001 /*
002 * Copyright (C) 2009 Google Inc.
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.GwtCompatible;
024 import com.google.common.annotations.GwtIncompatible;
025
026 import java.util.Iterator;
027 import java.util.NoSuchElementException;
028 import java.util.regex.Matcher;
029 import java.util.regex.Pattern;
030
031 /**
032 * An object that divides strings (or other instances of {@code CharSequence})
033 * into substrings, by recognizing a <i>separator</i> (a.k.a. "delimiter")
034 * which can be expressed as a single character, literal string, regular
035 * expression, {@code CharMatcher}, or by using a fixed substring length. This
036 * class provides the complementary functionality to {@link Joiner}.
037 *
038 * <p>Here is the most basic example of {@code Splitter} usage: <pre> {@code
039 *
040 * Splitter.on(',').split("foo,bar")}</pre>
041 *
042 * This invocation returns an {@code Iterable<String>} containing {@code "foo"}
043 * and {@code "bar"}, in that order.
044 *
045 * <p>By default {@code Splitter}'s behavior is very simplistic: <pre> {@code
046 *
047 * Splitter.on(',').split("foo,,bar, quux")}</pre>
048 *
049 * This returns an iterable containing {@code ["foo", "", "bar", " quux"]}.
050 * Notice that the splitter does not assume that you want empty strings removed,
051 * or that you wish to trim whitespace. If you want features like these, simply
052 * ask for them: <pre> {@code
053 *
054 * private static final Splitter MY_SPLITTER = Splitter.on(',')
055 * .trimResults()
056 * .omitEmptyStrings();}</pre>
057 *
058 * Now {@code MY_SPLITTER.split("foo, ,bar, quux,")} returns an iterable
059 * containing just {@code ["foo", "bar", "quux"]}. Note that the order in which
060 * the configuration methods are called is never significant; for instance,
061 * trimming is always applied first before checking for an empty result,
062 * regardless of the order in which the {@link #trimResults()} and
063 * {@link #omitEmptyStrings()} methods were invoked.
064 *
065 * <p><b>Warning: splitter instances are always immutable</b>; a configuration
066 * method such as {@code omitEmptyStrings} has no effect on the instance it
067 * is invoked on! You must store and use the new splitter instance returned by
068 * the method. This makes splitters thread-safe, and safe to store as {@code
069 * static final} constants (as illustrated above). <pre> {@code
070 *
071 * // Bad! Do not do this!
072 * Splitter splitter = Splitter.on('/');
073 * splitter.trimResults(); // does nothing!
074 * return splitter.split("wrong / wrong / wrong");}</pre>
075 *
076 * The separator recognized by the splitter does not have to be a single
077 * literal character as in the examples above. See the methods {@link
078 * #on(String)}, {@link #on(Pattern)} and {@link #on(CharMatcher)} for examples
079 * of other ways to specify separators.
080 *
081 * <p><b>Note:</b> this class does not mimic any of the quirky behaviors of
082 * similar JDK methods; for instance, it does not silently discard trailing
083 * separators, as does {@link String#split(String)}, nor does it have a default
084 * behavior of using five particular whitespace characters as separators, like
085 * {@link java.util.StringTokenizer}.
086 *
087 * @author Julien Silland
088 * @author Jesse Wilson
089 * @author Kevin Bourrillion
090 * @since 1
091 */
092 @GwtCompatible(emulated = true)
093 public final class Splitter {
094 private final CharMatcher trimmer;
095 private final boolean omitEmptyStrings;
096 private final Strategy strategy;
097 private final int limit;
098
099 private Splitter(Strategy strategy) {
100 this(strategy, false, CharMatcher.NONE, Integer.MAX_VALUE);
101 }
102
103 private Splitter(Strategy strategy, boolean omitEmptyStrings,
104 CharMatcher trimmer, int limit) {
105 this.strategy = strategy;
106 this.omitEmptyStrings = omitEmptyStrings;
107 this.trimmer = trimmer;
108 this.limit = limit;
109 }
110
111 /**
112 * Returns a splitter that uses the given single-character separator. For
113 * example, {@code Splitter.on(',').split("foo,,bar")} returns an iterable
114 * containing {@code ["foo", "", "bar"]}.
