1 |
/* |
2 |
* Written by Doug Lea with assistance from members of JCP JSR-166 |
3 |
* Expert Group and released to the public domain, as explained at |
4 |
* http://creativecommons.org/publicdomain/zero/1.0/ |
5 |
*/ |
6 |
|
7 |
package java.util.concurrent; |
8 |
import java.util.*; |
9 |
|
10 |
/** |
11 |
* A scalable concurrent {@link ConcurrentNavigableMap} implementation. |
12 |
* The map is sorted according to the {@linkplain Comparable natural |
13 |
* ordering} of its keys, or by a {@link Comparator} provided at map |
14 |
* creation time, depending on which constructor is used. |
15 |
* |
16 |
* <p>This class implements a concurrent variant of <a |
17 |
* href="http://en.wikipedia.org/wiki/Skip_list" target="_top">SkipLists</a> |
18 |
* providing expected average <i>log(n)</i> time cost for the |
19 |
* {@code containsKey}, {@code get}, {@code put} and |
20 |
* {@code remove} operations and their variants. Insertion, removal, |
21 |
* update, and access operations safely execute concurrently by |
22 |
* multiple threads. Iterators are <i>weakly consistent</i>, returning |
23 |
* elements reflecting the state of the map at some point at or since |
24 |
* the creation of the iterator. They do <em>not</em> throw {@link |
25 |
* ConcurrentModificationException}, and may proceed concurrently with |
26 |
* other operations. Ascending key ordered views and their iterators |
27 |
* are faster than descending ones. |
28 |
* |
29 |
* <p>All {@code Map.Entry} pairs returned by methods in this class |
30 |
* and its views represent snapshots of mappings at the time they were |
31 |
* produced. They do <em>not</em> support the {@code Entry.setValue} |
32 |
* method. (Note however that it is possible to change mappings in the |
33 |
* associated map using {@code put}, {@code putIfAbsent}, or |
34 |
* {@code replace}, depending on exactly which effect you need.) |
35 |
* |
36 |
* <p>Beware that, unlike in most collections, the {@code size} |
37 |
* method is <em>not</em> a constant-time operation. Because of the |
38 |
* asynchronous nature of these maps, determining the current number |
39 |
* of elements requires a traversal of the elements, and so may report |
40 |
* inaccurate results if this collection is modified during traversal. |
41 |
* Additionally, the bulk operations {@code putAll}, {@code equals}, |
42 |
* {@code toArray}, {@code containsValue}, and {@code clear} are |
43 |
* <em>not</em> guaranteed to be performed atomically. For example, an |
44 |
* iterator operating concurrently with a {@code putAll} operation |
45 |
* might view only some of the added elements. |
46 |
* |
47 |
* <p>This class and its views and iterators implement all of the |
48 |
* <em>optional</em> methods of the {@link Map} and {@link Iterator} |
49 |
* interfaces. Like most other concurrent collections, this class does |
50 |
* <em>not</em> permit the use of {@code null} keys or values because some |
51 |
* null return values cannot be reliably distinguished from the absence of |
52 |
* elements. |
53 |
* |
54 |
* <p>This class is a member of the |
55 |
* <a href="{@docRoot}/../technotes/guides/collections/index.html"> |
56 |
* Java Collections Framework</a>. |
57 |
* |
58 |
* @author Doug Lea |
59 |
* @param <K> the type of keys maintained by this map |
60 |
* @param <V> the type of mapped values |
61 |
* @since 1.6 |
62 |
*/ |
63 |
@SuppressWarnings("unchecked") |
64 |
public class ConcurrentSkipListMap<K,V> extends AbstractMap<K,V> |
65 |
implements ConcurrentNavigableMap<K,V>, |
66 |
Cloneable, |
67 |
java.io.Serializable { |
68 |
/* |
69 |
* This class implements a tree-like two-dimensionally linked skip |
70 |
* list in which the index levels are represented in separate |
71 |
* nodes from the base nodes holding data. There are two reasons |
72 |
* for taking this approach instead of the usual array-based |
73 |
* structure: 1) Array based implementations seem to encounter |
74 |
* more complexity and overhead 2) We can use cheaper algorithms |
75 |
* for the heavily-traversed index lists than can be used for the |
76 |
* base lists. Here's a picture of some of the basics for a |
77 |
* possible list with 2 levels of index: |
78 |
* |
79 |
* Head nodes Index nodes |
80 |
* +-+ right +-+ +-+ |
81 |
* |2|---------------->| |--------------------->| |->null |
82 |
* +-+ +-+ +-+ |
83 |
* | down | | |
84 |
* v v v |
85 |
* +-+ +-+ +-+ +-+ +-+ +-+ |
86 |
* |1|----------->| |->| |------>| |----------->| |------>| |->null |
87 |
* +-+ +-+ +-+ +-+ +-+ +-+ |
88 |
* v | | | | | |
89 |
* Nodes next v v v v v |
90 |
* +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ |
91 |
* | |->|A|->|B|->|C|->|D|->|E|->|F|->|G|->|H|->|I|->|J|->|K|->null |
92 |
* +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ |
93 |
* |
94 |
* The base lists use a variant of the HM linked ordered set |
95 |
* algorithm. See Tim Harris, "A pragmatic implementation of |
96 |
* non-blocking linked lists" |
97 |
* http://www.cl.cam.ac.uk/~tlh20/publications.html and Maged |
98 |
* Michael "High Performance Dynamic Lock-Free Hash Tables and |
99 |
* List-Based Sets" |
100 |
* http://www.research.ibm.com/people/m/michael/pubs.htm. The |
101 |
* basic idea in these lists is to mark the "next" pointers of |
102 |
* deleted nodes when deleting to avoid conflicts with concurrent |
103 |
* insertions, and when traversing to keep track of triples |
104 |
* (predecessor, node, successor) in order to detect when and how |
105 |
* to unlink these deleted nodes. |
106 |
* |
107 |
* Rather than using mark-bits to mark list deletions (which can |
108 |
* be slow and space-intensive using AtomicMarkedReference), nodes |
109 |
* use direct CAS'able next pointers. On deletion, instead of |
110 |
* marking a pointer, they splice in another node that can be |
111 |
* thought of as standing for a marked pointer (indicating this by |
112 |
* using otherwise impossible field values). Using plain nodes |
113 |
* acts roughly like "boxed" implementations of marked pointers, |
114 |
* but uses new nodes only when nodes are deleted, not for every |
115 |
* link. This requires less space and supports faster |
116 |
* traversal. Even if marked references were better supported by |
117 |
* JVMs, traversal using this technique might still be faster |
118 |
* because any search need only read ahead one more node than |
119 |
* otherwise required (to check for trailing marker) rather than |
120 |
* unmasking mark bits or whatever on each read. |
121 |
* |
122 |
* This approach maintains the essential property needed in the HM |
123 |
* algorithm of changing the next-pointer of a deleted node so |
124 |
* that any other CAS of it will fail, but implements the idea by |
125 |
* changing the pointer to point to a different node, not by |
126 |
* marking it. While it would be possible to further squeeze |
127 |
* space by defining marker nodes not to have key/value fields, it |
128 |
* isn't worth the extra type-testing overhead. The deletion |
129 |
* markers are rarely encountered during traversal and are |
130 |
* normally quickly garbage collected. (Note that this technique |
131 |
* would not work well in systems without garbage collection.) |
132 |
* |
133 |
* In addition to using deletion markers, the lists also use |
134 |
* nullness of value fields to indicate deletion, in a style |
135 |
* similar to typical lazy-deletion schemes. If a node's value is |
136 |
* null, then it is considered logically deleted and ignored even |
137 |
* though it is still reachable. This maintains proper control of |
138 |
* concurrent replace vs delete operations -- an attempted replace |
139 |
* must fail if a delete beat it by nulling field, and a delete |
140 |
* must return the last non-null value held in the field. (Note: |
141 |
* Null, rather than some special marker, is used for value fields |
142 |
* here because it just so happens to mesh with the Map API |
143 |
* requirement that method get returns null if there is no |
144 |
* mapping, which allows nodes to remain concurrently readable |
145 |
* even when deleted. Using any other marker value here would be |
146 |
* messy at best.) |
147 |
* |
148 |
* Here's the sequence of events for a deletion of node n with |
149 |
* predecessor b and successor f, initially: |
150 |
* |
151 |
* +------+ +------+ +------+ |
152 |
* ... | b |------>| n |----->| f | ... |
153 |
* +------+ +------+ +------+ |
154 |
* |
155 |
* 1. CAS n's value field from non-null to null. |
156 |
* From this point on, no public operations encountering |
157 |
* the node consider this mapping to exist. However, other |
158 |
* ongoing insertions and deletions might still modify |
159 |
* n's next pointer. |
160 |
* |
161 |
* 2. CAS n's next pointer to point to a new marker node. |
162 |
* From this point on, no other nodes can be appended to n. |
163 |
* which avoids deletion errors in CAS-based linked lists. |
164 |
* |
165 |
* +------+ +------+ +------+ +------+ |
166 |
* ... | b |------>| n |----->|marker|------>| f | ... |
167 |
* +------+ +------+ +------+ +------+ |
168 |
* |
169 |
* 3. CAS b's next pointer over both n and its marker. |
170 |
* From this point on, no new traversals will encounter n, |
171 |
* and it can eventually be GCed. |
172 |
* +------+ +------+ |
173 |
* ... | b |----------------------------------->| f | ... |
174 |
* +------+ +------+ |
175 |
* |
176 |
* A failure at step 1 leads to simple retry due to a lost race |
177 |
* with another operation. Steps 2-3 can fail because some other |
178 |
* thread noticed during a traversal a node with null value and |
179 |
* helped out by marking and/or unlinking. This helping-out |
180 |
* ensures that no thread can become stuck waiting for progress of |
181 |
* the deleting thread. The use of marker nodes slightly |
182 |
* complicates helping-out code because traversals must track |
183 |
* consistent reads of up to four nodes (b, n, marker, f), not |
184 |
* just (b, n, f), although the next field of a marker is |
185 |
* immutable, and once a next field is CAS'ed to point to a |
186 |
* marker, it never again changes, so this requires less care. |
187 |
* |
188 |
* Skip lists add indexing to this scheme, so that the base-level |
189 |
* traversals start close to the locations being found, inserted |
190 |
* or deleted -- usually base level traversals only traverse a few |
191 |
* nodes. This doesn't change the basic algorithm except for the |
192 |
* need to make sure base traversals start at predecessors (here, |
193 |
* b) that are not (structurally) deleted, otherwise retrying |
194 |
* after processing the deletion. |
195 |
* |
196 |
* Index levels are maintained as lists with volatile next fields, |
197 |
* using CAS to link and unlink. Races are allowed in index-list |
198 |
* operations that can (rarely) fail to link in a new index node |
199 |
* or delete one. (We can't do this of course for data nodes.) |
200 |
* However, even when this happens, the index lists remain sorted, |
201 |
* so correctly serve as indices. This can impact performance, |
202 |
* but since skip lists are probabilistic anyway, the net result |
203 |
* is that under contention, the effective "p" value may be lower |
204 |
* than its nominal value. And race windows are kept small enough |
205 |
* that in practice these failures are rare, even under a lot of |
206 |
* contention. |
207 |
* |
208 |
* The fact that retries (for both base and index lists) are |
209 |
* relatively cheap due to indexing allows some minor |
210 |
* simplifications of retry logic. Traversal restarts are |
211 |
* performed after most "helping-out" CASes. This isn't always |
212 |
* strictly necessary, but the implicit backoffs tend to help |
213 |
* reduce other downstream failed CAS's enough to outweigh restart |
214 |
* cost. This worsens the worst case, but seems to improve even |
215 |
* highly contended cases. |
216 |
* |
217 |
* Unlike most skip-list implementations, index insertion and |
218 |
* deletion here require a separate traversal pass occuring after |
219 |
* the base-level action, to add or remove index nodes. This adds |
220 |
* to single-threaded overhead, but improves contended |
221 |
* multithreaded performance by narrowing interference windows, |
222 |
* and allows deletion to ensure that all index nodes will be made |
223 |
* unreachable upon return from a public remove operation, thus |
224 |
* avoiding unwanted garbage retention. This is more important |
225 |
* here than in some other data structures because we cannot null |
226 |
* out node fields referencing user keys since they might still be |
227 |
* read by other ongoing traversals. |
228 |
* |
229 |
* Indexing uses skip list parameters that maintain good search |
230 |
* performance while using sparser-than-usual indices: The |
231 |
* hardwired parameters k=1, p=0.5 (see method randomLevel) mean |
232 |
* that about one-quarter of the nodes have indices. Of those that |
233 |
* do, half have one level, a quarter have two, and so on (see |
234 |
* Pugh's Skip List Cookbook, sec 3.4). The expected total space |
235 |
* requirement for a map is slightly less than for the current |
236 |
* implementation of java.util.TreeMap. |
237 |
* |
238 |
* Changing the level of the index (i.e, the height of the |
239 |
* tree-like structure) also uses CAS. The head index has initial |
240 |
* level/height of one. Creation of an index with height greater |
241 |
* than the current level adds a level to the head index by |
242 |
* CAS'ing on a new top-most head. To maintain good performance |
243 |
* after a lot of removals, deletion methods heuristically try to |
244 |
* reduce the height if the topmost levels appear to be empty. |
245 |
* This may encounter races in which it possible (but rare) to |
246 |
* reduce and "lose" a level just as it is about to contain an |
