source: josm/trunk/src/org/openstreetmap/josm/data/cache/JCSCachedTileLoaderJob.java

Last change on this file was 18832, checked in by stoecker, 7 months ago

JavaDoc
  • Property svn:eol-style set to native
File size: 24.7 KB
Line 
1// License: GPL. For details, see LICENSE file.
2package org.openstreetmap.josm.data.cache;
3
4import java.io.File;
5import java.io.FileNotFoundException;
6import java.io.IOException;
7import java.io.InputStream;
8import java.net.HttpURLConnection;
9import java.net.URL;
10import java.nio.file.Files;
11import java.security.SecureRandom;
12import java.util.Collections;
13import java.util.List;
14import java.util.Map;
15import java.util.Set;
16import java.util.concurrent.ConcurrentHashMap;
17import java.util.concurrent.ConcurrentMap;
18import java.util.concurrent.LinkedBlockingDeque;
19import java.util.concurrent.ThreadPoolExecutor;
20import java.util.concurrent.TimeUnit;
21import java.util.regex.Matcher;
22
23import org.openstreetmap.josm.data.cache.ICachedLoaderListener.LoadResult;
24import org.openstreetmap.josm.data.imagery.TileJobOptions;
25import org.openstreetmap.josm.data.preferences.IntegerProperty;
26import org.openstreetmap.josm.tools.CheckParameterUtil;
27import org.openstreetmap.josm.tools.HttpClient;
28import org.openstreetmap.josm.tools.Logging;
29import org.openstreetmap.josm.tools.Utils;
30
31import org.apache.commons.jcs3.access.behavior.ICacheAccess;
32import org.apache.commons.jcs3.engine.behavior.ICacheElement;
33
34/**
35 * Generic loader for HTTP based tiles. Uses custom attribute, to check, if entry has expired
36 * according to HTTP headers sent with tile. If so, it tries to verify using Etags
37 * or If-Modified-Since / Last-Modified.
38 *
39 * If the tile is not valid, it will try to download it from remote service and put it
40 * to cache. If remote server will fail it will try to use stale entry.
41 *
42 * This class will keep only one Job running for specified tile. All others will just finish, but
43 * listeners will be gathered and notified, once download job will be finished
44 *
45 * @author Wiktor Niesiobędzki
46 * @param <K> cache entry key type
47 * @param <V> cache value type
48 * @since 8168
49 */
50public abstract class JCSCachedTileLoaderJob<K, V extends CacheEntry> implements ICachedLoaderJob<K> {
51 protected static final long DEFAULT_EXPIRE_TIME = TimeUnit.DAYS.toMillis(7);
52 // Limit for the max-age value send by the server.
53 protected static final long EXPIRE_TIME_SERVER_LIMIT = TimeUnit.DAYS.toMillis(28);
54 // Absolute expire time limit. Cached tiles that are older will not be used,
55 // even if the refresh from the server fails.
56 protected static final long ABSOLUTE_EXPIRE_TIME_LIMIT = TimeUnit.DAYS.toMillis(365);
57
58 /**
59 * maximum download threads that will be started
60 */
61 public static final IntegerProperty THREAD_LIMIT = new IntegerProperty("cache.jcs.max_threads", 10);
62
63 /*
64 * ThreadPoolExecutor starts new threads, until THREAD_LIMIT is reached. Then it puts tasks into LinkedBlockingDeque.