115 *
116 * @param separator the character to recognize as a separator
117 * @return a splitter, with default settings, that recognizes that separator
118 */
119 public static Splitter on(char separator) {
120 return on(CharMatcher.is(separator));
121 }
122
123 /**
124 * Returns a splitter that considers any single character matched by the
125 * given {@code CharMatcher} to be a separator. For example, {@code
126 * Splitter.on(CharMatcher.anyOf(";,")).split("foo,;bar,quux")} returns an
127 * iterable containing {@code ["foo", "", "bar", "quux"]}.
128 *
129 * @param separatorMatcher a {@link CharMatcher} that determines whether a
130 * character is a separator
131 * @return a splitter, with default settings, that uses this matcher
132 */
133 public static Splitter on(final CharMatcher separatorMatcher) {
134 checkNotNull(separatorMatcher);
135
136 return new Splitter(new Strategy() {
137 @Override public SplittingIterator iterator(
138 Splitter splitter, final CharSequence toSplit) {
139 return new SplittingIterator(splitter, toSplit) {
140 @Override int separatorStart(int start) {
141 return separatorMatcher.indexIn(toSplit, start);
142 }
143
144 @Override int separatorEnd(int separatorPosition) {
145 return separatorPosition + 1;
146 }
147 };
148 }
149 });
150 }
151
152 /**
153 * Returns a splitter that uses the given fixed string as a separator. For
154 * example, {@code Splitter.on(", ").split("foo, bar, baz,qux")} returns an
155 * iterable containing {@code ["foo", "bar", "baz,qux"]}.
156 *
157 * @param separator the literal, nonempty string to recognize as a separator
158 * @return a splitter, with default settings, that recognizes that separator
159 */
160 public static Splitter on(final String separator) {
161 checkArgument(separator.length() != 0,
162 "The separator may not be the empty string.");
163
164 return new Splitter(new Strategy() {
165 @Override public SplittingIterator iterator(
166 Splitter splitter, CharSequence toSplit) {
167 return new SplittingIterator(splitter, toSplit) {
168 @Override public int separatorStart(int start) {
169 int delimeterLength = separator.length();
170
171 positions:
172 for (int p = start, last = toSplit.length() - delimeterLength;
173 p <= last; p++) {
174 for (int i = 0; i < delimeterLength; i++) {
175 if (toSplit.charAt(i + p) != separator.charAt(i)) {
176 continue positions;
177 }
178 }
179 return p;
180 }
181 return -1;
182 }
183
184 @Override public int separatorEnd(int separatorPosition) {
185 return separatorPosition + separator.length();
186 }
187 };
188 }
189 });
190 }
191
192 /**
193 * Returns a splitter that considers any subsequence matching {@code
194 * pattern} to be a separator. For example, {@code
195 * Splitter.on(Pattern.compile("\r?\n")).split(entireFile)} splits a string
196 * into lines whether it uses DOS-style or UNIX-style line terminators.
197 *
198 * @param separatorPattern the pattern that determines whether a subsequence
199 * is a separator. This pattern may not match the empty string.
200 * @return a splitter, with default settings, that uses this pattern
201 * @throws IllegalArgumentException if {@code separatorPattern} matches the
202 * empty string
203 */
204 @GwtIncompatible("java.util.regex")
205 public static Splitter on(final Pattern separatorPattern) {
206 checkNotNull(separatorPattern);
207 checkArgument(!separatorPattern.matcher("").matches(),
208 "The pattern may not match the empty string: %s", separatorPattern);
209
210 return new Splitter(new Strategy() {
211 @Override public SplittingIterator iterator(
212 final Splitter splitter, CharSequence toSplit) {
213 final Matcher matcher = separatorPattern.matcher(toSplit);
214 return new SplittingIterator(splitter, toSplit) {
215 @Override public int separatorStart(int start) {
216 return matcher.find(start) ? matcher.start() : -1;
217 }
218
219 @Override public int separatorEnd(int separatorPosition) {
220 return matcher.end();
221 }
222 };
223 }
224 });
225 }
226
227 /**
228 * Returns a splitter that considers any subsequence matching a given
229 * pattern (regular expression) to be a separator. For example, {@code
230 * Splitter.onPattern("\r?\n").split(entireFile)} splits a string into lines
231 * whether it uses DOS-style or UNIX-style line terminators. This is
232 * equivalent to {@code Splitter.on(Pattern.compile(pattern))}.
