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 * <p>See the Guava User Guide article on <a href=
092 * "http://code.google.com/p/guava-libraries/wiki/StringsExplained#Splitter">
093 * {@code Splitter}</a>.
094 *
095 * @author Julien Silland
096 * @author Jesse Wilson
097 * @author Kevin Bourrillion
098 * @author Louis Wasserman
099 * @since 1.0
100 */
101 @GwtCompatible(emulated = true)
102 public 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,
113 CharMatcher trimmer, int limit) {
114 this.strategy = strategy;
115 this.omitEmptyStrings = omitEmptyStrings;
116 this.trimmer = trimmer;
117 this.limit = limit;
118 }
119
120 /**
121 * Returns a splitter that uses the given single-character separator. For
122 * example, {@code Splitter.on(',').split("foo,,bar")} returns an iterable
123 * containing {@code ["foo", "", "bar"]}.
124 *
125 * @param separator the character to recognize as a separator
126 * @return a splitter, with default settings, that recognizes that separator
127 */
128 public static Splitter on(char separator) {
129 return on(CharMatcher.is(separator));
130 }
131
132 /**
133 * Returns a splitter that considers any single character matched by the
134 * given {@code CharMatcher} to be a separator. For example, {@code
135 * Splitter.on(CharMatcher.anyOf(";,")).split("foo,;bar,quux")} returns an
136 * iterable containing {@code ["foo", "", "bar", "quux"]}.
137 *
138 * @param separatorMatcher a {@link CharMatcher} that determines whether a
139 * character is a separator
140 * @return a splitter, with default settings, that uses this matcher
141 */
142 public static Splitter on(final CharMatcher separatorMatcher) {
143 checkNotNull(separatorMatcher);
144
145 return new Splitter(new Strategy() {
146 @Override public SplittingIterator iterator(
147 Splitter splitter, final CharSequence toSplit) {
148 return new SplittingIterator(splitter, toSplit) {
149 @Override int separatorStart(int start) {
150 return separatorMatcher.indexIn(toSplit, start);
151 }
152
153 @Override int separatorEnd(int separatorPosition) {
154 return separatorPosition + 1;
155 }
156 };
157 }
158 });
159 }
160
161 /**
162 * Returns a splitter that uses the given fixed string as a separator. For
163 * example, {@code Splitter.on(", ").split("foo, bar, baz,qux")} returns an
164 * iterable containing {@code ["foo", "bar", "baz,qux"]}.
165 *
166 * @param separator the literal, nonempty string to recognize as a separator
167 * @return a splitter, with default settings, that recognizes that separator
168 */
169 public static Splitter on(final String separator) {
170 checkArgument(separator.length() != 0,
171 "The separator may not be the empty string.");
172
173 return new Splitter(new Strategy() {
174 @Override public SplittingIterator iterator(
175 Splitter splitter, CharSequence toSplit) {
176 return new SplittingIterator(splitter, toSplit) {
177 @Override public int separatorStart(int start) {
178 int delimeterLength = separator.length();
179
180 positions:
181 for (int p = start, last = toSplit.length() - delimeterLength;
182 p <= last; p++) {
183 for (int i = 0; i < delimeterLength; i++) {
184 if (toSplit.charAt(i + p) != separator.charAt(i)) {
185 continue positions;
186 }
187 }
188 return p;
189 }
190 return -1;
191 }
192
193 @Override public int separatorEnd(int separatorPosition) {
194 return separatorPosition + separator.length();
195 }
196 };
197 }
198 });
199 }
200
201 /**
202 * Returns a splitter that considers any subsequence matching {@code
203 * pattern} to be a separator. For example, {@code
204 * Splitter.on(Pattern.compile("\r?\n")).split(entireFile)} splits a string
205 * into lines whether it uses DOS-style or UNIX-style line terminators.
206 *
207 * @param separatorPattern the pattern that determines whether a subsequence
208 * is a separator. This pattern may not match the empty string.
