Commit | Line | Data |
---|---|---|
08fe2e33 NR |
1 | package be.nikiroo.fanfix; |
2 | ||
3 | import java.io.BufferedOutputStream; | |
4 | import java.io.File; | |
5 | import java.io.FileInputStream; | |
6 | import java.io.FileOutputStream; | |
7 | import java.io.IOException; | |
8 | import java.io.InputStream; | |
9 | import java.net.CookieHandler; | |
10 | import java.net.CookieManager; | |
11 | import java.net.CookiePolicy; | |
12 | import java.net.CookieStore; | |
13 | import java.net.HttpCookie; | |
14 | import java.net.HttpURLConnection; | |
15 | import java.net.URISyntaxException; | |
16 | import java.net.URL; | |
17 | import java.net.URLConnection; | |
08fe2e33 NR |
18 | import java.util.Date; |
19 | import java.util.Map; | |
20 | import java.util.zip.GZIPInputStream; | |
21 | ||
22 | import javax.imageio.ImageIO; | |
23 | ||
24 | import be.nikiroo.fanfix.bundles.Config; | |
25 | import be.nikiroo.fanfix.supported.BasicSupport; | |
26 | import be.nikiroo.utils.IOUtils; | |
27 | import be.nikiroo.utils.MarkableFileInputStream; | |
28 | import be.nikiroo.utils.StringUtils; | |
29 | ||
30 | /** | |
31 | * This cache will manage Internet (and local) downloads, as well as put the | |
32 | * downloaded files into a cache. | |
33 | * <p> | |
34 | * As long the cached resource is not too old, it will use it instead of | |
35 | * retrieving the file again. | |
36 | * | |
37 | * @author niki | |
38 | */ | |
39 | public class Cache { | |
40 | private File dir; | |
41 | private String UA; | |
42 | private long tooOldChanging; | |
43 | private long tooOldStable; | |
44 | private CookieManager cookies; | |
45 | ||
46 | /** | |
47 | * Create a new {@link Cache} object. | |
48 | * | |
49 | * @param dir | |
50 | * the directory to use as cache | |
51 | * @param UA | |
52 | * the User-Agent to use to download the resources | |
53 | * @param hoursChanging | |
54 | * the number of hours after which a cached file that is thought | |
55 | * to change ~often is considered too old (or -1 for | |
56 | * "never too old") | |
57 | * @param hoursStable | |
58 | * the number of hours after which a LARGE cached file that is | |
59 | * thought to change rarely is considered too old (or -1 for | |
60 | * "never too old") | |
61 | * | |
62 | * @throws IOException | |
63 | * in case of I/O error | |
64 | */ | |
65 | public Cache(File dir, String UA, int hoursChanging, int hoursStable) | |
66 | throws IOException { | |
67 | this.dir = dir; | |
68 | this.UA = UA; | |
69 | this.tooOldChanging = 1000 * 60 * 60 * hoursChanging; | |
70 | this.tooOldStable = 1000 * 60 * 60 * hoursStable; | |
71 | ||
72 | if (dir != null) { | |
73 | if (!dir.exists()) { | |
74 | dir.mkdirs(); | |
75 | } | |
76 | } | |
77 | ||
78 | if (dir == null || !dir.exists()) { | |
79 | throw new IOException("Cannot create the cache directory: " | |
80 | + (dir == null ? "null" : dir.getAbsolutePath())); | |
81 | } | |
82 | ||
83 | cookies = new CookieManager(); | |
84 | cookies.setCookiePolicy(CookiePolicy.ACCEPT_ALL); | |
85 | CookieHandler.setDefault(cookies); | |
86 | } | |
87 | ||
88 | /** | |
89 | * Open a resource (will load it from the cache if possible, or save it into | |
90 | * the cache after downloading if not). | |
91 | * | |
92 | * @param url | |
93 | * the resource to open | |
94 | * @param support | |
95 | * the support to use to download the resource | |
96 | * @param stable | |
97 | * TRUE for more stable resources, FALSE when they often change | |
98 | * | |
99 | * @return the opened resource | |
