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/licenses/publicdomain |
5 |
*/ |
6 |
|
7 |
package jsr166x; |
8 |
|
9 |
import java.util.*; |
10 |
import java.util.concurrent.*; |
11 |
import java.util.concurrent.atomic.*; |
12 |
|
13 |
/** |
14 |
* A scalable {@link SortedMap} and {@link ConcurrentMap} |
15 |
* implementation. This class maintains a map in ascending key order, |
16 |
* sorted according to the <i>natural order</i> for the key's class |
17 |
* (see {@link Comparable}), or by the {@link Comparator} provided at |
18 |
* creation time, depending on which constructor is used. |
19 |
* |
20 |
* <p>This class implements a concurrent variant of <a |
21 |
* href="http://www.cs.umd.edu/~pugh/">SkipLists</a> providing |
22 |
* expected average <i>log(n)</i> time cost for the |
23 |
* <tt>containsKey</tt>, <tt>get</tt>, <tt>put</tt> and |
24 |
* <tt>remove</tt> operations and their variants. Insertion, removal, |
25 |
* update, and access operations safely execute concurrently by |
26 |
* multiple threads. Iterators are <i>weakly consistent</i>, returning |
27 |
* elements reflecting the state of the map at some point at or since |
28 |
* the creation of the iterator. They do <em>not</em> throw {@link |
29 |
* ConcurrentModificationException}, and may proceed concurrently with |
30 |
* other operations. |
31 |
* |
32 |
* <p>This class provides extended <tt>SortedMap</tt> methods |
33 |
* returning <tt>Map.Entry</tt> key-value pairs that may be useful in |
34 |
* searching for closest matches. Methods <tt>lowerEntry</tt>, |
35 |
* <tt>floorEntry</tt>, <tt>ceilingEntry</tt>, and |
36 |
* <tt>higherEntry</tt> return entries associated with keys |
37 |
* respectively less, less than or equal, greater than or equal, and |
38 |
* greater than a given key, returning null if there is no such key. |
39 |
* These methods are designed for locating, not traversing entries. To |
40 |
* traverse, use view iterators and/or <tt>submap</tt>. The class |
41 |
* additionally supports method <tt>removeFirstEntry</tt> that |
42 |
* atomically returns and removes the first mapping (i.e., with least |
43 |
* key), if one exists. |
44 |
* |
45 |
* <p> All <tt>Map.Entry</tt> pairs returned by methods in this class |
46 |
* and its views represent snapshots of mappings at the time they were |
47 |
* produced. They do <em>not</em> support the <tt>Entry.setValue</tt> |
48 |
* method. (Note however that it is possible to change mappings in the |
49 |
* associated map using <tt>put</tt>, <tt>putIfAbsent</tt>, or |
50 |
* <tt>replace</tt>, depending on exactly which effect you need.) |
51 |
* |
52 |
* <p>The {@link ConcurrentSkipListSubMap} objects returned by methods |
53 |
* <tt>submap</tt>, <tt>headMap</tt>, and <tt>tailMap</tt> support the |
54 |
* same extended set of operations as this class, but operate on their |
55 |
* designated subrange of mappings. |
56 |
* |
57 |
* <p>Beware that, unlike in most collections, the <tt>size</tt> |
58 |
* method is <em>not</em> a constant-time operation. Because of the |
59 |
* asynchronous nature of these maps, determining the current number |
60 |
* of elements requires a traversal of the elements. |
61 |
* |
62 |
* <p>This class and its views and iterators implement all of the |
63 |
* <em>optional</em> methods of the {@link Map} and {@link Iterator} |
64 |
* interfaces. Like most other concurrent collections, this class does |
65 |
* not permit the use of <tt>null</tt> keys or values because some |
66 |
* null return values cannot be reliably distinguished from the |
67 |
* absence of elements. |
68 |
* |
69 |
* @author Doug Lea |
70 |
* @param <K> the type of keys maintained by this map |
71 |
* @param <V> the type of mapped values |
72 |
*/ |
73 |
public class ConcurrentSkipListMap<K,V> extends AbstractMap<K,V> |
74 |
implements ConcurrentMap<K,V>, |
75 |
SortedMap<K,V>, |
76 |
Cloneable, |
77 |
java.io.Serializable { |
78 |
/* |
79 |
* This class implements a tree-like two-dimensionally linked skip |
80 |
* list in which the index levels are represented in separate |
81 |
* nodes from the base nodes holding data. There are two reasons |
82 |
* for taking this approach instead of the usual array-based |
83 |
* structure: 1) Array based implementations seem to encounter |
84 |
* more complexity and overhead 2) We can use cheaper algorithms |
85 |
* for the heavily-traversed index lists than can be used for the |
86 |
* base lists. Here's a picture of some of the basics for a |
87 |
* possible list with 2 levels of index: |
88 |
* |
89 |
* Head nodes Index nodes |
90 |
* +-+ right +-+ +-+ |
91 |
* |2|---------------->| |--------------------->| |->null |
92 |
* +-+ +-+ +-+ |
93 |
* | down | | |
94 |
* v v v |
95 |
* +-+ +-+ +-+ +-+ +-+ +-+ |
96 |
* |1|----------->| |->| |------>| |----------->| |------>| |->null |
97 |
* +-+ +-+ +-+ +-+ +-+ +-+ |
98 |
* | | | | | | |
99 |
* v Nodes v v v v v |
100 |
* +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ |
101 |
* | |->|A|->|B|->|C|->|D|->|E|->|F|->|G|->|H|->|I|->|J|->|K|->null |
102 |
* +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ +-+ |
103 |
* |
104 |
* The base lists use a variant of the HM linked ordered set |
105 |
* algorithm (See Tim Harris, "A pragmatic implementation of |
106 |
* non-blocking linked lists" |
107 |
* http://www.cl.cam.ac.uk/~tlh20/publications.html and Maged |
108 |
* Michael "High Performance Dynamic Lock-Free Hash Tables and |
109 |
* List-Based Sets" |
110 |
* http://www.research.ibm.com/people/m/michael/pubs.htm). The |
111 |
* basic idea in these lists is to mark pointers of deleted nodes |
112 |
* when deleting, and when traversing to keep track of triples |
113 |
* (predecessor, node, successor) in order to detect when and how |
114 |
* to unlink these deleted nodes. |
115 |
* |
116 |
* Rather than using mark-bits to mark list deletions (which can |
117 |
* be slow and space-intensive using AtomicMarkedReference), nodes |
118 |
* use direct CAS'able next pointers. On deletion, instead of |
119 |
* marking a pointer, they splice in another node that can be |
120 |
* thought of as standing for a marked pointer (indicating this by |
121 |
* using otherwise impossible field values). Using plain nodes |
122 |
* acts roughly like "boxed" implementations of marked pointers, |
123 |
* but uses new nodes only when nodes are deleted, not for every |
124 |
* link. This requires less space and supports faster |
125 |
* traversal. Even if marked references were better supported by |
126 |
* JVMs, traversal using this technique might still be faster |
127 |
* because any search need only read ahead one more node than |
128 |
* otherwise required (to check for trailing marker) rather than |
129 |
* unmasking mark bits or whatever on each read. |
130 |
* |
131 |
* This approach maintains the essential property needed in the HM |
132 |
* algorithm of changing the next-pointer of a deleted node so |
133 |
* that any other CAS of it will fail, but implements the idea by |
134 |
* changing the pointer to point to a different node, not by |
135 |
* marking it. While it would be possible to further squeeze |
136 |
* space by defining marker nodes not to have key/value fields, it |
137 |
* isn't worth the extra type-testing overhead. The deletion |
138 |
* markers are rarely encountered during traversal and are |
139 |
* normally quickly garbage collected. |
140 |
* |
141 |
* In addition to using deletion markers, the lists also use |
142 |
* nullness of value fields to indicate deletion, in a style |
143 |
* similar to typical lazy-deletion schemes. If a node's value is |
144 |
* null, then it is considered logically deleted and ignored even |
145 |
* though it is still reachable. This maintains proper control of |
146 |
* concurrent replace vs delete operations -- an attempted replace |
147 |
* must fail if a delete beat it by nulling field, and a delete |
148 |
* must return the last non-null value held in the field. (Note: |
149 |
* Null, rather than some special marker, is used for value fields |
150 |
* here because it just so happens to mesh with the Map API |
151 |
* requirement that method get returns null if there is no |
152 |
* mapping, which allows nodes to remain concurrently readable |
153 |
* even when deleted. Using any other marker value here would be |
154 |
* messy at best.) |
155 |
* |
156 |
* Here's the sequence of events for a deletion of node n with |
157 |
* predecessor b and successor f, initially: |
158 |
* |
159 |
* +------+ +------+ +------+ |
160 |
* ... | b |------>| n |----->| f | ... |
161 |
* +------+ +------+ +------+ |
162 |
* |
163 |
* 1. CAS n's value field from non-null to null. |
164 |
* From this point on, no public operations encountering |
165 |
* the node consider this mapping to exist. However, other |
166 |
* ongoing insertions and deletions might still modify |
167 |
* n's next pointer. |
168 |
* |
169 |
* 2. CAS n's next pointer to point to a new marker node. |
170 |
* From this point on, no other nodes can be appended to n. |
171 |
* which avoids deletion errors in CAS-based linked lists. |
172 |
* |
173 |
* +------+ +------+ +------+ +------+ |
174 |
* ... | b |------>| n |----->|marker|------>| f | ... |
175 |
* +------+ +------+ +------+ +------+ |
176 |
* |
177 |
* 3. CAS b's next pointer over both n and its marker. |
178 |
* From this point on, no new traversals will encounter n, |
179 |
* and it can eventually be GCed. |
180 |
* +------+ +------+ |
181 |
* ... | b |----------------------------------->| f | ... |
182 |
* +------+ +------+ |
183 |
* |
184 |
* A failure at step 1 leads to simple retry due to a lost race |
185 |
* with another operation. Steps 2-3 can fail because some other |
186 |
* thread noticed during a traversal a node with null value and |
187 |
* helped out by marking and/or unlinking. This helping-out |
188 |
* ensures that no thread can become stuck waiting for progress of |
189 |
* the deleting thread. The use of marker nodes slightly |
190 |
* complicates helping-out code because traversals must track |
191 |
* consistent reads of up to four nodes (b, n, marker, f), not |
192 |
* just (b, n, f), although the next field of a marker is |
193 |
* immutable, and once a next field is CAS'ed to point to a |
194 |
* marker, it never again changes, so this requires less care. |
195 |
* |
196 |
* Skip lists add indexing to this scheme, so that the base-level |
197 |
* traversals start close to the locations being found, inserted |
198 |
* or deleted -- usually base level traversals only traverse a few |
199 |
* nodes. This doesn't change the basic algorithm except for the |
200 |
* need to make sure base traversals start at predecessors (here, |
201 |
* b) that are not (structurally) deleted, otherwise retrying |
202 |
* after processing the deletion. |
203 |
* |
204 |
* Index levels are maintained as lists with volatile next fields, |
205 |
* using CAS to link and unlink. Races are allowed in index-list |
206 |
* operations that can (rarely) fail to link in a new index node |
207 |
* or delete one. (We can't do this of course for data nodes.) |
208 |
* However, even when this happens, the index lists remain sorted, |
209 |
* so correctly serve as indices. This can impact performance, |
210 |
* but since skip lists are probabilistic anyway, the net result |
211 |
* is that under contention, the effective "p" value may be lower |
212 |
* than its nominal value. And race windows are kept small enough |
213 |
* that in practice these failures are rare, even under a lot of |
214 |
* contention. |
215 |
* |
216 |
* The fact that retries (for both base and index lists) are |
217 |
* relatively cheap due to indexing allows some minor |
218 |
* simplifications of retry logic. Traversal restarts are |
219 |
* performed after most "helping-out" CASes. This isn't always |
220 |
* strictly necessary, but the implicit backoffs tend to help |
221 |
* reduce other downstream failed CAS's enough to outweigh restart |
222 |
* cost. This worsens the worst case, but seems to improve even |
223 |
* highly contended cases. |
224 |
* |
225 |
* Unlike most skip-list implementations, index insertion and |
226 |
* deletion here require a separate traversal pass occuring after |
227 |
* the base-level action, to add or remove index nodes. This adds |
228 |
