Commit | Line | Data |
---|---|---|
08fe2e33 NR |
1 | package be.nikiroo.fanfix.supported; |
2 | ||
68e370a4 | 3 | import java.io.BufferedReader; |
08fe2e33 NR |
4 | import java.io.ByteArrayInputStream; |
5 | import java.io.File; | |
6 | import java.io.IOException; | |
7 | import java.io.InputStream; | |
68e370a4 | 8 | import java.io.InputStreamReader; |
08fe2e33 NR |
9 | import java.net.MalformedURLException; |
10 | import java.net.URL; | |
08fe2e33 | 11 | import java.util.ArrayList; |
793f1071 | 12 | import java.util.Date; |
08fe2e33 NR |
13 | import java.util.HashMap; |
14 | import java.util.List; | |
15 | import java.util.Map; | |
16 | import java.util.Map.Entry; | |
17 | import java.util.Scanner; | |
18 | ||
19 | import be.nikiroo.fanfix.Instance; | |
20 | import be.nikiroo.fanfix.bundles.Config; | |
21 | import be.nikiroo.fanfix.bundles.StringId; | |
22 | import be.nikiroo.fanfix.data.Chapter; | |
23 | import be.nikiroo.fanfix.data.MetaData; | |
24 | import be.nikiroo.fanfix.data.Paragraph; | |
08fe2e33 | 25 | import be.nikiroo.fanfix.data.Paragraph.ParagraphType; |
9252c65e | 26 | import be.nikiroo.fanfix.data.Story; |
16a81ef7 | 27 | import be.nikiroo.utils.Image; |
3b2b638f | 28 | import be.nikiroo.utils.Progress; |
08fe2e33 NR |
29 | import be.nikiroo.utils.StringUtils; |
30 | ||
31 | /** | |
32 | * This class is the base class used by the other support classes. It can be | |
33 | * used outside of this package, and have static method that you can use to get | |
34 | * access to the correct support class. | |
35 | * <p> | |
36 | * It will be used with 'resources' (usually web pages or files). | |
37 | * | |
38 | * @author niki | |
39 | */ | |
40 | public abstract class BasicSupport { | |
41 | /** | |
42 | * The supported input types for which we can get a {@link BasicSupport} | |
43 | * object. | |
44 | * | |
45 | * @author niki | |
46 | */ | |
47 | public enum SupportType { | |
48 | /** EPUB files created with this program */ | |
49 | EPUB, | |
50 | /** Pure text file with some rules */ | |
51 | TEXT, | |
52 | /** TEXT but with associated .info file */ | |
53 | INFO_TEXT, | |
54 | /** My Little Pony fanfictions */ | |
55 | FIMFICTION, | |
56 | /** Fanfictions from a lot of different universes */ | |
57 | FANFICTION, | |
58 | /** Website with lots of Mangas */ | |
59 | MANGAFOX, | |
60 | /** Furry website with comics support */ | |
61 | E621, | |
a4143cd7 NR |
62 | /** Furry website with stories */ |
63 | YIFFSTAR, | |
f0608ab1 NR |
64 | /** Comics and images groups, mostly but not only NSFW */ |
65 | E_HENTAI, | |
08fe2e33 | 66 | /** CBZ files */ |
373da363 NR |
67 | CBZ, |
68 | /** HTML files */ | |
69 | HTML; | |
08fe2e33 NR |
70 | |
71 | /** | |
72 | * A description of this support type (more information than the | |
73 | * {@link BasicSupport#getSourceName()}). | |
74 | * | |
75 | * @return the description | |
76 | */ | |
77 | public String getDesc() { | |
78 | String desc = Instance.getTrans().getStringX(StringId.INPUT_DESC, | |
79 | this.name()); | |
80 | ||
81 | if (desc == null) { | |
82 | desc = Instance.getTrans().getString(StringId.INPUT_DESC, this); | |
83 | } | |
84 | ||
85 | return desc; | |
86 | } | |
87 | ||
88 | /** | |
89 | * The name of this support type (a short version). | |
90 | * | |
91 | * @return the name | |
92 | */ | |
93 | public String getSourceName() { | |
94 | BasicSupport support = BasicSupport.getSupport(this); | |
95 | if (support != null) { | |
96 | return support.getSourceName(); | |
97 | } | |
98 | ||
99 | return null; | |
100 | } | |
101 | ||
102 | @Override | |
103 | public String toString() { | |
104 | return super.toString().toLowerCase(); | |
105 | } | |
106 | ||