100 | * | |
101 | * @throws IOException | |
102 | * in case of I/O error | |
103 | */ | |
104 | public InputStream open(URL url, BasicSupport support, boolean stable) | |
105 | throws IOException { | |
106 | return open(url, support, stable, url); | |
107 | } | |
108 | ||
109 | /** | |
110 | * Open a resource (will load it from the cache if possible, or save it into | |
111 | * the cache after downloading if not). | |
112 | * <p> | |
113 | * The cached resource will be assimilated to the given original {@link URL} | |
114 | * | |
115 | * @param url | |
116 | * the resource to open | |
117 | * @param support | |
118 | * the support to use to download the resource | |
119 | * @param stable | |
120 | * TRUE for more stable resources, FALSE when they often change | |
121 | * @param originalUrl | |
122 | * the original {@link URL} used to locate the cached resource | |
123 | * | |
124 | * @return the opened resource | |
125 | * | |
126 | * @throws IOException | |
127 | * in case of I/O error | |
128 | */ | |
129 | public InputStream open(URL url, BasicSupport support, boolean stable, | |
130 | URL originalUrl) throws IOException { | |
131 | try { | |
132 | InputStream in = load(originalUrl, false, stable); | |
133 | if (in == null) { | |
134 | try { | |
135 | save(url, support, originalUrl); | |
136 | } catch (IOException e) { | |
137 | throw new IOException("Cannot save the url: " | |
138 | + (url == null ? "null" : url.toString()), e); | |
139 | } | |
140 | ||
141 | in = load(originalUrl, true, stable); | |
142 | } | |
143 | ||
144 | return in; | |
145 | } catch (IOException e) { | |
146 | throw new IOException("Cannot open the url: " | |
147 | + (url == null ? "null" : url.toString()), e); | |
148 | } | |
149 | } | |
150 | ||
151 | /** | |
152 | * Refresh the resource into cache if needed. | |
153 | * | |
154 | * @param url | |
155 | * the resource to open | |
156 | * @param support | |
157 | * the support to use to download the resource | |
158 | * @param stable | |
159 | * TRUE for more stable resources, FALSE when they often change | |
160 | * | |
161 | * @return TRUE if it was pre-downloaded | |
162 | * | |
163 | * @throws IOException | |
164 | * in case of I/O error | |
165 | */ | |
166 | public void refresh(URL url, BasicSupport support, boolean stable) | |
167 | throws IOException { | |
168 | File cached = getCached(url); | |
169 | if (cached.exists() && !isOld(cached, stable)) { | |
170 | return; | |
171 | } | |
172 | ||
173 | open(url, support, stable).close(); | |
174 | } | |
175 | ||
176 | /** | |
177 | * Check the resource to see if it is in the cache. | |
178 | * | |
179 | * @param url | |
180 | * the resource to check | |
181 | * | |
182 | * @return TRUE if it is | |
183 | * | |
184 | */ | |
185 | public boolean check(URL url) { | |
186 | return getCached(url).exists(); | |
187 | } | |
188 | ||
189 | /** | |
190 | * Open a resource (will load it from the cache if possible, or save it into | |
191 | * the cache after downloading if not) as an Image, then save it where | |
192 | * requested. | |
193 | * <p> | |
194 | * This version will not always work properly if the original file was not | |
195 | * downloaded before. | |
196 | * | |
197 | * @param url | |
198 | * the resource to open | |
199 | * | |
200 | * @return the opened resource image | |
201 | * | |
202 | * @throws IOException | |
203 | * in case of I/O error | |
204 | */ | |
205 | public void saveAsImage(URL url, File target) throws IOException { | |