247 |
* index (that will then never be encountered). This does no |
248 |
* structural harm, and in practice appears to be a better option |
249 |
* than allowing unrestrained growth of levels. |
250 |
* |
251 |
* The code for all this is more verbose than you'd like. Most |
252 |
* operations entail locating an element (or position to insert an |
253 |
* element). The code to do this can't be nicely factored out |
254 |
* because subsequent uses require a snapshot of predecessor |
255 |
* and/or successor and/or value fields which can't be returned |
256 |
* all at once, at least not without creating yet another object |
257 |
* to hold them -- creating such little objects is an especially |
258 |
* bad idea for basic internal search operations because it adds |
259 |
* to GC overhead. (This is one of the few times I've wished Java |
260 |
* had macros.) Instead, some traversal code is interleaved within |
261 |
* insertion and removal operations. The control logic to handle |
262 |
* all the retry conditions is sometimes twisty. Most search is |
263 |
* broken into 2 parts. findPredecessor() searches index nodes |
264 |
* only, returning a base-level predecessor of the key. findNode() |
265 |
* finishes out the base-level search. Even with this factoring, |
266 |
* there is a fair amount of near-duplication of code to handle |
267 |
* variants. |
268 |
* |
269 |
* For explanation of algorithms sharing at least a couple of |
270 |
* features with this one, see Mikhail Fomitchev's thesis |
271 |
* (http://www.cs.yorku.ca/~mikhail/), Keir Fraser's thesis |
272 |
* (http://www.cl.cam.ac.uk/users/kaf24/), and Hakan Sundell's |
273 |
* thesis (http://www.cs.chalmers.se/~phs/). |
274 |
* |
275 |
* Given the use of tree-like index nodes, you might wonder why |
276 |
* this doesn't use some kind of search tree instead, which would |
277 |
* support somewhat faster search operations. The reason is that |
278 |
* there are no known efficient lock-free insertion and deletion |
279 |
* algorithms for search trees. The immutability of the "down" |
280 |
* links of index nodes (as opposed to mutable "left" fields in |
281 |
* true trees) makes this tractable using only CAS operations. |
282 |
* |
283 |
* Notation guide for local variables |
284 |
* Node: b, n, f for predecessor, node, successor |
285 |
* Index: q, r, d for index node, right, down. |
286 |
* t for another index node |
287 |
* Head: h |
288 |
* Levels: j |
289 |
* Keys: k, key |
290 |
* Values: v, value |
291 |
* Comparisons: c |
292 |
*/ |
293 |
|
294 |
private static final long serialVersionUID = -8627078645895051609L; |
295 |
|
296 |
/** |
297 |
* Generates the initial random seed for the cheaper per-instance |
298 |
* random number generators used in randomLevel. |
299 |
*/ |
300 |
private static final Random seedGenerator = new Random(); |
301 |
|
302 |
/** |
303 |
* Special value used to identify base-level header |
304 |
*/ |
305 |
private static final Object BASE_HEADER = new Object(); |
306 |
|
307 |
/** |
308 |
* The topmost head index of the skiplist. |
309 |
*/ |
310 |
private transient volatile HeadIndex<K,V> head; |
311 |
|
312 |
/** |
313 |
* The comparator used to maintain order in this map, or null |
314 |
* if using natural ordering. |
315 |
* @serial |
316 |
*/ |
317 |
private final Comparator<? super K> comparator; |
318 |
|
319 |
/** |
320 |
* Seed for simple random number generator. Not volatile since it |
321 |
* doesn't matter too much if different threads don't see updates. |
322 |
*/ |
323 |
private transient int randomSeed; |
324 |
|
325 |
/** Lazily initialized key set */ |
326 |
private transient KeySet<K> keySet; |
327 |
/** Lazily initialized entry set */ |
328 |
private transient EntrySet<K,V> entrySet; |
329 |
/** Lazily initialized values collection */ |
330 |
private transient Values<V> values; |
331 |
/** Lazily initialized descending key set */ |
332 |
private transient ConcurrentNavigableMap<K,V> descendingMap; |
333 |
|
334 |
/** |
335 |
* Initializes or resets state. Needed by constructors, clone, |
336 |
* clear, readObject. and ConcurrentSkipListSet.clone. |
337 |
* (Note that comparator must be separately initialized.) |
338 |
*/ |
339 |
final void initialize() { |
340 |
keySet = null; |
341 |
entrySet = null; |
342 |
values = null; |
343 |
descendingMap = null; |
344 |
randomSeed = seedGenerator.nextInt() | 0x0100; // ensure nonzero |
345 |
head = new HeadIndex<K,V>(new Node<K,V>(null, BASE_HEADER, null), |
346 |
null, null, 1); |
347 |
} |
348 |
|
349 |
/** |
350 |
* compareAndSet head node |
351 |
*/ |
352 |
private boolean casHead(HeadIndex<K,V> cmp, HeadIndex<K,V> val) { |
353 |
return UNSAFE.compareAndSwapObject(this, headOffset, cmp, val); |
354 |
} |
355 |
|
356 |
/* ---------------- Nodes -------------- */ |
357 |
|
358 |
/** |
359 |
* Nodes hold keys and values, and are singly linked in sorted |
360 |
* order, possibly with some intervening marker nodes. The list is |
361 |
* headed by a dummy node accessible as head.node. The value field |
362 |
* is declared only as Object because it takes special non-V |
363 |
* values for marker and header nodes. |
364 |
*/ |
365 |
static final class Node<K,V> { |
366 |
final K key; |
367 |
volatile Object value; |
368 |
volatile Node<K,V> next; |
369 |
|
370 |
/** |
371 |
* Creates a new regular node. |
372 |
*/ |
373 |
Node(K key, Object value, Node<K,V> next) { |
374 |
this.key = key; |
375 |
this.value = value; |
376 |
this.next = next; |
377 |
} |
378 |
|
379 |
/** |
380 |
* Creates a new marker node. A marker is distinguished by |
381 |
* having its value field point to itself. Marker nodes also |
382 |
* have null keys, a fact that is exploited in a few places, |
383 |
* but this doesn't distinguish markers from the base-level |
384 |
* header node (head.node), which also has a null key. |
385 |
*/ |
386 |
Node(Node<K,V> next) { |
387 |
this.key = null; |
388 |
this.value = this; |
389 |
this.next = next; |
390 |
} |
391 |
|
392 |
/** |
393 |
* compareAndSet value field |
394 |
*/ |
395 |
boolean casValue(Object cmp, Object val) { |
396 |
return UNSAFE.compareAndSwapObject(this, valueOffset, cmp, val); |
397 |
} |
398 |
|
399 |
/** |
400 |
* compareAndSet next field |
401 |
*/ |
402 |
boolean casNext(Node<K,V> cmp, Node<K,V> val) { |
403 |
return UNSAFE.compareAndSwapObject(this, nextOffset, cmp, val); |
404 |
} |
405 |
|
406 |
/** |
407 |
* Returns true if this node is a marker. This method isn't |
408 |
* actually called in any current code checking for markers |
409 |
* because callers will have already read value field and need |
410 |
* to use that read (not another done here) and so directly |
411 |
* test if value points to node. |
412 |
* |
413 |
* @return true if this node is a marker node |
414 |
*/ |
415 |
boolean isMarker() { |
416 |
return value == this; |
417 |
} |
418 |
|
419 |
/** |
420 |
* Returns true if this node is the header of base-level list. |
421 |
* @return true if this node is header node |
422 |
*/ |
423 |
boolean isBaseHeader() { |
424 |
return value == BASE_HEADER; |
425 |
} |
426 |
|
427 |
/** |
428 |
* Tries to append a deletion marker to this node. |
429 |
* @param f the assumed current successor of this node |
430 |
* @return true if successful |
431 |
*/ |
432 |
boolean appendMarker(Node<K,V> f) { |
433 |
return casNext(f, new Node<K,V>(f)); |
434 |
} |
435 |
|
436 |
/** |
437 |
* Helps out a deletion by appending marker or unlinking from |
438 |
* predecessor. This is called during traversals when value |
439 |
* field seen to be null. |
440 |
* @param b predecessor |
441 |
* @param f successor |
442 |
*/ |
443 |
void helpDelete(Node<K,V> b, Node<K,V> f) { |
444 |
/* |
445 |
* Rechecking links and then doing only one of the |
446 |
* help-out stages per call tends to minimize CAS |
447 |
* interference among helping threads. |
448 |
*/ |
449 |
if (f == next && this == b.next) { |
450 |
if (f == null || f.value != f) // not already marked |
451 |
appendMarker(f); |
452 |
else |
453 |
b.casNext(this, f.next); |
454 |
} |
455 |
} |
456 |
|
457 |
/** |
458 |
* Returns value if this node contains a valid key-value pair, |
459 |
* else null. |
460 |
* @return this node's value if it isn't a marker or header or |
461 |
* is deleted, else null |
462 |
*/ |
463 |
V getValidValue() { |
464 |
Object v = value; |
465 |
if (v == this || v == BASE_HEADER) |
466 |
return null; |
467 |
return (V)v; |
468 |
} |
469 |
|
470 |
/** |
471 |
* Creates and returns a new SimpleImmutableEntry holding current |
472 |
* mapping if this node holds a valid value, else null. |
473 |
* @return new entry or null |
474 |
*/ |
475 |
AbstractMap.SimpleImmutableEntry<K,V> createSnapshot() { |
476 |
V v = getValidValue(); |
477 |
if (v == null) |
478 |
return null; |
479 |
return new AbstractMap.SimpleImmutableEntry<K,V>(key, v); |
480 |
} |
481 |
|
482 |
// UNSAFE mechanics |
483 |
|
484 |
private static final sun.misc.Unsafe UNSAFE; |
485 |
private static final long valueOffset; |
486 |
private static final long nextOffset; |
487 |
|
488 |
static { |
489 |
try { |
490 |
UNSAFE = sun.misc.Unsafe.getUnsafe(); |
491 |
Class<?> k = Node.class; |
492 |
valueOffset = UNSAFE.objectFieldOffset |
493 |
(k.getDeclaredField("value")); |
494 |
nextOffset = UNSAFE.objectFieldOffset |
495 |
(k.getDeclaredField("next")); |
496 |
} catch (Exception e) { |
497 |
throw new Error(e); |
498 |
} |
499 |
} |
500 |
} |
501 |
|
502 |
/* ---------------- Indexing -------------- */ |
503 |
|
504 |
/** |
505 |
* Index nodes represent the levels of the skip list. Note that |
506 |
* even though both Nodes and Indexes have forward-pointing |
507 |
* fields, they have different types and are handled in different |
508 |
* ways, that can't nicely be captured by placing field in a |
509 |
* shared abstract class. |
510 |
*/ |
511 |
static class Index<K,V> { |
512 |
final Node<K,V> node; |
513 |
final Index<K,V> down; |
514 |
volatile Index<K,V> right; |
515 |
|
516 |
/** |
517 |
* Creates index node with given values. |
518 |
*/ |
519 |
Index(Node<K,V> node, Index<K,V> down, Index<K,V> right) { |
520 |
this.node = node; |
521 |
this.down = down; |
522 |
this.right = right; |
523 |
} |
524 |
|
525 |
/** |
526 |
* compareAndSet right field |
527 |
*/ |
528 |
final boolean casRight(Index<K,V> cmp, Index<K,V> val) { |
529 |
return UNSAFE.compareAndSwapObject(this, rightOffset, cmp, val); |
530 |
} |
531 |
|
532 |
/** |
533 |
* Returns true if the node this indexes has been deleted. |
534 |
* @return true if indexed node is known to be deleted |
535 |
*/ |
536 |
final boolean indexesDeletedNode() { |
537 |
return node.value == null; |
538 |
} |
539 |
|
540 |
/** |
541 |
* Tries to CAS newSucc as successor. To minimize races with |
542 |
* unlink that may lose this index node, if the node being |
543 |
* indexed is known to be deleted, it doesn't try to link in. |
544 |
* @param succ the expected current successor |
545 |
* @param newSucc the new successor |
546 |
* @return true if successful |
547 |
*/ |
548 |
final boolean link(Index<K,V> succ, Index<K,V> newSucc) { |
549 |
Node<K,V> n = node; |
550 |
newSucc.right = succ; |
551 |
return n.value != null && casRight(succ, newSucc); |
552 |
} |
553 |
|
554 |
/** |
555 |
* Tries to CAS right field to skip over apparent successor |
556 |
* succ. Fails (forcing a retraversal by caller) if this node |
557 |
* is known to be deleted. |
558 |
* @param succ the expected current successor |
559 |
* @return true if successful |
560 |
*/ |
561 |
final boolean unlink(Index<K,V> succ) { |
562 |
return !indexesDeletedNode() && casRight(succ, succ.right); |
563 |
} |
564 |
|
565 |
// Unsafe mechanics |
566 |
private static final sun.misc.Unsafe UNSAFE; |
567 |
private static final long rightOffset; |
568 |
static { |
569 |
try { |
570 |
UNSAFE = sun.misc.Unsafe.getUnsafe(); |
571 |
Class<?> k = Index.class; |
572 |
rightOffset = UNSAFE.objectFieldOffset |
573 |
(k.getDeclaredField("right")); |
574 |
} catch (Exception e) { |
575 |
throw new Error(e); |
576 |
} |
577 |
} |
578 |
} |
579 |
|
580 |
/* ---------------- Head nodes -------------- */ |
581 |
|
582 |
/** |
583 |
* Nodes heading each level keep track of their level. |
584 |
*/ |
585 |
static final class HeadIndex<K,V> extends Index<K,V> { |
586 |
final int level; |
587 |
HeadIndex(Node<K,V> node, Index<K,V> down, Index<K,V> right, int level) { |
588 |
super(node, down, right); |
589 |
this.level = level; |
590 |
} |
591 |
} |
592 |
|
593 |
/* ---------------- Comparison utilities -------------- */ |
594 |
|
595 |
/** |
596 |
* Represents a key with a comparator as a Comparable. |
597 |
* |
598 |
* Because most sorted collections seem to use natural ordering on |
599 |
* Comparables (Strings, Integers, etc), most internal methods are |
600 |
* geared to use them. This is generally faster than checking |
601 |
* per-comparison whether to use comparator or comparable because |
602 |
* it doesn't require a (Comparable) cast for each comparison. |
603 |
* (Optimizers can only sometimes remove such redundant checks |
604 |
* themselves.) When Comparators are used, |
605 |
* ComparableUsingComparators are created so that they act in the |
606 |
* same way as natural orderings. This penalizes use of |
607 |
* Comparators vs Comparables, which seems like the right |
608 |
* tradeoff. |
609 |
*/ |
610 |
static final class ComparableUsingComparator<K> implements Comparable<K> { |
611 |
final K actualKey; |
612 |
final Comparator<? super K> cmp; |
613 |
ComparableUsingComparator(K key, Comparator<? super K> cmp) { |
614 |
this.actualKey = key; |
615 |
this.cmp = cmp; |
616 |
} |
617 |
public int compareTo(K k2) { |
618 |
return cmp.compare(actualKey, k2); |
619 |
} |
620 |
} |
621 |
|
622 |
/** |
623 |
* If using comparator, return a ComparableUsingComparator, else |
624 |