209 * @return a splitter, with default settings, that uses this pattern
210 * @throws IllegalArgumentException if {@code separatorPattern} matches the
211 * empty string
212 */
213 @GwtIncompatible("java.util.regex")
214 public static Splitter on(final Pattern separatorPattern) {
215 checkNotNull(separatorPattern);
216 checkArgument(!separatorPattern.matcher("").matches(),
217 "The pattern may not match the empty string: %s", separatorPattern);
218
219 return new Splitter(new Strategy() {
220 @Override public SplittingIterator iterator(
221 final Splitter splitter, CharSequence toSplit) {
222 final Matcher matcher = separatorPattern.matcher(toSplit);
223 return new SplittingIterator(splitter, toSplit) {
224 @Override public int separatorStart(int start) {
225 return matcher.find(start) ? matcher.start() : -1;
226 }
227
228 @Override public int separatorEnd(int separatorPosition) {
229 return matcher.end();
230 }
231 };
232 }
233 });
234 }
235
236 /**
237 * Returns a splitter that considers any subsequence matching a given
238 * pattern (regular expression) to be a separator. For example, {@code
239 * Splitter.onPattern("\r?\n").split(entireFile)} splits a string into lines
240 * whether it uses DOS-style or UNIX-style line terminators. This is
241 * equivalent to {@code Splitter.on(Pattern.compile(pattern))}.
242 *
243 * @param separatorPattern the pattern that determines whether a subsequence
244 * is a separator. This pattern may not match the empty string.
245 * @return a splitter, with default settings, that uses this pattern
246 * @throws java.util.regex.PatternSyntaxException if {@code separatorPattern}
247 * is a malformed expression
248 * @throws IllegalArgumentException if {@code separatorPattern} matches the
249 * empty string
250 */
251 @GwtIncompatible("java.util.regex")
252 public static Splitter onPattern(String separatorPattern) {
253 return on(Pattern.compile(separatorPattern));
254 }
255
256 /**
257 * Returns a splitter that divides strings into pieces of the given length.
258 * For example, {@code Splitter.fixedLength(2).split("abcde")} returns an
259 * iterable containing {@code ["ab", "cd", "e"]}. The last piece can be
260 * smaller than {@code length} but will never be empty.
261 *
262 * @param length the desired length of pieces after splitting
263 * @return a splitter, with default settings, that can split into fixed sized
264 * pieces
265 */
266 public static Splitter fixedLength(final int length) {
267 checkArgument(length > 0, "The length may not be less than 1");
268
269 return new Splitter(new Strategy() {
270 @Override public SplittingIterator iterator(
271 final Splitter splitter, CharSequence toSplit) {
272 return new SplittingIterator(splitter, toSplit) {
273 @Override public int separatorStart(int start) {
274 int nextChunkStart = start + length;
275 return (nextChunkStart < toSplit.length() ? nextChunkStart : -1);
276 }
277
278 @Override public int separatorEnd(int separatorPosition) {
279 return separatorPosition;
280 }
281 };
282 }
283 });
284 }
285
286 /**
287 * Returns a splitter that behaves equivalently to {@code this} splitter, but
288 * automatically omits empty strings from the results. For example, {@code
289 * Splitter.on(',').omitEmptyStrings().split(",a,,,b,c,,")} returns an
290 * iterable containing only {@code ["a", "b", "c"]}.
291 *
292 * <p>If either {@code trimResults} option is also specified when creating a
293 * splitter, that splitter always trims results first before checking for
294 * emptiness. So, for example, {@code
295 * Splitter.on(':').omitEmptyStrings().trimResults().split(": : : ")} returns
296 * an empty iterable.
297 *
298 * <p>Note that it is ordinarily not possible for {@link #split(CharSequence)}
299 * to return an empty iterable, but when using this option, it can (if the
300 * input sequence consists of nothing but separators).
301 *
302 * @return a splitter with the desired configuration
303 */
304 @CheckReturnValue
305 public Splitter omitEmptyStrings() {
306 return new Splitter(strategy, true, trimmer, limit);
307 }
308
309 /**
310 * Returns a splitter that behaves equivalently to {@code this} splitter but
311 * stops splitting after it reaches the limit.
312 * The limit defines the maximum number of items returned by the iterator.
313 *
314 * <p>For example,
315 * {@code Splitter.on(',').limit(3).split("a,b,c,d")} returns an iterable
316 * containing {@code ["a", "b", "c,d"]}. When omitting empty strings, the
317 * omitted strings do no count. Hence,
318 * {@code Splitter.on(',').limit(3).omitEmptyStrings().split("a,,,b,,,c,d")}
319 * returns an iterable containing {@code ["a", "b", "c,d"}.
320 * When trim is requested, all entries, including the last are trimmed. Hence
321 * {@code Splitter.on(',').limit(3).trimResults().split(" a , b , c , d ")}
322 * results in @{code ["a", "b", "c , d"]}.
