e621: woopsie
[fanfix.git] / supported / BasicSupport.java
CommitLineData
08fe2e33
NR
1package be.nikiroo.fanfix.supported;
2
08fe2e33
NR
3import java.io.IOException;
4import java.io.InputStream;
08fe2e33 5import java.net.URL;
08fe2e33 6import java.util.ArrayList;
793f1071 7import java.util.Date;
08fe2e33
NR
8import java.util.HashMap;
9import java.util.List;
10import java.util.Map;
11import java.util.Map.Entry;
0ffa4754
NR
12
13import org.jsoup.helper.DataUtil;
14import org.jsoup.nodes.Document;
15import org.jsoup.nodes.Element;
16import org.jsoup.nodes.Node;
08fe2e33
NR
17
18import be.nikiroo.fanfix.Instance;
08fe2e33
NR
19import be.nikiroo.fanfix.bundles.StringId;
20import be.nikiroo.fanfix.data.Chapter;
21import be.nikiroo.fanfix.data.MetaData;
9252c65e 22import be.nikiroo.fanfix.data.Story;
3b2b638f 23import be.nikiroo.utils.Progress;
08fe2e33
NR
24import be.nikiroo.utils.StringUtils;
25
26/**
27 * This class is the base class used by the other support classes. It can be
28 * used outside of this package, and have static method that you can use to get
29 * access to the correct support class.
30 * <p>
31 * It will be used with 'resources' (usually web pages or files).
32 *
33 * @author niki
34 */
35public abstract class BasicSupport {
0ffa4754
NR
36 private Document sourceNode;
37 private URL source;
08fe2e33 38 private SupportType type;
22848428 39 private URL currentReferer; // with only one 'r', as in 'HTTP'...
8d59ce07
NR
40
41 static protected BasicSupportHelper bsHelper = new BasicSupportHelper();
42 static protected BasicSupportImages bsImages = new BasicSupportImages();
43 static protected BasicSupportPara bsPara = new BasicSupportPara(new BasicSupportHelper(), new BasicSupportImages());
08fe2e33 44
08fe2e33
NR
45 /**
46 * Check if the given resource is supported by this {@link BasicSupport}.
47 *
48 * @param url
49 * the resource to check for
50 *
51 * @return TRUE if it is
52 */
53 protected abstract boolean supports(URL url);
54
55 /**
56 * Return TRUE if the support will return HTML encoded content values for
57 * the chapters content.
58 *
59 * @return TRUE for HTML
60 */
61 protected abstract boolean isHtml();
62
0efd25e3
NR
63 /**
64 * Return the {@link MetaData} of this story.
65 *
776ad3c6 66 * @return the associated {@link MetaData}, never NULL
0efd25e3
NR
67 *
68 * @throws IOException
69 * in case of I/O error
70 */
0ffa4754 71 protected abstract MetaData getMeta() throws IOException;
08fe2e33
NR
72
73 /**
74 * Return the story description.
75 *
08fe2e33
NR
76 * @return the description
77 *
78 * @throws IOException
79 * in case of I/O error
80 */
0ffa4754 81 protected abstract String getDesc() throws IOException;
08fe2e33 82
08fe2e33 83 /**
826e4569 84 * Return the list of chapters (name and resource).
0ffa4754
NR
85 * <p>
86 * Can be NULL if this {@link BasicSupport} do no use chapters.
08fe2e33 87 *
ed08c171
NR
88 * @param pg
89 * the optional progress reporter
08fe2e33 90 *
0ffa4754 91 * @return the chapters or NULL
08fe2e33
NR
92 *
93 * @throws IOException
94 * in case of I/O error
95 */
0ffa4754
NR
96 protected abstract List<Entry<String, URL>> getChapters(Progress pg)
97 throws IOException;
08fe2e33
NR
98
99 /**
100 * Return the content of the chapter (possibly HTML encoded, if
101 * {@link BasicSupport#isHtml()} is TRUE).
102 *
0ffa4754
NR
103 * @param chapUrl
104 * the chapter {@link URL}
08fe2e33
NR
105 * @param number
106 * the chapter number
ed08c171
NR
107 * @param pg
108 * the optional progress reporter
08fe2e33
NR
109 *
110 * @return the content
111 *
112 * @throws IOException
113 * in case of I/O error
114 */
0ffa4754
NR
115 protected abstract String getChapterContent(URL chapUrl, int number,
116 Progress pg) throws IOException;
6e06d2cc 117
08fe2e33
NR
118 /**
119 * Return the list of cookies (values included) that must be used to
120 * correctly fetch the resources.