* cast key as Comparable, which may cause ClassCastException, |
625 |
* which is propagated back to caller. |
626 |
*/ |
627 |
private Comparable<? super K> comparable(Object key) |
628 |
throws ClassCastException { |
629 |
if (key == null) |
630 |
throw new NullPointerException(); |
631 |
if (comparator != null) |
632 |
return new ComparableUsingComparator<K>((K)key, comparator); |
633 |
else |
634 |
return (Comparable<? super K>)key; |
635 |
} |
636 |
|
637 |
/** |
638 |
* Compares using comparator or natural ordering. Used when the |
639 |
* ComparableUsingComparator approach doesn't apply. |
640 |
*/ |
641 |
int compare(K k1, K k2) throws ClassCastException { |
642 |
Comparator<? super K> cmp = comparator; |
643 |
if (cmp != null) |
644 |
return cmp.compare(k1, k2); |
645 |
else |
646 |
return ((Comparable<? super K>)k1).compareTo(k2); |
647 |
} |
648 |
|
649 |
/** |
650 |
* Returns true if given key greater than or equal to least and |
651 |
* strictly less than fence, bypassing either test if least or |
652 |
* fence are null. Needed mainly in submap operations. |
653 |
*/ |
654 |
boolean inHalfOpenRange(K key, K least, K fence) { |
655 |
if (key == null) |
656 |
throw new NullPointerException(); |
657 |
return ((least == null || compare(key, least) >= 0) && |
658 |
(fence == null || compare(key, fence) < 0)); |
659 |
} |
660 |
|
661 |
/** |
662 |
* Returns true if given key greater than or equal to least and less |
663 |
* or equal to fence. Needed mainly in submap operations. |
664 |
*/ |
665 |
boolean inOpenRange(K key, K least, K fence) { |
666 |
if (key == null) |
667 |
throw new NullPointerException(); |
668 |
return ((least == null || compare(key, least) >= 0) && |
669 |
(fence == null || compare(key, fence) <= 0)); |
670 |
} |
671 |
|
672 |
/* ---------------- Traversal -------------- */ |
673 |
|
674 |
/** |
675 |
* Returns a base-level node with key strictly less than given key, |
676 |
* or the base-level header if there is no such node. Also |
677 |
* unlinks indexes to deleted nodes found along the way. Callers |
678 |
* rely on this side-effect of clearing indices to deleted nodes. |
679 |
* @param key the key |
680 |
* @return a predecessor of key |
681 |
*/ |
682 |
private Node<K,V> findPredecessor(Comparable<? super K> key) { |
683 |
if (key == null) |
684 |
throw new NullPointerException(); // don't postpone errors |
685 |
for (;;) { |
686 |
Index<K,V> q = head; |
687 |
Index<K,V> r = q.right; |
688 |
for (;;) { |
689 |
if (r != null) { |
690 |
Node<K,V> n = r.node; |
691 |
K k = n.key; |
692 |
if (n.value == null) { |
693 |
if (!q.unlink(r)) |
694 |
break; // restart |
695 |
r = q.right; // reread r |
696 |
continue; |
697 |
} |
698 |
if (key.compareTo(k) > 0) { |
699 |
q = r; |
700 |
r = r.right; |
701 |
continue; |
702 |
} |
703 |
} |
704 |
Index<K,V> d = q.down; |
705 |
if (d != null) { |
706 |
q = d; |
707 |
r = d.right; |
708 |
} else |
709 |
return q.node; |
710 |
} |
711 |
} |
712 |
} |
713 |
|
714 |
/** |
715 |
* Returns node holding key or null if no such, clearing out any |
716 |
* deleted nodes seen along the way. Repeatedly traverses at |
717 |
* base-level looking for key starting at predecessor returned |
718 |
* from findPredecessor, processing base-level deletions as |
719 |
* encountered. Some callers rely on this side-effect of clearing |
720 |
* deleted nodes. |
721 |
* |
722 |
* Restarts occur, at traversal step centered on node n, if: |
723 |
* |
724 |
* (1) After reading n's next field, n is no longer assumed |
725 |
* predecessor b's current successor, which means that |
726 |
* we don't have a consistent 3-node snapshot and so cannot |
727 |
* unlink any subsequent deleted nodes encountered. |
728 |
* |
729 |
* (2) n's value field is null, indicating n is deleted, in |
730 |
* which case we help out an ongoing structural deletion |
731 |
* before retrying. Even though there are cases where such |
732 |
* unlinking doesn't require restart, they aren't sorted out |
733 |
* here because doing so would not usually outweigh cost of |
734 |
* restarting. |
735 |
* |
736 |
* (3) n is a marker or n's predecessor's value field is null, |
737 |
* indicating (among other possibilities) that |
738 |
* findPredecessor returned a deleted node. We can't unlink |
739 |
* the node because we don't know its predecessor, so rely |
740 |
* on another call to findPredecessor to notice and return |
741 |
* some earlier predecessor, which it will do. This check is |
742 |
* only strictly needed at beginning of loop, (and the |
743 |
* b.value check isn't strictly needed at all) but is done |
744 |
* each iteration to help avoid contention with other |
745 |
* threads by callers that will fail to be able to change |
746 |
* links, and so will retry anyway. |
747 |
* |
748 |
* The traversal loops in doPut, doRemove, and findNear all |
749 |
* include the same three kinds of checks. And specialized |
750 |
* versions appear in findFirst, and findLast and their |
751 |
* variants. They can't easily share code because each uses the |
752 |
* reads of fields held in locals occurring in the orders they |
753 |
* were performed. |
754 |
* |
755 |
* @param key the key |
756 |
* @return node holding key, or null if no such |
757 |
*/ |
758 |
private Node<K,V> findNode(Comparable<? super K> key) { |
759 |
for (;;) { |
760 |
Node<K,V> b = findPredecessor(key); |
761 |
Node<K,V> n = b.next; |
762 |
for (;;) { |
763 |
if (n == null) |
764 |
return null; |
765 |
Node<K,V> f = n.next; |
766 |
if (n != b.next) // inconsistent read |
767 |
break; |
768 |
Object v = n.value; |
769 |
if (v == null) { // n is deleted |
770 |
n.helpDelete(b, f); |
771 |
break; |
772 |
} |
773 |
if (v == n || b.value == null) // b is deleted |
774 |
break; |
775 |
int c = key.compareTo(n.key); |
776 |
if (c == 0) |
777 |
return n; |
778 |
if (c < 0) |
779 |
return null; |
780 |
b = n; |
781 |
n = f; |
782 |
} |
783 |
} |
784 |
} |
785 |
|
786 |
/** |
787 |
* Gets value for key using findNode. |
788 |
* @param okey the key |
789 |
* @return the value, or null if absent |
790 |
*/ |
791 |
private V doGet(Object okey) { |
792 |
Comparable<? super K> key = comparable(okey); |
793 |
/* |
794 |
* Loop needed here and elsewhere in case value field goes |
795 |
* null just as it is about to be returned, in which case we |
796 |
* lost a race with a deletion, so must retry. |
797 |
*/ |
798 |
for (;;) { |
799 |
Node<K,V> n = findNode(key); |
800 |
if (n == null) |
801 |
return null; |
802 |
Object v = n.value; |
803 |
if (v != null) |
804 |
return (V)v; |
805 |
} |
806 |
} |
807 |
|
808 |
/* ---------------- Insertion -------------- */ |
809 |
|
810 |
/** |
811 |
* Main insertion method. Adds element if not present, or |
812 |
* replaces value if present and onlyIfAbsent is false. |
813 |
* @param kkey the key |
814 |
* @param value the value that must be associated with key |
815 |
* @param onlyIfAbsent if should not insert if already present |
816 |
* @return the old value, or null if newly inserted |
817 |
*/ |
818 |
private V doPut(K kkey, V value, boolean onlyIfAbsent) { |
819 |
Comparable<? super K> key = comparable(kkey); |
820 |
for (;;) { |
821 |
Node<K,V> b = findPredecessor(key); |
822 |
Node<K,V> n = b.next; |
823 |
for (;;) { |
824 |
if (n != null) { |
825 |
Node<K,V> f = n.next; |
826 |
if (n != b.next) // inconsistent read |
827 |
break; |
828 |
Object v = n.value; |
829 |
if (v == null) { // n is deleted |
830 |
n.helpDelete(b, f); |
831 |
break; |
832 |
} |
833 |
if (v == n || b.value == null) // b is deleted |
834 |
break; |
835 |
int c = key.compareTo(n.key); |
836 |
if (c > 0) { |
837 |
b = n; |
838 |
n = f; |
839 |
continue; |
840 |
} |
841 |
if (c == 0) { |
842 |
if (onlyIfAbsent || n.casValue(v, value)) |
843 |
return (V)v; |
844 |
else |
845 |
break; // restart if lost race to replace value |
846 |
} |
847 |
// else c < 0; fall through |
848 |
} |
849 |
|
850 |
Node<K,V> z = new Node<K,V>(kkey, value, n); |
851 |
if (!b.casNext(n, z)) |
852 |
break; // restart if lost race to append to b |
853 |
int level = randomLevel(); |
854 |
if (level > 0) |
855 |
insertIndex(z, level); |
856 |
return null; |
857 |
} |
858 |
} |
859 |
} |
860 |
|
861 |
/** |
862 |
* Returns a random level for inserting a new node. |
863 |
* Hardwired to k=1, p=0.5, max 31 (see above and |
864 |
* Pugh's "Skip List Cookbook", sec 3.4). |
865 |
* |
866 |
* This uses the simplest of the generators described in George |
867 |
* Marsaglia's "Xorshift RNGs" paper. This is not a high-quality |
868 |
* generator but is acceptable here. |
869 |
*/ |
870 |
private int randomLevel() { |
871 |
int x = randomSeed; |
872 |
x ^= x << 13; |
873 |
x ^= x >>> 17; |
874 |
randomSeed = x ^= x << 5; |
875 |
if ((x & 0x80000001) != 0) // test highest and lowest bits |
876 |
return 0; |
877 |
int level = 1; |
878 |
while (((x >>>= 1) & 1) != 0) ++level; |
879 |
return level; |
880 |
} |
881 |
|
882 |
/** |
883 |
* Creates and adds index nodes for the given node. |
884 |
* @param z the node |
885 |
* @param level the level of the index |
886 |
*/ |
887 |
private void insertIndex(Node<K,V> z, int level) { |
888 |
HeadIndex<K,V> h = head; |
889 |
int max = h.level; |
890 |
|
891 |
if (level <= max) { |
892 |
Index<K,V> idx = null; |
893 |
for (int i = 1; i <= level; ++i) |
894 |
idx = new Index<K,V>(z, idx, null); |
895 |
addIndex(idx, h, level); |
896 |
|
897 |
} else { // Add a new level |
898 |
/* |
899 |
* To reduce interference by other threads checking for |
900 |
* empty levels in tryReduceLevel, new levels are added |
901 |
* with initialized right pointers. Which in turn requires |
902 |
* keeping levels in an array to access them while |
903 |
* creating new head index nodes from the opposite |
904 |
* direction. |
905 |
*/ |
906 |
level = max + 1; |
907 |
Index<K,V>[] idxs = (Index<K,V>[])new Index<?,?>[level+1]; |
908 |
Index<K,V> idx = null; |
909 |
for (int i = 1; i <= level; ++i) |
910 |
idxs[i] = idx = new Index<K,V>(z, idx, null); |
911 |
|
912 |
HeadIndex<K,V> oldh; |
913 |
int k; |
914 |
for (;;) { |
915 |
oldh = head; |
916 |
int oldLevel = oldh.level; |
917 |
if (level <= oldLevel) { // lost race to add level |
918 |
k = level; |
919 |
break; |
920 |
} |
921 |
HeadIndex<K,V> newh = oldh; |
922 |
Node<K,V> oldbase = oldh.node; |
923 |
for (int j = oldLevel+1; j <= level; ++j) |
924 |
newh = new HeadIndex<K,V>(oldbase, newh, idxs[j], j); |
925 |
if (casHead(oldh, newh)) { |
926 |
k = oldLevel; |
927 |
break; |
928 |
} |
929 |
} |
930 |
addIndex(idxs[k], oldh, k); |
931 |
} |
932 |
} |
933 |
|
934 |
/** |
935 |
* Adds given index nodes from given level down to 1. |
936 |
* @param idx the topmost index node being inserted |
937 |
* @param h the value of head to use to insert. This must be |
938 |
* snapshotted by callers to provide correct insertion level. |
939 |
* @param indexLevel the level of the index |
940 |
*/ |
941 |
private void addIndex(Index<K,V> idx, HeadIndex<K,V> h, int indexLevel) { |
942 |
// Track next level to insert in case of retries |
943 |
int insertionLevel = indexLevel; |
944 |
Comparable<? super K> key = comparable(idx.node.key); |
945 |
if (key == null) throw new NullPointerException(); |
946 |
|
947 |
// Similar to findPredecessor, but adding index nodes along |
948 |
// path to key. |
949 |
for (;;) { |
950 |
int j = h.level; |
951 |
Index<K,V> q = h; |
952 |
Index<K,V> r = q.right; |
953 |
Index<K,V> t = idx; |
954 |
for (;;) { |
955 |
if (r != null) { |
956 |
Node<K,V> n = r.node; |
957 |
// compare before deletion check avoids needing recheck |
958 |
int c = key.compareTo(n.key); |
959 |
if (n.value == null) { |
960 |
if (!q.unlink(r)) |
961 |
break; |
962 |
r = q.right; |
963 |
continue; |
964 |
} |
965 |
if (c > 0) { |
966 |
q = r; |
967 |
r = r.right; |
968 |
continue; |
969 |
} |
970 |
} |
971 |
|
972 |
if (j == insertionLevel) { |
973 |
// Don't insert index if node already deleted |
974 |
if (t.indexesDeletedNode()) { |
975 |
findNode(key); // cleans up |
976 |
return; |
977 |
} |
978 |
if (!q.link(r, t)) |
979 |
break; // restart |
980 |
if (--insertionLevel == 0) { |
981 |
// need final deletion check before return |
982 |
if (t.indexesDeletedNode()) |
983 |
findNode(key); |
984 |
return; |
985 |
} |
986 |
} |
987 |
|
988 |
if (--j >= insertionLevel && j < indexLevel) |
989 |
t = t.down; |
990 |
q = q.down; |
991 |
r = q.right; |
992 |
} |
993 |
} |
994 |
} |
995 |
|
996 |
/* ---------------- Deletion -------------- */ |
997 |
|
998 |
/** |
999 |
* Main deletion method. Locates node, nulls value, appends a |
1000 |
* deletion marker, unlinks predecessor, removes associated index |
1001 |
* nodes, and possibly reduces head index level. |
1002 |
* |
1003 |
* Index nodes are cleared out simply by calling findPredecessor. |
1004 |
* which unlinks indexes to deleted nodes found along path to key, |
1005 |
* which will include the indexes to this node. This is done |
1006 |
* unconditionally. We can't check beforehand whether there are |
1007 |
* index nodes because it might be the case that some or all |
1008 |
* indexes hadn't been inserted yet for this node during initial |
1009 |
* search for it, and we'd like to ensure lack of garbage |
1010 |
* retention, so must call to be sure. |
1011 |
* |
1012 |
* @param okey the key |
1013 |
* @param value if non-null, the value that must be |
1014 |
* associated with key |
1015 |
* @return the node, or null if not found |
1016 |
*/ |
1017 |
final V doRemove(Object okey, Object value) { |
1018 |
Comparable<? super K> key = comparable(okey); |
1019 |
for (;;) { |
1020 |
Node<K,V> b = findPredecessor(key); |
1021 |
Node<K,V> n = b.next; |
1022 |
for (;;) { |
1023 |
if (n == null) |
1024 |
return null; |
1025 |
Node<K,V> f = n.next; |
1026 |
if (n != b.next) // inconsistent read |
1027 |
break; |
1028 |
Object v = n.value; |
1029 |
if (v == null) { // n is deleted |
1030 |
n.helpDelete(b, f); |