* to single-threaded overhead, but improves contended |
229 |
* multithreaded performance by narrowing interference windows, |
230 |
* and allows deletion to ensure that all index nodes will be made |
231 |
* unreachable upon return from a public remove operation, thus |
232 |
* avoiding unwanted garbage retention. This is more important |
233 |
* here than in some other data structures because we cannot null |
234 |
* out node fields referencing user keys since they might still be |
235 |
* read by other ongoing traversals. |
236 |
* |
237 |
* Indexing uses skip list parameters that maintain good search |
238 |
* performance while using sparser-than-usual indices: The |
239 |
* hardwired parameters k=1, p=0.5 (see method randomLevel) mean |
240 |
* that about one-quarter of the nodes have indices. Of those that |
241 |
* do, half have one level, a quarter have two, and so on (see |
242 |
* Pugh's Skip List Cookbook, sec 3.4). The expected total space |
243 |
* requirement for a map is slightly less than for the current |
244 |
* implementation of java.util.TreeMap. |
245 |
* |
246 |
* Changing the level of the index (i.e, the height of the |
247 |
* tree-like structure) also uses CAS. The head index has initial |
248 |
* level/height of one. Creation of an index with height greater |
249 |
* than the current level adds a level to the head index by |
250 |
* CAS'ing on a new top-most head. To maintain good performance |
251 |
* after a lot of removals, deletion methods heuristically try to |
252 |
* reduce the height if the topmost levels appear to be empty. |
253 |
* This may encounter races in which it possible (but rare) to |
254 |
* reduce and "lose" a level just as it is about to contain an |
255 |
* index (that will then never be encountered). This does no |
256 |
* structural harm, and in practice appears to be a better option |
257 |
* than allowing unrestrained growth of levels. |
258 |
* |
259 |
* The code for all this is more verbose than you'd like. Most |
260 |
* operations entail locating an element (or position to insert an |
261 |
* element). The code to do this can't be nicely factored out |
262 |
* because subsequent uses require a snapshot of predecessor |
263 |
* and/or successor and/or value fields which can't be returned |
264 |
* all at once, at least not without creating yet another object |
265 |
* to hold them -- creating such little objects is an especially |
266 |
* bad idea for basic internal search operations because it adds |
267 |
* to GC overhead. (This is one of the few times I've wished Java |
268 |
* had macros.) Instead, some traversal code is interleaved within |
269 |
* insertion and removal operations. The control logic to handle |
270 |
* all the retry conditions is sometimes twisty. Most search is |
271 |
* broken into 2 parts. findPredecessor() searches index nodes |
272 |
* only, returning a base-level predecessor of the key. findNode() |
273 |
* finishes out the base-level search. Even with this factoring, |
274 |
* there is a fair amount of near-duplication of code to handle |
275 |
* variants. |
276 |
* |
277 |
* For explanation of algorithms sharing at least a couple of |
278 |
* features with this one, see Mikhail Fomitchev's thesis |
279 |
* (http://www.cs.yorku.ca/~mikhail/) and Keir Fraser's thesis |
280 |
* (http://www.cl.cam.ac.uk/users/kaf24/). |
281 |
* |
282 |
* Given the use of tree-like index nodes, you might wonder why |
283 |
* this doesn't use some kind of search tree instead, which would |
284 |
* support somewhat faster search operations. The reason is that |
285 |
* there are no known efficient lock-free insertion and deletion |
286 |
* algorithms for search trees. The immutability of the "down" |
287 |
* links of index nodes (as opposed to mutable "left" fields in |
288 |
* true trees) makes this tractable using only CAS operations. |
289 |
* |
290 |
* Notation guide for local variables |
291 |
* Node: b, n, f for predecessor, node, successor |
292 |
* Index: q, r, d for index node, right, down. |
293 |
* t for another index node |
294 |
* Head: h |
295 |
* Levels: j |
296 |
* Keys: k, key |
297 |
* Values: v, value |
298 |
* Comparisons: c |
299 |
*/ |
300 |
|
301 |
private static final long serialVersionUID = -8627078645895051609L; |
302 |
|
303 |
/** |
304 |
* Special value used to identify base-level header |
305 |
*/ |
306 |
private static final Object BASE_HEADER = new Object(); |
307 |
|
308 |
/** |
309 |
* The topmost head index of the skiplist. |
310 |
*/ |
311 |
private transient volatile HeadIndex<K,V> head; |
312 |
|
313 |
/** |
314 |
* The Comparator used to maintain order in this Map, or null |
315 |
* if using natural order. |
316 |
* @serial |
317 |
*/ |
318 |
private final Comparator<? super K> comparator; |
319 |
|
320 |
/** |
321 |
* Seed for simple random number generator. Not volatile since it |
322 |
* doesn't matter too much if different threads don't see updates. |
323 |
*/ |
324 |
private transient int randomSeed; |
325 |
|
326 |
/** Lazily initialized key set */ |
327 |
private transient KeySet keySet; |
328 |
/** Lazily initialized entry set */ |
329 |
private transient EntrySet entrySet; |
330 |
/** Lazily initialized values collection */ |
331 |
private transient Values values; |
332 |
|
333 |
/** |
334 |
* Initialize or reset state. Needed by constructors, clone, |
335 |
* clear, readObject. and ConcurrentSkipListSet.clone. |
336 |
* (Note that comparator must be separately initialized.) |
337 |
*/ |
338 |
final void initialize() { |
339 |
keySet = null; |
340 |
entrySet = null; |
341 |
values = null; |
342 |
randomSeed = (int) System.nanoTime(); |
343 |
head = new HeadIndex<K,V>(new Node<K,V>(null, BASE_HEADER, null), |
344 |
null, null, 1); |
345 |
} |
346 |
|
347 |
/** Updater for casHead */ |
348 |
private static final |
349 |
AtomicReferenceFieldUpdater<ConcurrentSkipListMap, HeadIndex> |
350 |
headUpdater = AtomicReferenceFieldUpdater.newUpdater |
351 |
(ConcurrentSkipListMap.class, HeadIndex.class, "head"); |
352 |
|
353 |
/** |
354 |
* compareAndSet head node |
355 |
*/ |
356 |
private boolean casHead(HeadIndex<K,V> cmp, HeadIndex<K,V> val) { |
357 |
return headUpdater.compareAndSet(this, cmp, val); |
358 |
} |
359 |
|
360 |
/* ---------------- Nodes -------------- */ |
361 |
|
362 |
/** |
363 |
* Nodes hold keys and values, and are singly linked in sorted |
364 |
* order, possibly with some intervening marker nodes. The list is |
365 |
* headed by a dummy node accessible as head.node. The value field |
366 |
* is declared only as Object because it takes special non-V |
367 |
* values for marker and header nodes. |
368 |
*/ |
369 |
static final class Node<K,V> { |
370 |
final K key; |
371 |
volatile Object value; |
372 |
volatile Node<K,V> next; |
373 |
|
374 |
/** |
375 |
* Creates a new regular node. |
376 |
*/ |
377 |
Node(K key, Object value, Node<K,V> next) { |
378 |
this.key = key; |
379 |
this.value = value; |
380 |
this.next = next; |
381 |
} |
382 |
|
383 |
/** |
384 |
* Creates a new marker node. A marker is distinguished by |
385 |
* having its value field point to itself. Marker nodes also |
386 |
* have null keys, a fact that is exploited in a few places, |
387 |
* but this doesn't distinguish markers from the base-level |
388 |
* header node (head.node), which also has a null key. |
389 |
*/ |
390 |
Node(Node<K,V> next) { |
391 |
this.key = null; |
392 |
this.value = this; |
393 |
this.next = next; |
394 |
} |
395 |
|
396 |
/** Updater for casNext */ |
397 |
static final AtomicReferenceFieldUpdater<Node, Node> |
398 |
nextUpdater = AtomicReferenceFieldUpdater.newUpdater |
399 |
(Node.class, Node.class, "next"); |
400 |
|
401 |
/** Updater for casValue */ |
402 |
static final AtomicReferenceFieldUpdater<Node, Object> |
403 |
valueUpdater = AtomicReferenceFieldUpdater.newUpdater |
404 |
(Node.class, Object.class, "value"); |
405 |
|
406 |
|
407 |
/** |
408 |
* compareAndSet value field |
409 |
*/ |
410 |
boolean casValue(Object cmp, Object val) { |
411 |
return valueUpdater.compareAndSet(this, cmp, val); |
412 |
} |
413 |
|
414 |
/** |
415 |
* compareAndSet next field |
416 |
*/ |
417 |
boolean casNext(Node<K,V> cmp, Node<K,V> val) { |
418 |
return nextUpdater.compareAndSet(this, cmp, val); |
419 |
} |
420 |
|
421 |
/** |
422 |
* Return true if this node is a marker. This method isn't |
423 |
* actually called in an any current code checking for markers |
424 |
* because callers will have already read value field and need |
425 |
* to use that read (not another done here) and so directly |
426 |
* test if value points to node. |
427 |
* @param n a possibly null reference to a node |
428 |
* @return true if this node is a marker node |
429 |
*/ |
430 |
boolean isMarker() { |
431 |
return value == this; |
432 |
} |
433 |
|
434 |
/** |
435 |
* Return true if this node is the header of base-level list. |
436 |
* @return true if this node is header node |
437 |
*/ |
438 |
boolean isBaseHeader() { |
439 |
return value == BASE_HEADER; |
440 |
} |
441 |
|
442 |
/** |
443 |
* Tries to append a deletion marker to this node. |
444 |
* @param f the assumed current successor of this node |
445 |
* @return true if successful |
446 |
*/ |
447 |
boolean appendMarker(Node<K,V> f) { |
448 |
return casNext(f, new Node<K,V>(f)); |
449 |
} |
450 |
|
451 |
/** |
452 |
* Helps out a deletion by appending marker or unlinking from |
453 |
* predecessor. This is called during traversals when value |
454 |
* field seen to be null. |
455 |
* @param b predecessor |
456 |
* @param f successor |
457 |
*/ |
458 |
void helpDelete(Node<K,V> b, Node<K,V> f) { |
459 |
/* |
460 |
* Rechecking links and then doing only one of the |
461 |
* help-out stages per call tends to minimize CAS |
462 |
* interference among helping threads. |
463 |
*/ |
464 |
if (f == next && this == b.next) { |
465 |
if (f == null || f.value != f) // not already marked |
466 |
appendMarker(f); |
467 |
else |
468 |
b.casNext(this, f.next); |
469 |
} |
470 |
} |
471 |
|
472 |
/** |
473 |
* Return value if this node contains a valid key-value pair, |
474 |
* else null. |
475 |
* @return this node's value if it isn't a marker or header or |
476 |
* is deleted, else null. |
477 |
*/ |
478 |
V getValidValue() { |
479 |
Object v = value; |
480 |
if (v == this || v == BASE_HEADER) |
481 |
return null; |
482 |
return (V)v; |
483 |
} |
484 |
|
485 |
/** |
486 |
* Create and return a new SnapshotEntry holding current |
487 |
* mapping if this node holds a valid value, else null |
488 |
* @return new entry or null |
489 |
*/ |
490 |
SnapshotEntry<K,V> createSnapshot() { |
491 |
V v = getValidValue(); |
492 |
if (v == null) |
493 |
return null; |
494 |
return new SnapshotEntry(key, v); |
495 |
} |
496 |
} |
497 |
|
498 |
/* ---------------- Indexing -------------- */ |
499 |
|
500 |
/** |
501 |
* Index nodes represent the levels of the skip list. To improve |
502 |
* search performance, keys of the underlying nodes are cached. |
503 |
* Note that even though both Nodes and Indexes have |
504 |
* forward-pointing fields, they have different types and are |
505 |
* handled in different ways, that can't nicely be captured by |
506 |
* placing field in a shared abstract class. |
507 |
*/ |
508 |
static class Index<K,V> { |
509 |
final K key; |
510 |
final Node<K,V> node; |
511 |
final Index<K,V> down; |
512 |
volatile Index<K,V> right; |
513 |
|
514 |
/** |
515 |
* Creates index node with unknown right pointer |
516 |
*/ |
517 |
Index(Node<K,V> node, Index<K,V> down) { |
518 |
this.node = node; |
519 |
this.key = node.key; |
520 |
this.down = down; |
521 |
} |
522 |
|
523 |
/** |
524 |
* Creates index node with known right pointer |
525 |
*/ |
526 |
Index(Node<K,V> node, Index<K,V> down, Index<K,V> right) { |
527 |
this.node = node; |
528 |
this.key = node.key; |
529 |
this.down = down; |
530 |
this.right = right; |
531 |
} |
532 |
|
533 |
/** Updater for casRight */ |
534 |
static final AtomicReferenceFieldUpdater<Index, Index> |
535 |
rightUpdater = AtomicReferenceFieldUpdater.newUpdater |
536 |
(Index.class, Index.class, "right"); |
537 |
|
538 |
/** |
539 |
* compareAndSet right field |
540 |
*/ |
541 |
final boolean casRight(Index<K,V> cmp, Index<K,V> val) { |
542 |
return rightUpdater.compareAndSet(this, cmp, val); |
543 |
} |
544 |
|
545 |
/** |
546 |