233 *
234 * @param separatorPattern the pattern that determines whether a subsequence
235 * is a separator. This pattern may not match the empty string.
236 * @return a splitter, with default settings, that uses this pattern
237 * @throws java.util.regex.PatternSyntaxException if {@code separatorPattern}
238 * is a malformed expression
239 * @throws IllegalArgumentException if {@code separatorPattern} matches the
240 * empty string
241 */
242 @GwtIncompatible("java.util.regex")
243 public static Splitter onPattern(String separatorPattern) {
244 return on(Pattern.compile(separatorPattern));
245 }
246
247 /**
248 * Returns a splitter that divides strings into pieces of the given length.
249 * For example, {@code Splitter.fixedLength(2).split("abcde")} returns an
250 * iterable containing {@code ["ab", "cd", "e"]}. The last piece can be
251 * smaller than {@code length} but will never be empty.
252 *
253 * @param length the desired length of pieces after splitting
254 * @return a splitter, with default settings, that can split into fixed sized
255 * pieces
256 */
257 public static Splitter fixedLength(final int length) {
258 checkArgument(length > 0, "The length may not be less than 1");
259
260 return new Splitter(new Strategy() {
261 @Override public SplittingIterator iterator(
262 final Splitter splitter, CharSequence toSplit) {
263 return new SplittingIterator(splitter, toSplit) {
264 @Override public int separatorStart(int start) {
265 int nextChunkStart = start + length;
266 return (nextChunkStart < toSplit.length() ? nextChunkStart : -1);
267 }
268
269 @Override public int separatorEnd(int separatorPosition) {
270 return separatorPosition;
271 }
272 };
273 }
274 });
275 }
276
277 /**
278 * Returns a splitter that behaves equivalently to {@code this} splitter, but
279 * automatically omits empty strings from the results. For example, {@code
280 * Splitter.on(',').omitEmptyStrings().split(",a,,,b,c,,")} returns an
281 * iterable containing only {@code ["a", "b", "c"]}.
282 *
283 * <p>If either {@code trimResults} option is also specified when creating a
284 * splitter, that splitter always trims results first before checking for
285 * emptiness. So, for example, {@code
286 * Splitter.on(':').omitEmptyStrings().trimResults().split(": : : ")} returns
287 * an empty iterable.
288 *
289 * <p>Note that it is ordinarily not possible for {@link #split(CharSequence)}
290 * to return an empty iterable, but when using this option, it can (if the
291 * input sequence consists of nothing but separators).
292 *
293 * @return a splitter with the desired configuration
294 */
295 public Splitter omitEmptyStrings() {
296 return new Splitter(strategy, true, trimmer, limit);
297 }
298
299 /**
300 * Returns a splitter that behaves equivalently to {@code this} splitter, but
301 * automatically removes leading and trailing {@linkplain
302 * CharMatcher#WHITESPACE whitespace} from each returned substring; equivalent
303 * to {@code trimResults(CharMatcher.WHITESPACE)}. For example, {@code
304 * Splitter.on(',').trimResults().split(" a, b ,c ")} returns an iterable
305 * containing {@code ["a", "b", "c"]}.
306 *
307 * @return a splitter with the desired configuration
308 */
309 public Splitter trimResults() {
310 return trimResults(CharMatcher.WHITESPACE);
311 }
312
313 /**
314 * Returns a splitter that behaves equivalently to {@code this} splitter, but
315 * removes all leading or trailing characters matching the given {@code
316 * CharMatcher} from each returned substring. For example, {@code
317 * Splitter.on(',').trimResults(CharMatcher.is('_')).split("_a ,_b_ ,c__")}
318 * returns an iterable containing {@code ["a ", "b_ ", "c"]}.
319 *
320 * @param trimmer a {@link CharMatcher} that determines whether a character
321 * should be removed from the beginning/end of a subsequence
322 * @return a splitter with the desired configuration
323 */
324 // TODO(kevinb): throw if a trimmer was already specified!
