Commit | Line | Data |
---|---|---|
08fe2e33 NR |
1 | package be.nikiroo.fanfix.supported; |
2 | ||
3 | import java.io.IOException; | |
b5e9855b | 4 | import java.io.UnsupportedEncodingException; |
c4b18c94 | 5 | import java.net.MalformedURLException; |
08fe2e33 | 6 | import java.net.URL; |
b5e9855b | 7 | import java.net.URLDecoder; |
ce297a79 | 8 | import java.util.AbstractMap; |
08fe2e33 | 9 | import java.util.ArrayList; |
9b863b20 | 10 | import java.util.Collections; |
8ac3d099 | 11 | import java.util.Date; |
b5e9855b | 12 | import java.util.LinkedList; |
08fe2e33 NR |
13 | import java.util.List; |
14 | import java.util.Map.Entry; | |
8ac3d099 | 15 | |
5cf61f35 NR |
16 | import org.json.JSONArray; |
17 | import org.json.JSONException; | |
18 | import org.json.JSONObject; | |
8ac3d099 NR |
19 | import org.jsoup.helper.DataUtil; |
20 | import org.jsoup.nodes.Document; | |
21 | import org.jsoup.nodes.Element; | |
08fe2e33 NR |
22 | |
23 | import be.nikiroo.fanfix.Instance; | |
5cf61f35 | 24 | import be.nikiroo.fanfix.bundles.Config; |
68686a37 | 25 | import be.nikiroo.fanfix.data.MetaData; |
16a81ef7 | 26 | import be.nikiroo.utils.Image; |
3b2b638f | 27 | import be.nikiroo.utils.Progress; |
08fe2e33 | 28 | import be.nikiroo.utils.StringUtils; |
5cf61f35 | 29 | import be.nikiroo.utils.Version; |
08fe2e33 NR |
30 | |
31 | /** | |
8ac3d099 NR |
32 | * Support class for <a href="http://e621.net/">e621.net</a> and |
33 | * <a href="http://e926.net/">e926.net</a>, a Furry website supporting comics, | |
08fe2e33 NR |
34 | * including some of MLP. |
35 | * <p> | |
36 | * <a href="http://e926.net/">e926.net</a> only shows the "clean" images and | |
37 | * comics, but it can be difficult to browse. | |
38 | * | |
39 | * @author niki | |
40 | */ | |
8ac3d099 | 41 | class E621 extends BasicSupport { |
08fe2e33 NR |
42 | @Override |
43 | protected boolean supports(URL url) { | |
44 | String host = url.getHost(); | |
45 | if (host.startsWith("www.")) { | |
46 | host = host.substring("www.".length()); | |
47 | } | |
48 | ||
5cf61f35 NR |
49 | return ("e621.net".equals(host) || "e926.net".equals(host)) |
50 | && (isPool(url) || isSearchOrSet(url)); | |
08fe2e33 NR |
51 | } |
52 | ||
53 | @Override | |
54 | protected boolean isHtml() { | |
55 | return true; | |
56 | } | |
57 | ||
8ac3d099 NR |
58 | @Override |
59 | protected MetaData getMeta() throws IOException { | |
60 | MetaData meta = new MetaData(); | |
b5e9855b | 61 | |
8ac3d099 NR |
62 | meta.setTitle(getTitle()); |
63 | meta.setAuthor(getAuthor()); | |
bff19b54 | 64 | meta.setDate(bsHelper.formatDate(getDate())); |
8ac3d099 NR |
65 | meta.setTags(getTags()); |
66 | meta.setSource(getType().getSourceName()); | |
67 | meta.setUrl(getSource().toString()); | |
68 | meta.setPublisher(getType().getSourceName()); | |
69 | meta.setUuid(getSource().toString()); | |
70 | meta.setLuid(""); | |
71 | meta.setLang("en"); | |
72 | meta.setSubject("Furry"); | |
73 | meta.setType(getType().toString()); | |
74 | meta.setImageDocument(true); | |
75 | meta.setCover(getCover()); | |
76 | meta.setFakeCover(true); | |
595dfa7a | 77 | |
8ac3d099 | 78 | return meta; |
595dfa7a NR |