* Returns true if the node this indexes has been deleted. |
547 |
* @return true if indexed node is known to be deleted |
548 |
*/ |
549 |
final boolean indexesDeletedNode() { |
550 |
return node.value == null; |
551 |
} |
552 |
|
553 |
/** |
554 |
* Tries to CAS newSucc as successor. To minimize races with |
555 |
* unlink that may lose this index node, if the node being |
556 |
* indexed is known to be deleted, it doesn't try to link in. |
557 |
* @param succ the expected current successor |
558 |
* @param newSucc the new successor |
559 |
* @return true if successful |
560 |
*/ |
561 |
final boolean link(Index<K,V> succ, Index<K,V> newSucc) { |
562 |
Node<K,V> n = node; |
563 |
newSucc.right = succ; |
564 |
return n.value != null && casRight(succ, newSucc); |
565 |
} |
566 |
|
567 |
/** |
568 |
* Tries to CAS right field to skip over apparent successor |
569 |
* succ. Fails (forcing a retraversal by caller) if this node |
570 |
* is known to be deleted. |
571 |
* @param succ the expected current successor |
572 |
* @return true if successful |
573 |
*/ |
574 |
final boolean unlink(Index<K,V> succ) { |
575 |
return !indexesDeletedNode() && casRight(succ, succ.right); |
576 |
} |
577 |
} |
578 |
|
579 |
/* ---------------- Head nodes -------------- */ |
580 |
|
581 |
/** |
582 |
* Nodes heading each level keep track of their level. |
583 |
*/ |
584 |
static final class HeadIndex<K,V> extends Index<K,V> { |
585 |
final int level; |
586 |
HeadIndex(Node<K,V> node, Index<K,V> down, Index<K,V> right, |
587 |
int level) { |
588 |
super(node, down, right); |
589 |
this.level = level; |
590 |
} |
591 |
} |
592 |
|
593 |
/* ---------------- Map.Entry support -------------- */ |
594 |
|
595 |
/** |
596 |
* An immutable representation of a key-value mapping as it |
597 |
* existed at some point in time. This class does <em>not</em> |
598 |
* support the <tt>Map.Entry.setValue</tt> method. |
599 |
*/ |
600 |
static class SnapshotEntry<K,V> implements Map.Entry<K,V> { |
601 |
private final K key; |
602 |
private final V value; |
603 |
|
604 |
/** |
605 |
* Creates a new entry representing the given key and value. |
606 |
* @param key the key |
607 |
* @param value the value |
608 |
*/ |
609 |
SnapshotEntry(K key, V value) { |
610 |
this.key = key; |
611 |
this.value = value; |
612 |
} |
613 |
|
614 |
/** |
615 |
* Returns the key corresponding to this entry. |
616 |
* |
617 |
* @return the key corresponding to this entry. |
618 |
*/ |
619 |
public K getKey() { |
620 |
return key; |
621 |
} |
622 |
|
623 |
/** |
624 |
* Returns the value corresponding to this entry. |
625 |
* |
626 |
* @return the value corresponding to this entry. |
627 |
*/ |
628 |
public V getValue() { |
629 |
return value; |
630 |
} |
631 |
|
632 |
/** |
633 |
* Always fails, throwing <tt>UnsupportedOperationException</tt>. |
634 |
* @throws UnsupportedOperationException always. |
635 |
*/ |
636 |
public V setValue(V value) { |
637 |
throw new UnsupportedOperationException(); |
638 |
} |
639 |
|
640 |
// inherit javadoc |
641 |
public boolean equals(Object o) { |
642 |
if (!(o instanceof Map.Entry)) |
643 |
return false; |
644 |
Map.Entry e = (Map.Entry)o; |
645 |
// As mandated by Map.Entry spec: |
646 |
return ((key==null ? |
647 |
e.getKey()==null : key.equals(e.getKey())) && |
648 |
(value==null ? |
649 |
e.getValue()==null : value.equals(e.getValue()))); |
650 |
} |
651 |
|
652 |
|
653 |
// inherit javadoc |
654 |
public int hashCode() { |
655 |
// As mandated by Map.Entry spec: |
656 |
return ((key==null ? 0 : key.hashCode()) ^ |
657 |
(value==null ? 0 : value.hashCode())); |
658 |
} |
659 |
|
660 |
/** |
661 |
* Returns a String consisting of the key followed by an |
662 |
* equals sign (<tt>"="</tt>) followed by the associated |
663 |
* value. |
664 |
* @return a String representation of this entry. |
665 |
*/ |
666 |
public String toString() { |
667 |
return getKey() + "=" + getValue(); |
668 |
} |
669 |
} |
670 |
|
671 |
/* ---------------- Comparison utilities -------------- */ |
672 |
|
673 |
/** |
674 |
* Represents a key with a comparator as a Comparable. |
675 |
* |
676 |
* Because most sorted collections seem to use natural order on |
677 |
* Comparables (Strings, Integers, etc), most internal methods are |
678 |
* geared to use them. This is generally faster than checking |
679 |
* per-comparison whether to use comparator or comparable because |
680 |
* it doesn't require a (Comparable) cast for each comparison. |
681 |
* (Optimizers can only sometimes remove such redundant checks |
682 |
* themselves.) When Comparators are used, |
683 |
* ComparableUsingComparators are created so that they act in the |
684 |
* same way as natural orderings. This penalizes use of |
685 |
* Comparators vs Comparables, which seems like the right |
686 |
* tradeoff. |
687 |
*/ |
688 |
static final class ComparableUsingComparator<K> implements Comparable<K> { |
689 |
final K actualKey; |
690 |
final Comparator<? super K> cmp; |
691 |
ComparableUsingComparator(K key, Comparator<? super K> cmp) { |
692 |
this.actualKey = key; |
693 |
this.cmp = cmp; |
694 |
} |
695 |
public int compareTo(K k2) { |
696 |
return cmp.compare(actualKey, k2); |
697 |
} |
698 |
} |
699 |
|
700 |
/** |
701 |
* If using comparator, return a ComparableUsingComparator, else |
702 |
* cast key as Comparator, which may cause ClassCastException, |
703 |
* which is propagated back to caller. |
704 |
*/ |
705 |
private Comparable<K> comparable(Object key) throws ClassCastException { |
706 |
if (key == null) |
707 |
throw new NullPointerException(); |
708 |
return (comparator != null) |
709 |
? new ComparableUsingComparator(key, comparator) |
710 |
: (Comparable<K>)key; |
711 |
} |
712 |
|
713 |
/** |
714 |
* Compare using comparator or natural ordering. Used when the |
715 |
* ComparableUsingComparator approach doesn't apply. |
716 |
*/ |
717 |
int compare(K k1, K k2) throws ClassCastException { |
718 |
Comparator<? super K> cmp = comparator; |
719 |
if (cmp != null) |
720 |
return cmp.compare(k1, k2); |
721 |
else |
722 |
return ((Comparable<K>)k1).compareTo(k2); |
723 |
} |
724 |
|
725 |
/** |
726 |
* Return true if given key greater than or equal to least and |
727 |
* strictly less than fence. Needed mainly in submap operations. |
728 |
*/ |
729 |
boolean inHalfOpenRange(K key, K least, K fence) { |
730 |
if (key == null) |
731 |
throw new NullPointerException(); |
732 |
return ((least == null || compare(key, least) >= 0) && |
733 |
(fence == null || compare(key, fence) < 0)); |
734 |
} |
735 |
|
736 |
/** |
737 |
* Return true if given key greater than or equal to least and less |
738 |
* or equal to fence. Needed mainly in submap operations. |
739 |
*/ |
740 |
boolean inOpenRange(K key, K least, K fence) { |
741 |
if (key == null) |
742 |
throw new NullPointerException(); |
743 |
return ((least == null || compare(key, least) >= 0) && |
744 |
(fence == null || compare(key, fence) <= 0)); |
745 |
} |
746 |
|
747 |
/* ---------------- Traversal -------------- */ |
748 |
|
749 |
/** |
750 |
* Return a base-level node with key strictly less than given key, |
751 |
* or the base-level header if there is no such node. Also |
752 |
* unlinks indexes to deleted nodes found along the way. Callers |
753 |
* rely on this side-effect of clearing indices to deleted nodes. |
754 |
* @param key the key |
755 |
* @return a predecessor of key |
756 |
*/ |
757 |
private Node<K,V> findPredecessor(Comparable<K> key) { |
758 |
for (;;) { |
759 |
Index<K,V> q = head; |
760 |
for (;;) { |
761 |
Index<K,V> d, r; |
762 |
if ((r = q.right) != null) { |
763 |
if (r.indexesDeletedNode()) { |
764 |
if (q.unlink(r)) |
765 |
continue; // reread r |
766 |
else |
767 |
break; // restart |
768 |
} |
769 |
if (key.compareTo(r.key) > 0) { |
770 |
q = r; |
771 |
continue; |
772 |
} |
773 |
} |
774 |
if ((d = q.down) != null) |
775 |
q = d; |
776 |
else |
777 |
return q.node; |
778 |
} |
779 |
} |
780 |
} |
781 |
|
782 |
/** |
783 |
* Return node holding key or null if no such, clearing out any |
784 |
* deleted nodes seen along the way. Repeatedly traverses at |
785 |
* base-level looking for key starting at predecessor returned |
786 |
* from findPredecessor, processing base-level deletions as |
787 |
* encountered. Some callers rely on this side-effect of clearing |
788 |
* deleted nodes. |
789 |
* |
790 |
* Restarts occur, at traversal step centered on node n, if: |
791 |
* |
792 |
* (1) After reading n's next field, n is no longer assumed |
793 |
* predecessor b's current successor, which means that |
794 |
* we don't have a consistent 3-node snapshot and so cannot |
795 |
* unlink any subsequent deleted nodes encountered. |
796 |
* |
797 |
* (2) n's value field is null, indicating n is deleted, in |
798 |
* which case we help out an ongoing structural deletion |
799 |
* before retrying. Even though there are cases where such |
800 |
* unlinking doesn't require restart, they aren't sorted out |
801 |
* here because doing so would not usually outweigh cost of |
802 |
* restarting. |
803 |
* |
804 |
* (3) n is a marker or n's predecessor's value field is null, |
805 |
* indicating (among other possibilities) that |
806 |
* findPredecessor returned a deleted node. We can't unlink |
807 |
* the node because we don't know its predecessor, so rely |
808 |
* on another call to findPredecessor to notice and return |
809 |
* some earlier predecessor, which it will do. This check is |
810 |
* only strictly needed at beginning of loop, (and the |
811 |
* b.value check isn't strictly needed at all) but is done |
812 |
* each iteration to help avoid contention with other |
813 |
* threads by callers that will fail to be able to change |
814 |
* links, and so will retry anyway. |
815 |
* |
816 |
* The traversal loops in doPut, doRemove, and findNear all |
817 |
* include with the same three kinds of checks. And specialized |
818 |
* versions appear in doRemoveFirstEntry, findFirst, and |
819 |
* findLast. They can't easily share code because each uses the |
820 |
* reads of fields held in locals occurring in the orders they |
821 |
* were performed. |
822 |
* |
823 |
* @param key the key |
824 |
* @return node holding key, or null if no such. |
825 |
*/ |
826 |
private Node<K,V> findNode(Comparable<K> key) { |
827 |
for (;;) { |
828 |
Node<K,V> b = findPredecessor(key); |
829 |
Node<K,V> n = b.next; |
830 |
for (;;) { |
831 |
if (n == null) |
832 |
return null; |
833 |
Node<K,V> f = n.next; |
834 |
if (n != b.next) // inconsistent read |
835 |
break; |
836 |
Object v = n.value; |
837 |
if (v == null) { // n is deleted |
838 |
n.helpDelete(b, f); |
839 |
break; |
840 |
} |
841 |
if (v == n || b.value == null) // b is deleted |
842 |
break; |
843 |
int c = key.compareTo(n.key); |
844 |
if (c < 0) |
845 |
return null; |
846 |
if (c == 0) |
847 |
return n; |
848 |
b = n; |
849 |
n = f; |
850 |
} |
851 |
} |
852 |
} |
853 |
|
854 |
/** |
855 |
* Specialized variant of findNode to perform map.get. Does a weak |
856 |
* traversal, not bothering to fix any deleted index nodes, |
857 |
* returning early if it happens to see key in index, and passing |
858 |
* over any deleted base nodes, falling back to getUsingFindNode |
859 |
* only if it would otherwise return value from an ongoing |
860 |
* deletion. Also uses "bound" to eliminate need for some |
861 |
* comparisons (see Pugh Cookbook). Also folds uses of null checks |
862 |
* and node-skipping because markers have null keys. |
863 |
* @param okey the key |
864 |
* @return the value, or null if absent |
865 |
*/ |
866 |
private V doGet(Object okey) { |
867 |
Comparable<K> key = comparable(okey); |
868 |
K bound = null; |
869 |
Index<K,V> q = head; |
870 |
for (;;) { |
871 |
K rk; |
872 |
Index<K,V> d, r; |
873 |
if ((r = q.right) != null && |
874 |
(rk = r.key) != null && rk != bound) { |
875 |
int c = key.compareTo(rk); |
876 |
if (c > 0) { |
877 |
q = r; |
878 |
continue; |
879 |
} |
880 |
if (c == 0) { |
881 |
Object v = r.node.value; |
882 |
return (v != null)? (V)v : getUsingFindNode(key); |
883 |
} |
884 |
bound = rk; |
885 |
} |
886 |
if ((d = q.down) != null) |
887 |
q = d; |
888 |
else { |
889 |
for (Node<K,V> n = q.node.next; n != null; n = n.next) { |
890 |
K nk = n.key; |
891 |
if (nk != null) { |
892 |
int c = key.compareTo(nk); |
893 |
if (c == 0) { |
894 |