65 *
66 * The queue works FIFO, so one needs to take care about ordering of the entries submitted
67 *
68 * There is no point in canceling tasks, that are already taken by worker threads (if we made so much effort, we can at least cache
69 * the response, so later it could be used). We could actually cancel what is in LIFOQueue, but this is a tradeoff between simplicity
70 * and performance (we do want to have something to offer to worker threads before tasks will be resubmitted by class consumer)
71 */
72
73 private static final ThreadPoolExecutor DEFAULT_DOWNLOAD_JOB_DISPATCHER = new ThreadPoolExecutor(
74 1, // we have a small queue, so threads will be quickly started (threads are started only, when queue is full)
75 THREAD_LIMIT.get(), // do not this number of threads
76 30, // keepalive for thread
77 TimeUnit.SECONDS,
78 // make queue of LIFO type - so recently requested tiles will be loaded first (assuming that these are which user is waiting to see)
79 new LinkedBlockingDeque<Runnable>(),
80 Utils.newThreadFactory("JCS-downloader-%d", Thread.NORM_PRIORITY)
81 );
82
83 private static final ConcurrentMap<String, Set<ICachedLoaderListener>> inProgress = new ConcurrentHashMap<>();
84 private static final ConcurrentMap<String, Boolean> useHead = new ConcurrentHashMap<>();
85
86 protected final long now; // when the job started
87
88 protected final ICacheAccess<K, V> cache;
89 private ICacheElement<K, V> cacheElement;
90 protected V cacheData;
91 protected CacheEntryAttributes attributes;
92
93 // HTTP connection parameters
94 private final int connectTimeout;
95 private final int readTimeout;
96 private final Map<String, String> headers;
97 private final ThreadPoolExecutor downloadJobExecutor;
98 private Runnable finishTask;
99 private boolean force;
100 private final long minimumExpiryTime;
101
102 /**
103 * @param cache cache instance that we will work on
104 * @param options options of the request
105 * @param downloadJobExecutor that will be executing the jobs
106 */
107 protected JCSCachedTileLoaderJob(ICacheAccess<K, V> cache,
108 TileJobOptions options,
109 ThreadPoolExecutor downloadJobExecutor) {
110 CheckParameterUtil.ensureParameterNotNull(cache, "cache");
111 this.cache = cache;
112 this.now = System.currentTimeMillis();
113 this.connectTimeout = options.getConnectionTimeout();
114 this.readTimeout = options.getReadTimeout();
115 this.headers = options.getHeaders();
116 this.downloadJobExecutor = downloadJobExecutor;
117 this.minimumExpiryTime = TimeUnit.SECONDS.toMillis(options.getMinimumExpiryTime());
118 }
119
120 /**
121 * @param cache cache instance that we will work on
122 * @param options of the request
123 */
124 protected JCSCachedTileLoaderJob(ICacheAccess<K, V> cache,
125 TileJobOptions options) {
126 this(cache, options, DEFAULT_DOWNLOAD_JOB_DISPATCHER);
127 }
128
129 private void ensureCacheElement() {
130 if (cacheElement == null && getCacheKey() != null) {
131 cacheElement = cache.getCacheElement(getCacheKey());
132 if (cacheElement != null) {
133 attributes = (CacheEntryAttributes) cacheElement.getElementAttributes();
134 cacheData = cacheElement.getVal();
135 }
136 }
137 }
138
139 @Override
140 public V get() {
141 ensureCacheElement();
142 return cacheData;
143 }
144
145 @Override
146 public void submit(ICachedLoaderListener listener, boolean force) throws IOException {
147 this.force = force;
148 boolean first = false;
149 URL url = getUrl();
150 String deduplicationKey = null;
151 if (url != null) {
152 // url might be null, for example when Bing Attribution is not loaded yet
153 deduplicationKey = url.toString();
154 }
155 if (deduplicationKey == null) {
156 Logging.warn("No url returned for: {0}, skipping", getCacheKey());
157 throw new IllegalArgumentException("No url returned");
158 }
159 synchronized (this) {
160 first = !inProgress.containsKey(deduplicationKey);
161 }
162 inProgress.computeIfAbsent(deduplicationKey, k -> ConcurrentHashMap.newKeySet()).add(listener);
163
164 if (first || force) {
165 // submit all jobs to separate thread, so calling thread is not blocked with IO when loading from disk
166 Logging.debug("JCS - Submitting job for execution for url: {0}", getUrlNoException());
167 downloadJobExecutor.execute(this);
168 }
169 }
170
171 /**
172 * This method is run when job has finished
173 */
174 protected void executionFinished() {
175 if (finishTask != null) {
176 finishTask.run();
177 }
178 }
179
180 /**
181 * Checks if object from cache has sufficient data to be returned.