107 | /** | |
0efd25e3 NR |
108 | * Call {@link SupportType#valueOf(String)} after conversion to upper |
109 | * case. | |
08fe2e33 NR |
110 | * |
111 | * @param typeName | |
112 | * the possible type name | |
113 | * | |
114 | * @return NULL or the type | |
115 | */ | |
116 | public static SupportType valueOfUC(String typeName) { | |
117 | return SupportType.valueOf(typeName == null ? null : typeName | |
118 | .toUpperCase()); | |
119 | } | |
120 | ||
121 | /** | |
0efd25e3 NR |
122 | * Call {@link SupportType#valueOf(String)} after conversion to upper |
123 | * case but return NULL for NULL instead of raising exception. | |
08fe2e33 NR |
124 | * |
125 | * @param typeName | |
126 | * the possible type name | |
127 | * | |
128 | * @return NULL or the type | |
129 | */ | |
130 | public static SupportType valueOfNullOkUC(String typeName) { | |
131 | if (typeName == null) { | |
132 | return null; | |
133 | } | |
134 | ||
135 | return SupportType.valueOfUC(typeName); | |
136 | } | |
137 | ||
138 | /** | |
0efd25e3 NR |
139 | * Call {@link SupportType#valueOf(String)} after conversion to upper |
140 | * case but return NULL in case of error instead of raising an | |
141 | * exception. | |
08fe2e33 NR |
142 | * |
143 | * @param typeName | |
144 | * the possible type name | |
145 | * | |
146 | * @return NULL or the type | |
147 | */ | |
148 | public static SupportType valueOfAllOkUC(String typeName) { | |
149 | try { | |
150 | return SupportType.valueOfUC(typeName); | |
151 | } catch (Exception e) { | |
152 | return null; | |
153 | } | |
154 | } | |
155 | } | |
156 | ||
08fe2e33 NR |
157 | private InputStream in; |
158 | private SupportType type; | |
22848428 | 159 | private URL currentReferer; // with only one 'r', as in 'HTTP'... |
08fe2e33 NR |
160 | |
161 | // quote chars | |
e8eeea0a | 162 | private char openQuote = Instance.getTrans().getCharacter( |
08fe2e33 | 163 | StringId.OPEN_SINGLE_QUOTE); |
e8eeea0a | 164 | private char closeQuote = Instance.getTrans().getCharacter( |
08fe2e33 | 165 | StringId.CLOSE_SINGLE_QUOTE); |
e8eeea0a | 166 | private char openDoubleQuote = Instance.getTrans().getCharacter( |
08fe2e33 | 167 | StringId.OPEN_DOUBLE_QUOTE); |
e8eeea0a | 168 | private char closeDoubleQuote = Instance.getTrans().getCharacter( |
08fe2e33 NR |
169 | StringId.CLOSE_DOUBLE_QUOTE); |
170 | ||
171 | /** | |
172 | * The name of this support class. | |
173 | * | |
174 | * @return the name | |
175 | */ | |
176 | protected abstract String getSourceName(); | |
177 | ||
178 | /** | |
179 | * Check if the given resource is supported by this {@link BasicSupport}. | |
180 | * | |
181 | * @param url | |
182 | * the resource to check for | |
183 | * | |
184 | * @return TRUE if it is | |
185 | */ | |
186 | protected abstract boolean supports(URL url); | |
187 | ||
188 | /** | |
189 | * Return TRUE if the support will return HTML encoded content values for | |
190 | * the chapters content. | |
191 | * | |
192 | * @return TRUE for HTML | |
193 | */ | |
194 | protected abstract boolean isHtml(); | |
195 | ||
0efd25e3 NR |
196 | /** |
197 | * Return the {@link MetaData} of this story. | |
198 | * | |
199 | * @param source | |
200 | * the source of the story | |
201 | * @param in | |
202 | * the input (the main resource) | |
203 | * | |
776ad3c6 | 204 | * @return the associated {@link MetaData}, never NULL |
0efd25e3 NR |
205 | * |
206 | * @throws IOException | |
207 | * in case of I/O error | |
208 | */ | |
68686a37 | 209 | protected abstract MetaData getMeta(URL source, InputStream in) |
08fe2e33 NR |
210 | throws IOException; |
211 | ||
212 | /** | |
213 | * Return the story description. | |
214 | * | |
215 | * @param source | |