Object v = n.value; |
895 |
return (v != null)? (V)v : getUsingFindNode(key); |
896 |
} |
897 |
if (c < 0) |
898 |
return null; |
899 |
} |
900 |
} |
901 |
return null; |
902 |
} |
903 |
} |
904 |
} |
905 |
|
906 |
/** |
907 |
* Perform map.get via findNode. Used as a backup if doGet |
908 |
* encounters an in-progress deletion. |
909 |
* @param key the key |
910 |
* @return the value, or null if absent |
911 |
*/ |
912 |
private V getUsingFindNode(Comparable<K> key) { |
913 |
// Loop needed here and elsewhere to protect against value |
914 |
// field going null just as it is about to be returned. |
915 |
for (;;) { |
916 |
Node<K,V> n = findNode(key); |
917 |
if (n == null) |
918 |
return null; |
919 |
Object v = n.value; |
920 |
if (v != null) |
921 |
return (V)v; |
922 |
} |
923 |
} |
924 |
|
925 |
/* ---------------- Insertion -------------- */ |
926 |
|
927 |
/** |
928 |
* Main insertion method. Adds element if not present, or |
929 |
* replaces value if present and onlyIfAbsent is false. |
930 |
* @param kkey the key |
931 |
* @param value the value that must be associated with key |
932 |
* @param onlyIfAbsent if should not insert if already present |
933 |
* @return the old value, or null if newly inserted |
934 |
*/ |
935 |
private V doPut(K kkey, V value, boolean onlyIfAbsent) { |
936 |
Comparable<K> key = comparable(kkey); |
937 |
for (;;) { |
938 |
Node<K,V> b = findPredecessor(key); |
939 |
Node<K,V> n = b.next; |
940 |
for (;;) { |
941 |
if (n != null) { |
942 |
Node<K,V> f = n.next; |
943 |
if (n != b.next) // inconsistent read |
944 |
break;; |
945 |
Object v = n.value; |
946 |
if (v == null) { // n is deleted |
947 |
n.helpDelete(b, f); |
948 |
break; |
949 |
} |
950 |
if (v == n || b.value == null) // b is deleted |
951 |
break; |
952 |
int c = key.compareTo(n.key); |
953 |
if (c > 0) { |
954 |
b = n; |
955 |
n = f; |
956 |
continue; |
957 |
} |
958 |
if (c == 0) { |
959 |
if (onlyIfAbsent || n.casValue(v, value)) |
960 |
return (V)v; |
961 |
else |
962 |
break; // restart if lost race to replace value |
963 |
} |
964 |
// else c < 0; fall through |
965 |
} |
966 |
|
967 |
Node<K,V> z = new Node<K,V>(kkey, value, n); |
968 |
if (!b.casNext(n, z)) |
969 |
break; // restart if lost race to append to b |
970 |
int level = randomLevel(); |
971 |
if (level > 0) |
972 |
insertIndex(z, level); |
973 |
return null; |
974 |
} |
975 |
} |
976 |
} |
977 |
|
978 |
/** |
979 |
* Return a random level for inserting a new node. |
980 |
* Hardwired to k=1, p=0.5, max 31. |
981 |
* |
982 |
* This uses a cheap pseudo-random function that according to |
983 |
* http://home1.gte.net/deleyd/random/random4.html was used in |
984 |
* Turbo Pascal. It seems the fastest usable one here. The low |
985 |
* bits are apparently not very random (the original used only |
986 |
* upper 16 bits) so we traverse from highest bit down (i.e., test |
987 |
* sign), thus hardly ever use lower bits. |
988 |
*/ |
989 |
private int randomLevel() { |
990 |
int level = 0; |
991 |
int r = randomSeed; |
992 |
randomSeed = r * 134775813 + 1; |
993 |
if (r < 0) { |
994 |
while ((r <<= 1) > 0) |
995 |
++level; |
996 |
} |
997 |
return level; |
998 |
} |
999 |
|
1000 |
/** |
1001 |
* Create and add index nodes for given node. |
1002 |
* @param z the node |
1003 |
* @param level the level of the index |
1004 |
*/ |
1005 |
private void insertIndex(Node<K,V> z, int level) { |
1006 |
HeadIndex<K,V> h = head; |
1007 |
int max = h.level; |
1008 |
|
1009 |
if (level <= max) { |
1010 |
Index<K,V> idx = null; |
1011 |
for (int i = 1; i <= level; ++i) |
1012 |
idx = new Index<K,V>(z, idx); |
1013 |
addIndex(idx, h, level); |
1014 |
|
1015 |
} else { // Add a new level |
1016 |
/* |
1017 |
* To reduce interference by other threads checking for |
1018 |
* empty levels in tryReduceLevel, new levels are added |
1019 |
* with initialized right pointers. Which in turn requires |
1020 |
* keeping levels in an array to access them while |
1021 |
* creating new head index nodes from the opposite |
1022 |
* direction. |
1023 |
*/ |
1024 |
level = max + 1; |
1025 |
Index<K,V>[] idxs = (Index<K,V>[])new Index[level+1]; |
1026 |
Index<K,V> idx = null; |
1027 |
for (int i = 1; i <= level; ++i) |
1028 |
idxs[i] = idx = new Index<K,V>(z, idx); |
1029 |
|
1030 |
HeadIndex<K,V> oldh; |
1031 |
int k; |
1032 |
for (;;) { |
1033 |
oldh = head; |
1034 |
int oldLevel = oldh.level; |
1035 |
if (level <= oldLevel) { // lost race to add level |
1036 |
k = level; |
1037 |
break; |
1038 |
} |
1039 |
HeadIndex<K,V> newh = oldh; |
1040 |
Node<K,V> oldbase = oldh.node; |
1041 |
for (int j = oldLevel+1; j <= level; ++j) |
1042 |
newh = new HeadIndex<K,V>(oldbase, newh, idxs[j], j); |
1043 |
if (casHead(oldh, newh)) { |
1044 |
k = oldLevel; |
1045 |
break; |
1046 |
} |
1047 |
} |
1048 |
addIndex(idxs[k], oldh, k); |
1049 |
} |
1050 |
} |
1051 |
|
1052 |
/** |
1053 |
* Add given index nodes from given level down to 1. |
1054 |
* @param idx the topmost index node being inserted |
1055 |
* @param h the value of head to use to insert. This must be |
1056 |
* snapshotted by callers to provide correct insertion level |
1057 |
* @param indexLevel the level of the index |
1058 |
*/ |
1059 |
private void addIndex(Index<K,V> idx, HeadIndex<K,V> h, int indexLevel) { |
1060 |
// Track next level to insert in case of retries |
1061 |
int insertionLevel = indexLevel; |
1062 |
Comparable<K> key = comparable(idx.key); |
1063 |
|
1064 |
// Similar to findPredecessor, but adding index nodes along |
1065 |
// path to key. |
1066 |
for (;;) { |
1067 |
Index<K,V> q = h; |
1068 |
Index<K,V> t = idx; |
1069 |
int j = h.level; |
1070 |
for (;;) { |
1071 |
Index<K,V> r = q.right; |
1072 |
if (r != null) { |
1073 |
// compare before deletion check avoids needing recheck |
1074 |
int c = key.compareTo(r.key); |
1075 |
if (r.indexesDeletedNode()) { |
1076 |
if (q.unlink(r)) |
1077 |
continue; |
1078 |
else |
1079 |
break; |
1080 |
} |
1081 |
if (c > 0) { |
1082 |
q = r; |
1083 |
continue; |
1084 |
} |
1085 |
} |
1086 |
|
1087 |
if (j == insertionLevel) { |
1088 |
// Don't insert index if node already deleted |
1089 |
if (t.indexesDeletedNode()) { |
1090 |
findNode(key); // cleans up |
1091 |
return; |
1092 |
} |
1093 |
if (!q.link(r, t)) |
1094 |
break; // restart |
1095 |
if (--insertionLevel == 0) { |
1096 |
// need final deletion check before return |
1097 |
if (t.indexesDeletedNode()) |
1098 |
findNode(key); |
1099 |
return; |
1100 |
} |
1101 |
} |
1102 |
|
1103 |
if (j > insertionLevel && j <= indexLevel) |
1104 |
t = t.down; |
1105 |
q = q.down; |
1106 |
--j; |
1107 |
} |
1108 |
} |
1109 |
} |
1110 |
|
1111 |
/* ---------------- Deletion -------------- */ |
1112 |
|
1113 |
/** |
1114 |
* Main deletion method. Locates node, nulls value, appends a |
1115 |
* deletion marker, unlinks predecessor, removes associated index |
1116 |
* nodes, and possibly reduces head index level. |
1117 |
* |
1118 |
* Index node are cleared out simply by calling findPredecessor. |
1119 |
* which unlinks indexes to deleted nodes found along path to key, |
1120 |
* which will include the indexes to this node. This is done |
1121 |
* unconditionally. We can't check beforehand whether there are |
1122 |
* index nodes because it might be the case that some or all |
1123 |
* indexes hadn't been inserted yet for this node during initial |
1124 |
* search for it, and we'd like to ensure lack of garbage |
1125 |
* retention, so must call to be sure. |
1126 |
* |
1127 |
* @param okey the key |
1128 |
* @param value if non-null, the value that must be |
1129 |
* associated with key |
1130 |
* @return the node, or null if not found |
1131 |
*/ |
1132 |
private V doRemove(Object okey, Object value) { |
1133 |
Comparable<K> key = comparable(okey); |
1134 |
for (;;) { |
1135 |
Node<K,V> b = findPredecessor(key); |
1136 |
Node<K,V> n = b.next; |
1137 |
for (;;) { |
1138 |
if (n == null) |
1139 |
return null; |
1140 |
Node<K,V> f = n.next; |
1141 |
if (n != b.next) // inconsistent read |
1142 |
break; |
1143 |
Object v = n.value; |
1144 |
if (v == null) { // n is deleted |
1145 |
n.helpDelete(b, f); |
1146 |
break; |
1147 |
} |
1148 |
if (v == n || b.value == null) // b is deleted |
1149 |
break; |
1150 |
int c = key.compareTo(n.key); |
1151 |
if (c < 0) |
1152 |
return null; |
1153 |
if (c > 0) { |
1154 |
b = n; |
1155 |
n = f; |
1156 |
continue; |
1157 |
} |
1158 |
if (value != null && !value.equals(v)) |
1159 |
return null; |
1160 |
if (!n.casValue(v, null)) |
1161 |
break; |
1162 |
if (!n.appendMarker(f) || !b.casNext(n, f)) |
1163 |
findNode(key); // Retry via findNode |
1164 |
else { |
1165 |
findPredecessor(key); // Clean index |
1166 |
if (head.right == null) |
1167 |
tryReduceLevel(); |
1168 |
} |
1169 |
return (V)v; |
1170 |
} |
1171 |
} |
1172 |
} |
1173 |
|
1174 |
/** |
1175 |
* Possibly reduce head level if it has no nodes. This method can |
1176 |
* (rarely) make mistakes, in which case levels can disappear even |
1177 |
* though they are about to contain index nodes. This impacts |
1178 |
* performance, not correctness. To minimize mistakes and also to |
1179 |
* reduce hysteresis, the level is reduced by one only if the |
1180 |
* topmost three levels look empty. Also, if the removed level |
1181 |
* looks non-empty after CAS, we try to change it back quick |
1182 |
* before anyone notices our mistake! (This trick works pretty |
1183 |
* well because this method will practically never make mistakes |
1184 |
* unless current thread stalls immediately before first CAS, in |
1185 |
* which case it is very unlikely to stall again immediately |
1186 |
* afterwards, so will recover.) |
1187 |
* |
1188 |
* We put up with all this rather than just let levels grow |
1189 |
* because otherwise, even a small map that has undergone a large |
1190 |
* number of insertions and removals will have a lot of levels, |
1191 |
* slowing down access more than would an occasional unwanted |
1192 |
* reduction. |
1193 |
*/ |
1194 |
private void tryReduceLevel() { |
1195 |
HeadIndex<K,V> h = head; |
1196 |
HeadIndex<K,V> d; |
1197 |
HeadIndex<K,V> e; |
1198 |
if (h.level > 3 && |
1199 |
(d = (HeadIndex<K,V>)h.down) != null && |
1200 |
(e = (HeadIndex<K,V>)d.down) != null && |
1201 |
e.right == null && |
1202 |
d.right == null && |
1203 |
h.right == null && |
1204 |
casHead(h, d) && // try to set |
1205 |
h.right != null) // recheck |
1206 |
casHead(d, h); // try to backout |
1207 |
} |
1208 |
|
1209 |
|
1210 |
/* ---------------- Positional operations -------------- */ |
1211 |
|
1212 |
/** |
1213 |
* Specialized version of find to get first valid node |
1214 |
* @return first node or null if empty |
1215 |
*/ |
1216 |
Node<K,V> findFirst() { |
1217 |
for (;;) { |
1218 |
// cheaper checks because we know head is never deleted |
1219 |
Node<K,V> b = head.node; |
1220 |
Node<K,V> n = b.next; |
1221 |
if (n == null) |
1222 |
return null; |
1223 |
if (n.value != null) |
1224 |
return n; |
1225 |
n.helpDelete(b, n.next); |
1226 |
} |
1227 |
} |
1228 |
|
1229 |
/** |
1230 |
* Remove first entry; return its key or null if empty. |
1231 |
* Used by ConcurrentSkipListSet |
1232 |
*/ |
1233 |
K removeFirstKey() { |
1234 |
for (;;) { |
1235 |
Node<K,V> b = head.node; |
1236 |
Node<K,V> n = b.next; |
1237 |
if (n == null) |
1238 |
return null; |
1239 |
Node<K,V> f = n.next; |
1240 |
if (n != b.next) |
1241 |
continue; |
1242 |
Object v = n.value; |
1243 |
if (v == null) { |
1244 |
n.helpDelete(b, f); |
1245 |
continue; |
1246 |
} |
1247 |
if (!n.casValue(v, null)) |
1248 |
continue; |
1249 |
if (!n.appendMarker(f) || !b.casNext(n, f)) |
1250 |
findFirst(); // retry |
1251 |
clearIndexToFirst(); |
1252 |
return n.key; |
1253 |
} |
1254 |
} |
1255 |
|
1256 |
/** |
1257 |
* Remove first entry; return SnapshotEntry or null if empty. |
1258 |
*/ |
1259 |
private SnapshotEntry<K,V> doRemoveFirstEntry() { |
1260 |
/* |
1261 |
* This must be mostly duplicated from removeFirstKey because we |
1262 |
* need to save the last value read before it is nulled out |
1263 |
*/ |
1264 |
for (;;) { |
1265 |
Node<K,V> b = head.node; |
1266 |
Node<K,V> n = b.next; |
1267 |
if (n == null) |
1268 |
return null; |
1269 |
Node<K,V> f = n.next; |
1270 |
if (n != b.next) |
1271 |
continue; |
1272 |
Object v = n.value; |
1273 |
if (v == null) { |
1274 |
n.helpDelete(b, f); |
1275 |