1031 |
break; |
1032 |
} |
1033 |
if (v == n || b.value == null) // b is deleted |
1034 |
break; |
1035 |
int c = key.compareTo(n.key); |
1036 |
if (c < 0) |
1037 |
return null; |
1038 |
if (c > 0) { |
1039 |
b = n; |
1040 |
n = f; |
1041 |
continue; |
1042 |
} |
1043 |
if (value != null && !value.equals(v)) |
1044 |
return null; |
1045 |
if (!n.casValue(v, null)) |
1046 |
break; |
1047 |
if (!n.appendMarker(f) || !b.casNext(n, f)) |
1048 |
findNode(key); // Retry via findNode |
1049 |
else { |
1050 |
findPredecessor(key); // Clean index |
1051 |
if (head.right == null) |
1052 |
tryReduceLevel(); |
1053 |
} |
1054 |
return (V)v; |
1055 |
} |
1056 |
} |
1057 |
} |
1058 |
|
1059 |
/** |
1060 |
* Possibly reduce head level if it has no nodes. This method can |
1061 |
* (rarely) make mistakes, in which case levels can disappear even |
1062 |
* though they are about to contain index nodes. This impacts |
1063 |
* performance, not correctness. To minimize mistakes as well as |
1064 |
* to reduce hysteresis, the level is reduced by one only if the |
1065 |
* topmost three levels look empty. Also, if the removed level |
1066 |
* looks non-empty after CAS, we try to change it back quick |
1067 |
* before anyone notices our mistake! (This trick works pretty |
1068 |
* well because this method will practically never make mistakes |
1069 |
* unless current thread stalls immediately before first CAS, in |
1070 |
* which case it is very unlikely to stall again immediately |
1071 |
* afterwards, so will recover.) |
1072 |
* |
1073 |
* We put up with all this rather than just let levels grow |
1074 |
* because otherwise, even a small map that has undergone a large |
1075 |
* number of insertions and removals will have a lot of levels, |
1076 |
* slowing down access more than would an occasional unwanted |
1077 |
* reduction. |
1078 |
*/ |
1079 |
private void tryReduceLevel() { |
1080 |
HeadIndex<K,V> h = head; |
1081 |
HeadIndex<K,V> d; |
1082 |
HeadIndex<K,V> e; |
1083 |
if (h.level > 3 && |
1084 |
(d = (HeadIndex<K,V>)h.down) != null && |
1085 |
(e = (HeadIndex<K,V>)d.down) != null && |
1086 |
e.right == null && |
1087 |
d.right == null && |
1088 |
h.right == null && |
1089 |
casHead(h, d) && // try to set |
1090 |
h.right != null) // recheck |
1091 |
casHead(d, h); // try to backout |
1092 |
} |
1093 |
|
1094 |
/* ---------------- Finding and removing first element -------------- */ |
1095 |
|
1096 |
/** |
1097 |
* Specialized variant of findNode to get first valid node. |
1098 |
* @return first node or null if empty |
1099 |
*/ |
1100 |
Node<K,V> findFirst() { |
1101 |
for (;;) { |
1102 |
Node<K,V> b = head.node; |
1103 |
Node<K,V> n = b.next; |
1104 |
if (n == null) |
1105 |
return null; |
1106 |
if (n.value != null) |
1107 |
return n; |
1108 |
n.helpDelete(b, n.next); |
1109 |
} |
1110 |
} |
1111 |
|
1112 |
/** |
1113 |
* Removes first entry; returns its snapshot. |
1114 |
* @return null if empty, else snapshot of first entry |
1115 |
*/ |
1116 |
Map.Entry<K,V> doRemoveFirstEntry() { |
1117 |
for (;;) { |
1118 |
Node<K,V> b = head.node; |
1119 |
Node<K,V> n = b.next; |
1120 |
if (n == null) |
1121 |
return null; |
1122 |
Node<K,V> f = n.next; |
1123 |
if (n != b.next) |
1124 |
continue; |
1125 |
Object v = n.value; |
1126 |
if (v == null) { |
1127 |
n.helpDelete(b, f); |
1128 |
continue; |
1129 |
} |
1130 |
if (!n.casValue(v, null)) |
1131 |
continue; |
1132 |
if (!n.appendMarker(f) || !b.casNext(n, f)) |
1133 |
findFirst(); // retry |
1134 |
clearIndexToFirst(); |
1135 |
return new AbstractMap.SimpleImmutableEntry<K,V>(n.key, (V)v); |
1136 |
} |
1137 |
} |
1138 |
|
1139 |
/** |
1140 |
* Clears out index nodes associated with deleted first entry. |
1141 |
*/ |
1142 |
private void clearIndexToFirst() { |
1143 |
for (;;) { |
1144 |
Index<K,V> q = head; |
1145 |
for (;;) { |
1146 |
Index<K,V> r = q.right; |
1147 |
if (r != null && r.indexesDeletedNode() && !q.unlink(r)) |
1148 |
break; |
1149 |
if ((q = q.down) == null) { |
1150 |
if (head.right == null) |
1151 |
tryReduceLevel(); |
1152 |
return; |
1153 |
} |
1154 |
} |
1155 |
} |
1156 |
} |
1157 |
|
1158 |
|
1159 |
/* ---------------- Finding and removing last element -------------- */ |
1160 |
|
1161 |
/** |
1162 |
* Specialized version of find to get last valid node. |
1163 |
* @return last node or null if empty |
1164 |
*/ |
1165 |
Node<K,V> findLast() { |
1166 |
/* |
1167 |
* findPredecessor can't be used to traverse index level |
1168 |
* because this doesn't use comparisons. So traversals of |
1169 |
* both levels are folded together. |
1170 |
*/ |
1171 |
Index<K,V> q = head; |
1172 |
for (;;) { |
1173 |
Index<K,V> d, r; |
1174 |
if ((r = q.right) != null) { |
1175 |
if (r.indexesDeletedNode()) { |
1176 |
q.unlink(r); |
1177 |
q = head; // restart |
1178 |
} |
1179 |
else |
1180 |
q = r; |
1181 |
} else if ((d = q.down) != null) { |
1182 |
q = d; |
1183 |
} else { |
1184 |
Node<K,V> b = q.node; |
1185 |
Node<K,V> n = b.next; |
1186 |
for (;;) { |
1187 |
if (n == null) |
1188 |
return b.isBaseHeader() ? null : b; |
1189 |
Node<K,V> f = n.next; // inconsistent read |
1190 |
if (n != b.next) |
1191 |
break; |
1192 |
Object v = n.value; |
1193 |
if (v == null) { // n is deleted |
1194 |
n.helpDelete(b, f); |
1195 |
break; |
1196 |
} |
1197 |
if (v == n || b.value == null) // b is deleted |
1198 |
break; |
1199 |
b = n; |
1200 |
n = f; |
1201 |
} |
1202 |
q = head; // restart |
1203 |
} |
1204 |
} |
1205 |
} |
1206 |
|
1207 |
/** |
1208 |
* Specialized variant of findPredecessor to get predecessor of last |
1209 |
* valid node. Needed when removing the last entry. It is possible |
1210 |
* that all successors of returned node will have been deleted upon |
1211 |
* return, in which case this method can be retried. |
1212 |
* @return likely predecessor of last node |
1213 |
*/ |
1214 |
private Node<K,V> findPredecessorOfLast() { |
1215 |
for (;;) { |
1216 |
Index<K,V> q = head; |
1217 |
for (;;) { |
1218 |
Index<K,V> d, r; |
1219 |
if ((r = q.right) != null) { |
1220 |
if (r.indexesDeletedNode()) { |
1221 |
q.unlink(r); |
1222 |
break; // must restart |
1223 |
} |
1224 |
// proceed as far across as possible without overshooting |
1225 |
if (r.node.next != null) { |
1226 |
q = r; |
1227 |
continue; |
1228 |
} |
1229 |
} |
1230 |
if ((d = q.down) != null) |
1231 |
q = d; |
1232 |
else |
1233 |
return q.node; |
1234 |
} |
1235 |
} |
1236 |
} |
1237 |
|
1238 |
/** |
1239 |
* Removes last entry; returns its snapshot. |
1240 |
* Specialized variant of doRemove. |
1241 |
* @return null if empty, else snapshot of last entry |
1242 |
*/ |
1243 |
Map.Entry<K,V> doRemoveLastEntry() { |
1244 |
for (;;) { |
1245 |
Node<K,V> b = findPredecessorOfLast(); |
1246 |
Node<K,V> n = b.next; |
1247 |
if (n == null) { |
1248 |
if (b.isBaseHeader()) // empty |
1249 |
return null; |
1250 |
else |
1251 |
continue; // all b's successors are deleted; retry |
1252 |
} |
1253 |
for (;;) { |
1254 |
Node<K,V> f = n.next; |
1255 |
if (n != b.next) // inconsistent read |
1256 |
break; |
1257 |
Object v = n.value; |
1258 |
if (v == null) { // n is deleted |
1259 |
n.helpDelete(b, f); |
1260 |
break; |
1261 |
} |
1262 |
if (v == n || b.value == null) // b is deleted |
1263 |
break; |
1264 |
if (f != null) { |
1265 |
b = n; |
1266 |
n = f; |
1267 |
continue; |
1268 |
} |
1269 |
if (!n.casValue(v, null)) |
1270 |
break; |
1271 |
K key = n.key; |
1272 |
Comparable<? super K> ck = comparable(key); |
1273 |
if (!n.appendMarker(f) || !b.casNext(n, f)) |
1274 |
findNode(ck); // Retry via findNode |
1275 |
else { |
1276 |
findPredecessor(ck); // Clean index |
1277 |
if (head.right == null) |
1278 |
tryReduceLevel(); |
1279 |
} |
1280 |
return new AbstractMap.SimpleImmutableEntry<K,V>(key, (V)v); |
1281 |
} |
1282 |
} |
1283 |
} |
1284 |
|
1285 |
/* ---------------- Relational operations -------------- */ |
1286 |
|
1287 |
// Control values OR'ed as arguments to findNear |
1288 |
|
1289 |
private static final int EQ = 1; |
1290 |
private static final int LT = 2; |
1291 |
private static final int GT = 0; // Actually checked as !LT |
1292 |
|
1293 |
/** |
1294 |
* Utility for ceiling, floor, lower, higher methods. |
1295 |
* @param kkey the key |
1296 |
* @param rel the relation -- OR'ed combination of EQ, LT, GT |
1297 |
* @return nearest node fitting relation, or null if no such |
1298 |
*/ |
1299 |
Node<K,V> findNear(K kkey, int rel) { |
1300 |
Comparable<? super K> key = comparable(kkey); |
1301 |
for (;;) { |
1302 |
Node<K,V> b = findPredecessor(key); |
1303 |
Node<K,V> n = b.next; |
1304 |
for (;;) { |
1305 |
if (n == null) |
1306 |
return ((rel & LT) == 0 || b.isBaseHeader()) ? null : b; |
1307 |
Node<K,V> f = n.next; |
1308 |
if (n != b.next) // inconsistent read |
1309 |
break; |
1310 |
Object v = n.value; |
1311 |
if (v == null) { // n is deleted |
1312 |
n.helpDelete(b, f); |
1313 |
break; |
1314 |
} |
1315 |
if (v == n || b.value == null) // b is deleted |
1316 |
break; |
1317 |
int c = key.compareTo(n.key); |
1318 |
if ((c == 0 && (rel & EQ) != 0) || |
1319 |
(c < 0 && (rel & LT) == 0)) |
1320 |
return n; |
1321 |
if ( c <= 0 && (rel & LT) != 0) |
1322 |
return b.isBaseHeader() ? null : b; |
1323 |
b = n; |
1324 |
n = f; |
1325 |
} |
1326 |
} |
1327 |
} |
1328 |
|
1329 |
/** |
1330 |
* Returns SimpleImmutableEntry for results of findNear. |
1331 |
* @param key the key |
1332 |
* @param rel the relation -- OR'ed combination of EQ, LT, GT |
1333 |
* @return Entry fitting relation, or null if no such |
1334 |
*/ |
1335 |
AbstractMap.SimpleImmutableEntry<K,V> getNear(K key, int rel) { |
1336 |
for (;;) { |
1337 |
Node<K,V> n = findNear(key, rel); |
1338 |
if (n == null) |
1339 |
return null; |
1340 |
AbstractMap.SimpleImmutableEntry<K,V> e = n.createSnapshot(); |
1341 |
if (e != null) |
1342 |
return e; |
1343 |
} |
1344 |
} |
1345 |
|
1346 |
|
1347 |
/* ---------------- Constructors -------------- */ |
1348 |
|
1349 |
/** |
1350 |
* Constructs a new, empty map, sorted according to the |
1351 |
* {@linkplain Comparable natural ordering} of the keys. |
1352 |
*/ |
1353 |
public ConcurrentSkipListMap() { |
1354 |
this.comparator = null; |
1355 |
initialize(); |
1356 |
} |
1357 |
|
1358 |
/** |
1359 |
* Constructs a new, empty map, sorted according to the specified |
1360 |
* comparator. |
1361 |
* |
1362 |
* @param comparator the comparator that will be used to order this map. |
1363 |
* If {@code null}, the {@linkplain Comparable natural |
1364 |
* ordering} of the keys will be used. |
1365 |
*/ |
1366 |
public ConcurrentSkipListMap(Comparator<? super K> comparator) { |
1367 |
this.comparator = comparator; |
1368 |
initialize(); |
1369 |
} |
1370 |
|
1371 |
/** |
1372 |
* Constructs a new map containing the same mappings as the given map, |
1373 |
* sorted according to the {@linkplain Comparable natural ordering} of |
1374 |
* the keys. |
1375 |
* |
1376 |
* @param m the map whose mappings are to be placed in this map |
1377 |
* @throws ClassCastException if the keys in {@code m} are not |
1378 |
* {@link Comparable}, or are not mutually comparable |
1379 |
* @throws NullPointerException if the specified map or any of its keys |
1380 |
* or values are null |
1381 |
*/ |
1382 |
public ConcurrentSkipListMap(Map<? extends K, ? extends V> m) { |
1383 |
this.comparator = null; |
1384 |
initialize(); |
1385 |
putAll(m); |
1386 |
} |
1387 |
|
1388 |
/** |
1389 |
* Constructs a new map containing the same mappings and using the |
1390 |
* same ordering as the specified sorted map. |
1391 |
* |
1392 |
* @param m the sorted map whose mappings are to be placed in this |
1393 |
* map, and whose comparator is to be used to sort this map |
1394 |
* @throws NullPointerException if the specified sorted map or any of |
1395 |
* its keys or values are null |
1396 |
*/ |
1397 |
public ConcurrentSkipListMap(SortedMap<K, ? extends V> m) { |
1398 |
this.comparator = m.comparator(); |
1399 |
initialize(); |
1400 |
buildFromSorted(m); |
1401 |
} |
1402 |
|
1403 |
/** |
1404 |
* Returns a shallow copy of this {@code ConcurrentSkipListMap} |
1405 |
* instance. (The keys and values themselves are not cloned.) |
1406 |
* |
1407 |
* @return a shallow copy of this map |
1408 |
*/ |
1409 |
public ConcurrentSkipListMap<K,V> clone() { |
1410 |
try { |
1411 |
@SuppressWarnings("unchecked") |
1412 |
ConcurrentSkipListMap<K,V> clone = |
1413 |
(ConcurrentSkipListMap<K,V>) super.clone(); |
1414 |
clone.initialize(); |
1415 |
clone.buildFromSorted(this); |
1416 |
return clone; |
1417 |
} catch (CloneNotSupportedException e) { |
1418 |
throw new InternalError(); |
1419 |
} |
1420 |
} |
1421 |
|
1422 |
/** |
1423 |
* Streamlined bulk insertion to initialize from elements of |
1424 |
* given sorted map. Call only from constructor or clone |
1425 |
* method. |
1426 |
*/ |
1427 |
private void buildFromSorted(SortedMap<K, ? extends V> map) { |
1428 |
if (map == null) |
1429 |
throw new NullPointerException(); |
1430 |
|
1431 |
HeadIndex<K,V> h = head; |
1432 |
Node<K,V> basepred = h.node; |
1433 |
|
1434 |
// Track the current rightmost node at each level. Uses an |
1435 |
// ArrayList to avoid committing to initial or maximum level. |
1436 |
ArrayList<Index<K,V>> preds = new ArrayList<Index<K,V>>(); |
1437 |
|
1438 |
// initialize |
1439 |
for (int i = 0; i <= h.level; ++i) |
1440 |
preds.add(null); |
1441 |
Index<K,V> q = h; |
1442 |
for (int i = h.level; i > 0; --i) { |
1443 |
preds.set(i, q); |
1444 |
q = q.down; |
1445 |
} |
1446 |
|
1447 |
Iterator<? extends Map.Entry<? extends K, ? extends V>> it = |
1448 |
map.entrySet().iterator(); |
1449 |
while (it.hasNext()) { |
1450 |
Map.Entry<? extends K, ? extends V> e = it.next(); |
1451 |
int j = randomLevel(); |
1452 |
if (j > h.level) j = h.level + 1; |
1453 |
K k = e.getKey(); |
1454 |
V v = e.getValue(); |
1455 |
if (k == null || v == null) |
1456 |
throw new NullPointerException(); |