325 public Splitter trimResults(CharMatcher trimmer) {
326 checkNotNull(trimmer);
327 return new Splitter(strategy, omitEmptyStrings, trimmer, limit);
328 }
329
330 /**
331 * Splits {@code sequence} into string components and makes them available
332 * through an {@link Iterator}, which may be lazily evaluated.
333 *
334 * @param sequence the sequence of characters to split
335 * @return an iteration over the segments split from the parameter.
336 */
337 public Iterable<String> split(final CharSequence sequence) {
338 checkNotNull(sequence);
339
340 return new Iterable<String>() {
341 @Override public Iterator<String> iterator() {
342 return strategy.iterator(Splitter.this, sequence);
343 }
344 };
345 }
346
347 private interface Strategy {
348 Iterator<String> iterator(Splitter splitter, CharSequence toSplit);
349 }
350
351 private abstract static class SplittingIterator
352 extends AbstractIterator<String> {
353 final CharSequence toSplit;
354 final CharMatcher trimmer;
355 final boolean omitEmptyStrings;
356
357 /**
358 * Returns the first index in {@code toSplit} at or after {@code start}
359 * that contains the separator.
360 */
361 abstract int separatorStart(int start);
362
363 /**
364 * Returns the first index in {@code toSplit} after {@code
365 * separatorPosition} that does not contain a separator. This method is only
366 * invoked after a call to {@code separatorStart}.
367 */
368 abstract int separatorEnd(int separatorPosition);
369
370 int offset = 0;
371 int limit;
372
373 protected SplittingIterator(Splitter splitter, CharSequence toSplit) {
374 this.trimmer = splitter.trimmer;
375 this.omitEmptyStrings = splitter.omitEmptyStrings;
376 this.limit = splitter.limit;
377 this.toSplit = toSplit;
378 }
379
380 @Override protected String computeNext() {
381 while (offset != -1) {
382 int start = offset;
383 int end;
384
385 int separatorPosition = separatorStart(offset);
386 if (separatorPosition == -1) {
387 end = toSplit.length();
388 offset = -1;
389 } else {
390 end = separatorPosition;
391 offset = separatorEnd(separatorPosition);
392 }
393
394 while (start < end && trimmer.matches(toSplit.charAt(start))) {
395 start++;
396 }
397 while (end > start && trimmer.matches(toSplit.charAt(end - 1))) {
398 end--;
399 }
400
401 if (omitEmptyStrings && start == end) {
402 continue;
403 }
404
405 if (limit == 1) {
406 // The limit has been reached, return the rest of the string as the
407 // final item. This is tested after empty string removal so that
408 // empty strings do not count towards the limit.
409 end = toSplit.length();
410 offset = -1;
411 // Since we may have changed the end, we need to trim it again.
412 while (end > start && trimmer.matches(toSplit.charAt(end - 1))) {
413 end--;
414 }
415 } else {
416 limit--;
417 }
418
419 return toSplit.subSequence(start, end).toString();
420 }
421 return endOfData();
422 }
423 }
424
425 /*
426 * Copied from common.collect.AbstractIterator. TODO(kevinb): un-fork if these
427 * packages are ever combined into a single library.
428 */
429 private abstract static class AbstractIterator<T> implements Iterator<T> {
430 State state = State.NOT_READY;
431
432 enum State {
433 READY, NOT_READY, DONE, FAILED,
434 }
435
436 T next;
437
438 protected abstract T computeNext();
439
440 protected final T endOfData() {
441 state = State.DONE;
442 return null;
443 }
444
445 public final boolean hasNext() {
446 checkState(state != State.FAILED);
447 switch (state) {
448 case DONE:
449 return false;
450 case READY:
451 return true;
452 default:
453 }
454 return tryToComputeNext();
455 }
456
457 boolean tryToComputeNext() {
458 state = State.FAILED; // temporary pessimism
459 next = computeNext();
460 if (state != State.DONE) {
461 state = State.READY;
462 return true;
463 }
464 return false;
465 }
466
467 public final T next() {
468 if (!hasNext()) {
469 throw new NoSuchElementException();
470 }
471 state = State.NOT_READY;
472 return next;
473 }
474
475 @Override public void remove() {
476 throw new UnsupportedOperationException();
477 }
478 }
479 }