continue; |
1276 |
} |
1277 |
if (!n.casValue(v, null)) |
1278 |
continue; |
1279 |
if (!n.appendMarker(f) || !b.casNext(n, f)) |
1280 |
findFirst(); // retry |
1281 |
clearIndexToFirst(); |
1282 |
return new SnapshotEntry<K,V>(n.key, (V)v); |
1283 |
} |
1284 |
} |
1285 |
|
1286 |
/** |
1287 |
* Clear out index nodes associated with deleted first entry. |
1288 |
* Needed by removeFirstKey and removeFirstEntry |
1289 |
*/ |
1290 |
private void clearIndexToFirst() { |
1291 |
for (;;) { |
1292 |
Index<K,V> q = head; |
1293 |
for (;;) { |
1294 |
Index<K,V> r = q.right; |
1295 |
if (r != null && r.indexesDeletedNode() && !q.unlink(r)) |
1296 |
break; |
1297 |
if ((q = q.down) == null) { |
1298 |
if (head.right == null) |
1299 |
tryReduceLevel(); |
1300 |
return; |
1301 |
} |
1302 |
} |
1303 |
} |
1304 |
} |
1305 |
|
1306 |
/** |
1307 |
* Specialized version of find to get last valid node |
1308 |
* @return last node or null if empty |
1309 |
*/ |
1310 |
Node<K,V> findLast() { |
1311 |
/* |
1312 |
* findPredecessor can't be used to traverse index level |
1313 |
* because this doesn't use comparisons. So traversals of |
1314 |
* both levels are folded together. |
1315 |
*/ |
1316 |
Index<K,V> q = head; |
1317 |
for (;;) { |
1318 |
Index<K,V> d, r; |
1319 |
if ((r = q.right) != null) { |
1320 |
if (r.indexesDeletedNode()) { |
1321 |
q.unlink(r); |
1322 |
q = head; // restart |
1323 |
} |
1324 |
else |
1325 |
q = r; |
1326 |
} else if ((d = q.down) != null) { |
1327 |
q = d; |
1328 |
} else { |
1329 |
Node<K,V> b = q.node; |
1330 |
Node<K,V> n = b.next; |
1331 |
for (;;) { |
1332 |
if (n == null) |
1333 |
return (b.isBaseHeader())? null : b; |
1334 |
Node<K,V> f = n.next; // inconsistent read |
1335 |
if (n != b.next) |
1336 |
break; |
1337 |
Object v = n.value; |
1338 |
if (v == null) { // n is deleted |
1339 |
n.helpDelete(b, f); |
1340 |
break; |
1341 |
} |
1342 |
if (v == n || b.value == null) // b is deleted |
1343 |
break; |
1344 |
b = n; |
1345 |
n = f; |
1346 |
} |
1347 |
q = head; // restart |
1348 |
} |
1349 |
} |
1350 |
} |
1351 |
|
1352 |
/* ---------------- Relational operations -------------- */ |
1353 |
|
1354 |
// Control values OR'ed as arguments to findNear |
1355 |
|
1356 |
private static final int EQ = 1; |
1357 |
private static final int LT = 2; |
1358 |
private static final int GT = 0; |
1359 |
|
1360 |
/** |
1361 |
* Utility for ceiling, floor, lower, higher methods. |
1362 |
* @param kkey the key |
1363 |
* @param rel the relation -- OR'ed combination of EQ, LT, GT |
1364 |
* @return nearest node fitting relation, or null if no such |
1365 |
*/ |
1366 |
Node<K,V> findNear(K kkey, int rel) { |
1367 |
Comparable<K> key = comparable(kkey); |
1368 |
for (;;) { |
1369 |
Node<K,V> b = findPredecessor(key); |
1370 |
Node<K,V> n = b.next; |
1371 |
for (;;) { |
1372 |
if (n == null) |
1373 |
return ((rel & LT) == 0 || b.isBaseHeader())? null : b; |
1374 |
Node<K,V> f = n.next; |
1375 |
if (n != b.next) // inconsistent read |
1376 |
break; |
1377 |
Object v = n.value; |
1378 |
if (v == null) { // n is deleted |
1379 |
n.helpDelete(b, f); |
1380 |
break; |
1381 |
} |
1382 |
if (v == n || b.value == null) // b is deleted |
1383 |
break; |
1384 |
int c = key.compareTo(n.key); |
1385 |
if ((c == 0 && (rel & EQ) != 0) || |
1386 |
(c < 0 && (rel & LT) == 0)) |
1387 |
return n; |
1388 |
if ( c <= 0 && (rel & LT) != 0) |
1389 |
return (b.isBaseHeader())? null : b; |
1390 |
b = n; |
1391 |
n = f; |
1392 |
} |
1393 |
} |
1394 |
} |
1395 |
|
1396 |
/** |
1397 |
* Return SnapshotEntry for results of findNear. |
1398 |
* @param kkey the key |
1399 |
* @param rel the relation -- OR'ed combination of EQ, LT, GT |
1400 |
* @return Entry fitting relation, or null if no such |
1401 |
*/ |
1402 |
SnapshotEntry<K,V> getNear(K kkey, int rel) { |
1403 |
for (;;) { |
1404 |
Node<K,V> n = findNear(kkey, rel); |
1405 |
if (n == null) |
1406 |
return null; |
1407 |
SnapshotEntry<K,V> e = n.createSnapshot(); |
1408 |
if (e != null) |
1409 |
return e; |
1410 |
} |
1411 |
} |
1412 |
|
1413 |
/* ---------------- Constructors -------------- */ |
1414 |
|
1415 |
/** |
1416 |
* Constructs a new empty map, sorted according to the keys' natural |
1417 |
* order. |
1418 |
*/ |
1419 |
public ConcurrentSkipListMap() { |
1420 |
this.comparator = null; |
1421 |
initialize(); |
1422 |
} |
1423 |
|
1424 |
/** |
1425 |
* Constructs a new empty map, sorted according to the given comparator. |
1426 |
* |
1427 |
* @param c the comparator that will be used to sort this map. A |
1428 |
* <tt>null</tt> value indicates that the keys' <i>natural |
1429 |
* ordering</i> should be used. |
1430 |
*/ |
1431 |
public ConcurrentSkipListMap(Comparator<? super K> c) { |
1432 |
this.comparator = c; |
1433 |
initialize(); |
1434 |
} |
1435 |
|
1436 |
/** |
1437 |
* Constructs a new map containing the same mappings as the given map, |
1438 |
* sorted according to the keys' <i>natural order</i>. |
1439 |
* |
1440 |
* @param m the map whose mappings are to be placed in this map. |
1441 |
* @throws ClassCastException if the keys in m are not Comparable, or |
1442 |
* are not mutually comparable. |
1443 |
* @throws NullPointerException if the specified map is null. |
1444 |
*/ |
1445 |
public ConcurrentSkipListMap(Map<? extends K, ? extends V> m) { |
1446 |
this.comparator = null; |
1447 |
initialize(); |
1448 |
putAll(m); |
1449 |
} |
1450 |
|
1451 |
/** |
1452 |
* Constructs a new map containing the same mappings as the given |
1453 |
* <tt>SortedMap</tt>, sorted according to the same ordering. |
1454 |
* @param m the sorted map whose mappings are to be placed in this map, |
1455 |
* and whose comparator is to be used to sort this map. |
1456 |
* @throws NullPointerException if the specified sorted map is null. |
1457 |
*/ |
1458 |
public ConcurrentSkipListMap(SortedMap<K, ? extends V> m) { |
1459 |
this.comparator = m.comparator(); |
1460 |
initialize(); |
1461 |
buildFromSorted(m); |
1462 |
} |
1463 |
|
1464 |
/** |
1465 |
* Returns a shallow copy of this <tt>Map</tt> instance. (The keys and |
1466 |
* values themselves are not cloned.) |
1467 |
* |
1468 |
* @return a shallow copy of this Map. |
1469 |
*/ |
1470 |
public Object clone() { |
1471 |
ConcurrentSkipListMap<K,V> clone = null; |
1472 |
try { |
1473 |
clone = (ConcurrentSkipListMap<K,V>) super.clone(); |
1474 |
} catch (CloneNotSupportedException e) { |
1475 |
throw new InternalError(); |
1476 |
} |
1477 |
|
1478 |
clone.initialize(); |
1479 |
clone.buildFromSorted(this); |
1480 |
return clone; |
1481 |
} |
1482 |
|
1483 |
/** |
1484 |
* Streamlined bulk insertion to initialize from elements of |
1485 |
* given sorted map. Call only from constructor or clone |
1486 |
* method. |
1487 |
*/ |
1488 |
private void buildFromSorted(SortedMap<K, ? extends V> map) { |
1489 |
if (map == null) |
1490 |
throw new NullPointerException(); |
1491 |
|
1492 |
HeadIndex<K,V> h = head; |
1493 |
Node<K,V> basepred = h.node; |
1494 |
|
1495 |
// Track the current rightmost node at each level. Uses an |
1496 |
// ArrayList to avoid committing to initial or maximum level. |
1497 |
ArrayList<Index<K,V>> preds = new ArrayList<Index<K,V>>(); |
1498 |
|
1499 |
// initialize |
1500 |
for (int i = 0; i <= h.level; ++i) |
1501 |
preds.add(null); |
1502 |
Index<K,V> q = h; |
1503 |
for (int i = h.level; i > 0; --i) { |
1504 |
preds.set(i, q); |
1505 |
q = q.down; |
1506 |
} |
1507 |
|
1508 |
Iterator<? extends Map.Entry<? extends K, ? extends V>> it = |
1509 |
map.entrySet().iterator(); |
1510 |
while (it.hasNext()) { |
1511 |
Map.Entry<? extends K, ? extends V> e = it.next(); |
1512 |
int j = randomLevel(); |
1513 |
if (j > h.level) j = h.level + 1; |
1514 |
K k = e.getKey(); |
1515 |
V v = e.getValue(); |
1516 |
if (k == null || v == null) |
1517 |
throw new NullPointerException(); |
1518 |
Node<K,V> z = new Node<K,V>(k, v, null); |
1519 |
basepred.next = z; |
1520 |
basepred = z; |
1521 |
if (j > 0) { |
1522 |
Index<K,V> idx = null; |
1523 |
for (int i = 1; i <= j; ++i) { |
1524 |
idx = new Index<K,V>(z, idx); |
1525 |
if (i > h.level) |
1526 |
h = new HeadIndex<K,V>(h.node, h, idx, i); |
1527 |
|
1528 |
if (i < preds.size()) { |
1529 |
preds.get(i).right = idx; |
1530 |
preds.set(i, idx); |
1531 |
} else |
1532 |
preds.add(idx); |
1533 |
} |
1534 |
} |
1535 |
} |
1536 |
head = h; |
1537 |
} |
1538 |
|
1539 |
/* ---------------- Serialization -------------- */ |
1540 |
|
1541 |
/** |
1542 |
* Save the state of the <tt>Map</tt> instance to a stream. |
1543 |
* |
1544 |
* @serialData The key (Object) and value (Object) for each |
1545 |
* key-value mapping represented by the Map, followed by |
1546 |
* <tt>null</tt>. The key-value mappings are emitted in key-order |
1547 |
* (as determined by the Comparator, or by the keys' natural |
1548 |
* ordering if no Comparator). |
1549 |
*/ |
1550 |
private void writeObject(java.io.ObjectOutputStream s) |
1551 |
throws java.io.IOException { |
1552 |
// Write out the Comparator and any hidden stuff |
1553 |
s.defaultWriteObject(); |
1554 |
|
1555 |
// Write out keys and values (alternating) |
1556 |
for (Node<K,V> n = findFirst(); n != null; n = n.next) { |
1557 |
V v = n.getValidValue(); |
1558 |
if (v != null) { |
1559 |
s.writeObject(n.key); |
1560 |
s.writeObject(v); |
1561 |
} |
1562 |
} |
1563 |
s.writeObject(null); |
1564 |
} |
1565 |
|
1566 |
/** |
1567 |
* Reconstitute the <tt>Map</tt> instance from a stream. |
1568 |
*/ |
1569 |
private void readObject(final java.io.ObjectInputStream s) |
1570 |
throws java.io.IOException, ClassNotFoundException { |
1571 |
// Read in the Comparator and any hidden stuff |
1572 |
s.defaultReadObject(); |
1573 |
// Reset transients |
1574 |
initialize(); |
1575 |
|
1576 |
/* |
1577 |
* This is basically identical to buildFromSorted, but is |
1578 |
* distinct because readObject calls can't be nicely adapted |
1579 |
* as the kind of iterator needed by buildFromSorted. (They |
1580 |
* can be, but doing so requires type cheats and/or creation |
1581 |
* of adaptor classes.) It is simpler to just adapt the code. |
1582 |
*/ |
1583 |
|
1584 |
HeadIndex<K,V> h = head; |
1585 |
Node<K,V> basepred = h.node; |
1586 |
ArrayList<Index<K,V>> preds = new ArrayList<Index<K,V>>(); |
1587 |
for (int i = 0; i <= h.level; ++i) |
1588 |
preds.add(null); |
1589 |
Index<K,V> q = h; |
1590 |
for (int i = h.level; i > 0; --i) { |
1591 |
preds.set(i, q); |
1592 |
q = q.down; |
1593 |
} |
1594 |
|
1595 |
for (;;) { |
1596 |
Object k = s.readObject(); |
1597 |
if (k == null) |
1598 |
break; |
1599 |
Object v = s.readObject(); |
1600 |
if (v == null) |
1601 |
throw new NullPointerException(); |
1602 |
K key = (K) k; |
1603 |
V val = (V) v; |
1604 |
int j = randomLevel(); |
1605 |
if (j > h.level) j = h.level + 1; |
1606 |
Node<K,V> z = new Node<K,V>(key, val, null); |
1607 |
basepred.next = z; |
1608 |
basepred = z; |
1609 |
if (j > 0) { |
1610 |
Index<K,V> idx = null; |
1611 |
for (int i = 1; i <= j; ++i) { |
1612 |
idx = new Index<K,V>(z, idx); |
1613 |
if (i > h.level) |
1614 |
h = new HeadIndex<K,V>(h.node, h, idx, i); |
1615 |
|
1616 |
if (i < preds.size()) { |
1617 |
preds.get(i).right = idx; |
1618 |
preds.set(i, idx); |
1619 |
} else |
1620 |
preds.add(idx); |
1621 |
} |
1622 |
} |
1623 |
} |
1624 |
head = h; |
1625 |
} |
1626 |
|
1627 |
/* ------ Map API methods ------ */ |
1628 |
|
1629 |
/** |
1630 |
* Returns <tt>true</tt> if this map contains a mapping for the specified |
1631 |
* key. |
1632 |
* @param key key whose presence in this map is to be tested. |
1633 |
* @return <tt>true</tt> if this map contains a mapping for the |
1634 |
* specified key. |
1635 |
* @throws ClassCastException if the key cannot be compared with the keys |
1636 |
* currently in the map. |
1637 |
* @throws NullPointerException if the key is <tt>null</tt>. |
1638 |
*/ |
1639 |
public boolean containsKey(Object key) { |
1640 |
return doGet(key) != null; |
1641 |
} |
1642 |
|
1643 |
/** |
1644 |
* Returns the value to which this map maps the specified key. Returns |
1645 |
* <tt>null</tt> if the map contains no mapping for this key. |
1646 |
* |
1647 |
* @param key key whose associated value is to be returned. |
1648 |
* @return the value to which this map maps the specified key, or |
1649 |
* <tt>null</tt> if the map contains no mapping for the key. |
1650 |