206 | URL cachedUrl = new URL(url.toString() | |
207 | + "." | |
208 | + Instance.getConfig().getString(Config.IMAGE_FORMAT_CONTENT) | |
209 | .toLowerCase()); | |
210 | File cached = getCached(cachedUrl); | |
211 | ||
212 | if (!cached.exists() || isOld(cached, true)) { | |
213 | InputStream imageIn = Instance.getCache().open(url, null, true); | |
214 | ImageIO.write(StringUtils.toImage(imageIn), Instance.getConfig() | |
215 | .getString(Config.IMAGE_FORMAT_CONTENT).toLowerCase(), | |
216 | cached); | |
217 | } | |
218 | ||
219 | IOUtils.write(new FileInputStream(cached), target); | |
220 | } | |
221 | ||
222 | /** | |
223 | * Manually add this item to the cache. | |
224 | * | |
225 | * @param in | |
226 | * the input data | |
227 | * @param uniqueID | |
228 | * a unique ID for this resource | |
229 | * | |
9252c65e | 230 | * @return the resulting {@link File} |
08fe2e33 NR |
231 | * |
232 | * @throws IOException | |
233 | * in case of I/O error | |
234 | */ | |
235 | public File addToCache(InputStream in, String uniqueID) throws IOException { | |
236 | File file = getCached(new File(uniqueID).toURI().toURL()); | |
237 | IOUtils.write(in, file); | |
238 | return file; | |
239 | } | |
240 | ||
241 | /** | |
242 | * Clean the cache (delete the cached items). | |
243 | * | |
244 | * @param onlyOld | |
245 | * only clean the files that are considered too old | |
246 | * | |
247 | * @return the number of cleaned items | |
248 | */ | |
249 | public int cleanCache(boolean onlyOld) { | |
250 | int num = 0; | |
251 | for (File file : dir.listFiles()) { | |
252 | if (!onlyOld || isOld(file, true)) { | |
253 | if (file.delete()) { | |
254 | num++; | |
255 | } else { | |
256 | System.err.println("Cannot delete temporary file: " | |
257 | + file.getAbsolutePath()); | |
258 | } | |
259 | } | |
260 | } | |
261 | return num; | |
262 | } | |
263 | ||
264 | /** | |
265 | * Open a resource from the cache if it exists. | |
266 | * | |
267 | * @param url | |
268 | * the resource to open | |
269 | * @return the opened resource | |
270 | * @throws IOException | |
271 | * in case of I/O error | |
272 | */ | |
273 | private InputStream load(URL url, boolean allowOld, boolean stable) | |
274 | throws IOException { | |
275 | File cached = getCached(url); | |
276 | if (cached.exists() && !isOld(cached, stable)) { | |
277 | return new MarkableFileInputStream(new FileInputStream(cached)); | |
278 | } | |
279 | ||
280 | return null; | |
281 | } | |
282 | ||
283 | /** | |
284 | * Save the given resource to the cache. | |
285 | * | |
286 | * @param url | |
287 | * the resource | |
288 | * @param support | |
289 | * the {@link BasicSupport} used to download it | |
290 | * @param originalUrl | |
291 | * the original {@link URL} used to locate the cached resource | |
292 | * | |
293 | * @throws IOException | |
294 | * in case of I/O error | |
295 | * @throws URISyntaxException | |
296 | */ | |
297 | private void save(URL url, BasicSupport support, URL originalUrl) | |
298 | throws IOException { | |
299 | URLConnection conn = url.openConnection(); | |
300 | ||
301 | conn.setRequestProperty("User-Agent", UA); | |
302 | conn.setRequestProperty("Cookie", generateCookies(support)); | |
303 | conn.setRequestProperty("Accept-Encoding", "gzip"); | |
89cb07a6 | 304 | if (support != null && support.getCurrentReferer() != null) { |
08fe2e33 NR |
305 | conn.setRequestProperty("Referer", support.getCurrentReferer() |
306 | .toString()); | |
307 | conn.setRequestProperty("Host", support.getCurrentReferer() | |