182 * @return {@code true} if object from cache has sufficient data to be returned
183 */
184 protected boolean isObjectLoadable() {
185 if (cacheData == null) {
186 return false;
187 }
188 return cacheData.getContent().length > 0;
189 }
190
191 /**
192 * Simple implementation. All errors should be cached as empty. Though some JDK (JDK8 on Windows for example)
193 * doesn't return 4xx error codes, instead they do throw an FileNotFoundException or IOException
194 * @param headerFields headers sent by server
195 * @param responseCode http status code
196 *
197 * @return true if we should put empty object into cache, regardless of what remote resource has returned
198 */
199 protected boolean cacheAsEmpty(Map<String, List<String>> headerFields, int responseCode) {
200 return attributes.getResponseCode() < 500;
201 }
202
203 /**
204 * Returns key under which discovered server settings will be kept.
205 * @return key under which discovered server settings will be kept
206 */
207 protected String getServerKey() {
208 try {
209 return getUrl().getHost();
210 } catch (IOException e) {
211 Logging.trace(e);
212 return null;
213 }
214 }
215
216 @Override
217 public void run() {
218 final Thread currentThread = Thread.currentThread();
219 final String oldName = currentThread.getName();
220 currentThread.setName("JCS Downloading: " + getUrlNoException());
221 Logging.debug("JCS - starting fetch of url: {0} ", getUrlNoException());
222 ensureCacheElement();
223 try {
224 // try to fetch from cache
225 if (!force && cacheElement != null && isCacheElementValid() && isObjectLoadable()) {
226 // we got something in cache, and it's valid, so lets return it
227 Logging.debug("JCS - Returning object from cache: {0}", getCacheKey());
228 finishLoading(LoadResult.SUCCESS);
229 return;
230 }
231
232 // try to load object from remote resource
233 if (loadObject()) {
234 finishLoading(LoadResult.SUCCESS);
235 } else {
236 // if loading failed - check if we can return stale entry
237 if (isObjectLoadable()) {
238 // try to get stale entry in cache
239 finishLoading(LoadResult.SUCCESS);
240 Logging.debug("JCS - found stale object in cache: {0}", getUrlNoException());
241 } else {
242 // failed completely
243 finishLoading(LoadResult.FAILURE);
244 }
245 }
246 } finally {
247 executionFinished();
248 currentThread.setName(oldName);
249 }
250 }
251
252 private void finishLoading(LoadResult result) {
253 Set<ICachedLoaderListener> listeners;
254 try {
255 listeners = inProgress.remove(getUrl().toString());
256 } catch (IOException e) {
257 listeners = null;
258 Logging.trace(e);
259 }
260 if (listeners == null) {
261 Logging.warn("Listener not found for URL: {0}. Listener not notified!", getUrlNoException());
262 return;
263 }
264 for (ICachedLoaderListener l: listeners) {
265 l.loadingFinished(cacheData, attributes, result);
266 }
267 }
268
269 protected boolean isCacheElementValid() {
270 long expires = attributes.getExpirationTime();
271
272 // check by expire date set by server
273 if (expires != 0L) {
274 // put a limit to the expire time (some servers send a value
275 // that is too large)
276 expires = Math.min(expires, attributes.getCreateTime() + Math.max(EXPIRE_TIME_SERVER_LIMIT, minimumExpiryTime));
277 if (now > expires) {
278 Logging.debug("JCS - Object {0} has expired -> valid to {1}, now is: {2}",
279 getUrlNoException(), Long.toString(expires), Long.toString(now));
280 return false;
281 }
282 } else if (attributes.getLastModification() > 0 &&
283 now - attributes.getLastModification() > Math.max(DEFAULT_EXPIRE_TIME, minimumExpiryTime)) {