79 | } |
80 | ||
8ac3d099 NR |
81 | @Override |
82 | protected String getDesc() throws IOException { | |
83 | if (isSearchOrSet(getSource())) { | |
b5e9855b | 84 | StringBuilder builder = new StringBuilder(); |
5cf61f35 NR |
85 | builder.append("A collection of images from ") |
86 | .append(getSource().getHost()).append("\n") // | |
87 | .append("\tTime of creation: " | |
88 | + StringUtils.fromTime(new Date().getTime())) | |
89 | .append("\n") // | |
8ac3d099 NR |
90 | .append("\tTags: ");// |
91 | for (String tag : getTags()) { | |
92 | builder.append("\t\t").append(tag); | |
b5e9855b NR |
93 | } |
94 | ||
95 | return builder.toString(); | |
96 | } | |
97 | ||
8ac3d099 NR |
98 | if (isPool(getSource())) { |
99 | Element el = getSourceNode().getElementById("description"); | |
100 | if (el != null) { | |
101 | return el.text(); | |
08fe2e33 NR |
102 | } |
103 | } | |
104 | ||
105 | return null; | |
106 | } | |
107 | ||
08fe2e33 | 108 | @Override |
5cf61f35 | 109 | protected List<Entry<String, URL>> getChapters(Progress pg) |
8ac3d099 | 110 | throws IOException { |
5cf61f35 NR |
111 | int i = 1; |
112 | String jsonUrl = getJsonUrl(); | |
113 | if (jsonUrl != null) { | |
114 | for (i = 1; true; i++) { | |
115 | if (i > 1) { | |
116 | try { | |
117 | // The API does not accept more than 2 request per sec, | |
118 | // and asks us to limit at one per sec when possible | |
119 | Thread.sleep(1000); | |
120 | } catch (InterruptedException e) { | |
121 | } | |
122 | } | |
b5e9855b | 123 | |
b5e9855b | 124 | try { |
5cf61f35 NR |
125 | JSONObject json = getJson(jsonUrl + "&page=" + i, false); |
126 | if (!json.has("posts")) | |
b5e9855b | 127 | break; |
5cf61f35 NR |
128 | JSONArray posts = json.getJSONArray("posts"); |
129 | if (posts.isEmpty()) | |
130 | break; | |
131 | } catch (Exception e) { | |
132 | e.printStackTrace(); | |
b5e9855b | 133 | } |
b5e9855b | 134 | } |
5cf61f35 NR |
135 | |
136 | // The last page was empty: | |
137 | i--; | |
138 | } | |
139 | ||
140 | // The pages and images are in reverse order on /posts/ | |
141 | List<Entry<String, URL>> chapters = new LinkedList<Entry<String, URL>>(); | |
142 | for (int page = i; page > 0; page--) { | |
143 | chapters.add(new AbstractMap.SimpleEntry<String, URL>( | |
144 | "Page " + Integer.toString(i - page + 1), | |
145 | new URL(jsonUrl + "&page=" + page))); | |
b5e9855b NR |
146 | } |
147 | ||
5cf61f35 | 148 | return chapters; |
b5e9855b NR |
149 | } |
150 | ||
8ac3d099 | 151 | @Override |
5cf61f35 NR |
152 | protected String getChapterContent(URL chapUrl, int number, Progress pg) |
153 | throws IOException { | |
8ac3d099 | 154 | StringBuilder builder = new StringBuilder(); |
5cf61f35 NR |
155 | |
156 | JSONObject json = getJson(chapUrl, false); | |
157 | JSONArray postsArr = json.getJSONArray("posts"); | |
158 | ||
159 | // The pages and images are in reverse order on /posts/ | |
160 | List<JSONObject> posts = new ArrayList<JSONObject>(postsArr.length()); | |
161 | for (int i = postsArr.length() - 1; i >= 0; i--) { | |
162 | Object o = postsArr.get(i); | |
163 | if (o instanceof JSONObject) | |
164 | posts.add((JSONObject) o); | |