1457 |
Node<K,V> z = new Node<K,V>(k, v, null); |
1458 |
basepred.next = z; |
1459 |
basepred = z; |
1460 |
if (j > 0) { |
1461 |
Index<K,V> idx = null; |
1462 |
for (int i = 1; i <= j; ++i) { |
1463 |
idx = new Index<K,V>(z, idx, null); |
1464 |
if (i > h.level) |
1465 |
h = new HeadIndex<K,V>(h.node, h, idx, i); |
1466 |
|
1467 |
if (i < preds.size()) { |
1468 |
preds.get(i).right = idx; |
1469 |
preds.set(i, idx); |
1470 |
} else |
1471 |
preds.add(idx); |
1472 |
} |
1473 |
} |
1474 |
} |
1475 |
head = h; |
1476 |
} |
1477 |
|
1478 |
/* ---------------- Serialization -------------- */ |
1479 |
|
1480 |
/** |
1481 |
* Saves this map to a stream (that is, serializes it). |
1482 |
* |
1483 |
* @serialData The key (Object) and value (Object) for each |
1484 |
* key-value mapping represented by the map, followed by |
1485 |
* {@code null}. The key-value mappings are emitted in key-order |
1486 |
* (as determined by the Comparator, or by the keys' natural |
1487 |
* ordering if no Comparator). |
1488 |
*/ |
1489 |
private void writeObject(java.io.ObjectOutputStream s) |
1490 |
throws java.io.IOException { |
1491 |
// Write out the Comparator and any hidden stuff |
1492 |
s.defaultWriteObject(); |
1493 |
|
1494 |
// Write out keys and values (alternating) |
1495 |
for (Node<K,V> n = findFirst(); n != null; n = n.next) { |
1496 |
V v = n.getValidValue(); |
1497 |
if (v != null) { |
1498 |
s.writeObject(n.key); |
1499 |
s.writeObject(v); |
1500 |
} |
1501 |
} |
1502 |
s.writeObject(null); |
1503 |
} |
1504 |
|
1505 |
/** |
1506 |
* Reconstitutes this map from a stream (that is, deserializes it). |
1507 |
*/ |
1508 |
private void readObject(final java.io.ObjectInputStream s) |
1509 |
throws java.io.IOException, ClassNotFoundException { |
1510 |
// Read in the Comparator and any hidden stuff |
1511 |
s.defaultReadObject(); |
1512 |
// Reset transients |
1513 |
initialize(); |
1514 |
|
1515 |
/* |
1516 |
* This is nearly identical to buildFromSorted, but is |
1517 |
* distinct because readObject calls can't be nicely adapted |
1518 |
* as the kind of iterator needed by buildFromSorted. (They |
1519 |
* can be, but doing so requires type cheats and/or creation |
1520 |
* of adaptor classes.) It is simpler to just adapt the code. |
1521 |
*/ |
1522 |
|
1523 |
HeadIndex<K,V> h = head; |
1524 |
Node<K,V> basepred = h.node; |
1525 |
ArrayList<Index<K,V>> preds = new ArrayList<Index<K,V>>(); |
1526 |
for (int i = 0; i <= h.level; ++i) |
1527 |
preds.add(null); |
1528 |
Index<K,V> q = h; |
1529 |
for (int i = h.level; i > 0; --i) { |
1530 |
preds.set(i, q); |
1531 |
q = q.down; |
1532 |
} |
1533 |
|
1534 |
for (;;) { |
1535 |
Object k = s.readObject(); |
1536 |
if (k == null) |
1537 |
break; |
1538 |
Object v = s.readObject(); |
1539 |
if (v == null) |
1540 |
throw new NullPointerException(); |
1541 |
K key = (K) k; |
1542 |
V val = (V) v; |
1543 |
int j = randomLevel(); |
1544 |
if (j > h.level) j = h.level + 1; |
1545 |
Node<K,V> z = new Node<K,V>(key, val, null); |
1546 |
basepred.next = z; |
1547 |
basepred = z; |
1548 |
if (j > 0) { |
1549 |
Index<K,V> idx = null; |
1550 |
for (int i = 1; i <= j; ++i) { |
1551 |
idx = new Index<K,V>(z, idx, null); |
1552 |
if (i > h.level) |
1553 |
h = new HeadIndex<K,V>(h.node, h, idx, i); |
1554 |
|
1555 |
if (i < preds.size()) { |
1556 |
preds.get(i).right = idx; |
1557 |
preds.set(i, idx); |
1558 |
} else |
1559 |
preds.add(idx); |
1560 |
} |
1561 |
} |
1562 |
} |
1563 |
head = h; |
1564 |
} |
1565 |
|
1566 |
/* ------ Map API methods ------ */ |
1567 |
|
1568 |
/** |
1569 |
* Returns {@code true} if this map contains a mapping for the specified |
1570 |
* key. |
1571 |
* |
1572 |
* @param key key whose presence in this map is to be tested |
1573 |
* @return {@code true} if this map contains a mapping for the specified key |
1574 |
* @throws ClassCastException if the specified key cannot be compared |
1575 |
* with the keys currently in the map |
1576 |
* @throws NullPointerException if the specified key is null |
1577 |
*/ |
1578 |
public boolean containsKey(Object key) { |
1579 |
return doGet(key) != null; |
1580 |
} |
1581 |
|
1582 |
/** |
1583 |
* Returns the value to which the specified key is mapped, |
1584 |
* or {@code null} if this map contains no mapping for the key. |
1585 |
* |
1586 |
* <p>More formally, if this map contains a mapping from a key |
1587 |
* {@code k} to a value {@code v} such that {@code key} compares |
1588 |
* equal to {@code k} according to the map's ordering, then this |
1589 |
* method returns {@code v}; otherwise it returns {@code null}. |
1590 |
* (There can be at most one such mapping.) |
1591 |
* |
1592 |
* @throws ClassCastException if the specified key cannot be compared |
1593 |
* with the keys currently in the map |
1594 |
* @throws NullPointerException if the specified key is null |
1595 |
*/ |
1596 |
public V get(Object key) { |
1597 |
return doGet(key); |
1598 |
} |
1599 |
|
1600 |
/** |
1601 |
* Associates the specified value with the specified key in this map. |
1602 |
* If the map previously contained a mapping for the key, the old |
1603 |
* value is replaced. |
1604 |
* |
1605 |
* @param key key with which the specified value is to be associated |
1606 |
* @param value value to be associated with the specified key |
1607 |
* @return the previous value associated with the specified key, or |
1608 |
* {@code null} if there was no mapping for the key |
1609 |
* @throws ClassCastException if the specified key cannot be compared |
1610 |
* with the keys currently in the map |
1611 |
* @throws NullPointerException if the specified key or value is null |
1612 |
*/ |
1613 |
public V put(K key, V value) { |
1614 |
if (value == null) |
1615 |
throw new NullPointerException(); |
1616 |
return doPut(key, value, false); |
1617 |
} |
1618 |
|
1619 |
/** |
1620 |
* Removes the mapping for the specified key from this map if present. |
1621 |
* |
1622 |
* @param key key for which mapping should be removed |
1623 |
* @return the previous value associated with the specified key, or |
1624 |
* {@code null} if there was no mapping for the key |
1625 |
* @throws ClassCastException if the specified key cannot be compared |
1626 |
* with the keys currently in the map |
1627 |
* @throws NullPointerException if the specified key is null |
1628 |
*/ |
1629 |
public V remove(Object key) { |
1630 |
return doRemove(key, null); |
1631 |
} |
1632 |
|
1633 |
/** |
1634 |
* Returns {@code true} if this map maps one or more keys to the |
1635 |
* specified value. This operation requires time linear in the |
1636 |
* map size. Additionally, it is possible for the map to change |
1637 |
* during execution of this method, in which case the returned |
1638 |
* result may be inaccurate. |
1639 |
* |
1640 |
* @param value value whose presence in this map is to be tested |
1641 |
* @return {@code true} if a mapping to {@code value} exists; |
1642 |
* {@code false} otherwise |
1643 |
* @throws NullPointerException if the specified value is null |
1644 |
*/ |
1645 |
public boolean containsValue(Object value) { |
1646 |
if (value == null) |
1647 |
throw new NullPointerException(); |
1648 |
for (Node<K,V> n = findFirst(); n != null; n = n.next) { |
1649 |
V v = n.getValidValue(); |
1650 |
if (v != null && value.equals(v)) |
1651 |
return true; |
1652 |
} |
1653 |
return false; |
1654 |
} |
1655 |
|
1656 |
/** |
1657 |
* Returns the number of key-value mappings in this map. If this map |
1658 |
* contains more than {@code Integer.MAX_VALUE} elements, it |
1659 |
* returns {@code Integer.MAX_VALUE}. |
1660 |
* |
1661 |
* <p>Beware that, unlike in most collections, this method is |
1662 |
* <em>NOT</em> a constant-time operation. Because of the |
1663 |
* asynchronous nature of these maps, determining the current |
1664 |
* number of elements requires traversing them all to count them. |
1665 |
* Additionally, it is possible for the size to change during |
1666 |
* execution of this method, in which case the returned result |
1667 |
* will be inaccurate. Thus, this method is typically not very |
1668 |
* useful in concurrent applications. |
1669 |
* |
1670 |
* @return the number of elements in this map |
1671 |
*/ |
1672 |
public int size() { |
1673 |
long count = 0; |
1674 |
for (Node<K,V> n = findFirst(); n != null; n = n.next) { |
1675 |
if (n.getValidValue() != null) |
1676 |
++count; |
1677 |
} |
1678 |
return (count >= Integer.MAX_VALUE) ? Integer.MAX_VALUE : (int) count; |
1679 |
} |
1680 |
|
1681 |
/** |
1682 |
* Returns {@code true} if this map contains no key-value mappings. |
1683 |
* @return {@code true} if this map contains no key-value mappings |
1684 |
*/ |
1685 |
public boolean isEmpty() { |
1686 |
return findFirst() == null; |
1687 |
} |
1688 |
|
1689 |
/** |
1690 |
* Removes all of the mappings from this map. |
1691 |
*/ |
1692 |
public void clear() { |
1693 |
initialize(); |
1694 |
} |
1695 |
|
1696 |
/* ---------------- View methods -------------- */ |
1697 |
|
1698 |
/* |
1699 |
* Note: Lazy initialization works for views because view classes |
1700 |
* are stateless/immutable so it doesn't matter wrt correctness if |
1701 |
* more than one is created (which will only rarely happen). Even |
1702 |
* so, the following idiom conservatively ensures that the method |
1703 |
* returns the one it created if it does so, not one created by |
1704 |
* another racing thread. |
1705 |
*/ |
1706 |
|
1707 |
/** |
1708 |
* Returns a {@link NavigableSet} view of the keys contained in this map. |
1709 |
* The set's iterator returns the keys in ascending order. |
1710 |
* The set is backed by the map, so changes to the map are |
1711 |
* reflected in the set, and vice-versa. The set supports element |
1712 |
* removal, which removes the corresponding mapping from the map, |
1713 |
* via the {@code Iterator.remove}, {@code Set.remove}, |
1714 |
* {@code removeAll}, {@code retainAll}, and {@code clear} |
1715 |
* operations. It does not support the {@code add} or {@code addAll} |
1716 |
* operations. |
1717 |
* |
1718 |
* <p>The view's {@code iterator} is a "weakly consistent" iterator |
1719 |
* that will never throw {@link ConcurrentModificationException}, |
1720 |
* and guarantees to traverse elements as they existed upon |
1721 |
* construction of the iterator, and may (but is not guaranteed to) |
1722 |
* reflect any modifications subsequent to construction. |
1723 |
* |
1724 |
* <p>This method is equivalent to method {@code navigableKeySet}. |
1725 |
* |
1726 |
* @return a navigable set view of the keys in this map |
1727 |
*/ |
1728 |
public NavigableSet<K> keySet() { |
1729 |
KeySet<K> ks = keySet; |
1730 |
return (ks != null) ? ks : (keySet = new KeySet<K>(this)); |
1731 |
} |
1732 |
|
1733 |
public NavigableSet<K> navigableKeySet() { |
1734 |
KeySet<K> ks = keySet; |
1735 |
return (ks != null) ? ks : (keySet = new KeySet<K>(this)); |
1736 |
} |
1737 |
|
1738 |
/** |
1739 |
* Returns a {@link Collection} view of the values contained in this map. |
1740 |
* The collection's iterator returns the values in ascending order |
1741 |
* of the corresponding keys. |
1742 |
* The collection is backed by the map, so changes to the map are |
1743 |
* reflected in the collection, and vice-versa. The collection |
1744 |
* supports element removal, which removes the corresponding |
1745 |
* mapping from the map, via the {@code Iterator.remove}, |
1746 |
* {@code Collection.remove}, {@code removeAll}, |
1747 |
* {@code retainAll} and {@code clear} operations. It does not |
1748 |
* support the {@code add} or {@code addAll} operations. |
1749 |
* |
1750 |
* <p>The view's {@code iterator} is a "weakly consistent" iterator |
1751 |
* that will never throw {@link ConcurrentModificationException}, |
1752 |
* and guarantees to traverse elements as they existed upon |
1753 |
* construction of the iterator, and may (but is not guaranteed to) |
1754 |
* reflect any modifications subsequent to construction. |
1755 |
*/ |
1756 |
public Collection<V> values() { |
1757 |
Values<V> vs = values; |
1758 |
return (vs != null) ? vs : (values = new Values<V>(this)); |
1759 |
} |
1760 |
|
1761 |
/** |
1762 |
* Returns a {@link Set} view of the mappings contained in this map. |
1763 |
* The set's iterator returns the entries in ascending key order. |
1764 |
* The set is backed by the map, so changes to the map are |
1765 |
* reflected in the set, and vice-versa. The set supports element |
1766 |
* removal, which removes the corresponding mapping from the map, |
1767 |
* via the {@code Iterator.remove}, {@code Set.remove}, |
1768 |
* {@code removeAll}, {@code retainAll} and {@code clear} |
1769 |
* operations. It does not support the {@code add} or |
1770 |
* {@code addAll} operations. |
1771 |
* |
1772 |
* <p>The view's {@code iterator} is a "weakly consistent" iterator |
1773 |
* that will never throw {@link ConcurrentModificationException}, |
1774 |
* and guarantees to traverse elements as they existed upon |
1775 |
* construction of the iterator, and may (but is not guaranteed to) |
1776 |
* reflect any modifications subsequent to construction. |
1777 |
* |
1778 |
* <p>The {@code Map.Entry} elements returned by |
1779 |
* {@code iterator.next()} do <em>not</em> support the |
1780 |
* {@code setValue} operation. |
1781 |
* |
1782 |
* @return a set view of the mappings contained in this map, |
1783 |
* sorted in ascending key order |
1784 |
*/ |
1785 |
public Set<Map.Entry<K,V>> entrySet() { |
1786 |
EntrySet<K,V> es = entrySet; |
1787 |
return (es != null) ? es : (entrySet = new EntrySet<K,V>(this)); |
1788 |
} |
1789 |
|
1790 |
public ConcurrentNavigableMap<K,V> descendingMap() { |
1791 |
ConcurrentNavigableMap<K,V> dm = descendingMap; |
1792 |
return (dm != null) ? dm : (descendingMap = new SubMap<K,V> |
1793 |
(this, null, false, null, false, true)); |
1794 |
} |
1795 |
|
1796 |