284 // check by file modification date
285 Logging.debug("JCS - Object has expired, maximum file age reached {0}", getUrlNoException());
286 return false;
287 } else if (now - attributes.getCreateTime() > Math.max(DEFAULT_EXPIRE_TIME, minimumExpiryTime)) {
288 Logging.debug("JCS - Object has expired, maximum time since object creation reached {0}", getUrlNoException());
289 return false;
290 }
291 return true;
292 }
293
294 /**
295 * Load an cache object
296 * @return {@code true} if object was successfully downloaded, false, if there was a loading failure
297 * @since 18831
298 */
299 protected boolean loadObject() {
300 if (attributes == null) {
301 attributes = new CacheEntryAttributes();
302 }
303 final URL url = this.getUrlNoException();
304 if (url == null) {
305 return false;
306 }
307
308 if (url.getProtocol().contains("http")) {
309 return loadObjectHttp();
310 }
311 if (url.getProtocol().contains("file")) {
312 return loadObjectFile(url);
313 }
314
315 return false;
316 }
317
318 private boolean loadObjectFile(URL url) {
319 String fileName = url.toExternalForm();
320 File file = new File(fileName.substring("file:/".length() - 1));
321 if (!file.exists()) {
322 file = new File(fileName.substring("file://".length() - 1));
323 }
324 try (InputStream fileInputStream = Files.newInputStream(file.toPath())) {
325 cacheData = createCacheEntry(Utils.readBytesFromStream(fileInputStream));
326 cache.put(getCacheKey(), cacheData, attributes);
327 return true;
328 } catch (IOException e) {
329 Logging.error(e);
330 attributes.setError(e);
331 attributes.setException(e);
332 }
333 return false;
334 }
335
336 /**
337 * Load an cache object via HTTP
338 * @return {@code true} if object was successfully downloaded via http, false, if there was a loading failure
339 */
340 private boolean loadObjectHttp() {
341 try {
342 // if we have object in cache, and host doesn't support If-Modified-Since nor If-None-Match
343 // then just use HEAD request and check returned values
344 if (isObjectLoadable() &&
345 Boolean.TRUE.equals(useHead.get(getServerKey())) &&
346 isCacheValidUsingHead()) {
347 Logging.debug("JCS - cache entry verified using HEAD request: {0}", getUrl());
348 return true;
349 }
350
351 Logging.debug("JCS - starting HttpClient GET request for URL: {0}", getUrl());
352 final HttpClient request = getRequest("GET");
353
354 if (isObjectLoadable() &&
355 (now - attributes.getLastModification()) <= ABSOLUTE_EXPIRE_TIME_LIMIT) {
356 request.setIfModifiedSince(attributes.getLastModification());
357 }
358 if (isObjectLoadable() && attributes.getEtag() != null) {
359 request.setHeader("If-None-Match", attributes.getEtag());
360 }
361
362 final HttpClient.Response urlConn = request.connect();
363
364 if (urlConn.getResponseCode() == 304) {
365 // If isModifiedSince or If-None-Match has been set
366 // and the server answers with a HTTP 304 = "Not Modified"
367 Logging.debug("JCS - If-Modified-Since/ETag test: local version is up to date: {0}", getUrl());
368 // update cache attributes
369 attributes = parseHeaders(urlConn);
370 cache.put(getCacheKey(), cacheData, attributes);
371 return true;
372 } else if (isObjectLoadable() // we have an object in cache, but we haven't received 304 response code
373 && (
374 (attributes.getEtag() != null && attributes.getEtag().equals(urlConn.getHeaderField("ETag"))) ||
375 attributes.getLastModification() == urlConn.getLastModified())
376 ) {
377 // we sent ETag or If-Modified-Since, but didn't get 304 response code
378 // for further requests - use HEAD
379 String serverKey = getServerKey();