216 | * the source of the story | |
217 | * @param in | |
218 | * the input (the main resource) | |
219 | * | |
220 | * @return the description | |
221 | * | |
222 | * @throws IOException | |
223 | * in case of I/O error | |
224 | */ | |
225 | protected abstract String getDesc(URL source, InputStream in) | |
226 | throws IOException; | |
227 | ||
08fe2e33 NR |
228 | /** |
229 | * Return the list of chapters (name and resource). | |
230 | * | |
231 | * @param source | |
232 | * the source of the story | |
233 | * @param in | |
234 | * the input (the main resource) | |
ed08c171 NR |
235 | * @param pg |
236 | * the optional progress reporter | |
08fe2e33 NR |
237 | * |
238 | * @return the chapters | |
239 | * | |
240 | * @throws IOException | |
241 | * in case of I/O error | |
242 | */ | |
243 | protected abstract List<Entry<String, URL>> getChapters(URL source, | |
ed08c171 | 244 | InputStream in, Progress pg) throws IOException; |
08fe2e33 NR |
245 | |
246 | /** | |
247 | * Return the content of the chapter (possibly HTML encoded, if | |
248 | * {@link BasicSupport#isHtml()} is TRUE). | |
249 | * | |
250 | * @param source | |
251 | * the source of the story | |
252 | * @param in | |
253 | * the input (the main resource) | |
254 | * @param number | |
255 | * the chapter number | |
ed08c171 NR |
256 | * @param pg |
257 | * the optional progress reporter | |
08fe2e33 NR |
258 | * |
259 | * @return the content | |
260 | * | |
261 | * @throws IOException | |
262 | * in case of I/O error | |
263 | */ | |
264 | protected abstract String getChapterContent(URL source, InputStream in, | |
ed08c171 | 265 | int number, Progress pg) throws IOException; |
08fe2e33 | 266 | |
6e06d2cc NR |
267 | /** |
268 | * Log into the support (can be a no-op depending upon the support). | |
269 | * | |
270 | * @throws IOException | |
271 | * in case of I/O error | |
272 | */ | |
315f14ae | 273 | @SuppressWarnings("unused") |
6e06d2cc | 274 | public void login() throws IOException { |
6e06d2cc NR |
275 | } |
276 | ||
08fe2e33 NR |
277 | /** |
278 | * Return the list of cookies (values included) that must be used to | |
279 | * correctly fetch the resources. | |
280 | * <p> | |
281 | * You are expected to call the super method implementation if you override | |
282 | * it. | |
283 | * | |
284 | * @return the cookies | |
285 | */ | |
315f14ae | 286 | public Map<String, String> getCookies() { |
08fe2e33 NR |
287 | return new HashMap<String, String>(); |
288 | } | |
289 | ||
315f14ae NR |
290 | /** |
291 | * OAuth authorisation (aka, "bearer XXXXXXX"). | |
292 | * | |
293 | * @return the OAuth string | |
294 | */ | |
295 | public String getOAuth() { | |
296 | return null; | |
297 | } | |
298 | ||
a4143cd7 NR |
299 | /** |
300 | * Return the canonical form of the main {@link URL}. | |
301 | * | |
302 | * @param source | |
303 | * the source {@link URL} | |
304 | * | |
305 | * @return the canonical form of this {@link URL} | |
306 | * | |
307 | * @throws IOException | |
308 | * in case of I/O error | |
309 | */ | |
315f14ae | 310 | @SuppressWarnings("unused") |
a4143cd7 NR |
311 | public URL getCanonicalUrl(URL source) throws IOException { |
312 | return source; | |
313 | } | |
314 | ||
08fe2e33 NR |
315 | /** |
316 | * Process the given story resource into a partially filled {@link Story} | |
317 | * object containing the name and metadata, except for the description. | |
318 | * | |
319 | * @param url | |
320 | * the story resource | |
321 | * | |
322 | * @return the {@link Story} | |
323 | * | |
324 | * @throws IOException | |
325 | * in case of I/O error | |
326 | */ | |
327 | public Story processMeta(URL url) throws IOException { | |
