322       * @since 9.0
323       */
324      @CheckReturnValue
325      public Splitter limit(int limit) {
326        checkArgument(limit > 0, "must be greater than zero: %s", limit);
327        return new Splitter(strategy, omitEmptyStrings, trimmer, limit);
328      }
329    
330      /**
331       * Returns a splitter that behaves equivalently to {@code this} splitter, but
332       * automatically removes leading and trailing {@linkplain
333       * CharMatcher#WHITESPACE whitespace} from each returned substring; equivalent
334       * to {@code trimResults(CharMatcher.WHITESPACE)}. For example, {@code
335       * Splitter.on(',').trimResults().split(" a, b ,c ")} returns an iterable
336       * containing {@code ["a", "b", "c"]}.
337       *
338       * @return a splitter with the desired configuration
339       */
340      @CheckReturnValue
341      public Splitter trimResults() {
342        return trimResults(CharMatcher.WHITESPACE);
343      }
344    
345      /**
346       * Returns a splitter that behaves equivalently to {@code this} splitter, but
347       * removes all leading or trailing characters matching the given {@code
348       * CharMatcher} from each returned substring. For example, {@code
349       * Splitter.on(',').trimResults(CharMatcher.is('_')).split("_a ,_b_ ,c__")}
350       * returns an iterable containing {@code ["a ", "b_ ", "c"]}.
351       *
352       * @param trimmer a {@link CharMatcher} that determines whether a character
353       *     should be removed from the beginning/end of a subsequence
354       * @return a splitter with the desired configuration
355       */
356      // TODO(kevinb): throw if a trimmer was already specified!
357      @CheckReturnValue
358      public Splitter trimResults(CharMatcher trimmer) {
359        checkNotNull(trimmer);
360        return new Splitter(strategy, omitEmptyStrings, trimmer, limit);
361      }
362    
363      /**
364       * Splits {@code sequence} into string components and makes them available
365       * through an {@link Iterator}, which may be lazily evaluated.
366       *
367       * @param sequence the sequence of characters to split
368       * @return an iteration over the segments split from the parameter.
369       */
370      public Iterable<String> split(final CharSequence sequence) {
371        checkNotNull(sequence);
372    
373        return new Iterable<String>() {
374          @Override public Iterator<String> iterator() {
375            return spliterator(sequence);
376          }
377        };
378      }
379    
380      private Iterator<String> spliterator(CharSequence sequence) {
381        return strategy.iterator(this, sequence);
382      }
383    
384      /**
385       * Returns a {@code MapSplitter} which splits entries based on this splitter,
386       * and splits entries into keys and values using the specified separator.
387       *
388       * @since 10.0
389       */
390      @CheckReturnValue
391      @Beta
392      public MapSplitter withKeyValueSeparator(String separator) {
393        return withKeyValueSeparator(on(separator));
394      }
395    
396      /**
397       * Returns a {@code MapSplitter} which splits entries based on this splitter,
398       * and splits entries into keys and values using the specified key-value
399       * splitter.
400       *
401       * @since 10.0
402       */
403      @CheckReturnValue
404      @Beta
405      public MapSplitter withKeyValueSeparator(Splitter keyValueSplitter) {
406        return new MapSplitter(this, keyValueSplitter);
407      }
408    
409      /**
410       * An object that splits strings into maps as {@code Splitter} splits
411       * iterables and lists. Like {@code Splitter}, it is thread-safe and
412       * immutable.