public NavigableSet<K> descendingKeySet() { |
1797 |
return descendingMap().navigableKeySet(); |
1798 |
} |
1799 |
|
1800 |
/* ---------------- AbstractMap Overrides -------------- */ |
1801 |
|
1802 |
/** |
1803 |
* Compares the specified object with this map for equality. |
1804 |
* Returns {@code true} if the given object is also a map and the |
1805 |
* two maps represent the same mappings. More formally, two maps |
1806 |
* {@code m1} and {@code m2} represent the same mappings if |
1807 |
* {@code m1.entrySet().equals(m2.entrySet())}. This |
1808 |
* operation may return misleading results if either map is |
1809 |
* concurrently modified during execution of this method. |
1810 |
* |
1811 |
* @param o object to be compared for equality with this map |
1812 |
* @return {@code true} if the specified object is equal to this map |
1813 |
*/ |
1814 |
public boolean equals(Object o) { |
1815 |
if (o == this) |
1816 |
return true; |
1817 |
if (!(o instanceof Map)) |
1818 |
return false; |
1819 |
Map<?,?> m = (Map<?,?>) o; |
1820 |
try { |
1821 |
for (Map.Entry<K,V> e : this.entrySet()) |
1822 |
if (! e.getValue().equals(m.get(e.getKey()))) |
1823 |
return false; |
1824 |
for (Map.Entry<?,?> e : m.entrySet()) { |
1825 |
Object k = e.getKey(); |
1826 |
Object v = e.getValue(); |
1827 |
if (k == null || v == null || !v.equals(get(k))) |
1828 |
return false; |
1829 |
} |
1830 |
return true; |
1831 |
} catch (ClassCastException unused) { |
1832 |
return false; |
1833 |
} catch (NullPointerException unused) { |
1834 |
return false; |
1835 |
} |
1836 |
} |
1837 |
|
1838 |
/* ------ ConcurrentMap API methods ------ */ |
1839 |
|
1840 |
/** |
1841 |
* {@inheritDoc} |
1842 |
* |
1843 |
* @return the previous value associated with the specified key, |
1844 |
* or {@code null} if there was no mapping for the key |
1845 |
* @throws ClassCastException if the specified key cannot be compared |
1846 |
* with the keys currently in the map |
1847 |
* @throws NullPointerException if the specified key or value is null |
1848 |
*/ |
1849 |
public V putIfAbsent(K key, V value) { |
1850 |
if (value == null) |
1851 |
throw new NullPointerException(); |
1852 |
return doPut(key, value, true); |
1853 |
} |
1854 |
|
1855 |
/** |
1856 |
* {@inheritDoc} |
1857 |
* |
1858 |
* @throws ClassCastException if the specified key cannot be compared |
1859 |
* with the keys currently in the map |
1860 |
* @throws NullPointerException if the specified key is null |
1861 |
*/ |
1862 |
public boolean remove(Object key, Object value) { |
1863 |
if (key == null) |
1864 |
throw new NullPointerException(); |
1865 |
if (value == null) |
1866 |
return false; |
1867 |
return doRemove(key, value) != null; |
1868 |
} |
1869 |
|
1870 |
/** |
1871 |
* {@inheritDoc} |
1872 |
* |
1873 |
* @throws ClassCastException if the specified key cannot be compared |
1874 |
* with the keys currently in the map |
1875 |
* @throws NullPointerException if any of the arguments are null |
1876 |
*/ |
1877 |
public boolean replace(K key, V oldValue, V newValue) { |
1878 |
if (oldValue == null || newValue == null) |
1879 |
throw new NullPointerException(); |
1880 |
Comparable<? super K> k = comparable(key); |
1881 |
for (;;) { |
1882 |
Node<K,V> n = findNode(k); |
1883 |
if (n == null) |
1884 |
return false; |
1885 |
Object v = n.value; |
1886 |
if (v != null) { |
1887 |
if (!oldValue.equals(v)) |
1888 |
return false; |
1889 |
if (n.casValue(v, newValue)) |
1890 |
return true; |
1891 |
} |
1892 |
} |
1893 |
} |
1894 |
|
1895 |
/** |
1896 |
* {@inheritDoc} |
1897 |
* |
1898 |
* @return the previous value associated with the specified key, |
1899 |
* or {@code null} if there was no mapping for the key |
1900 |
* @throws ClassCastException if the specified key cannot be compared |
1901 |
* with the keys currently in the map |
1902 |
* @throws NullPointerException if the specified key or value is null |
1903 |
*/ |
1904 |
public V replace(K key, V value) { |
1905 |
if (value == null) |
1906 |
throw new NullPointerException(); |
1907 |
Comparable<? super K> k = comparable(key); |
1908 |
for (;;) { |
1909 |
Node<K,V> n = findNode(k); |
1910 |
if (n == null) |
1911 |
return null; |
1912 |
Object v = n.value; |
1913 |
if (v != null && n.casValue(v, value)) |
1914 |
return (V)v; |
1915 |
} |
1916 |
} |
1917 |
|
1918 |
/* ------ SortedMap API methods ------ */ |
1919 |
|
1920 |
public Comparator<? super K> comparator() { |
1921 |
return comparator; |
1922 |
} |
1923 |
|
1924 |
/** |
1925 |
* @throws NoSuchElementException {@inheritDoc} |
1926 |
*/ |
1927 |
public K firstKey() { |
1928 |
Node<K,V> n = findFirst(); |
1929 |
if (n == null) |
1930 |
throw new NoSuchElementException(); |
1931 |
return n.key; |
1932 |
} |
1933 |
|
1934 |
/** |
1935 |
* @throws NoSuchElementException {@inheritDoc} |
1936 |
*/ |
1937 |
public K lastKey() { |
1938 |
Node<K,V> n = findLast(); |
1939 |
if (n == null) |
1940 |
throw new NoSuchElementException(); |
1941 |
return n.key; |
1942 |
} |
1943 |
|
1944 |
/** |
1945 |
* @throws ClassCastException {@inheritDoc} |
1946 |
* @throws NullPointerException if {@code fromKey} or {@code toKey} is null |
1947 |
* @throws IllegalArgumentException {@inheritDoc} |
1948 |
*/ |
1949 |
public ConcurrentNavigableMap<K,V> subMap(K fromKey, |
1950 |
boolean fromInclusive, |
1951 |
K toKey, |
1952 |
boolean toInclusive) { |
1953 |
if (fromKey == null || toKey == null) |
1954 |
throw new NullPointerException(); |
1955 |
return new SubMap<K,V> |
1956 |
(this, fromKey, fromInclusive, toKey, toInclusive, false); |
1957 |
} |
1958 |
|
1959 |
/** |
1960 |
* @throws ClassCastException {@inheritDoc} |
1961 |
* @throws NullPointerException if {@code toKey} is null |
1962 |
* @throws IllegalArgumentException {@inheritDoc} |
1963 |
*/ |
1964 |
public ConcurrentNavigableMap<K,V> headMap(K toKey, |
1965 |
boolean inclusive) { |
1966 |
if (toKey == null) |
1967 |
throw new NullPointerException(); |
1968 |
return new SubMap<K,V> |
1969 |
(this, null, false, toKey, inclusive, false); |
1970 |
} |
1971 |
|
1972 |
/** |
1973 |
* @throws ClassCastException {@inheritDoc} |
1974 |
* @throws NullPointerException if {@code fromKey} is null |
1975 |
* @throws IllegalArgumentException {@inheritDoc} |
1976 |
*/ |
1977 |
public ConcurrentNavigableMap<K,V> tailMap(K fromKey, |
1978 |
boolean inclusive) { |
1979 |
if (fromKey == null) |
1980 |
throw new NullPointerException(); |
1981 |
return new SubMap<K,V> |
1982 |
(this, fromKey, inclusive, null, false, false); |
1983 |
} |
1984 |
|
1985 |
/** |
1986 |
* @throws ClassCastException {@inheritDoc} |
1987 |
* @throws NullPointerException if {@code fromKey} or {@code toKey} is null |
1988 |
* @throws IllegalArgumentException {@inheritDoc} |
1989 |
*/ |
1990 |
public ConcurrentNavigableMap<K,V> subMap(K fromKey, K toKey) { |
1991 |
return subMap(fromKey, true, toKey, false); |
1992 |
} |
1993 |
|
1994 |
/** |
1995 |
* @throws ClassCastException {@inheritDoc} |
1996 |
* @throws NullPointerException if {@code toKey} is null |
1997 |
* @throws IllegalArgumentException {@inheritDoc} |
1998 |
*/ |
1999 |
public ConcurrentNavigableMap<K,V> headMap(K toKey) { |
2000 |
return headMap(toKey, false); |
2001 |
} |
2002 |
|
2003 |
/** |
2004 |
* @throws ClassCastException {@inheritDoc} |
2005 |
* @throws NullPointerException if {@code fromKey} is null |
2006 |
* @throws IllegalArgumentException {@inheritDoc} |
2007 |
*/ |
2008 |
public ConcurrentNavigableMap<K,V> tailMap(K fromKey) { |
2009 |
return tailMap(fromKey, true); |
2010 |
} |
2011 |
|
2012 |
/* ---------------- Relational operations -------------- */ |
2013 |
|
2014 |
/** |
2015 |
* Returns a key-value mapping associated with the greatest key |
2016 |
* strictly less than the given key, or {@code null} if there is |
2017 |
* no such key. The returned entry does <em>not</em> support the |
2018 |
* {@code Entry.setValue} method. |
2019 |
* |
2020 |
* @throws ClassCastException {@inheritDoc} |
2021 |
* @throws NullPointerException if the specified key is null |
2022 |
*/ |
2023 |
public Map.Entry<K,V> lowerEntry(K key) { |
2024 |
return getNear(key, LT); |
2025 |
} |
2026 |
|
2027 |
/** |
2028 |
* @throws ClassCastException {@inheritDoc} |
2029 |
* @throws NullPointerException if the specified key is null |
2030 |
*/ |
2031 |
public K lowerKey(K key) { |
2032 |
Node<K,V> n = findNear(key, LT); |
2033 |
return (n == null) ? null : n.key; |
2034 |
} |
2035 |
|
2036 |
/** |
2037 |
* Returns a key-value mapping associated with the greatest key |
2038 |
* less than or equal to the given key, or {@code null} if there |
2039 |
* is no such key. The returned entry does <em>not</em> support |
2040 |
* the {@code Entry.setValue} method. |
2041 |
* |
2042 |
* @param key the key |
2043 |
* @throws ClassCastException {@inheritDoc} |
2044 |
* @throws NullPointerException if the specified key is null |
2045 |
*/ |
2046 |
public Map.Entry<K,V> floorEntry(K key) { |
2047 |
return getNear(key, LT|EQ); |
2048 |
} |
2049 |
|
2050 |
/** |
2051 |
* @param key the key |
2052 |
* @throws ClassCastException {@inheritDoc} |
2053 |
* @throws NullPointerException if the specified key is null |
2054 |
*/ |
2055 |
public K floorKey(K key) { |
2056 |
Node<K,V> n = findNear(key, LT|EQ); |
2057 |
return (n == null) ? null : n.key; |
2058 |
} |
2059 |
|
2060 |
/** |
2061 |
* Returns a key-value mapping associated with the least key |
2062 |
* greater than or equal to the given key, or {@code null} if |
2063 |
* there is no such entry. The returned entry does <em>not</em> |
2064 |
* support the {@code Entry.setValue} method. |
2065 |
* |
2066 |
* @throws ClassCastException {@inheritDoc} |
2067 |
* @throws NullPointerException if the specified key is null |
2068 |
*/ |
2069 |
public Map.Entry<K,V> ceilingEntry(K key) { |
2070 |
return getNear(key, GT|EQ); |
2071 |
} |
2072 |
|
2073 |
/** |
2074 |
* @throws ClassCastException {@inheritDoc} |
2075 |
* @throws NullPointerException if the specified key is null |
2076 |
*/ |
2077 |
public K ceilingKey(K key) { |
2078 |
Node<K,V> n = findNear(key, GT|EQ); |
2079 |
return (n == null) ? null : n.key; |
2080 |
} |
2081 |
|
2082 |
/** |
2083 |
* Returns a key-value mapping associated with the least key |
2084 |
* strictly greater than the given key, or {@code null} if there |
2085 |
* is no such key. The returned entry does <em>not</em> support |
2086 |
* the {@code Entry.setValue} method. |
2087 |
* |
2088 |
* @param key the key |
2089 |
* @throws ClassCastException {@inheritDoc} |
2090 |
* @throws NullPointerException if the specified key is null |
2091 |
*/ |
2092 |
public Map.Entry<K,V> higherEntry(K key) { |
2093 |
return getNear(key, GT); |
2094 |
} |
2095 |
|
2096 |
/** |
2097 |
* @param key the key |
2098 |
* @throws ClassCastException {@inheritDoc} |
2099 |
* @throws NullPointerException if the specified key is null |
2100 |
*/ |
2101 |
public K higherKey(K key) { |
2102 |
Node<K,V> n = findNear(key, GT); |
2103 |
return (n == null) ? null : n.key; |
2104 |
} |
2105 |
|
2106 |
/** |
2107 |
* Returns a key-value mapping associated with the least |
2108 |
* key in this map, or {@code null} if the map is empty. |
2109 |
* The returned entry does <em>not</em> support |
2110 |
* the {@code Entry.setValue} method. |
2111 |
*/ |
2112 |
public Map.Entry<K,V> firstEntry() { |
2113 |
for (;;) { |
2114 |
Node<K,V> n = findFirst(); |
2115 |
if (n == null) |
2116 |
return null; |
2117 |
AbstractMap.SimpleImmutableEntry<K,V> e = n.createSnapshot(); |
2118 |
if (e != null) |
2119 |
return e; |
2120 |
} |
2121 |
} |
2122 |
|
2123 |
/** |
2124 |
* Returns a key-value mapping associated with the greatest |
2125 |
* key in this map, or {@code null} if the map is empty. |
2126 |
* The returned entry does <em>not</em> support |
2127 |
* the {@code Entry.setValue} method. |
2128 |
*/ |
2129 |
public Map.Entry<K,V> lastEntry() { |
2130 |
for (;;) { |
2131 |
Node<K,V> n = findLast(); |
2132 |
if (n == null) |
2133 |
return null; |
2134 |
AbstractMap.SimpleImmutableEntry<K,V> e = n.createSnapshot(); |
2135 |
if (e != null) |
2136 |
return e; |
2137 |
} |
2138 |
} |
2139 |
|
2140 |
/** |
2141 |
* Removes and returns a key-value mapping associated with |
2142 |
* the least key in this map, or {@code null} if the map is empty. |
2143 |
* The returned entry does <em>not</em> support |
2144 |
* the {@code Entry.setValue} method. |
2145 |
*/ |
2146 |
public Map.Entry<K,V> pollFirstEntry() { |
2147 |
return doRemoveFirstEntry(); |
2148 |
} |
2149 |
|
2150 |
/** |
2151 |
* Removes and returns a key-value mapping associated with |
2152 |
* the greatest key in this map, or {@code null} if the map is empty. |
2153 |
* The returned entry does <em>not</em> support |
2154 |
* the {@code Entry.setValue} method. |
2155 |
*/ |
2156 |
public Map.Entry<K,V> pollLastEntry() { |
2157 |
return doRemoveLastEntry(); |
2158 |
} |
2159 |
|
2160 |
|
2161 |
/* ---------------- Iterators -------------- */ |
2162 |
|
2163 |
/** |
2164 |
* Base of iterator classes: |
2165 |
*/ |
2166 |
abstract class Iter<T> implements Iterator<T> { |
2167 |
/** the last node returned by next() */ |
2168 |
Node<K,V> lastReturned; |
2169 |
/** the next node to return from next(); */ |
2170 |
Node<K,V> next; |
2171 |
/** Cache of next value field to maintain weak consistency */ |
2172 |
V nextValue; |
2173 |
|
2174 |
/** Initializes ascending iterator for entire range. */ |
2175 |
Iter() { |
2176 |
for (;;) { |
2177 |
next = findFirst(); |
2178 |
if (next == null) |
2179 |
break; |
2180 |
Object x = next.value; |
2181 |
if (x != null && x != next) { |
2182 |
nextValue = (V) x; |
2183 |
break; |
2184 |
} |
2185 |
} |
2186 |
} |
2187 |
|
2188 |
public final boolean hasNext() { |
2189 |
return next != null; |
2190 |
} |
2191 |
|
2192 |
/** Advances next to higher entry. */ |
2193 |
final void advance() { |
2194 |
if (next == null) |
2195 |
throw new NoSuchElementException(); |