121 * <p>
122 * You are expected to call the super method implementation if you override
123 * it.
124 *
125 * @return the cookies
126 */
315f14ae 127 public Map<String, String> getCookies() {
08fe2e33
NR
128 return new HashMap<String, String>();
129 }
130
315f14ae
NR
131 /**
132 * OAuth authorisation (aka, "bearer XXXXXXX").
133 *
134 * @return the OAuth string
135 */
136 public String getOAuth() {
137 return null;
138 }
139
a4143cd7
NR
140 /**
141 * Return the canonical form of the main {@link URL}.
142 *
143 * @param source
0ffa4754
NR
144 * the source {@link URL}, which can be NULL
145 *
146 * @return the canonical form of this {@link URL} or NULL if the source was
147 * NULL
148 */
149 protected URL getCanonicalUrl(URL source) {
150 return source;
151 }
152
153 /**
154 * The main {@link Node} for this {@link Story}.
155 *
156 * @return the node
157 */
158 protected Element getSourceNode() {
159 return sourceNode;
160 }
161
162 /**
163 * The main {@link URL} for this {@link Story}.
164 *
165 * @return the URL
166 */
167 protected URL getSource() {
168 return source;
169 }
170
171 /**
172 * The current referer {@link URL} (only one 'r', as in 'HTML'...), i.e.,
173 * the current {@link URL} we work on.
174 *
175 * @return the referer
176 */
177 public URL getCurrentReferer() {
178 return currentReferer;
179 }
180
181 /**
182 * The current referer {@link URL} (only one 'r', as in 'HTML'...), i.e.,
183 * the current {@link URL} we work on.
184 *
185 * @param currentReferer
186 * the new referer
187 */
188 protected void setCurrentReferer(URL currentReferer) {
189 this.currentReferer = currentReferer;
190 }
191
192 /**
193 * The support type.
194 *
195 * @return the type
196 */
197 public SupportType getType() {
198 return type;
199 }
200
201 /**
202 * The support type.
203 *
204 * @param type
205 * the new type
206 */
207 protected void setType(SupportType type) {
208 this.type = type;
209 }
210
211 /**
212 * Open an input link that will be used for the support.
213 * <p>
7445f856
NR
214 * Can return NULL, in which case you are supposed to work without a source
215 * node.
0ffa4754
NR
216 *
217 * @param source
a4143cd7
NR
218 * the source {@link URL}
219 *
0ffa4754
NR
220 * @return the {@link InputStream}
221 *
222 * @throws IOException
223 * in case of I/O error
224 */
225 protected Document loadDocument(URL source) throws IOException {
226 String url = getCanonicalUrl(source).toString();
d66deb8d 227 return DataUtil.load(Instance.getInstance().getCache().open(source, this, false), "UTF-8", url.toString());
0ffa4754
NR
228 }
229
230 /**
231 * Log into the support (can be a no-op depending upon the support).
a4143cd7
NR
232 *
233 * @throws IOException
234 * in case of I/O error
235 */
0ffa4754
NR
236 protected void login() throws IOException {
237 }
238
0ffa4754
NR
239 /**
240 * Now that we have processed the {@link Story}, close the resources if any.
241 */
242 protected void close() {
243 setCurrentReferer(null);
a4143cd7
NR
244 }
245
08fe2e33
NR
246 /**
247 * Process the given story resource into a partially filled {@link Story}
248 * object containing the name and metadata.
249 *
0efd25e3
NR
250 * @param getDesc
251 * retrieve the description of the story, or not
ed08c171
NR
252 * @param pg
253 * the optional progress reporter
08fe2e33 254 *
776ad3c6 255 * @return the {@link Story}, never NULL
08fe2e33
NR
256 *
257 * @throws IOException
258 * in case of I/O error
259 */
0ffa4754
NR
260 protected Story processMeta(boolean getDesc, Progress pg)
261 throws IOException {
ed08c171
NR
262 if (pg == null) {
263 pg = new Progress();
264 } else {
265 pg.setMinMax(0, 100);
266 }
267
0ffa4754 268 pg.setProgress(30);
ed08c171 269
0ffa4754
NR
270 Story story = new Story();
271 MetaData meta = getMeta();
272 if (meta.getCreationDate() == null || meta.getCreationDate().isEmpty()) {
273 meta.setCreationDate(StringUtils.fromTime(new Date().getTime()));
274 }
275 story.setMeta(meta);
920af1c7 276 pg.put("meta", meta);
ed08c171 277
0ffa4754 278 pg.setProgress(50);
08fe2e33 279
0ffa4754 280 if (meta.getCover() == null) {
8d59ce07 281 meta.setCover(bsHelper.getDefaultCover(meta.getSubject()));
0ffa4754 282 }
08fe2e33 283
0ffa4754 284 pg.setProgress(60);
a4143cd7 285
0ffa4754 286 if (getDesc) {
d66deb8d
NR
287 String descChapterName = Instance.getInstance().getTrans().getString(StringId.DESCRIPTION);
288 story.getMeta().setResume(bsPara.makeChapter(this, source, 0, descChapterName, //
289 getDesc(), isHtml(), null));
08fe2e33 290 }
0ffa4754 291
fdc55375 292 pg.done();
0ffa4754 293 return story;
08fe2e33
NR
294 }
295
9005532f 296 /**
826e4569
NR
297 * Process the given story resource into a fully filled {@link Story}
298 * object.