323 *
324 * @param limit the maximum number of items returns
325 * @return a splitter with the desired configuration
326 * @since 9.0
327 */
328 @CheckReturnValue
329 public Splitter limit(int limit) {
330 checkArgument(limit > 0, "must be greater than zero: %s", limit);
331 return new Splitter(strategy, omitEmptyStrings, trimmer, limit);
332 }
333
334 /**
335 * Returns a splitter that behaves equivalently to {@code this} splitter, but
336 * automatically removes leading and trailing {@linkplain
337 * CharMatcher#WHITESPACE whitespace} from each returned substring; equivalent
338 * to {@code trimResults(CharMatcher.WHITESPACE)}. For example, {@code
339 * Splitter.on(',').trimResults().split(" a, b ,c ")} returns an iterable
340 * containing {@code ["a", "b", "c"]}.
341 *
342 * @return a splitter with the desired configuration
343 */
344 @CheckReturnValue
345 public Splitter trimResults() {
346 return trimResults(CharMatcher.WHITESPACE);
347 }
348
349 /**
350 * Returns a splitter that behaves equivalently to {@code this} splitter, but
351 * removes all leading or trailing characters matching the given {@code
352 * CharMatcher} from each returned substring. For example, {@code
353 * Splitter.on(',').trimResults(CharMatcher.is('_')).split("_a ,_b_ ,c__")}
354 * returns an iterable containing {@code ["a ", "b_ ", "c"]}.
355 *
356 * @param trimmer a {@link CharMatcher} that determines whether a character
357 * should be removed from the beginning/end of a subsequence
358 * @return a splitter with the desired configuration
359 */
360 // TODO(kevinb): throw if a trimmer was already specified!
361 @CheckReturnValue
362 public Splitter trimResults(CharMatcher trimmer) {
363 checkNotNull(trimmer);
364 return new Splitter(strategy, omitEmptyStrings, trimmer, limit);
365 }
366
367 /**
368 * Splits {@code sequence} into string components and makes them available
369 * through an {@link Iterator}, which may be lazily evaluated.
370 *
371 * @param sequence the sequence of characters to split
372 * @return an iteration over the segments split from the parameter.
373 */
374 public Iterable<String> split(final CharSequence sequence) {
375 checkNotNull(sequence);
376
377 return new Iterable<String>() {
378 @Override public Iterator<String> iterator() {
379 return spliterator(sequence);
380 }
381 @Override public String toString() {
382 return Joiner.on(", ")
383 .appendTo(new StringBuilder().append('['), this)
384 .append(']')
385 .toString();
386 }
387 };
388 }
389
390 private Iterator<String> spliterator(CharSequence sequence) {
391 return strategy.iterator(this, sequence);
392 }
393
394 /**
395 * Returns a {@code MapSplitter} which splits entries based on this splitter,
396 * and splits entries into keys and values using the specified separator.
397 *
398 * @since 10.0
399 */
400 @CheckReturnValue
401 @Beta
402 public MapSplitter withKeyValueSeparator(String separator) {
403 return withKeyValueSeparator(on(separator));
404 }
405
406 /**
407 * Returns a {@code MapSplitter} which splits entries based on this splitter,
408 * and splits entries into keys and values using the specified key-value
409 * splitter.
410 *
411 * @since 10.0
412 */
413 @CheckReturnValue
414 @Beta
415 public MapSplitter withKeyValueSeparator(Splitter keyValueSplitter) {
416 return new MapSplitter(this, keyValueSplitter);
417 }
418
419 /**
420 * An object that splits strings into maps as {@code Splitter} splits
421 * iterables and lists. Like {@code Splitter}, it is thread-safe and
422 * immutable.
423 *
424 * @since 10.0
425 */
426 @Beta
427 public static final class MapSplitter {
428 private static final String INVALID_ENTRY_MESSAGE =
429 "Chunk [%s] is not a valid entry";
430 private final Splitter outerSplitter;
431 private final Splitter entrySplitter;
432
433 private MapSplitter(Splitter outerSplitter, Splitter entrySplitter) {
434 this.outerSplitter = outerSplitter; // only "this" is passed
435 this.entrySplitter = checkNotNull(entrySplitter);
436 }
437
438 /**
439 * Splits {@code sequence} into substrings, splits each substring into
440 * an entry, and returns an unmodifiable map with each of the entries. For
441 * example, <code>
442 * Splitter.on(';').trimResults().withKeyValueSeparator("=>")
443 * .split("a=>b ; c=>b")
444 * </code> will return a mapping from {@code "a"} to {@code "b"} and
445 * {@code "c"} to {@code b}.