308 | .getHost()); | |
309 | } | |
310 | ||
311 | conn.connect(); | |
312 | ||
313 | // Check if redirect | |
314 | if (conn instanceof HttpURLConnection | |
315 | && ((HttpURLConnection) conn).getResponseCode() / 100 == 3) { | |
316 | String newUrl = conn.getHeaderField("Location"); | |
317 | save(new URL(newUrl), support, originalUrl); | |
318 | return; | |
319 | } | |
320 | ||
321 | InputStream in = conn.getInputStream(); | |
322 | if ("gzip".equals(conn.getContentEncoding())) { | |
323 | in = new GZIPInputStream(in); | |
324 | } | |
325 | ||
326 | try { | |
327 | File cached = getCached(originalUrl); | |
328 | BufferedOutputStream out = new BufferedOutputStream( | |
329 | new FileOutputStream(cached)); | |
330 | try { | |
331 | byte[] buf = new byte[4096]; | |
332 | int len; | |
333 | while ((len = in.read(buf)) > 0) { | |
334 | out.write(buf, 0, len); | |
335 | } | |
336 | } finally { | |
337 | out.close(); | |
338 | } | |
339 | } finally { | |
340 | in.close(); | |
341 | } | |
342 | } | |
343 | ||
344 | /** | |
345 | * Check if the {@link File} is too old according to | |
346 | * {@link Cache#tooOldChanging}. | |
347 | * | |
348 | * @param file | |
349 | * the file to check | |
350 | * @param stable | |
351 | * TRUE to denote files that are not supposed to change too often | |
352 | * | |
353 | * @return TRUE if it is | |
354 | */ | |
355 | private boolean isOld(File file, boolean stable) { | |
356 | long max = tooOldChanging; | |
357 | if (stable) { | |
358 | max = tooOldStable; | |
359 | } | |
360 | ||
361 | if (max < 0) { | |
362 | return false; | |
363 | } | |
364 | ||
365 | long time = new Date().getTime() - file.lastModified(); | |
366 | if (time < 0) { | |
367 | System.err.println("Timestamp in the future for file: " | |
368 | + file.getAbsolutePath()); | |
369 | } | |
370 | ||
371 | return time < 0 || time > max; | |
372 | } | |
373 | ||
374 | /** | |
375 | * Get the cache resource from the cache if it is present for this | |
376 | * {@link URL}. | |
377 | * | |
378 | * @param url | |
379 | * the url | |
380 | * @return the cached version if present, NULL if not | |
381 | */ | |
382 | private File getCached(URL url) { | |
383 | String name = url.getHost(); | |
384 | if (name == null || name.length() == 0) { | |
385 | name = url.getFile(); | |
386 | } else { | |
387 | name = url.toString(); | |
388 | } | |
389 | ||
390 | name = name.replace('/', '_').replace(':', '_'); | |
391 | ||
392 | return new File(dir, name); | |
393 | } | |
394 | ||
395 | /** | |
396 | * Generate the cookie {@link String} from the local {@link CookieStore} so | |
397 | * it is ready to be passed. | |
398 | * | |
399 | * @return the cookie | |
400 | */ | |
401 | private String generateCookies(BasicSupport support) { | |
402 | StringBuilder builder = new StringBuilder(); | |
403 | for (HttpCookie cookie : cookies.getCookieStore().getCookies()) { | |
404 | if (builder.length() > 0) { | |
405 | builder.append(';'); | |
406 | } | |
407 | ||
408 | // TODO: check if format is ok | |
409 | builder.append(cookie.toString()); | |
410 | } | |
411 | ||
412 | if (support != null) { | |
413 | for (Map.Entry<String, String> set : support.getCookies() | |
414 | .entrySet()) { | |
415 | if (builder.length() > 0) { | |
416 | builder.append(';'); | |
417 | } | |
418 | builder.append(set.getKey()); | |
419 | builder.append('='); | |
420 | builder.append(set.getValue()); | |
421 | } | |
422 | } | |
423 | ||
424 | return builder.toString(); | |
425 | } | |
426 | } |