9005532f
NR
299 *
300 * @param pg
301 * the optional progress reporter
302 *
303 * @return the {@link Story}, never NULL
304 *
305 * @throws IOException
306 * in case of I/O error
307 */
6569afb4 308 // TODO: ADD final when BasicSupport_Deprecated is gone
9005532f
NR
309 public Story process(Progress pg) throws IOException {
310 setCurrentReferer(source);
311 login();
312 sourceNode = loadDocument(source);
313
314 try {
315 return doProcess(pg);
316 } finally {
317 close();
318 }
319 }
320
08fe2e33 321 /**
826e4569
NR
322 * Actual processing step, without the calls to other methods.
323 * <p>
324 * Will convert the story resource into a fully filled {@link Story} object.
08fe2e33 325 *
92fb0719
NR
326 * @param pg
327 * the optional progress reporter
08fe2e33 328 *
776ad3c6 329 * @return the {@link Story}, never NULL
08fe2e33
NR
330 *
331 * @throws IOException
332 * in case of I/O error
333 */
826e4569 334 protected Story doProcess(Progress pg) throws IOException {
92fb0719
NR
335 if (pg == null) {
336 pg = new Progress();
337 } else {
338 pg.setMinMax(0, 100);
339 }
3b039231
NR
340
341 pg.setName("Initialising");
92fb0719 342
92fb0719 343 pg.setProgress(1);
9005532f
NR
344 Progress pgMeta = new Progress();
345 pg.addProgress(pgMeta, 10);
346 Story story = processMeta(true, pgMeta);
68328e17 347 pgMeta.done(); // 10%
920af1c7 348 pg.put("meta", story.getMeta());
ed08c171 349
9005532f
NR
350 Progress pgGetChapters = new Progress();
351 pg.addProgress(pgGetChapters, 10);
352 story.setChapters(new ArrayList<Chapter>());
353 List<Entry<String, URL>> chapters = getChapters(pgGetChapters);
68328e17 354 pgGetChapters.done(); // 20%
9005532f
NR
355
356 if (chapters != null) {
357 Progress pgChaps = new Progress("Extracting chapters", 0,
358 chapters.size() * 300);
359 pg.addProgress(pgChaps, 80);
360
361 long words = 0;
362 int i = 1;
363 for (Entry<String, URL> chap : chapters) {
364 pgChaps.setName("Extracting chapter " + i);
365 URL chapUrl = chap.getValue();
366 String chapName = chap.getKey();
367 if (chapUrl != null) {
368 setCurrentReferer(chapUrl);
369 }
370
371 pgChaps.setProgress(i * 100);
372 Progress pgGetChapterContent = new Progress();
373 Progress pgMakeChapter = new Progress();
374 pgChaps.addProgress(pgGetChapterContent, 100);
375 pgChaps.addProgress(pgMakeChapter, 100);
376
377 String content = getChapterContent(chapUrl, i,
378 pgGetChapterContent);
68328e17 379 pgGetChapterContent.done();
8d59ce07 380 Chapter cc = bsPara.makeChapter(this, chapUrl, i,
9005532f 381 chapName, content, isHtml(), pgMakeChapter);
68328e17 382 pgMakeChapter.done();
ed08c171 383
9005532f
NR
384 words += cc.getWords();
385 story.getChapters().add(cc);
386 story.getMeta().setWords(words);
387
388 i++;
08fe2e33
NR
389 }
390
9005532f 391 pgChaps.setName("Extracting chapters");
fdc55375 392 pgChaps.done();
08fe2e33 393 }
9005532f 394
68328e17
NR
395 pg.done();
396
9005532f 397 return story;
08fe2e33
NR
398 }
399
99d71bd7
NR
400 /**
401 * Create a chapter from the given data.