446 *
447 * <p>The returned map preserves the order of the entries from
448 * {@code sequence}.
449 *
450 * @throws IllegalArgumentException if the specified sequence does not split
451 * into valid map entries, or if there are duplicate keys
452 */
453 public Map<String, String> split(CharSequence sequence) {
454 Map<String, String> map = new LinkedHashMap<String, String>();
455 for (String entry : outerSplitter.split(sequence)) {
456 Iterator<String> entryFields = entrySplitter.spliterator(entry);
457
458 checkArgument(entryFields.hasNext(), INVALID_ENTRY_MESSAGE, entry);
459 String key = entryFields.next();
460 checkArgument(!map.containsKey(key), "Duplicate key [%s] found.", key);
461
462 checkArgument(entryFields.hasNext(), INVALID_ENTRY_MESSAGE, entry);
463 String value = entryFields.next();
464 map.put(key, value);
465
466 checkArgument(!entryFields.hasNext(), INVALID_ENTRY_MESSAGE, entry);
467 }
468 return Collections.unmodifiableMap(map);
469 }
470 }
471
472 private interface Strategy {
473 Iterator<String> iterator(Splitter splitter, CharSequence toSplit);
474 }
475
476 private abstract static class SplittingIterator extends AbstractIterator<String> {
477 final CharSequence toSplit;
478 final CharMatcher trimmer;
479 final boolean omitEmptyStrings;
480
481 /**
482 * Returns the first index in {@code toSplit} at or after {@code start}
483 * that contains the separator.
484 */
485 abstract int separatorStart(int start);
486
487 /**
488 * Returns the first index in {@code toSplit} after {@code
489 * separatorPosition} that does not contain a separator. This method is only
490 * invoked after a call to {@code separatorStart}.
491 */
492 abstract int separatorEnd(int separatorPosition);
493
494 int offset = 0;
495 int limit;
496
497 protected SplittingIterator(Splitter splitter, CharSequence toSplit) {
498 this.trimmer = splitter.trimmer;
499 this.omitEmptyStrings = splitter.omitEmptyStrings;
500 this.limit = splitter.limit;
501 this.toSplit = toSplit;
502 }
503
504 @Override protected String computeNext() {
505 /*
506 * The returned string will be from the end of the last match to the
507 * beginning of the next one. nextStart is the start position of the
508 * returned substring, while offset is the place to start looking for a
509 * separator.
510 */
511 int nextStart = offset;
512 while (offset != -1) {
513 int start = nextStart;
514 int end;
515
516 int separatorPosition = separatorStart(offset);
517 if (separatorPosition == -1) {
518 end = toSplit.length();
519 offset = -1;
520 } else {
521 end = separatorPosition;
522 offset = separatorEnd(separatorPosition);
523 }
524 if (offset == nextStart) {
525 /*
526 * This occurs when some pattern has an empty match, even if it
527 * doesn't match the empty string -- for example, if it requires
528 * lookahead or the like. The offset must be increased to look for
529 * separators beyond this point, without changing the start position
530 * of the next returned substring -- so nextStart stays the same.
531 */
532 offset++;
533 if (offset >= toSplit.length()) {
534 offset = -1;
535 }
536 continue;
537 }
538
539 while (start < end && trimmer.matches(toSplit.charAt(start))) {
540 start++;
541 }
542 while (end > start && trimmer.matches(toSplit.charAt(end - 1))) {
543 end--;
544 }
545
546 if (omitEmptyStrings && start == end) {
547 // Don't include the (unused) separator in next split string.
548 nextStart = offset;
549 continue;
550 }
551
552 if (limit == 1) {
553 // The limit has been reached, return the rest of the string as the
554 // final item. This is tested after empty string removal so that
555 // empty strings do not count towards the limit.
556 end = toSplit.length();
557 offset = -1;
558 // Since we may have changed the end, we need to trim it again.
559 while (end > start && trimmer.matches(toSplit.charAt(end - 1))) {
560 end--;
561 }
562 } else {
563 limit--;
564 }
565
566 return toSplit.subSequence(start, end).toString();
567 }
568 return endOfData();
569 }
570 }
571 }