75002fcc | 165 | } |
5cf61f35 NR |
166 | |
167 | for (JSONObject post : posts) { | |
168 | if (!post.has("file")) | |
169 | continue; | |
170 | JSONObject file = post.getJSONObject("file"); | |
171 | if (!file.has("url")) | |
172 | continue; | |
173 | ||
174 | try { | |
175 | String url = file.getString("url"); | |
176 | builder.append("["); | |
177 | builder.append(url); | |
178 | builder.append("]<br/>"); | |
179 | } catch (JSONException e) { | |
180 | // Can be NULL if filtered | |
181 | // When the value is NULL, we get an exception | |
182 | // but the "has" method still returns true | |
a3d0728c NR |
183 | Instance.getInstance().getTraceHandler() |
184 | .error("Cannot get image for chapter " + number + " of " | |
185 | + getSource()); | |
5cf61f35 | 186 | } |
8ac3d099 NR |
187 | } |
188 | ||
189 | return builder.toString(); | |
190 | } | |
191 | ||
192 | @Override | |
193 | protected URL getCanonicalUrl(URL source) { | |
8fbfa934 NR |
194 | // Convert search-pools into proper pools |
195 | if (source.getPath().equals("/posts") && source.getQuery() != null | |
196 | && source.getQuery().startsWith("tags=pool%3A")) { | |
197 | String poolNumber = source.getQuery() | |
198 | .substring("tags=pool%3A".length()); | |
199 | try { | |
200 | Integer.parseInt(poolNumber); | |
201 | String base = source.getProtocol() + "://" + source.getHost(); | |
202 | if (source.getPort() != -1) { | |
203 | base = base + ":" + source.getPort(); | |
204 | } | |
42cdf6f0 | 205 | source = new URL(base + "/pools/" + poolNumber); |
8fbfa934 | 206 | } catch (NumberFormatException e) { |
36c35b92 | 207 | // Not a simple pool, skip |
8fbfa934 NR |
208 | } catch (MalformedURLException e) { |
209 | // Cannot happen | |
210 | } | |
211 | } | |
5cf61f35 | 212 | |
8ac3d099 NR |
213 | if (isSetOriginalUrl(source)) { |
214 | try { | |
5cf61f35 NR |
215 | Document doc = DataUtil.load(Instance.getInstance().getCache() |
216 | .open(source, this, false), "UTF-8", source.toString()); | |
217 | for (Element shortname : doc | |
218 | .getElementsByClass("set-shortname")) { | |
8ac3d099 NR |
219 | for (Element el : shortname.getElementsByTag("a")) { |
220 | if (!el.attr("href").isEmpty()) | |
221 | return new URL(el.absUrl("href")); | |
08fe2e33 NR |
222 | } |
223 | } | |
8ac3d099 | 224 | } catch (IOException e) { |
d66deb8d | 225 | Instance.getInstance().getTraceHandler().error(e); |
08fe2e33 NR |
226 | } |
227 | } | |
228 | ||
c4b18c94 NR |
229 | if (isPool(source)) { |
230 | try { | |
5cf61f35 NR |
231 | return new URL( |
232 | source.toString().replace("/pool/show/", "/pools/")); | |
c4b18c94 NR |
233 | } catch (MalformedURLException e) { |
234 | } | |
235 | } | |
236 | ||
8ac3d099 NR |
237 | return super.getCanonicalUrl(source); |
238 | } | |
239 | ||
8ac3d099 NR |
240 | private String getTitle() { |
241 | String title = ""; | |
242 | ||
243 | Element el = getSourceNode().getElementsByTag("title").first(); | |
244 | if (el != null) { | |
245 | title = el.text().trim(); | |
08fe2e33 NR |
246 | } |
247 | ||
36c35b92 | 248 | for (String s : new String[] { "e621", "-", "e621", "Pool", "-" }) { |
8ac3d099 NR |
249 | if (title.startsWith(s)) { |