402 *
403 * @param source
404 * the source URL for this content, which can be used to try and
405 * find images if images are present in the format [image-url]
406 * @param number
407 * the chapter number (0 = description)
408 * @param name
409 * the chapter name
410 * @param content
411 * the content of the chapter
412 * @return the {@link Chapter}
413 *
414 * @throws IOException
415 * in case of I/O error
416 */
417 public Chapter makeChapter(URL source, int number, String name,
418 String content) throws IOException {
8d59ce07 419 return bsPara.makeChapter(this, source, number, name,
99d71bd7
NR
420 content, isHtml(), null);
421 }
422
08fe2e33 423 /**
0ffa4754
NR
424 * Return a {@link BasicSupport} implementation supporting the given
425 * resource if possible.
08fe2e33 426 *
0ffa4754
NR
427 * @param url
428 * the story resource
08fe2e33 429 *
0ffa4754 430 * @return an implementation that supports it, or NULL
08fe2e33 431 */
0ffa4754
NR
432 public static BasicSupport getSupport(URL url) {
433 if (url == null) {
434 return null;
435 }
08fe2e33 436
0ffa4754
NR
437 // TEXT and INFO_TEXT always support files (not URLs though)
438 for (SupportType type : SupportType.values()) {
439 if (type != SupportType.TEXT && type != SupportType.INFO_TEXT) {
440 BasicSupport support = getSupport(type, url);
441 if (support != null && support.supports(url)) {
442 return support;
443 }
444 }
445 }
08fe2e33 446
0ffa4754
NR
447 for (SupportType type : new SupportType[] { SupportType.INFO_TEXT,
448 SupportType.TEXT }) {
449 BasicSupport support = getSupport(type, url);
450 if (support != null && support.supports(url)) {
451 return support;
452 }
453 }
454
455 return null;
08fe2e33
NR
456 }
457
458 /**
0ffa4754 459 * Return a {@link BasicSupport} implementation supporting the given type.
08fe2e33 460 *
0ffa4754 461 * @param type
99d71bd7 462 * the type, must not be NULL
0ffa4754
NR
463 * @param url
464 * the {@link URL} to support (can be NULL to get an
727108fe
NR
465 * "abstract support"; if not NULL, will be used as the source
466 * URL)
08fe2e33 467 *
0ffa4754 468 * @return an implementation that supports it, or NULL
08fe2e33 469 */
0ffa4754
NR
470 public static BasicSupport getSupport(SupportType type, URL url) {
471 BasicSupport support = null;
08fe2e33 472
08fe2e33
NR
473 switch (type) {
474 case EPUB:
0ffa4754
NR
475 support = new Epub();
476 break;
08fe2e33 477 case INFO_TEXT:
0ffa4754
NR
478 support = new InfoText();
479 break;
08fe2e33 480 case FIMFICTION:
315f14ae
NR
481 try {
482 // Can fail if no client key or NO in options
0ffa4754 483 support = new FimfictionApi();
315f14ae 484 } catch (IOException e) {
0ffa4754 485 support = new Fimfiction();
315f14ae 486 }
0ffa4754 487 break;
08fe2e33 488 case FANFICTION:
0ffa4754
NR
489 support = new Fanfiction();
490 break;
08fe2e33 491 case TEXT:
0ffa4754
NR
492 support = new Text();
493 break;
413bcc29
NR
494 case MANGAHUB:
495 support = new MangaHub();
0ffa4754 496 break;
08fe2e33 497 case E621:
0ffa4754
NR
498 support = new E621();
499 break;
a4143cd7 500 case YIFFSTAR:
0ffa4754
NR
501 support = new YiffStar();
502 break;
f0608ab1 503 case E_HENTAI:
0ffa4754
NR
504 support = new EHentai();
505 break;
af1f506f
NR
506 case MANGA_LEL:
507 support = new MangaLel();
508 break;
08fe2e33 509 case CBZ:
0ffa4754
NR
510 support = new Cbz();
511 break;
373da363 512 case HTML:
0ffa4754
NR
513 support = new Html();
514 break;
68686a37
NR
515 }
516
0ffa4754
NR
517 if (support != null) {
518 support.setType(type);
519 support.source = support.getCanonicalUrl(url);
315f14ae
NR
520 }
521
0ffa4754 522 return support;
315f14ae 523 }
08fe2e33 524}