* @throws ClassCastException if the key cannot be compared with the keys |
1651 |
* currently in the map. |
1652 |
* @throws NullPointerException if the key is <tt>null</tt>. |
1653 |
*/ |
1654 |
public V get(Object key) { |
1655 |
return doGet(key); |
1656 |
} |
1657 |
|
1658 |
/** |
1659 |
* Associates the specified value with the specified key in this map. |
1660 |
* If the map previously contained a mapping for this key, the old |
1661 |
* value is replaced. |
1662 |
* |
1663 |
* @param key key with which the specified value is to be associated. |
1664 |
* @param value value to be associated with the specified key. |
1665 |
* |
1666 |
* @return previous value associated with specified key, or <tt>null</tt> |
1667 |
* if there was no mapping for key. |
1668 |
* @throws ClassCastException if the key cannot be compared with the keys |
1669 |
* currently in the map. |
1670 |
* @throws NullPointerException if the key or value are <tt>null</tt>. |
1671 |
*/ |
1672 |
public V put(K key, V value) { |
1673 |
if (value == null) |
1674 |
throw new NullPointerException(); |
1675 |
return doPut(key, value, false); |
1676 |
} |
1677 |
|
1678 |
/** |
1679 |
* Removes the mapping for this key from this Map if present. |
1680 |
* |
1681 |
* @param key key for which mapping should be removed |
1682 |
* @return previous value associated with specified key, or <tt>null</tt> |
1683 |
* if there was no mapping for key. |
1684 |
* |
1685 |
* @throws ClassCastException if the key cannot be compared with the keys |
1686 |
* currently in the map. |
1687 |
* @throws NullPointerException if the key is <tt>null</tt>. |
1688 |
*/ |
1689 |
public V remove(Object key) { |
1690 |
return doRemove(key, null); |
1691 |
} |
1692 |
|
1693 |
/** |
1694 |
* Returns <tt>true</tt> if this map maps one or more keys to the |
1695 |
* specified value. This operation requires time linear in the |
1696 |
* Map size. |
1697 |
* |
1698 |
* @param value value whose presence in this Map is to be tested. |
1699 |
* @return <tt>true</tt> if a mapping to <tt>value</tt> exists; |
1700 |
* <tt>false</tt> otherwise. |
1701 |
* @throws NullPointerException if the value is <tt>null</tt>. |
1702 |
*/ |
1703 |
public boolean containsValue(Object value) { |
1704 |
if (value == null) |
1705 |
throw new NullPointerException(); |
1706 |
for (Node<K,V> n = findFirst(); n != null; n = n.next) { |
1707 |
V v = n.getValidValue(); |
1708 |
if (v != null && value.equals(v)) |
1709 |
return true; |
1710 |
} |
1711 |
return false; |
1712 |
} |
1713 |
|
1714 |
/** |
1715 |
* Returns the number of elements in this map. If this map |
1716 |
* contains more than <tt>Integer.MAX_VALUE</tt> elements, it |
1717 |
* returns <tt>Integer.MAX_VALUE</tt>. |
1718 |
* |
1719 |
* <p>Beware that, unlike in most collections, this method is |
1720 |
* <em>NOT</em> a constant-time operation. Because of the |
1721 |
* asynchronous nature of these maps, determining the current |
1722 |
* number of elements requires traversing them all to count them. |
1723 |
* Additionally, it is possible for the size to change during |
1724 |
* execution of this method, in which case the returned result |
1725 |
* will be inaccurate. Thus, this method is typically not very |
1726 |
* useful in concurrent applications. |
1727 |
* |
1728 |
* @return the number of elements in this map. |
1729 |
*/ |
1730 |
public int size() { |
1731 |
long count = 0; |
1732 |
for (Node<K,V> n = findFirst(); n != null; n = n.next) { |
1733 |
if (n.getValidValue() != null) |
1734 |
++count; |
1735 |
} |
1736 |
return (count >= Integer.MAX_VALUE)? Integer.MAX_VALUE : (int)count; |
1737 |
} |
1738 |
|
1739 |
/** |
1740 |
* Returns <tt>true</tt> if this map contains no key-value mappings. |
1741 |
* @return <tt>true</tt> if this map contains no key-value mappings. |
1742 |
*/ |
1743 |
public boolean isEmpty() { |
1744 |
return findFirst() == null; |
1745 |
} |
1746 |
|
1747 |
/** |
1748 |
* Removes all mappings from this map. |
1749 |
*/ |
1750 |
public void clear() { |
1751 |
initialize(); |
1752 |
} |
1753 |
|
1754 |
/** |
1755 |
* Returns a set view of the keys contained in this map. The set is |
1756 |
* backed by the map, so changes to the map are reflected in the set, and |
1757 |
* vice-versa. The set supports element removal, which removes the |
1758 |
* corresponding mapping from this map, via the <tt>Iterator.remove</tt>, |
1759 |
* <tt>Set.remove</tt>, <tt>removeAll</tt>, <tt>retainAll</tt>, and |
1760 |
* <tt>clear</tt> operations. It does not support the <tt>add</tt> or |
1761 |
* <tt>addAll</tt> operations. |
1762 |
* The view's <tt>iterator</tt> is a "weakly consistent" iterator that |
1763 |
* will never throw {@link java.util.ConcurrentModificationException}, |
1764 |
* and guarantees to traverse elements as they existed upon |
1765 |
* construction of the iterator, and may (but is not guaranteed to) |
1766 |
* reflect any modifications subsequent to construction. |
1767 |
* |
1768 |
* @return a set view of the keys contained in this map. |
1769 |
*/ |
1770 |
public Set<K> keySet() { |
1771 |
/* |
1772 |
* Note: Lazy intialization works here and for other views |
1773 |
* because view classes are stateless/immutable so it doesn't |
1774 |
* matter wrt correctness if more than one is created (which |
1775 |
* will only rarely happen). Even so, the following idiom |
1776 |
* conservatively ensures that the method returns the one it |
1777 |
* created if it does so, not one created by another racing |
1778 |
* thread. |
1779 |
*/ |
1780 |
KeySet ks = keySet; |
1781 |
return (ks != null) ? ks : (keySet = new KeySet()); |
1782 |
} |
1783 |
|
1784 |
/** |
1785 |
* Returns a collection view of the values contained in this map. |
1786 |
* The collection is backed by the map, so changes to the map are |
1787 |
* reflected in the collection, and vice-versa. The collection |
1788 |
* supports element removal, which removes the corresponding |
1789 |
* mapping from this map, via the <tt>Iterator.remove</tt>, |
1790 |
* <tt>Collection.remove</tt>, <tt>removeAll</tt>, |
1791 |
* <tt>retainAll</tt>, and <tt>clear</tt> operations. It does not |
1792 |
* support the <tt>add</tt> or <tt>addAll</tt> operations. The |
1793 |
* view's <tt>iterator</tt> is a "weakly consistent" iterator that |
1794 |
* will never throw {@link |
1795 |
* java.util.ConcurrentModificationException}, and guarantees to |
1796 |
* traverse elements as they existed upon construction of the |
1797 |
* iterator, and may (but is not guaranteed to) reflect any |
1798 |
* modifications subsequent to construction. |
1799 |
* |
1800 |
* @return a collection view of the values contained in this map. |
1801 |
*/ |
1802 |
public Collection<V> values() { |
1803 |
Values vs = values; |
1804 |
return (vs != null) ? vs : (values = new Values()); |
1805 |
} |
1806 |
|
1807 |
/** |
1808 |
* Returns a collection view of the mappings contained in this |
1809 |
* map. Each element in the returned collection is a |
1810 |
* <tt>Map.Entry</tt>. The collection is backed by the map, so |
1811 |
* changes to the map are reflected in the collection, and |
1812 |
* vice-versa. The collection supports element removal, which |
1813 |
* removes the corresponding mapping from the map, via the |
1814 |
* <tt>Iterator.remove</tt>, <tt>Collection.remove</tt>, |
1815 |
* <tt>removeAll</tt>, <tt>retainAll</tt>, and <tt>clear</tt> |
1816 |
* operations. It does not support the <tt>add</tt> or |
1817 |
* <tt>addAll</tt> operations. The view's <tt>iterator</tt> is a |
1818 |
* "weakly consistent" iterator that will never throw {@link |
1819 |
* java.util.ConcurrentModificationException}, and guarantees to |
1820 |
* traverse elements as they existed upon construction of the |
1821 |
* iterator, and may (but is not guaranteed to) reflect any |
1822 |
* modifications subsequent to construction. The |
1823 |
* <tt>Map.Entry</tt> elements returned by |
1824 |
* <tt>iterator.next()</tt> do <em>not</em> support the |
1825 |
* <tt>setValue</tt> operation. |
1826 |
* |
1827 |
* @return a collection view of the mappings contained in this map. |
1828 |
*/ |
1829 |
public Set<Map.Entry<K,V>> entrySet() { |
1830 |
EntrySet es = entrySet; |
1831 |
return (es != null) ? es : (entrySet = new EntrySet()); |
1832 |
} |
1833 |
|
1834 |
/* ------ ConcurrentMap API methods ------ */ |
1835 |
|
1836 |
/** |
1837 |
* If the specified key is not already associated |
1838 |
* with a value, associate it with the given value. |
1839 |
* This is equivalent to |
1840 |
* <pre> |
1841 |
* if (!map.containsKey(key)) |
1842 |
* return map.put(key, value); |
1843 |
* else |
1844 |
* return map.get(key); |
1845 |
* </pre> |
1846 |
* Except that the action is performed atomically. |
1847 |
* @param key key with which the specified value is to be associated. |
1848 |
* @param value value to be associated with the specified key. |
1849 |
* @return previous value associated with specified key, or <tt>null</tt> |
1850 |
* if there was no mapping for key. |
1851 |
* |
1852 |
* @throws ClassCastException if the key cannot be compared with the keys |
1853 |
* currently in the map. |
1854 |
* @throws NullPointerException if the key or value are <tt>null</tt>. |
1855 |
*/ |
1856 |
public V putIfAbsent(K key, V value) { |
1857 |
if (value == null) |
1858 |
throw new NullPointerException(); |
1859 |
return doPut(key, value, true); |
1860 |
} |
1861 |
|
1862 |
/** |
1863 |
* Remove entry for key only if currently mapped to given value. |
1864 |
* Acts as |
1865 |
* <pre> |
1866 |
* if ((map.containsKey(key) && map.get(key).equals(value)) { |
1867 |
* map.remove(key); |
1868 |
* return true; |
1869 |
* } else return false; |
1870 |
* </pre> |
1871 |
* except that the action is performed atomically. |
1872 |
* @param key key with which the specified value is associated. |
1873 |
* @param value value associated with the specified key. |
1874 |
* @return true if the value was removed, false otherwise |
1875 |
* @throws ClassCastException if the key cannot be compared with the keys |
1876 |
* currently in the map. |
1877 |
* @throws NullPointerException if the key or value are <tt>null</tt>. |
1878 |
*/ |
1879 |
public boolean remove(Object key, Object value) { |
1880 |
if (value == null) |
1881 |
throw new NullPointerException(); |
1882 |
return doRemove(key, value) != null; |
1883 |
} |
1884 |
|
1885 |
/** |
1886 |
* Replace entry for key only if currently mapped to given value. |
1887 |
* Acts as |
1888 |
* <pre> |
1889 |
* if ((map.containsKey(key) && map.get(key).equals(oldValue)) { |
1890 |
* map.put(key, newValue); |
1891 |
* return true; |
1892 |
* } else return false; |
1893 |
* </pre> |
1894 |
* except that the action is performed atomically. |
1895 |
* @param key key with which the specified value is associated. |
1896 |
* @param oldValue value expected to be associated with the specified key. |
1897 |
* @param newValue value to be associated with the specified key. |
1898 |
* @return true if the value was replaced |
1899 |
* @throws ClassCastException if the key cannot be compared with the keys |
1900 |
* currently in the map. |
1901 |
* @throws NullPointerException if key, oldValue or newValue are |
1902 |
* <tt>null</tt>. |
1903 |
*/ |
1904 |
public boolean replace(K key, V oldValue, V newValue) { |
1905 |
if (oldValue == null || newValue == null) |
1906 |
throw new NullPointerException(); |
1907 |
Comparable<K> k = comparable(key); |
1908 |
for (;;) { |
1909 |
Node<K,V> n = findNode(k); |
1910 |
if (n == null) |
1911 |
return false; |
1912 |
Object v = n.value; |
1913 |
if (v != null) { |
1914 |
if (!oldValue.equals(v)) |
1915 |
return false; |
1916 |
if (n.casValue(v, newValue)) |
1917 |
return true; |
1918 |
} |
1919 |
} |
1920 |
} |
1921 |
|
1922 |
/** |
1923 |
* Replace entry for key only if currently mapped to some value. |
1924 |
* Acts as |
1925 |
* <pre> |
1926 |
* if ((map.containsKey(key)) { |
1927 |
* return map.put(key, value); |
1928 |
* } else return null; |
1929 |
* </pre> |
1930 |
* except that the action is performed atomically. |
1931 |
* @param key key with which the specified value is associated. |
1932 |
* @param value value to be associated with the specified key. |
1933 |
* @return previous value associated with specified key, or <tt>null</tt> |
1934 |
* if there was no mapping for key. |
1935 |