250 | title = title.substring(s.length()).trim(); | |
08fe2e33 | 251 | } |
8ac3d099 NR |
252 | if (title.endsWith(s)) { |
253 | title = title.substring(0, title.length() - s.length()).trim(); | |
254 | } | |
08fe2e33 NR |
255 | } |
256 | ||
8ac3d099 NR |
257 | if (isSearchOrSet(getSource())) { |
258 | title = title.isEmpty() ? "e621" : "[e621] " + title; | |
259 | } | |
5cf61f35 | 260 | |
8ac3d099 | 261 | return title; |
08fe2e33 | 262 | } |
b5e9855b | 263 | |
5cf61f35 NR |
264 | private String getAuthor() { |
265 | List<String> list = new ArrayList<String>(); | |
266 | String jsonUrl = getJsonUrl(); | |
267 | if (jsonUrl != null) { | |
268 | try { | |
269 | JSONObject json = getJson(jsonUrl, false); | |
270 | JSONArray posts = json.getJSONArray("posts"); | |
271 | for (Object obj : posts) { | |
272 | if (!(obj instanceof JSONObject)) | |
273 | continue; | |
274 | ||
275 | JSONObject post = (JSONObject) obj; | |
276 | if (!post.has("tags")) | |
277 | continue; | |
278 | ||
279 | JSONObject tags = post.getJSONObject("tags"); | |
280 | if (!tags.has("artist")) | |
281 | continue; | |
282 | ||
283 | JSONArray artists = tags.getJSONArray("artist"); | |
284 | for (Object artist : artists) { | |
285 | if (list.contains(artist.toString())) | |
286 | continue; | |
287 | ||
288 | list.add(artist.toString()); | |
8ac3d099 | 289 | } |
9948521d | 290 | } |
5cf61f35 NR |
291 | } catch (Exception e) { |
292 | e.printStackTrace(); | |
8ac3d099 NR |
293 | } |
294 | } | |
295 | ||
5cf61f35 NR |
296 | StringBuilder builder = new StringBuilder(); |
297 | for (String artist : list) { | |
298 | if (builder.length() > 0) { | |
299 | builder.append(", "); | |
300 | } | |
301 | builder.append(artist); | |
302 | } | |
8ac3d099 | 303 | |
5cf61f35 NR |
304 | return builder.toString(); |
305 | } | |
8ac3d099 | 306 | |
5cf61f35 NR |
307 | private String getDate() { |
308 | String jsonUrl = getJsonUrl(); | |
309 | if (jsonUrl != null) { | |
310 | try { | |
311 | JSONObject json = getJson(jsonUrl, false); | |
312 | JSONArray posts = json.getJSONArray("posts"); | |
313 | for (Object obj : posts) { | |
314 | if (!(obj instanceof JSONObject)) | |
315 | continue; | |
8d1a4fd2 | 316 | |
5cf61f35 NR |
317 | JSONObject post = (JSONObject) obj; |
318 | if (!post.has("created_at")) | |
319 | continue; | |
320 | ||
321 | return post.getString("created_at"); | |
8d1a4fd2 | 322 | } |
5cf61f35 NR |
323 | } catch (Exception e) { |
324 | e.printStackTrace(); | |
8d1a4fd2 | 325 | } |
8ac3d099 | 326 | } |
9948521d | 327 | |
5cf61f35 | 328 | return ""; |
8ac3d099 NR |
329 | } |
330 | ||
331 | // no tags for pools | |
332 | private List<String> getTags() { | |
333 | List<String> tags = new ArrayList<String>(); | |
334 | if (isSearchOrSet(getSource())) { | |
335 | String str = getTagsFromUrl(getSource()); | |
336 | for (String tag : str.split("\\+")) { | |
9b863b20 | 337 | try { |
8ac3d099 NR |
338 | tags.add(URLDecoder.decode(tag.trim(), "UTF-8").trim()); |
339 | } catch (UnsupportedEncodingException e) { | |
9b863b20 NR |
340 | } |
341 | } | |
342 | } | |
9948521d | 343 | |
8ac3d099 NR |
344 | return tags; |
345 | } | |
346 | ||