ed08c171 | 328 | return processMeta(url, true, false, null); |
08fe2e33 NR |
329 | } |
330 | ||
331 | /** | |
332 | * Process the given story resource into a partially filled {@link Story} | |
333 | * object containing the name and metadata. | |
334 | * | |
335 | * @param url | |
336 | * the story resource | |
08fe2e33 NR |
337 | * @param close |
338 | * close "this" and "in" when done | |
0efd25e3 NR |
339 | * @param getDesc |
340 | * retrieve the description of the story, or not | |
ed08c171 NR |
341 | * @param pg |
342 | * the optional progress reporter | |
08fe2e33 | 343 | * |
776ad3c6 | 344 | * @return the {@link Story}, never NULL |
08fe2e33 NR |
345 | * |
346 | * @throws IOException | |
347 | * in case of I/O error | |
348 | */ | |
ed08c171 NR |
349 | protected Story processMeta(URL url, boolean close, boolean getDesc, |
350 | Progress pg) throws IOException { | |
351 | if (pg == null) { | |
352 | pg = new Progress(); | |
353 | } else { | |
354 | pg.setMinMax(0, 100); | |
355 | } | |
356 | ||
6e06d2cc | 357 | login(); |
ed08c171 | 358 | pg.setProgress(10); |
6e06d2cc | 359 | |
a4143cd7 NR |
360 | url = getCanonicalUrl(url); |
361 | ||
362 | setCurrentReferer(url); | |
363 | ||
315f14ae | 364 | in = openInput(url); // NULL allowed here |
08fe2e33 | 365 | try { |
68686a37 | 366 | preprocess(url, getInput()); |
ed08c171 | 367 | pg.setProgress(30); |
08fe2e33 NR |
368 | |
369 | Story story = new Story(); | |
68686a37 | 370 | MetaData meta = getMeta(url, getInput()); |
793f1071 NR |
371 | if (meta.getCreationDate() == null |
372 | || meta.getCreationDate().isEmpty()) { | |
373 | meta.setCreationDate(StringUtils.fromTime(new Date().getTime())); | |
374 | } | |
68686a37 NR |
375 | story.setMeta(meta); |
376 | ||
ed08c171 NR |
377 | pg.setProgress(50); |
378 | ||
211f7ddb | 379 | if (meta.getCover() == null) { |
68686a37 NR |
380 | meta.setCover(getDefaultCover(meta.getSubject())); |
381 | } | |
08fe2e33 | 382 | |
ed08c171 NR |
383 | pg.setProgress(60); |
384 | ||
08fe2e33 NR |
385 | if (getDesc) { |
386 | String descChapterName = Instance.getTrans().getString( | |
387 | StringId.DESCRIPTION); | |
388 | story.getMeta().setResume( | |
389 | makeChapter(url, 0, descChapterName, | |
ed08c171 | 390 | getDesc(url, getInput()), null)); |
08fe2e33 NR |
391 | } |
392 | ||
ed08c171 | 393 | pg.setProgress(100); |
08fe2e33 NR |
394 | return story; |
395 | } finally { | |
396 | if (close) { | |
397 | try { | |
398 | close(); | |
399 | } catch (IOException e) { | |
62c63b07 | 400 | Instance.getTraceHandler().error(e); |
08fe2e33 NR |
401 | } |
402 | ||
403 | if (in != null) { | |
404 | in.close(); | |
405 | } | |
406 | } | |
a4143cd7 NR |
407 | |
408 | setCurrentReferer(null); | |
08fe2e33 NR |
409 | } |
410 | } | |
411 | ||
412 | /** | |
413 | * Process the given story resource into a fully filled {@link Story} | |
414 | * object. | |
415 | * | |
416 | * @param url | |
417 | * the story resource | |
92fb0719 NR |
418 | * @param pg |
419 | * the optional progress reporter | |
08fe2e33 | 420 | * |
776ad3c6 | 421 | * @return the {@link Story}, never NULL |
08fe2e33 NR |
422 | * |
423 | * @throws IOException | |
424 | * in case of I/O error | |
425 | */ | |
92fb0719 NR |
426 | public Story process(URL url, Progress pg) throws IOException { |
427 | if (pg == null) { | |
428 | pg = new Progress(); | |
429 | } else { | |
430 | pg.setMinMax(0, 100); | |
431 | } | |
432 | ||
a4143cd7 | 433 | url = getCanonicalUrl(url); |
92fb0719 | 434 | pg.setProgress(1); |
08fe2e33 | 435 | try { |
ed08c171 NR |
436 | Progress pgMeta = new Progress(); |
437 | pg.addProgress(pgMeta, 10); | |
438 | Story story = processMeta(url, false, true, pgMeta); | |
439 | if (!pgMeta.isDone()) { | |