* @throws ClassCastException if the key cannot be compared with the keys |
1936 |
* currently in the map. |
1937 |
* @throws NullPointerException if the key or value are <tt>null</tt>. |
1938 |
*/ |
1939 |
public V replace(K key, V value) { |
1940 |
if (value == null) |
1941 |
throw new NullPointerException(); |
1942 |
Comparable<K> k = comparable(key); |
1943 |
for (;;) { |
1944 |
Node<K,V> n = findNode(k); |
1945 |
if (n == null) |
1946 |
return null; |
1947 |
Object v = n.value; |
1948 |
if (v != null && n.casValue(v, value)) |
1949 |
return (V)v; |
1950 |
} |
1951 |
} |
1952 |
|
1953 |
/* ------ SortedMap API methods ------ */ |
1954 |
|
1955 |
/** |
1956 |
* Returns the comparator used to order this map, or <tt>null</tt> |
1957 |
* if this map uses its keys' natural order. |
1958 |
* |
1959 |
* @return the comparator associated with this map, or |
1960 |
* <tt>null</tt> if it uses its keys' natural sort method. |
1961 |
*/ |
1962 |
public Comparator<? super K> comparator() { |
1963 |
return comparator; |
1964 |
} |
1965 |
|
1966 |
/** |
1967 |
* Returns the first (lowest) key currently in this map. |
1968 |
* |
1969 |
* @return the first (lowest) key currently in this map. |
1970 |
* @throws NoSuchElementException Map is empty. |
1971 |
*/ |
1972 |
public K firstKey() { |
1973 |
Node<K,V> n = findFirst(); |
1974 |
if (n == null) |
1975 |
throw new NoSuchElementException(); |
1976 |
return n.key; |
1977 |
} |
1978 |
|
1979 |
/** |
1980 |
* Returns the last (highest) key currently in this map. |
1981 |
* |
1982 |
* @return the last (highest) key currently in this map. |
1983 |
* @throws NoSuchElementException Map is empty. |
1984 |
*/ |
1985 |
public K lastKey() { |
1986 |
Node<K,V> n = findLast(); |
1987 |
if (n == null) |
1988 |
throw new NoSuchElementException(); |
1989 |
return n.key; |
1990 |
} |
1991 |
|
1992 |
/** |
1993 |
* Returns a view of the portion of this map whose keys range from |
1994 |
* <tt>fromKey</tt>, inclusive, to <tt>toKey</tt>, exclusive. (If |
1995 |
* <tt>fromKey</tt> and <tt>toKey</tt> are equal, the returned sorted map |
1996 |
* is empty.) The returned sorted map is backed by this map, so changes |
1997 |
* in the returned sorted map are reflected in this map, and vice-versa. |
1998 |
|
1999 |
* @param fromKey low endpoint (inclusive) of the subMap. |
2000 |
* @param toKey high endpoint (exclusive) of the subMap. |
2001 |
* |
2002 |
* @return a view of the portion of this map whose keys range from |
2003 |
* <tt>fromKey</tt>, inclusive, to <tt>toKey</tt>, exclusive. |
2004 |
* |
2005 |
* @throws ClassCastException if <tt>fromKey</tt> and <tt>toKey</tt> |
2006 |
* cannot be compared to one another using this map's comparator |
2007 |
* (or, if the map has no comparator, using natural ordering). |
2008 |
* @throws IllegalArgumentException if <tt>fromKey</tt> is greater than |
2009 |
* <tt>toKey</tt>. |
2010 |
* @throws NullPointerException if <tt>fromKey</tt> or <tt>toKey</tt> is |
2011 |
* <tt>null</tt>. |
2012 |
*/ |
2013 |
public ConcurrentSkipListSubMap<K,V> subMap(K fromKey, K toKey) { |
2014 |
if (fromKey == null || toKey == null) |
2015 |
throw new NullPointerException(); |
2016 |
return new ConcurrentSkipListSubMap(this, fromKey, toKey); |
2017 |
} |
2018 |
|
2019 |
/** |
2020 |
* Returns a view of the portion of this map whose keys are strictly less |
2021 |
* than <tt>toKey</tt>. The returned sorted map is backed by this map, so |
2022 |
* changes in the returned sorted map are reflected in this map, and |
2023 |
* vice-versa. |
2024 |
* @param toKey high endpoint (exclusive) of the headMap. |
2025 |
* @return a view of the portion of this map whose keys are strictly |
2026 |
* less than <tt>toKey</tt>. |
2027 |
* |
2028 |
* @throws ClassCastException if <tt>toKey</tt> is not compatible |
2029 |
* with this map's comparator (or, if the map has no comparator, |
2030 |
* if <tt>toKey</tt> does not implement <tt>Comparable</tt>). |
2031 |
* @throws NullPointerException if <tt>toKey</tt> is <tt>null</tt>. |
2032 |
*/ |
2033 |
public ConcurrentSkipListSubMap<K,V> headMap(K toKey) { |
2034 |
if (toKey == null) |
2035 |
throw new NullPointerException(); |
2036 |
return new ConcurrentSkipListSubMap(this, null, toKey); |
2037 |
} |
2038 |
|
2039 |
/** |
2040 |
* Returns a view of the portion of this map whose keys are |
2041 |
* greater than or equal to <tt>fromKey</tt>. The returned sorted |
2042 |
* map is backed by this map, so changes in the returned sorted |
2043 |
* map are reflected in this map, and vice-versa. |
2044 |
* @param fromKey low endpoint (inclusive) of the tailMap. |
2045 |
* @return a view of the portion of this map whose keys are greater |
2046 |
* than or equal to <tt>fromKey</tt>. |
2047 |
* @throws ClassCastException if <tt>fromKey</tt> is not compatible |
2048 |
* with this map's comparator (or, if the map has no comparator, |
2049 |
* if <tt>fromKey</tt> does not implement <tt>Comparable</tt>). |
2050 |
* @throws NullPointerException if <tt>fromKey</tt> is <tt>null</tt>. |
2051 |
*/ |
2052 |
public ConcurrentSkipListSubMap<K,V> tailMap(K fromKey) { |
2053 |
if (fromKey == null) |
2054 |
throw new NullPointerException(); |
2055 |
return new ConcurrentSkipListSubMap(this, fromKey, null); |
2056 |
} |
2057 |
|
2058 |
/* ---------------- Relational operations -------------- */ |
2059 |
|
2060 |
/** |
2061 |
* Returns a key-value mapping associated with the least key |
2062 |
* greater than or equal to the given key, or null if there is |
2063 |
* no such entry. The returned entry does <em>not</em> support |
2064 |
* the <tt>Entry.setValue</tt> method. |
2065 |
* |
2066 |
* @param key the key. |
2067 |
* @return an Entry associated with ceiling of given key, or null |
2068 |
* if there is no such Entry. |
2069 |
* @throws ClassCastException if key cannot be compared with the keys |
2070 |
* currently in the map. |
2071 |
* @throws NullPointerException if key is <tt>null</tt>. |
2072 |
*/ |
2073 |
public Map.Entry<K,V> ceilingEntry(K key) { |
2074 |
return getNear(key, GT|EQ); |
2075 |
} |
2076 |
|
2077 |
/** |
2078 |
* Returns a key-value mapping associated with the greatest |
2079 |
* key strictly less than the given key, or null if there is no |
2080 |
* such entry. The returned entry does <em>not</em> support |
2081 |
* the <tt>Entry.setValue</tt> method. |
2082 |
* |
2083 |
* @param key the key. |
2084 |
* @return an Entry with greatest key less than the given |
2085 |
* key, or null if there is no such Entry. |
2086 |
* @throws ClassCastException if key cannot be compared with the keys |
2087 |
* currently in the map. |
2088 |
* @throws NullPointerException if key is <tt>null</tt>. |
2089 |
*/ |
2090 |
public Map.Entry<K,V> lowerEntry(K key) { |
2091 |
return getNear(key, LT); |
2092 |
} |
2093 |
|
2094 |
/** |
2095 |
* Returns a key-value mapping associated with the greatest |
2096 |
* key less than or equal to the given key, or null if there is no |
2097 |
* such entry. The returned entry does <em>not</em> support |
2098 |
* the <tt>Entry.setValue</tt> method. |
2099 |
* |
2100 |
* @param key the key. |
2101 |
* @return an Entry associated with floor of given key, or null |
2102 |
* if there is no such Entry. |
2103 |
* @throws ClassCastException if key cannot be compared with the keys |
2104 |
* currently in the map. |
2105 |
* @throws NullPointerException if key is <tt>null</tt>. |
2106 |
*/ |
2107 |
public Map.Entry<K,V> floorEntry(K key) { |
2108 |
return getNear(key, LT|EQ); |
2109 |
} |
2110 |
|
2111 |
/** |
2112 |
* Returns a key-value mapping associated with the least |
2113 |
* key strictly greater than the given key, or null if there is no |
2114 |
* such entry. The returned entry does <em>not</em> support |
2115 |
* the <tt>Entry.setValue</tt> method. |
2116 |
* |
2117 |
* @param key the key. |
2118 |
* @return an Entry with least key greater than the given key, or |
2119 |
* null if there is no such Entry. |
2120 |
* @throws ClassCastException if key cannot be compared with the keys |
2121 |
* currently in the map. |
2122 |
* @throws NullPointerException if key is <tt>null</tt>. |
2123 |
*/ |
2124 |
public Map.Entry<K,V> higherEntry(K key) { |
2125 |
return getNear(key, GT); |
2126 |
} |
2127 |
|
2128 |
/** |
2129 |
* Returns a key-value mapping associated with the least |
2130 |
* key in this map, or null if the map is empty. |
2131 |
* The returned entry does <em>not</em> support |
2132 |
* the <tt>Entry.setValue</tt> method. |
2133 |
* |
2134 |
* @return an Entry with least key, or null |
2135 |
* if the map is empty. |
2136 |
*/ |
2137 |
public Map.Entry<K,V> firstEntry() { |
2138 |
for (;;) { |
2139 |
Node<K,V> n = findFirst(); |
2140 |
if (n == null) |
2141 |
return null; |
2142 |
SnapshotEntry<K,V> e = n.createSnapshot(); |
2143 |
if (e != null) |
2144 |
return e; |
2145 |
} |
2146 |
} |
2147 |
|
2148 |
/** |
2149 |
* Returns a key-value mapping associated with the greatest |
2150 |
* key in this map, or null if the map is empty. |
2151 |
* The returned entry does <em>not</em> support |
2152 |
* the <tt>Entry.setValue</tt> method. |
2153 |
* |
2154 |
* @return an Entry with greatest key, or null |
2155 |
* if the map is empty. |
2156 |
*/ |
2157 |
public Map.Entry<K,V> lastEntry() { |
2158 |
for (;;) { |
2159 |
Node<K,V> n = findLast(); |
2160 |
if (n == null) |
2161 |
return null; |
2162 |
SnapshotEntry<K,V> e = n.createSnapshot(); |
2163 |
if (e != null) |
2164 |
return e; |
2165 |
} |
2166 |
} |
2167 |
|
2168 |
/** |
2169 |
* Removes and returns a key-value mapping associated with |
2170 |
* the least key in this map, or null if the map is empty. |
2171 |
* The returned entry does <em>not</em> support |
2172 |
* the <tt>Entry.setValue</tt> method. |
2173 |
* |
2174 |
* @return the removed first entry of this map, or null |
2175 |
* if the map is empty. |
2176 |
*/ |
2177 |
public Map.Entry<K,V> removeFirstEntry() { |
2178 |
return doRemoveFirstEntry(); |
2179 |
} |
2180 |
|
2181 |
/* ---------------- Iterators -------------- */ |
2182 |
|
2183 |
/** |
2184 |
* Base of iterator classes. |
2185 |
* (Six kinds: {key, value, entry} X {map, submap}) |
2186 |
*/ |
2187 |
abstract class ConcurrentSkipListMapIterator { |
2188 |
/** the last node returned by next() */ |
2189 |
Node<K,V> last; |
2190 |
/** the next node to return from next(); */ |
2191 |
Node<K,V> next; |
2192 |
/** Cache of next value field to maintain weak consistency */ |
2193 |
Object nextValue; |
2194 |
|
2195 |
/** Create normal iterator for entire range */ |
2196 |
ConcurrentSkipListMapIterator() { |
2197 |
for (;;) { |
2198 |
next = findFirst(); |
2199 |
if (next == null) |
2200 |
break; |
2201 |
nextValue = next.value; |
2202 |
if (nextValue != null && nextValue != next) |
2203 |
break; |
2204 |
} |
2205 |
} |
2206 |
|
2207 |
/** |
2208 |
* Create a submap iterator starting at given least key, or |
2209 |
* first node if least is null, but not greater or equal to |
2210 |
* fence, or end if fence is null. |
2211 |
*/ |
2212 |
ConcurrentSkipListMapIterator(K least, K fence) { |
2213 |
for (;;) { |
2214 |
next = findCeiling(least); |
2215 |
if (next == null) |
2216 |
break; |
2217 |
nextValue = next.value; |
2218 |
if (nextValue != null && nextValue != next) { |
2219 |
if (fence != null && compare(fence, next.key) <= 0) { |
2220 |
next = null; |
2221 |
nextValue = null; |
2222 |
} |
2223 |
break; |
2224 |
} |
2225 |
} |
2226 |
} |
2227 |
|
2228 |
public final boolean hasNext() { |
2229 |
return next != null; |
2230 |
} |
2231 |
|
2232 |
final void advance() { |
2233 |
if ((last = next) == null) |
2234 |
throw new NoSuchElementException(); |
2235 |
for (;;) { |
2236 |
next = next.next; |
2237 |
if (next == null) |
2238 |
break; |
2239 |
nextValue = next.value; |
2240 |
if (nextValue != null && nextValue != next) |
2241 |
break; |
2242 |
} |
2243 |
} |
2244 |
|
2245 |
/** |
2246 |
* Version of advance for submaps to stop at fence |
2247 |
*/ |
2248 |
final void advance(K fence) { |
2249 |
if ((last = next) == null) |
2250 |
throw new NoSuchElementException(); |
2251 |
for (;;) { |
2252 |
next = next.next; |
2253 |
if (next == null) |
2254 |
break; |
2255 |
nextValue = next.value; |
2256 |
if (nextValue != null && nextValue != next) { |
2257 |
if (fence != null && compare(fence, next.key) <= 0) { |
2258 |
next = null; |
2259 |
nextValue = null; |
2260 |
} |
2261 |
break; |
2262 |
} |