5cf61f35 NR |
347 | // returns "xxx+ddd+ggg" if "tags=xxx+ddd+ggg" was present in the query |
348 | private String getTagsFromUrl(URL url) { | |
349 | String tags = url == null ? "" : url.getQuery(); | |
350 | int pos = tags.indexOf("tags="); | |
351 | ||
352 | if (pos >= 0) { | |
353 | tags = tags.substring(pos).substring("tags=".length()); | |
354 | } else { | |
355 | return ""; | |
356 | } | |
357 | ||
358 | pos = tags.indexOf('&'); | |
359 | if (pos > 0) { | |
360 | tags = tags.substring(0, pos); | |
361 | } | |
362 | pos = tags.indexOf('/'); | |
363 | if (pos > 0) { | |
364 | tags = tags.substring(0, pos); | |
365 | } | |
366 | ||
367 | return tags; | |
368 | } | |
369 | ||
8ac3d099 NR |
370 | private Image getCover() throws IOException { |
371 | Image image = null; | |
372 | List<Entry<String, URL>> chapters = getChapters(null); | |
373 | if (!chapters.isEmpty()) { | |
12c180fc NR |
374 | URL chap1Url = chapters.get(0).getValue(); |
375 | String imgsChap1 = getChapterContent(chap1Url, 1, null); | |
376 | if (!imgsChap1.isEmpty()) { | |
377 | imgsChap1 = imgsChap1.split("]")[0].substring(1).trim(); | |
378 | image = bsImages.getImage(this, new URL(imgsChap1)); | |
379 | } | |
8ac3d099 NR |
380 | } |
381 | ||
382 | return image; | |
383 | } | |
384 | ||
5cf61f35 NR |
385 | // always /posts.json/ url |
386 | private String getJsonUrl() { | |
387 | String url = null; | |
388 | if (isSearchOrSet(getSource())) { | |
389 | url = getSource().toString().replace("/posts", "/posts.json"); | |
390 | } | |
391 | ||
392 | if (isPool(getSource())) { | |
393 | String poolNumber = getSource().getPath() | |
394 | .substring("/pools/".length()); | |
395 | url = "https://e621.net/posts.json" + "?tags=pool%3A" + poolNumber; | |
396 | } | |
397 | ||
398 | if (url != null) { | |
399 | // Note: one way to override the blacklist | |
400 | String login = Instance.getInstance().getConfig() | |
401 | .getString(Config.LOGIN_E621_LOGIN); | |
402 | String apk = Instance.getInstance().getConfig() | |
403 | .getString(Config.LOGIN_E621_APIKEY); | |
404 | ||
405 | if (login != null && !login.isEmpty() && apk != null | |
406 | && !apk.isEmpty()) { | |
407 | url = String.format("%s&login=%s&api_key=%s&_client=%s", url, | |
408 | login, apk, "fanfix-" + Version.getCurrentVersion()); | |
409 | } | |
410 | } | |
411 | ||
412 | return url; | |
413 | } | |
414 | ||
8ac3d099 NR |
415 | // note: will be removed at getCanonicalUrl() |
416 | private boolean isSetOriginalUrl(URL originalUrl) { | |
417 | return originalUrl.getPath().startsWith("/post_sets/"); | |
9b863b20 NR |
418 | } |
419 | ||
b5e9855b | 420 | private boolean isPool(URL url) { |
5cf61f35 NR |
421 | return url.getPath().startsWith("/pools/") |
422 | || url.getPath().startsWith("/pool/show/"); | |
b5e9855b NR |
423 | } |
424 | ||
8ac3d099 NR |
425 | // set will be renamed into search by canonical url |
426 | private boolean isSearchOrSet(URL url) { | |
427 | return | |
428 | // search: | |
429 | (url.getPath().equals("/posts") && url.getQuery().contains("tags=")) | |
430 | // or set: | |
431 | || isSetOriginalUrl(url); | |
b5e9855b | 432 | } |
08fe2e33 | 433 | } |