2196 |
lastReturned = next; |
2197 |
for (;;) { |
2198 |
next = next.next; |
2199 |
if (next == null) |
2200 |
break; |
2201 |
Object x = next.value; |
2202 |
if (x != null && x != next) { |
2203 |
nextValue = (V) x; |
2204 |
break; |
2205 |
} |
2206 |
} |
2207 |
} |
2208 |
|
2209 |
public void remove() { |
2210 |
Node<K,V> l = lastReturned; |
2211 |
if (l == null) |
2212 |
throw new IllegalStateException(); |
2213 |
// It would not be worth all of the overhead to directly |
2214 |
// unlink from here. Using remove is fast enough. |
2215 |
ConcurrentSkipListMap.this.remove(l.key); |
2216 |
lastReturned = null; |
2217 |
} |
2218 |
|
2219 |
} |
2220 |
|
2221 |
final class ValueIterator extends Iter<V> { |
2222 |
public V next() { |
2223 |
V v = nextValue; |
2224 |
advance(); |
2225 |
return v; |
2226 |
} |
2227 |
} |
2228 |
|
2229 |
final class KeyIterator extends Iter<K> { |
2230 |
public K next() { |
2231 |
Node<K,V> n = next; |
2232 |
advance(); |
2233 |
return n.key; |
2234 |
} |
2235 |
} |
2236 |
|
2237 |
final class EntryIterator extends Iter<Map.Entry<K,V>> { |
2238 |
public Map.Entry<K,V> next() { |
2239 |
Node<K,V> n = next; |
2240 |
V v = nextValue; |
2241 |
advance(); |
2242 |
return new AbstractMap.SimpleImmutableEntry<K,V>(n.key, v); |
2243 |
} |
2244 |
} |
2245 |
|
2246 |
// Factory methods for iterators needed by ConcurrentSkipListSet etc |
2247 |
|
2248 |
Iterator<K> keyIterator() { |
2249 |
return new KeyIterator(); |
2250 |
} |
2251 |
|
2252 |
Iterator<V> valueIterator() { |
2253 |
return new ValueIterator(); |
2254 |
} |
2255 |
|
2256 |
Iterator<Map.Entry<K,V>> entryIterator() { |
2257 |
return new EntryIterator(); |
2258 |
} |
2259 |
|
2260 |
/* ---------------- View Classes -------------- */ |
2261 |
|
2262 |
/* |
2263 |
* View classes are static, delegating to a ConcurrentNavigableMap |
2264 |
* to allow use by SubMaps, which outweighs the ugliness of |
2265 |
* needing type-tests for Iterator methods. |
2266 |
*/ |
2267 |
|
2268 |
static final <E> List<E> toList(Collection<E> c) { |
2269 |
// Using size() here would be a pessimization. |
2270 |
ArrayList<E> list = new ArrayList<E>(); |
2271 |
for (E e : c) |
2272 |
list.add(e); |
2273 |
return list; |
2274 |
} |
2275 |
|
2276 |
static final class KeySet<E> |
2277 |
extends AbstractSet<E> implements NavigableSet<E> { |
2278 |
private final ConcurrentNavigableMap<E,?> m; |
2279 |
KeySet(ConcurrentNavigableMap<E,?> map) { m = map; } |
2280 |
public int size() { return m.size(); } |
2281 |
public boolean isEmpty() { return m.isEmpty(); } |
2282 |
public boolean contains(Object o) { return m.containsKey(o); } |
2283 |
public boolean remove(Object o) { return m.remove(o) != null; } |
2284 |
public void clear() { m.clear(); } |
2285 |
public E lower(E e) { return m.lowerKey(e); } |
2286 |
public E floor(E e) { return m.floorKey(e); } |
2287 |
public E ceiling(E e) { return m.ceilingKey(e); } |
2288 |
public E higher(E e) { return m.higherKey(e); } |
2289 |
public Comparator<? super E> comparator() { return m.comparator(); } |
2290 |
public E first() { return m.firstKey(); } |
2291 |
public E last() { return m.lastKey(); } |
2292 |
public E pollFirst() { |
2293 |
Map.Entry<E,?> e = m.pollFirstEntry(); |
2294 |
return (e == null) ? null : e.getKey(); |
2295 |
} |
2296 |
public E pollLast() { |
2297 |
Map.Entry<E,?> e = m.pollLastEntry(); |
2298 |
return (e == null) ? null : e.getKey(); |
2299 |
} |
2300 |
public Iterator<E> iterator() { |
2301 |
if (m instanceof ConcurrentSkipListMap) |
2302 |
return ((ConcurrentSkipListMap<E,Object>)m).keyIterator(); |
2303 |
else |
2304 |
return ((ConcurrentSkipListMap.SubMap<E,Object>)m).keyIterator(); |
2305 |
} |
2306 |
public boolean equals(Object o) { |
2307 |
if (o == this) |
2308 |
return true; |
2309 |
if (!(o instanceof Set)) |
2310 |
return false; |
2311 |
Collection<?> c = (Collection<?>) o; |
2312 |
try { |
2313 |
return containsAll(c) && c.containsAll(this); |
2314 |
} catch (ClassCastException unused) { |
2315 |
return false; |
2316 |
} catch (NullPointerException unused) { |
2317 |
return false; |
2318 |
} |
2319 |
} |
2320 |
public Object[] toArray() { return toList(this).toArray(); } |
2321 |
public <T> T[] toArray(T[] a) { return toList(this).toArray(a); } |
2322 |
public Iterator<E> descendingIterator() { |
2323 |
return descendingSet().iterator(); |
2324 |
} |
2325 |
public NavigableSet<E> subSet(E fromElement, |
2326 |
boolean fromInclusive, |
2327 |
E toElement, |
2328 |
boolean toInclusive) { |
2329 |
return new KeySet<E>(m.subMap(fromElement, fromInclusive, |
2330 |
toElement, toInclusive)); |
2331 |
} |
2332 |
public NavigableSet<E> headSet(E toElement, boolean inclusive) { |
2333 |
return new KeySet<E>(m.headMap(toElement, inclusive)); |
2334 |
} |
2335 |
public NavigableSet<E> tailSet(E fromElement, boolean inclusive) { |
2336 |
return new KeySet<E>(m.tailMap(fromElement, inclusive)); |
2337 |
} |
2338 |
public NavigableSet<E> subSet(E fromElement, E toElement) { |
2339 |
return subSet(fromElement, true, toElement, false); |
2340 |
} |
2341 |
public NavigableSet<E> headSet(E toElement) { |
2342 |
return headSet(toElement, false); |
2343 |
} |
2344 |
public NavigableSet<E> tailSet(E fromElement) { |
2345 |
return tailSet(fromElement, true); |
2346 |
} |
2347 |
public NavigableSet<E> descendingSet() { |
2348 |
return new KeySet<E>(m.descendingMap()); |
2349 |
} |
2350 |
} |
2351 |
|
2352 |
static final class Values<E> extends AbstractCollection<E> { |
2353 |
private final ConcurrentNavigableMap<?,E> m; |
2354 |
Values(ConcurrentNavigableMap<?,E> map) { |
2355 |
m = map; |
2356 |
} |
2357 |
public Iterator<E> iterator() { |
2358 |
if (m instanceof ConcurrentSkipListMap) |
2359 |
return ((ConcurrentSkipListMap<?,E>)m).valueIterator(); |
2360 |
else |
2361 |
return ((SubMap<?,E>)m).valueIterator(); |
2362 |
} |
2363 |
public boolean isEmpty() { |
2364 |
return m.isEmpty(); |
2365 |
} |
2366 |
public int size() { |
2367 |
return m.size(); |
2368 |
} |
2369 |
public boolean contains(Object o) { |
2370 |
return m.containsValue(o); |
2371 |
} |
2372 |
public void clear() { |
2373 |
m.clear(); |
2374 |
} |
2375 |
public Object[] toArray() { return toList(this).toArray(); } |
2376 |
public <T> T[] toArray(T[] a) { return toList(this).toArray(a); } |
2377 |
} |
2378 |
|
2379 |
static final class EntrySet<K1,V1> extends AbstractSet<Map.Entry<K1,V1>> { |
2380 |
private final ConcurrentNavigableMap<K1, V1> m; |
2381 |
EntrySet(ConcurrentNavigableMap<K1, V1> map) { |
2382 |
m = map; |
2383 |
} |
2384 |
|
2385 |
public Iterator<Map.Entry<K1,V1>> iterator() { |
2386 |
if (m instanceof ConcurrentSkipListMap) |
2387 |
return ((ConcurrentSkipListMap<K1,V1>)m).entryIterator(); |
2388 |
else |
2389 |
return ((SubMap<K1,V1>)m).entryIterator(); |
2390 |
} |
2391 |
|
2392 |
public boolean contains(Object o) { |
2393 |
if (!(o instanceof Map.Entry)) |
2394 |
return false; |
2395 |
Map.Entry<?,?> e = (Map.Entry<?,?>)o; |
2396 |
V1 v = m.get(e.getKey()); |
2397 |
return v != null && v.equals(e.getValue()); |
2398 |
} |
2399 |
public boolean remove(Object o) { |
2400 |
if (!(o instanceof Map.Entry)) |
2401 |
return false; |
2402 |
Map.Entry<?,?> e = (Map.Entry<?,?>)o; |
2403 |
return m.remove(e.getKey(), |
2404 |
e.getValue()); |
2405 |
} |
2406 |
public boolean isEmpty() { |
2407 |
return m.isEmpty(); |
2408 |
} |
2409 |
public int size() { |
2410 |
return m.size(); |
2411 |
} |
2412 |
public void clear() { |
2413 |
m.clear(); |
2414 |
} |
2415 |
public boolean equals(Object o) { |
2416 |
if (o == this) |
2417 |
return true; |
2418 |
if (!(o instanceof Set)) |
2419 |
return false; |
2420 |
Collection<?> c = (Collection<?>) o; |
2421 |
try { |
2422 |
return containsAll(c) && c.containsAll(this); |
2423 |
} catch (ClassCastException unused) { |
2424 |
return false; |
2425 |
} catch (NullPointerException unused) { |
2426 |
return false; |
2427 |
} |
2428 |
} |
2429 |
public Object[] toArray() { return toList(this).toArray(); } |
2430 |
public <T> T[] toArray(T[] a) { return toList(this).toArray(a); } |
2431 |
} |
2432 |
|
2433 |
/** |
2434 |
* Submaps returned by {@link ConcurrentSkipListMap} submap operations |
2435 |
* represent a subrange of mappings of their underlying |
2436 |
* maps. Instances of this class support all methods of their |
2437 |
* underlying maps, differing in that mappings outside their range are |
2438 |
* ignored, and attempts to add mappings outside their ranges result |
2439 |
* in {@link IllegalArgumentException}. Instances of this class are |
2440 |
* constructed only using the {@code subMap}, {@code headMap}, and |
2441 |
* {@code tailMap} methods of their underlying maps. |
2442 |
* |
2443 |
* @serial include |
2444 |
*/ |
2445 |
static final class SubMap<K,V> extends AbstractMap<K,V> |
2446 |
implements ConcurrentNavigableMap<K,V>, Cloneable, |
2447 |
java.io.Serializable { |
2448 |
private static final long serialVersionUID = -7647078645895051609L; |
2449 |
|
2450 |
/** Underlying map */ |
2451 |
private final ConcurrentSkipListMap<K,V> m; |
2452 |
/** lower bound key, or null if from start */ |
2453 |
private final K lo; |
2454 |
/** upper bound key, or null if to end */ |
2455 |
private final K hi; |
2456 |
/** inclusion flag for lo */ |
2457 |
private final boolean loInclusive; |
2458 |
/** inclusion flag for hi */ |
2459 |
private final boolean hiInclusive; |
2460 |
/** direction */ |
2461 |
private final boolean isDescending; |
2462 |
|
2463 |
// Lazily initialized view holders |
2464 |
private transient KeySet<K> keySetView; |
2465 |
private transient Set<Map.Entry<K,V>> entrySetView; |
2466 |
private transient Collection<V> valuesView; |
2467 |
|
2468 |
/** |
2469 |
* Creates a new submap, initializing all fields. |
2470 |
*/ |
2471 |
SubMap(ConcurrentSkipListMap<K,V> map, |
2472 |
K fromKey, boolean fromInclusive, |
2473 |
K toKey, boolean toInclusive, |
2474 |
boolean isDescending) { |
2475 |
if (fromKey != null && toKey != null && |
2476 |
map.compare(fromKey, toKey) > 0) |
2477 |
throw new IllegalArgumentException("inconsistent range"); |
2478 |
this.m = map; |
2479 |
this.lo = fromKey; |
2480 |
this.hi = toKey; |
2481 |
this.loInclusive = fromInclusive; |
2482 |
this.hiInclusive = toInclusive; |
2483 |
this.isDescending = isDescending; |
2484 |
} |
2485 |
|
2486 |
/* ---------------- Utilities -------------- */ |
2487 |
|
2488 |
private boolean tooLow(K key) { |
2489 |
if (lo != null) { |
2490 |
int c = m.compare(key, lo); |
2491 |
if (c < 0 || (c == 0 && !loInclusive)) |
2492 |
return true; |
2493 |
} |
2494 |
return false; |
2495 |
} |
2496 |
|
2497 |
private boolean tooHigh(K key) { |
2498 |
if (hi != null) { |
2499 |
int c = m.compare(key, hi); |
2500 |
if (c > 0 || (c == 0 && !hiInclusive)) |
2501 |
return true; |
2502 |
} |
2503 |
return false; |
2504 |
} |
2505 |
|
2506 |
private boolean inBounds(K key) { |
2507 |
return !tooLow(key) && !tooHigh(key); |
2508 |
} |
2509 |
|
2510 |
private void checkKeyBounds(K key) throws IllegalArgumentException { |
2511 |
if (key == null) |
2512 |
throw new NullPointerException(); |
2513 |
if (!inBounds(key)) |
2514 |
throw new IllegalArgumentException("key out of range"); |
2515 |
} |
2516 |
|
2517 |
/** |
2518 |
* Returns true if node key is less than upper bound of range. |
2519 |
*/ |
2520 |
private boolean isBeforeEnd(ConcurrentSkipListMap.Node<K,V> n) { |
2521 |
if (n == null) |
2522 |
return false; |
2523 |
if (hi == null) |
2524 |
return true; |
2525 |
K k = n.key; |
2526 |
if (k == null) // pass by markers and headers |
2527 |
return true; |
2528 |
int c = m.compare(k, hi); |
2529 |
if (c > 0 || (c == 0 && !hiInclusive)) |
2530 |
return false; |
2531 |
return true; |
2532 |
} |
2533 |
|
2534 |
/** |
2535 |
* Returns lowest node. This node might not be in range, so |
2536 |
* most usages need to check bounds. |
2537 |
*/ |
2538 |
private ConcurrentSkipListMap.Node<K,V> loNode() { |
2539 |
if (lo == null) |
2540 |
return m.findFirst(); |
2541 |
else if (loInclusive) |
2542 |
return m.findNear(lo, GT|EQ); |
2543 |
else |
2544 |
return m.findNear(lo, GT); |
2545 |
} |
2546 |
|
2547 |
/** |
2548 |
* Returns highest node. This node might not be in range, so |
2549 |
* most usages need to check bounds. |
2550 |
*/ |
2551 |
private ConcurrentSkipListMap.Node<K,V> hiNode() { |
2552 |
if (hi == null) |
2553 |
return m.findLast(); |
2554 |
else if (hiInclusive) |
2555 |
return m.findNear(hi, LT|EQ); |
2556 |
else |
2557 |
return m.findNear(hi, LT); |
2558 |
} |
2559 |
|
2560 |
/** |
2561 |
* Returns lowest absolute key (ignoring directonality). |
2562 |
*/ |
2563 |
private K lowestKey() { |
2564 |
ConcurrentSkipListMap.Node<K,V> n = loNode(); |
2565 |
if (isBeforeEnd(n)) |
2566 |
return n.key; |
2567 |
else |
2568 |
throw new NoSuchElementException(); |
2569 |
} |
2570 |
|
2571 |
/** |
2572 |
* Returns highest absolute key (ignoring directonality). |
2573 |
*/ |
2574 |
private K highestKey() { |
2575 |
ConcurrentSkipListMap.Node<K,V> n = hiNode(); |
2576 |
if (n != null) { |
2577 |
K last = n.key; |
2578 |
if (inBounds(last)) |
2579 |
return last; |
2580 |
} |
2581 |
throw new NoSuchElementException(); |
2582 |
} |
2583 |
|
2584 |
private Map.Entry<K,V> lowestEntry() { |
2585 |
for (;;) { |
2586 |
ConcurrentSkipListMap.Node<K,V> n = loNode(); |
2587 |
if (!isBeforeEnd(n)) |
2588 |
return null; |
2589 |
Map.Entry<K,V> e = n.createSnapshot(); |
2590 |
if (e != null) |
2591 |
return e; |
2592 |
} |
2593 |
} |
2594 |
|
2595 |
private Map.Entry<K,V> highestEntry() { |
2596 |
for (;;) { |
2597 |
ConcurrentSkipListMap.Node<K,V> n = hiNode(); |
2598 |
if (n == null || !inBounds(n.key)) |
2599 |
return null; |
2600 |
Map.Entry<K,V> e = n.createSnapshot(); |
2601 |
if (e != null) |
2602 |
return e; |
2603 |
} |
2604 |
} |
2605 |
|
2606 |
private Map.Entry<K,V> removeLowest() { |
2607 |
for (;;) { |
2608 |
Node<K,V> n = loNode(); |