2263 |
} |
2264 |
} |
2265 |
|
2266 |
public void remove() { |
2267 |
Node<K,V> l = last; |
2268 |
if (l == null) |
2269 |
throw new IllegalStateException(); |
2270 |
// It would not be worth all of the overhead to directly |
2271 |
// unlink from here. Using remove is fast enough. |
2272 |
ConcurrentSkipListMap.this.remove(l.key); |
2273 |
} |
2274 |
} |
2275 |
|
2276 |
final class ValueIterator extends ConcurrentSkipListMapIterator |
2277 |
implements Iterator<V> { |
2278 |
public V next() { |
2279 |
Object v = nextValue; |
2280 |
advance(); |
2281 |
return (V)v; |
2282 |
} |
2283 |
} |
2284 |
|
2285 |
final class KeyIterator extends ConcurrentSkipListMapIterator |
2286 |
implements Iterator<K> { |
2287 |
public K next() { |
2288 |
Node<K,V> n = next; |
2289 |
advance(); |
2290 |
return n.key; |
2291 |
} |
2292 |
} |
2293 |
|
2294 |
/** |
2295 |
* Entry iterators use the same trick as in ConcurrentHashMap and |
2296 |
* elsewhere of using the iterator itself to represent entries, |
2297 |
* thus avoiding having to create entry objects in next(). |
2298 |
*/ |
2299 |
class EntryIterator extends ConcurrentSkipListMapIterator |
2300 |
implements Map.Entry<K,V>, Iterator<Map.Entry<K,V>> { |
2301 |
/** Cache of last value returned */ |
2302 |
Object lastValue; |
2303 |
|
2304 |
EntryIterator() { |
2305 |
super(); |
2306 |
} |
2307 |
|
2308 |
EntryIterator(K least, K fence) { |
2309 |
super(least, fence); |
2310 |
} |
2311 |
|
2312 |
public Map.Entry<K,V> next() { |
2313 |
lastValue = nextValue; |
2314 |
advance(); |
2315 |
return this; |
2316 |
} |
2317 |
|
2318 |
public K getKey() { |
2319 |
Node<K,V> l = last; |
2320 |
if (l == null) |
2321 |
throw new IllegalStateException(); |
2322 |
return l.key; |
2323 |
} |
2324 |
|
2325 |
public V getValue() { |
2326 |
Object v = lastValue; |
2327 |
if (last == null || v == null) |
2328 |
throw new IllegalStateException(); |
2329 |
return (V)v; |
2330 |
} |
2331 |
|
2332 |
public V setValue(V value) { |
2333 |
throw new UnsupportedOperationException(); |
2334 |
} |
2335 |
|
2336 |
public boolean equals(Object o) { |
2337 |
// If not acting as entry, just use default. |
2338 |
if (last == null) |
2339 |
return super.equals(o); |
2340 |
if (!(o instanceof Map.Entry)) |
2341 |
return false; |
2342 |
Map.Entry e = (Map.Entry)o; |
2343 |
return (getKey().equals(e.getKey()) && |
2344 |
getValue().equals(e.getValue())); |
2345 |
} |
2346 |
|
2347 |
public int hashCode() { |
2348 |
// If not acting as entry, just use default. |
2349 |
if (last == null) |
2350 |
return super.hashCode(); |
2351 |
return getKey().hashCode() ^ getValue().hashCode(); |
2352 |
} |
2353 |
|
2354 |
public String toString() { |
2355 |
// If not acting as entry, just use default. |
2356 |
if (last == null) |
2357 |
return super.toString(); |
2358 |
return getKey() + "=" + getValue(); |
2359 |
} |
2360 |
} |
2361 |
|
2362 |
/** |
2363 |
* Submap iterators start at given starting point at beginning of |
2364 |
* submap range, and advance until they are at end of range. |
2365 |
*/ |
2366 |
class SubMapEntryIterator extends EntryIterator { |
2367 |
final K fence; |
2368 |
SubMapEntryIterator(K least, K fence) { |
2369 |
super(least, fence); |
2370 |
this.fence = fence; |
2371 |
} |
2372 |
|
2373 |
public Map.Entry<K,V> next() { |
2374 |
lastValue = nextValue; |
2375 |
advance(fence); |
2376 |
return this; |
2377 |
} |
2378 |
} |
2379 |
|
2380 |
class SubMapValueIterator extends ConcurrentSkipListMapIterator |
2381 |
implements Iterator<V> { |
2382 |
final K fence; |
2383 |
SubMapValueIterator(K least, K fence) { |
2384 |
super(least, fence); |
2385 |
this.fence = fence; |
2386 |
} |
2387 |
|
2388 |
public V next() { |
2389 |
Object v = nextValue; |
2390 |
advance(fence); |
2391 |
return (V)v; |
2392 |
} |
2393 |
} |
2394 |
|
2395 |
class SubMapKeyIterator extends ConcurrentSkipListMapIterator |
2396 |
implements Iterator<K> { |
2397 |
final K fence; |
2398 |
SubMapKeyIterator(K least, K fence) { |
2399 |
super(least, fence); |
2400 |
this.fence = fence; |
2401 |
} |
2402 |
|
2403 |
public K next() { |
2404 |
Node<K,V> n = next; |
2405 |
advance(fence); |
2406 |
return n.key; |
2407 |
} |
2408 |
} |
2409 |
|
2410 |
/* ---------------- Utilities for views, sets, submaps -------------- */ |
2411 |
|
2412 |
// Factory methods for iterators needed by submaps and/or |
2413 |
// ConcurrentSkipListSet |
2414 |
|
2415 |
Iterator<K> keyIterator() { |
2416 |
return new KeyIterator(); |
2417 |
} |
2418 |
|
2419 |
SubMapEntryIterator subMapEntryIterator(K least, K fence) { |
2420 |
return new SubMapEntryIterator(least, fence); |
2421 |
} |
2422 |
|
2423 |
SubMapKeyIterator subMapKeyIterator(K least, K fence) { |
2424 |
return new SubMapKeyIterator(least, fence); |
2425 |
} |
2426 |
|
2427 |
SubMapValueIterator subMapValueIterator(K least, K fence) { |
2428 |
return new SubMapValueIterator(least, fence); |
2429 |
} |
2430 |
|
2431 |
|
2432 |
/** |
2433 |
* Version of remove with boolean return. Needed by |
2434 |
* view classes and ConcurrentSkipListSet |
2435 |
*/ |
2436 |
boolean removep(Object key) { |
2437 |
return doRemove(key, null) != null; |
2438 |
} |
2439 |
|
2440 |
/** |
2441 |
* Return SnapshotEntry for results of findNear ofter screening |
2442 |
* to ensure result is in given range. Needed by submaps. |
2443 |
* @param kkey the key |
2444 |
* @param rel the relation -- OR'ed combination of EQ, LT, GT |
2445 |
* @param least minimum allowed key value |
2446 |
* @param fence key greater than maximum allowed key value |
2447 |
* @return Entry fitting relation, or null if no such |
2448 |
*/ |
2449 |
SnapshotEntry<K,V> getNear(K kkey, int rel, K least, K fence) { |
2450 |
K key = kkey; |
2451 |
// Don't return keys less than least |
2452 |
if ((rel & LT) == 0) { |
2453 |
if (compare(key, least) < 0) { |
2454 |
key = least; |
2455 |
rel = rel | EQ; |
2456 |
} |
2457 |
} |
2458 |
|
2459 |
for (;;) { |
2460 |
Node<K,V> n = findNear(key, rel); |
2461 |
if (n == null || !inHalfOpenRange(n.key, least, fence)) |
2462 |
return null; |
2463 |
SnapshotEntry<K,V> e = n.createSnapshot(); |
2464 |
if (e != null) |
2465 |
return e; |
2466 |
} |
2467 |
} |
2468 |
|
2469 |
// Methods expanding out relational operations for submaps |
2470 |
|
2471 |
/** |
2472 |
* Return ceiling, or first node if key is null |
2473 |
*/ |
2474 |
Node<K,V> findCeiling(K key) { |
2475 |
return (key == null)? findFirst() : findNear(key, GT|EQ); |
2476 |
} |
2477 |
|
2478 |
/** |
2479 |
* Return lower node, or last node if key is null |
2480 |
*/ |
2481 |
Node<K,V> findLower(K key) { |
2482 |
return (key == null)? findLast() : findNear(key, LT); |
2483 |
} |
2484 |
|
2485 |
/** |
2486 |
* Find and remove least element of subrange. |
2487 |
*/ |
2488 |
SnapshotEntry<K,V> removeFirstEntryOfSubrange(K least, K fence) { |
2489 |
for (;;) { |
2490 |
Node<K,V> n = findCeiling(least); |
2491 |
if (n == null) |
2492 |
return null; |
2493 |
K k = n.key; |
2494 |
if (fence != null && compare(k, fence) >= 0) |
2495 |
return null; |
2496 |
V v = doRemove(k, null); |
2497 |
if (v != null) |
2498 |
return new SnapshotEntry<K,V>(k,v); |
2499 |
} |
2500 |
} |
2501 |
|
2502 |
SnapshotEntry<K,V> getCeiling(K key, K least, K fence) { |
2503 |
return getNear(key, GT|EQ, least, fence); |
2504 |
} |
2505 |
|
2506 |
SnapshotEntry<K,V> getLower(K key, K least, K fence) { |
2507 |
return getNear(key, LT, least, fence); |
2508 |
} |
2509 |
|
2510 |
SnapshotEntry<K,V> getFloor(K key, K least, K fence) { |
2511 |
return getNear(key, LT|EQ, least, fence); |
2512 |
} |
2513 |
|
2514 |
SnapshotEntry<K,V> getHigher(K key, K least, K fence) { |
2515 |
return getNear(key, GT, least, fence); |
2516 |
} |
2517 |
|
2518 |
// Key-returning relational methods for ConcurrentSkipListSet |
2519 |
|
2520 |
K ceilingKey(K key) { |
2521 |
Node<K,V> n = findNear(key, GT|EQ); |
2522 |
return (n == null)? null : n.key; |
2523 |
} |
2524 |
|
2525 |
K lowerKey(K key) { |
2526 |
Node<K,V> n = findNear(key, LT); |
2527 |
return (n == null)? null : n.key; |
2528 |
} |
2529 |
|
2530 |
K floorKey(K key) { |
2531 |
Node<K,V> n = findNear(key, LT|EQ); |
2532 |
return (n == null)? null : n.key; |
2533 |
} |
2534 |
|
2535 |
K higherKey(K key) { |
2536 |
Node<K,V> n = findNear(key, GT); |
2537 |
return (n == null)? null : n.key; |
2538 |
} |
2539 |
|
2540 |
K lowestKey() { |
2541 |
Node<K,V> n = findFirst(); |
2542 |
return (n == null)? null : n.key; |
2543 |
} |
2544 |
|
2545 |
K highestKey() { |
2546 |
Node<K,V> n = findLast(); |
2547 |
return (n == null)? null : n.key; |
2548 |
} |
2549 |
|
2550 |
/* ---------------- Views -------------- */ |
2551 |
|
2552 |
final class KeySet extends AbstractSet<K> { |
2553 |
public Iterator<K> iterator() { |
2554 |
return new KeyIterator(); |
2555 |
} |
2556 |
public boolean isEmpty() { |
2557 |
return ConcurrentSkipListMap.this.isEmpty(); |
2558 |
} |
2559 |
public int size() { |
2560 |
return ConcurrentSkipListMap.this.size(); |
2561 |
} |
2562 |
public boolean contains(Object o) { |
2563 |
return ConcurrentSkipListMap.this.containsKey(o); |
2564 |
} |
2565 |
public boolean remove(Object o) { |
2566 |
return ConcurrentSkipListMap.this.removep(o); |
2567 |
} |
2568 |
public void clear() { |
2569 |
ConcurrentSkipListMap.this.clear(); |
2570 |
} |
2571 |
public Object[] toArray() { |
2572 |
Collection<K> c = new ArrayList<K>(); |
2573 |
for (Iterator<K> i = iterator(); i.hasNext(); ) |
2574 |
c.add(i.next()); |
2575 |
return c.toArray(); |
2576 |
} |
2577 |
public <T> T[] toArray(T[] a) { |
2578 |
Collection<K> c = new ArrayList<K>(); |
2579 |
for (Iterator<K> i = iterator(); i.hasNext(); ) |
2580 |
c.add(i.next()); |
2581 |
return c.toArray(a); |
2582 |
} |
2583 |
} |
2584 |
|
2585 |
|
2586 |
final class Values extends AbstractCollection<V> { |
2587 |
public Iterator<V> iterator() { |
2588 |
return new ValueIterator(); |
2589 |
} |
2590 |
public boolean isEmpty() { |
2591 |
return ConcurrentSkipListMap.this.isEmpty(); |
2592 |
} |
2593 |
public int size() { |
2594 |
return ConcurrentSkipListMap.this.size(); |
2595 |
} |
2596 |
public boolean contains(Object o) { |
2597 |
return ConcurrentSkipListMap.this.containsValue(o); |
2598 |
} |
2599 |
public void clear() { |
2600 |
ConcurrentSkipListMap.this.clear(); |
2601 |
} |
2602 |
public Object[] toArray() { |
2603 |
Collection<V> c = new ArrayList<V>(); |
2604 |
for (Iterator<V> i = iterator(); i.hasNext(); ) |
2605 |
c.add(i.next()); |
2606 |
return c.toArray(); |
2607 |
} |
2608 |
public <T> T[] toArray(T[] a) { |
2609 |
Collection<V> c = new ArrayList<V>(); |
2610 |
for (Iterator<V> i = iterator(); i.hasNext(); ) |
2611 |
c.add(i.next()); |
2612 |
return c.toArray(a); |
2613 |
} |
2614 |
} |
2615 |
|
2616 |
final class EntrySet extends AbstractSet<Map.Entry<K,V>> { |
2617 |
public Iterator<Map.Entry<K,V>> iterator() { |
2618 |
return new EntryIterator(); |
2619 |
} |
2620 |
public boolean contains(Object o) { |
2621 |
if (!(o instanceof Map.Entry)) |
2622 |
return false; |
2623 |
Map.Entry<K,V> e = (Map.Entry<K,V>)o; |
2624 |
V v = ConcurrentSkipListMap.this.get(e.getKey()); |
2625 |
return v != null && v.equals(e.getValue()); |
2626 |
} |
2627 |
public boolean remove(Object o) { |
2628 |
if (!(o instanceof Map.Entry)) |
2629 |
return false; |
2630 |
Map.Entry<K,V> e = (Map.Entry<K,V>)o; |
2631 |
return ConcurrentSkipListMap.this.remove(e.getKey(), e.getValue()); |
2632 |
} |
2633 |
public boolean isEmpty() { |
2634 |
return ConcurrentSkipListMap.this.isEmpty(); |
2635 |
} |
2636 |
public int size() { |
2637 |
return ConcurrentSkipListMap.this.size(); |
2638 |
} |
2639 |
public void clear() { |
2640 |
ConcurrentSkipListMap.this.clear(); |
2641 |
} |
2642 |
|
2643 |
public Object[] toArray() { |
2644 |
Collection<Map.Entry<K,V>> c = new ArrayList<Map.Entry<K,V>>(); |
2645 |
for (Node<K,V> n = findFirst(); n != null; n = n.next) { |
2646 |
Map.Entry<K,V> e = n.createSnapshot(); |
2647 |
if (e != null) |
2648 |
c.add(e); |
2649 |
} |
2650 |
return c.toArray(); |
2651 |
} |
2652 |
public <T> T[] toArray(T[] a) { |
2653 |
Collection<Map.Entry<K,V>> c = new ArrayList<Map.Entry<K,V>>(); |
2654 |
for (Node<K,V> n = findFirst(); n != null; n = n.next) { |
2655 |
Map.Entry<K,V> e = n.createSnapshot(); |
2656 |
if (e != null) |
2657 |
c.add(e); |
2658 |
} |
2659 |
return c.toArray(a); |
2660 |
} |
2661 |
} |
2662 |
} |