440 | pgMeta.setProgress(pgMeta.getMax()); // 10% | |
441 | } | |
442 | ||
754a5bc2 NR |
443 | pg.setName("Retrieving " + story.getMeta().getTitle()); |
444 | ||
a4143cd7 NR |
445 | setCurrentReferer(url); |
446 | ||
ed08c171 NR |
447 | Progress pgGetChapters = new Progress(); |
448 | pg.addProgress(pgGetChapters, 10); | |
08fe2e33 | 449 | story.setChapters(new ArrayList<Chapter>()); |
ed08c171 NR |
450 | List<Entry<String, URL>> chapters = getChapters(url, getInput(), |
451 | pgGetChapters); | |
452 | if (!pgGetChapters.isDone()) { | |
453 | pgGetChapters.setProgress(pgGetChapters.getMax()); // 20% | |
454 | } | |
08fe2e33 | 455 | |
08fe2e33 | 456 | if (chapters != null) { |
ed08c171 NR |
457 | Progress pgChaps = new Progress("Extracting chapters", 0, |
458 | chapters.size() * 300); | |
92fb0719 NR |
459 | pg.addProgress(pgChaps, 80); |
460 | ||
793f1071 | 461 | long words = 0; |
ed08c171 | 462 | int i = 1; |
08fe2e33 | 463 | for (Entry<String, URL> chap : chapters) { |
ed08c171 | 464 | pgChaps.setName("Extracting chapter " + i); |
315f14ae NR |
465 | InputStream chapIn = null; |
466 | if (chap.getValue() != null) { | |
467 | setCurrentReferer(chap.getValue()); | |
468 | chapIn = Instance.getCache().open(chap.getValue(), | |
469 | this, true); | |
470 | } | |
ed08c171 | 471 | pgChaps.setProgress(i * 100); |
08fe2e33 | 472 | try { |
ed08c171 NR |
473 | Progress pgGetChapterContent = new Progress(); |
474 | Progress pgMakeChapter = new Progress(); | |
475 | pgChaps.addProgress(pgGetChapterContent, 100); | |
476 | pgChaps.addProgress(pgMakeChapter, 100); | |
477 | ||
478 | String content = getChapterContent(url, chapIn, i, | |
479 | pgGetChapterContent); | |
480 | if (!pgGetChapterContent.isDone()) { | |
481 | pgGetChapterContent.setProgress(pgGetChapterContent | |
482 | .getMax()); | |
483 | } | |
484 | ||
793f1071 | 485 | Chapter cc = makeChapter(url, i, chap.getKey(), |
ed08c171 NR |
486 | content, pgMakeChapter); |
487 | if (!pgMakeChapter.isDone()) { | |
488 | pgMakeChapter.setProgress(pgMakeChapter.getMax()); | |
489 | } | |
490 | ||
793f1071 NR |
491 | words += cc.getWords(); |
492 | story.getChapters().add(cc); | |
776ad3c6 | 493 | story.getMeta().setWords(words); |
08fe2e33 | 494 | } finally { |
315f14ae NR |
495 | if (chapIn != null) { |
496 | chapIn.close(); | |
497 | } | |
08fe2e33 | 498 | } |
a6395bef | 499 | |
ed08c171 | 500 | i++; |
08fe2e33 | 501 | } |
ed08c171 NR |
502 | |
503 | pgChaps.setName("Extracting chapters"); | |
92fb0719 | 504 | } else { |
ed08c171 | 505 | pg.setProgress(80); |
08fe2e33 NR |
506 | } |
507 | ||
508 | return story; | |
509 | ||
510 | } finally { | |
511 | try { | |
512 | close(); | |
513 | } catch (IOException e) { | |
62c63b07 | 514 | Instance.getTraceHandler().error(e); |
08fe2e33 NR |
515 | } |
516 | ||
517 | if (in != null) { | |
518 | in.close(); | |
519 | } | |
520 | ||
a4143cd7 | 521 | setCurrentReferer(null); |
08fe2e33 NR |
522 | } |
523 | } | |
524 | ||
525 | /** | |
a4143cd7 | 526 | * The support type. |
08fe2e33 NR |
527 | * |
528 | * @return the type | |
529 | */ | |
530 | public SupportType getType() { | |
531 | return type; | |
532 | } | |
533 | ||
534 | /** | |
535 | * The current referer {@link URL} (only one 'r', as in 'HTML'...), i.e., | |
536 | * the current {@link URL} we work on. | |
537 | * | |
538 | * @return the referer | |
539 | */ | |
540 | public URL getCurrentReferer() { | |
541 | return currentReferer; | |
542 | } | |
543 | ||
544 | /** | |
545 | * The current referer {@link URL} (only one 'r', as in 'HTML'...), i.e., | |
546 | * the current {@link URL} we work on. | |
547 | * | |
548 | * @param currentReferer | |
549 | * the new referer | |