380 Logging.info("JCS - Host: {0} found not to return 304 codes for If-Modified-Since or If-None-Match headers",
381 serverKey);
382 useHead.put(serverKey, Boolean.TRUE);
383 }
384
385 attributes = parseHeaders(urlConn);
386
387 for (int i = 0; i < 5; ++i) {
388 if (urlConn.getResponseCode() == HttpURLConnection.HTTP_UNAVAILABLE) {
389 Thread.sleep(5000L+new SecureRandom().nextInt(5000));
390 continue;
391 }
392
393 attributes.setResponseCode(urlConn.getResponseCode());
394 byte[] raw;
395 if (urlConn.getResponseCode() == HttpURLConnection.HTTP_OK) {
396 raw = Utils.readBytesFromStream(urlConn.getContent());
397 } else {
398 raw = new byte[]{};
399 try {
400 String data = urlConn.fetchContent();
401 if (!data.isEmpty()) {
402 String detectErrorMessage = detectErrorMessage(data);
403 if (detectErrorMessage != null) {
404 attributes.setErrorMessage(detectErrorMessage);
405 }
406 }
407 } catch (IOException e) {
408 Logging.warn(e);
409 }
410 }
411
412 if (isResponseLoadable(urlConn.getHeaderFields(), urlConn.getResponseCode(), raw)) {
413 // we need to check cacheEmpty, so for cases, when data is returned, but we want to store
414 // as empty (eg. empty tile images) to save some space
415 cacheData = createCacheEntry(raw);
416 cache.put(getCacheKey(), cacheData, attributes);
417 Logging.debug("JCS - downloaded key: {0}, length: {1}, url: {2}",
418 getCacheKey(), raw.length, getUrl());
419 return true;
420 } else if (cacheAsEmpty(urlConn.getHeaderFields(), urlConn.getResponseCode())) {
421 cacheData = createCacheEntry(new byte[]{});
422 cache.put(getCacheKey(), cacheData, attributes);
423 Logging.debug("JCS - Caching empty object {0}", getUrl());
424 return true;
425 } else {
426 Logging.debug("JCS - failure during load - response is not loadable nor cached as empty");
427 return false;
428 }
429 }
430 } catch (FileNotFoundException e) {
431 Logging.debug("JCS - Caching empty object as server returned 404 for: {0}", getUrlNoException());
432 attributes.setResponseCode(404);
433 attributes.setError(e);
434 attributes.setException(e);
435 boolean doCache = isResponseLoadable(null, 404, null) || cacheAsEmpty(Collections.emptyMap(), 404);
436 if (doCache) {
437 cacheData = createCacheEntry(new byte[]{});
438 cache.put(getCacheKey(), cacheData, attributes);
439 }
440 return doCache;
441 } catch (IOException e) {
442 Logging.debug("JCS - IOException during communication with server for: {0}", getUrlNoException());
443 if (isObjectLoadable()) {
444 return true;
445 } else {
446 attributes.setError(e);
447 attributes.setException(e);
448 attributes.setResponseCode(599); // set dummy error code, greater than 500 so it will be not cached
449 return false;
450 }
451
452 } catch (InterruptedException e) {
453 attributes.setError(e);
454 attributes.setException(e);
455 Logging.logWithStackTrace(Logging.LEVEL_WARN, e, "JCS - Exception during download {0}", getUrlNoException());
456 Thread.currentThread().interrupt();
457 }
458 Logging.warn("JCS - Silent failure during download: {0}", getUrlNoException());
459 return false;
460 }
461
462 /**
463 * Tries do detect an error message from given string.
464 * @param data string to analyze
465 * @return error message if detected, or null
466 * @since 14535
467 */
468 public String detectErrorMessage(String data) {
469 Matcher m = HttpClient.getTomcatErrorMatcher(data);
470 return m.matches() ? m.group(1).replace("'", "''") : null;
471 }
472
473 /**
474 * Check if the object is loadable. This means, if the data will be parsed, and if this response
475 * will finish as successful retrieve.