413       *
414       * @since 10.0
415       */
416      @Beta
417      public static final class MapSplitter {
418        private static final String INVALID_ENTRY_MESSAGE =
419            "Chunk [%s] is not a valid entry";
420        private final Splitter outerSplitter;
421        private final Splitter entrySplitter;
422    
423        private MapSplitter(Splitter outerSplitter, Splitter entrySplitter) {
424          this.outerSplitter = outerSplitter; // only "this" is passed
425          this.entrySplitter = checkNotNull(entrySplitter);
426        }
427    
428        /**
429         * Splits {@code sequence} into substrings, splits each substring into
430         * an entry, and returns an unmodifiable map with each of the entries. For
431         * example, <code>
432         * Splitter.on(';').trimResults().withKeyValueSeparator("=>")
433         * .split("a=>b ; c=>b")
434         * </code> will return a mapping from {@code "a"} to {@code "b"} and
435         * {@code "c"} to {@code b}.
436         *
437         * <p>The returned map preserves the order of the entries from
438         * {@code sequence}.
439         *
440         * @throws IllegalArgumentException if the specified sequence does not split
441         *         into valid map entries, or if there are duplicate keys
442         */
443        public Map<String, String> split(CharSequence sequence) {
444          Map<String, String> map = new LinkedHashMap<String, String>();
445          for (String entry : outerSplitter.split(sequence)) {
446            Iterator<String> entryFields = entrySplitter.spliterator(entry);
447    
448            checkArgument(entryFields.hasNext(), INVALID_ENTRY_MESSAGE, entry);
449            String key = entryFields.next();
450            checkArgument(!map.containsKey(key), "Duplicate key [%s] found.", key);
451    
452            checkArgument(entryFields.hasNext(), INVALID_ENTRY_MESSAGE, entry);
453            String value = entryFields.next();
454            map.put(key, value);
455    
456            checkArgument(!entryFields.hasNext(), INVALID_ENTRY_MESSAGE, entry);
457          }
458          return Collections.unmodifiableMap(map);
459        }
460      }
461    
462      private interface Strategy {
463        Iterator<String> iterator(Splitter splitter, CharSequence toSplit);
464      }
465    
466      private abstract static class SplittingIterator
467          extends AbstractIterator<String> {
468        final CharSequence toSplit;
469        final CharMatcher trimmer;
470        final boolean omitEmptyStrings;
471    
472        /**
473         * Returns the first index in {@code toSplit} at or after {@code start}
474         * that contains the separator.
475         */
476        abstract int separatorStart(int start);
477    
478        /**
479         * Returns the first index in {@code toSplit} after {@code
480         * separatorPosition} that does not contain a separator. This method is only
481         * invoked after a call to {@code separatorStart}.
482         */
483        abstract int separatorEnd(int separatorPosition);
484    
485        int offset = 0;
486        int limit;
487    
488        protected SplittingIterator(Splitter splitter, CharSequence toSplit) {
489          this.trimmer = splitter.trimmer;
490          this.omitEmptyStrings = splitter.omitEmptyStrings;
491          this.limit = splitter.limit;
492          this.toSplit = toSplit;
493        }
494    
495        @Override protected String computeNext() {
496          while (offset != -1) {
497            int start = offset;
498            int end;
499    
500            int separatorPosition = separatorStart(offset);
501            if (separatorPosition == -1) {
502              end = toSplit.length();
503              offset = -1;
504            } else {
505              end = separatorPosition;
506              offset = separatorEnd(separatorPosition);
507            }
508    
509            while (start < end && trimmer.matches(toSplit.charAt(start))) {
510              start++;
511            }
512            while (end > start && trimmer.matches(toSplit.charAt(end - 1))) {
513              end--;
514            }
515    
516            if (omitEmptyStrings && start == end) {
517              continue;
518            }
519    
520            if (limit == 1) {
521              // The limit has been reached, return the rest of the string as the
522              // final item.  This is tested after empty string removal so that
523              // empty strings do not count towards the limit.
524              end = toSplit.length();
525              offset = -1;
526              // Since we may have changed the end, we need to trim it again.
527              while (end > start && trimmer.matches(toSplit.charAt(end - 1))) {
528                end--;
529              }
530            } else {
531              limit--;
532            }
533    
534            return toSplit.subSequence(start, end).toString();
535          }
536          return endOfData();
537        }
538      }
539    }