2609 |
if (n == null) |
2610 |
return null; |
2611 |
K k = n.key; |
2612 |
if (!inBounds(k)) |
2613 |
return null; |
2614 |
V v = m.doRemove(k, null); |
2615 |
if (v != null) |
2616 |
return new AbstractMap.SimpleImmutableEntry<K,V>(k, v); |
2617 |
} |
2618 |
} |
2619 |
|
2620 |
private Map.Entry<K,V> removeHighest() { |
2621 |
for (;;) { |
2622 |
Node<K,V> n = hiNode(); |
2623 |
if (n == null) |
2624 |
return null; |
2625 |
K k = n.key; |
2626 |
if (!inBounds(k)) |
2627 |
return null; |
2628 |
V v = m.doRemove(k, null); |
2629 |
if (v != null) |
2630 |
return new AbstractMap.SimpleImmutableEntry<K,V>(k, v); |
2631 |
} |
2632 |
} |
2633 |
|
2634 |
/** |
2635 |
* Submap version of ConcurrentSkipListMap.getNearEntry |
2636 |
*/ |
2637 |
private Map.Entry<K,V> getNearEntry(K key, int rel) { |
2638 |
if (isDescending) { // adjust relation for direction |
2639 |
if ((rel & LT) == 0) |
2640 |
rel |= LT; |
2641 |
else |
2642 |
rel &= ~LT; |
2643 |
} |
2644 |
if (tooLow(key)) |
2645 |
return ((rel & LT) != 0) ? null : lowestEntry(); |
2646 |
if (tooHigh(key)) |
2647 |
return ((rel & LT) != 0) ? highestEntry() : null; |
2648 |
for (;;) { |
2649 |
Node<K,V> n = m.findNear(key, rel); |
2650 |
if (n == null || !inBounds(n.key)) |
2651 |
return null; |
2652 |
K k = n.key; |
2653 |
V v = n.getValidValue(); |
2654 |
if (v != null) |
2655 |
return new AbstractMap.SimpleImmutableEntry<K,V>(k, v); |
2656 |
} |
2657 |
} |
2658 |
|
2659 |
// Almost the same as getNearEntry, except for keys |
2660 |
private K getNearKey(K key, int rel) { |
2661 |
if (isDescending) { // adjust relation for direction |
2662 |
if ((rel & LT) == 0) |
2663 |
rel |= LT; |
2664 |
else |
2665 |
rel &= ~LT; |
2666 |
} |
2667 |
if (tooLow(key)) { |
2668 |
if ((rel & LT) == 0) { |
2669 |
ConcurrentSkipListMap.Node<K,V> n = loNode(); |
2670 |
if (isBeforeEnd(n)) |
2671 |
return n.key; |
2672 |
} |
2673 |
return null; |
2674 |
} |
2675 |
if (tooHigh(key)) { |
2676 |
if ((rel & LT) != 0) { |
2677 |
ConcurrentSkipListMap.Node<K,V> n = hiNode(); |
2678 |
if (n != null) { |
2679 |
K last = n.key; |
2680 |
if (inBounds(last)) |
2681 |
return last; |
2682 |
} |
2683 |
} |
2684 |
return null; |
2685 |
} |
2686 |
for (;;) { |
2687 |
Node<K,V> n = m.findNear(key, rel); |
2688 |
if (n == null || !inBounds(n.key)) |
2689 |
return null; |
2690 |
K k = n.key; |
2691 |
V v = n.getValidValue(); |
2692 |
if (v != null) |
2693 |
return k; |
2694 |
} |
2695 |
} |
2696 |
|
2697 |
/* ---------------- Map API methods -------------- */ |
2698 |
|
2699 |
public boolean containsKey(Object key) { |
2700 |
if (key == null) throw new NullPointerException(); |
2701 |
K k = (K)key; |
2702 |
return inBounds(k) && m.containsKey(k); |
2703 |
} |
2704 |
|
2705 |
public V get(Object key) { |
2706 |
if (key == null) throw new NullPointerException(); |
2707 |
K k = (K)key; |
2708 |
return (!inBounds(k)) ? null : m.get(k); |
2709 |
} |
2710 |
|
2711 |
public V put(K key, V value) { |
2712 |
checkKeyBounds(key); |
2713 |
return m.put(key, value); |
2714 |
} |
2715 |
|
2716 |
public V remove(Object key) { |
2717 |
K k = (K)key; |
2718 |
return (!inBounds(k)) ? null : m.remove(k); |
2719 |
} |
2720 |
|
2721 |
public int size() { |
2722 |
long count = 0; |
2723 |
for (ConcurrentSkipListMap.Node<K,V> n = loNode(); |
2724 |
isBeforeEnd(n); |
2725 |
n = n.next) { |
2726 |
if (n.getValidValue() != null) |
2727 |
++count; |
2728 |
} |
2729 |
return count >= Integer.MAX_VALUE ? Integer.MAX_VALUE : (int)count; |
2730 |
} |
2731 |
|
2732 |
public boolean isEmpty() { |
2733 |
return !isBeforeEnd(loNode()); |
2734 |
} |
2735 |
|
2736 |
public boolean containsValue(Object value) { |
2737 |
if (value == null) |
2738 |
throw new NullPointerException(); |
2739 |
for (ConcurrentSkipListMap.Node<K,V> n = loNode(); |
2740 |
isBeforeEnd(n); |
2741 |
n = n.next) { |
2742 |
V v = n.getValidValue(); |
2743 |
if (v != null && value.equals(v)) |
2744 |
return true; |
2745 |
} |
2746 |
return false; |
2747 |
} |
2748 |
|
2749 |
public void clear() { |
2750 |
for (ConcurrentSkipListMap.Node<K,V> n = loNode(); |
2751 |
isBeforeEnd(n); |
2752 |
n = n.next) { |
2753 |
if (n.getValidValue() != null) |
2754 |
m.remove(n.key); |
2755 |
} |
2756 |
} |
2757 |
|
2758 |
/* ---------------- ConcurrentMap API methods -------------- */ |
2759 |
|
2760 |
public V putIfAbsent(K key, V value) { |
2761 |
checkKeyBounds(key); |
2762 |
return m.putIfAbsent(key, value); |
2763 |
} |
2764 |
|
2765 |
public boolean remove(Object key, Object value) { |
2766 |
K k = (K)key; |
2767 |
return inBounds(k) && m.remove(k, value); |
2768 |
} |
2769 |
|
2770 |
public boolean replace(K key, V oldValue, V newValue) { |
2771 |
checkKeyBounds(key); |
2772 |
return m.replace(key, oldValue, newValue); |
2773 |
} |
2774 |
|
2775 |
public V replace(K key, V value) { |
2776 |
checkKeyBounds(key); |
2777 |
return m.replace(key, value); |
2778 |
} |
2779 |
|
2780 |
/* ---------------- SortedMap API methods -------------- */ |
2781 |
|
2782 |
public Comparator<? super K> comparator() { |
2783 |
Comparator<? super K> cmp = m.comparator(); |
2784 |
if (isDescending) |
2785 |
return Collections.reverseOrder(cmp); |
2786 |
else |
2787 |
return cmp; |
2788 |
} |
2789 |
|
2790 |
/** |
2791 |
* Utility to create submaps, where given bounds override |
2792 |
* unbounded(null) ones and/or are checked against bounded ones. |
2793 |
*/ |
2794 |
private SubMap<K,V> newSubMap(K fromKey, |
2795 |
boolean fromInclusive, |
2796 |
K toKey, |
2797 |
boolean toInclusive) { |
2798 |
if (isDescending) { // flip senses |
2799 |
K tk = fromKey; |
2800 |
fromKey = toKey; |
2801 |
toKey = tk; |
2802 |
boolean ti = fromInclusive; |
2803 |
fromInclusive = toInclusive; |
2804 |
toInclusive = ti; |
2805 |
} |
2806 |
if (lo != null) { |
2807 |
if (fromKey == null) { |
2808 |
fromKey = lo; |
2809 |
fromInclusive = loInclusive; |
2810 |
} |
2811 |
else { |
2812 |
int c = m.compare(fromKey, lo); |
2813 |
if (c < 0 || (c == 0 && !loInclusive && fromInclusive)) |
2814 |
throw new IllegalArgumentException("key out of range"); |
2815 |
} |
2816 |
} |
2817 |
if (hi != null) { |
2818 |
if (toKey == null) { |
2819 |
toKey = hi; |
2820 |
toInclusive = hiInclusive; |
2821 |
} |
2822 |
else { |
2823 |
int c = m.compare(toKey, hi); |
2824 |
if (c > 0 || (c == 0 && !hiInclusive && toInclusive)) |
2825 |
throw new IllegalArgumentException("key out of range"); |
2826 |
} |
2827 |
} |
2828 |
return new SubMap<K,V>(m, fromKey, fromInclusive, |
2829 |
toKey, toInclusive, isDescending); |
2830 |
} |
2831 |
|
2832 |
public SubMap<K,V> subMap(K fromKey, |
2833 |
boolean fromInclusive, |
2834 |
K toKey, |
2835 |
boolean toInclusive) { |
2836 |
if (fromKey == null || toKey == null) |
2837 |
throw new NullPointerException(); |
2838 |
return newSubMap(fromKey, fromInclusive, toKey, toInclusive); |
2839 |
} |
2840 |
|
2841 |
public SubMap<K,V> headMap(K toKey, |
2842 |
boolean inclusive) { |
2843 |
if (toKey == null) |
2844 |
throw new NullPointerException(); |
2845 |
return newSubMap(null, false, toKey, inclusive); |
2846 |
} |
2847 |
|
2848 |
public SubMap<K,V> tailMap(K fromKey, |
2849 |
boolean inclusive) { |
2850 |
if (fromKey == null) |
2851 |
throw new NullPointerException(); |
2852 |
return newSubMap(fromKey, inclusive, null, false); |
2853 |
} |
2854 |
|
2855 |
public SubMap<K,V> subMap(K fromKey, K toKey) { |
2856 |
return subMap(fromKey, true, toKey, false); |
2857 |
} |
2858 |
|
2859 |
public SubMap<K,V> headMap(K toKey) { |
2860 |
return headMap(toKey, false); |
2861 |
} |
2862 |
|
2863 |
public SubMap<K,V> tailMap(K fromKey) { |
2864 |
return tailMap(fromKey, true); |
2865 |
} |
2866 |
|
2867 |
public SubMap<K,V> descendingMap() { |
2868 |
return new SubMap<K,V>(m, lo, loInclusive, |
2869 |
hi, hiInclusive, !isDescending); |
2870 |
} |
2871 |
|
2872 |
/* ---------------- Relational methods -------------- */ |
2873 |
|
2874 |
public Map.Entry<K,V> ceilingEntry(K key) { |
2875 |
return getNearEntry(key, GT|EQ); |
2876 |
} |
2877 |
|
2878 |
public K ceilingKey(K key) { |
2879 |
return getNearKey(key, GT|EQ); |
2880 |
} |
2881 |
|
2882 |
public Map.Entry<K,V> lowerEntry(K key) { |
2883 |
return getNearEntry(key, LT); |
2884 |
} |
2885 |
|
2886 |
public K lowerKey(K key) { |
2887 |
return getNearKey(key, LT); |
2888 |
} |
2889 |
|
2890 |
public Map.Entry<K,V> floorEntry(K key) { |
2891 |
return getNearEntry(key, LT|EQ); |
2892 |
} |
2893 |
|
2894 |
public K floorKey(K key) { |
2895 |
return getNearKey(key, LT|EQ); |
2896 |
} |
2897 |
|
2898 |
public Map.Entry<K,V> higherEntry(K key) { |
2899 |
return getNearEntry(key, GT); |
2900 |
} |
2901 |
|
2902 |
public K higherKey(K key) { |
2903 |
return getNearKey(key, GT); |
2904 |
} |
2905 |
|
2906 |
public K firstKey() { |
2907 |
return isDescending ? highestKey() : lowestKey(); |
2908 |
} |
2909 |
|
2910 |
public K lastKey() { |
2911 |
return isDescending ? lowestKey() : highestKey(); |
2912 |
} |
2913 |
|
2914 |
public Map.Entry<K,V> firstEntry() { |
2915 |
return isDescending ? highestEntry() : lowestEntry(); |
2916 |
} |
2917 |
|
2918 |
public Map.Entry<K,V> lastEntry() { |
2919 |
return isDescending ? lowestEntry() : highestEntry(); |
2920 |
} |
2921 |
|
2922 |
public Map.Entry<K,V> pollFirstEntry() { |
2923 |
return isDescending ? removeHighest() : removeLowest(); |
2924 |
} |
2925 |
|
2926 |
public Map.Entry<K,V> pollLastEntry() { |
2927 |
return isDescending ? removeLowest() : removeHighest(); |
2928 |
} |
2929 |
|
2930 |
/* ---------------- Submap Views -------------- */ |
2931 |
|
2932 |
public NavigableSet<K> keySet() { |
2933 |
KeySet<K> ks = keySetView; |
2934 |
return (ks != null) ? ks : (keySetView = new KeySet<K>(this)); |
2935 |
} |
2936 |
|
2937 |
public NavigableSet<K> navigableKeySet() { |
2938 |
KeySet<K> ks = keySetView; |
2939 |
return (ks != null) ? ks : (keySetView = new KeySet<K>(this)); |
2940 |
} |
2941 |
|
2942 |
public Collection<V> values() { |
2943 |
Collection<V> vs = valuesView; |
2944 |
return (vs != null) ? vs : (valuesView = new Values<V>(this)); |
2945 |
} |
2946 |
|
2947 |
public Set<Map.Entry<K,V>> entrySet() { |
2948 |
Set<Map.Entry<K,V>> es = entrySetView; |
2949 |
return (es != null) ? es : (entrySetView = new EntrySet<K,V>(this)); |
2950 |
} |
2951 |
|
2952 |
public NavigableSet<K> descendingKeySet() { |
2953 |
return descendingMap().navigableKeySet(); |
2954 |
} |
2955 |
|
2956 |
Iterator<K> keyIterator() { |
2957 |
return new SubMapKeyIterator(); |
2958 |
} |
2959 |
|
2960 |
Iterator<V> valueIterator() { |
2961 |
return new SubMapValueIterator(); |
2962 |
} |
2963 |
|
2964 |
Iterator<Map.Entry<K,V>> entryIterator() { |
2965 |
return new SubMapEntryIterator(); |
2966 |
} |
2967 |
|
2968 |
/** |
2969 |
* Variant of main Iter class to traverse through submaps. |
2970 |
*/ |
2971 |
abstract class SubMapIter<T> implements Iterator<T> { |
2972 |
/** the last node returned by next() */ |
2973 |
Node<K,V> lastReturned; |
2974 |
/** the next node to return from next(); */ |
2975 |
Node<K,V> next; |
2976 |
/** Cache of next value field to maintain weak consistency */ |
2977 |
V nextValue; |
2978 |
|
2979 |
SubMapIter() { |
2980 |
for (;;) { |
2981 |
next = isDescending ? hiNode() : loNode(); |
2982 |
if (next == null) |
2983 |
break; |
2984 |
Object x = next.value; |
2985 |
if (x != null && x != next) { |
2986 |
if (! inBounds(next.key)) |
2987 |
next = null; |
2988 |
else |
2989 |
nextValue = (V) x; |
2990 |
break; |
2991 |
} |
2992 |
} |
2993 |
} |
2994 |
|
2995 |
public final boolean hasNext() { |
2996 |
return next != null; |
2997 |
} |
2998 |
|
2999 |
final void advance() { |
3000 |
if (next == null) |
3001 |
throw new NoSuchElementException(); |
3002 |
lastReturned = next; |
3003 |
if (isDescending) |
3004 |
descend(); |
3005 |
else |
3006 |
ascend(); |
3007 |
} |
3008 |
|
3009 |
private void ascend() { |
3010 |
for (;;) { |
3011 |
next = next.next; |
3012 |
if (next == null) |
3013 |
break; |
3014 |
Object x = next.value; |
3015 |
if (x != null && x != next) { |
3016 |
if (tooHigh(next.key)) |
3017 |
next = null; |
3018 |
else |
3019 |
nextValue = (V) x; |
3020 |
break; |
3021 |
} |
3022 |
} |
3023 |
} |
3024 |
|
3025 |
private void descend() { |
3026 |
for (;;) { |
3027 |
next = m.findNear(lastReturned.key, LT); |
3028 |
if (next == null) |
3029 |
break; |
3030 |
Object x = next.value; |
3031 |
if (x != null && x != next) { |
3032 |
if (tooLow(next.key)) |
3033 |
next = null; |
3034 |
else |
3035 |
nextValue = (V) x; |
3036 |
break; |
3037 |
} |
3038 |
} |
3039 |
} |
3040 |
|
3041 |
public void remove() { |
3042 |
Node<K,V> l = lastReturned; |
3043 |
if (l == null) |
3044 |
throw new IllegalStateException(); |
3045 |
m.remove(l.key); |
3046 |
lastReturned = null; |
3047 |
} |
3048 |
|
3049 |
} |
3050 |
|
3051 |
final class SubMapValueIterator extends SubMapIter<V> { |
3052 |
public V next() { |
3053 |
V v = nextValue; |
3054 |
advance(); |
3055 |
return v; |
3056 |
} |
3057 |
} |
3058 |
|
3059 |
final class SubMapKeyIterator extends SubMapIter<K> { |
3060 |
public K next() { |
3061 |
Node<K,V> n = next; |
3062 |
advance(); |
3063 |
return n.key; |
3064 |
} |
3065 |
} |
3066 |
|
3067 |
final class SubMapEntryIterator extends SubMapIter<Map.Entry<K,V>> { |
3068 |
public Map.Entry<K,V> next() { |
3069 |
Node<K,V> n = next; |
3070 |
V v = nextValue; |
3071 |
advance(); |
3072 |
return new AbstractMap.SimpleImmutableEntry<K,V>(n.key, v); |
3073 |
} |
3074 |
} |
3075 |
} |
3076 |
|
3077 |
// Unsafe mechanics |
3078 |
private static final sun.misc.Unsafe UNSAFE; |
3079 |
private static final long headOffset; |
3080 |
static { |
3081 |
try { |
3082 |
UNSAFE = sun.misc.Unsafe.getUnsafe(); |
3083 |
Class<?> k = ConcurrentSkipListMap.class; |
3084 |
headOffset = UNSAFE.objectFieldOffset |
3085 |
(k.getDeclaredField("head")); |
3086 |
} catch (Exception e) { |
3087 |
throw new Error(e); |
3088 |
} |
3089 |
} |
3090 |
} |