476 *
477 * This simple implementation doesn't load empty response, nor client (4xx) and server (5xx) errors
478 *
479 * @param headerFields headers sent by server
480 * @param responseCode http status code
481 * @param raw data read from server
482 * @return true if object should be cached and returned to listener
483 */
484 protected boolean isResponseLoadable(Map<String, List<String>> headerFields, int responseCode, byte[] raw) {
485 return raw != null && raw.length != 0 && responseCode < 400;
486 }
487
488 protected abstract V createCacheEntry(byte[] content);
489
490 protected CacheEntryAttributes parseHeaders(HttpClient.Response urlConn) {
491 CacheEntryAttributes ret = new CacheEntryAttributes();
492
493 /*
494 * according to https://www.ietf.org/rfc/rfc2616.txt Cache-Control takes precedence over max-age
495 * max-age is for private caches, s-max-age is for shared caches. We take any value that is larger
496 */
497 Long expiration = 0L;
498 String cacheControl = urlConn.getHeaderField("Cache-Control");
499 if (cacheControl != null) {
500 for (String token: cacheControl.split(",", -1)) {
501 try {
502 if (token.startsWith("max-age=")) {
503 expiration = Math.max(expiration,
504 TimeUnit.SECONDS.toMillis(Long.parseLong(token.substring("max-age=".length())))
505 + System.currentTimeMillis()
506 );
507 }
508 if (token.startsWith("s-max-age=")) {
509 expiration = Math.max(expiration,
510 TimeUnit.SECONDS.toMillis(Long.parseLong(token.substring("s-max-age=".length())))
511 + System.currentTimeMillis()
512 );
513 }
514 } catch (NumberFormatException e) {
515 // ignore malformed Cache-Control headers
516 Logging.trace(e);
517 }
518 }
519 }
520
521 if (expiration.equals(0L)) {
522 expiration = urlConn.getExpiration();
523 }
524
525 // if nothing is found - set default
526 if (expiration.equals(0L)) {
527 expiration = System.currentTimeMillis() + DEFAULT_EXPIRE_TIME;
528 }
529
530 ret.setExpirationTime(Math.max(minimumExpiryTime + System.currentTimeMillis(), expiration));
531 ret.setLastModification(now);
532 ret.setEtag(urlConn.getHeaderField("ETag"));
533
534 return ret;
535 }
536
537 private HttpClient getRequest(String requestMethod) throws IOException {
538 final HttpClient urlConn = HttpClient.create(getUrl(), requestMethod);
539 urlConn.setAccept("text/html, image/png, image/jpeg, image/gif, */*");
540 urlConn.setReadTimeout(readTimeout); // 30 seconds read timeout
541 urlConn.setConnectTimeout(connectTimeout);
542 if (headers != null) {
543 urlConn.setHeaders(headers);
544 }
545
546 final boolean noCache = force
547 // To remove when switching to Java 11
548 // Workaround for https://bugs.openjdk.java.net/browse/JDK-8146450
549 || (Utils.getJavaVersion() == 8 && Utils.isRunningJavaWebStart());
550 urlConn.useCache(!noCache);
551
552 return urlConn;
553 }
554
555 private boolean isCacheValidUsingHead() throws IOException {
556 final HttpClient.Response urlConn = getRequest("HEAD").connect();
557 long lastModified = urlConn.getLastModified();
558 boolean ret = (attributes.getEtag() != null && attributes.getEtag().equals(urlConn.getHeaderField("ETag"))) ||
559 (lastModified != 0 && lastModified <= attributes.getLastModification());
560 if (ret) {
561 // update attributes
562 attributes = parseHeaders(urlConn);
563 cache.put(getCacheKey(), cacheData, attributes);
564 }
565 return ret;
566 }
567
568 /**
569 * TODO: move to JobFactory
570 * cancels all outstanding tasks in the queue.
571 */
572 public void cancelOutstandingTasks() {
573 for (Runnable r: downloadJobExecutor.getQueue()) {
574 if (downloadJobExecutor.remove(r) && r instanceof JCSCachedTileLoaderJob) {
575 ((JCSCachedTileLoaderJob<?, ?>) r).handleJobCancellation();
576 }
577 }
578 }
579
580 /**
581 * Sets a job, that will be run, when job will finish execution
582 * @param runnable that will be executed
583 */
584 public void setFinishedTask(Runnable runnable) {
585 this.finishTask = runnable;
586
587 }
588
589 /**
590 * Marks this job as canceled
591 */
592 public void handleJobCancellation() {
593 finishLoading(LoadResult.CANCELED);
594 }
595
596 private URL getUrlNoException() {
597 try {
598 return getUrl();
599 } catch (IOException e) {
600 Logging.trace(e);
601 return null;
602 }
603 }
604}
Note: See TracBrowser for help on using the repository browser.