550 | */ | |
551 | protected void setCurrentReferer(URL currentReferer) { | |
552 | this.currentReferer = currentReferer; | |
553 | } | |
554 | ||
555 | /** | |
556 | * The support type. | |
557 | * | |
558 | * @param type | |
559 | * the new type | |
560 | * | |
561 | * @return this | |
562 | */ | |
563 | protected BasicSupport setType(SupportType type) { | |
564 | this.type = type; | |
565 | return this; | |
566 | } | |
567 | ||
568 | /** | |
68686a37 | 569 | * Prepare the support if needed before processing. |
08fe2e33 NR |
570 | * |
571 | * @param source | |
572 | * the source of the story | |
573 | * @param in | |
574 | * the input (the main resource) | |
575 | * | |
08fe2e33 NR |
576 | * @throws IOException |
577 | * on I/O error | |
578 | */ | |
211f7ddb | 579 | @SuppressWarnings("unused") |
68686a37 | 580 | protected void preprocess(URL source, InputStream in) throws IOException { |
08fe2e33 NR |
581 | } |
582 | ||
583 | /** | |
584 | * Now that we have processed the {@link Story}, close the resources if any. | |
585 | * | |
586 | * @throws IOException | |
587 | * on I/O error | |
588 | */ | |
315f14ae | 589 | @SuppressWarnings("unused") |
08fe2e33 NR |
590 | protected void close() throws IOException { |
591 | } | |
592 | ||
593 | /** | |
594 | * Create a {@link Chapter} object from the given information, formatting | |
595 | * the content as it should be. | |
596 | * | |
0efd25e3 NR |
597 | * @param source |
598 | * the source of the story | |
08fe2e33 NR |
599 | * @param number |
600 | * the chapter number | |
601 | * @param name | |
602 | * the chapter name | |
603 | * @param content | |
604 | * the chapter content | |
ed08c171 NR |
605 | * @param pg |
606 | * the optional progress reporter | |
08fe2e33 NR |
607 | * |
608 | * @return the {@link Chapter} | |
609 | * | |
610 | * @throws IOException | |
611 | * in case of I/O error | |
612 | */ | |
613 | protected Chapter makeChapter(URL source, int number, String name, | |
ed08c171 | 614 | String content, Progress pg) throws IOException { |
08fe2e33 | 615 | // Chapter name: process it correctly, then remove the possible |
f60df2f1 NR |
616 | // redundant "Chapter x: " in front of it, or "-" (as in |
617 | // "Chapter 5: - Fun!" after the ": " was automatically added) | |
08fe2e33 NR |
618 | String chapterName = processPara(name).getContent().trim(); |
619 | for (String lang : Instance.getConfig().getString(Config.CHAPTER) | |
620 | .split(",")) { | |
621 | String chapterWord = Instance.getConfig().getStringX( | |
622 | Config.CHAPTER, lang); | |
623 | if (chapterName.startsWith(chapterWord)) { | |
624 | chapterName = chapterName.substring(chapterWord.length()) | |
625 | .trim(); | |
626 | break; | |
627 | } | |
628 | } | |
629 | ||
630 | if (chapterName.startsWith(Integer.toString(number))) { | |
631 | chapterName = chapterName.substring( | |
632 | Integer.toString(number).length()).trim(); | |
633 | } | |
634 | ||
f60df2f1 | 635 | while (chapterName.startsWith(":") || chapterName.startsWith("-")) { |
08fe2e33 NR |
636 | chapterName = chapterName.substring(1).trim(); |
637 | } | |
638 | // | |
639 | ||
640 | Chapter chap = new Chapter(number, chapterName); | |
641 | ||
68e370a4 | 642 | if (content != null) { |
ed08c171 | 643 | List<Paragraph> paras = makeParagraphs(source, content, pg); |
793f1071 NR |
644 | long words = 0; |
645 | for (Paragraph para : paras) { | |
646 | words += para.getWords(); | |
647 | } | |
648 | chap.setParagraphs(paras); | |
649 | chap.setWords(words); | |
08fe2e33 NR |
650 | } |
651 | ||
68e370a4 NR |
652 | return chap; |
653 | ||
654 | } | |
655 | ||
656 | /** | |
657 | * Convert the given content into {@link Paragraph}s. | |
658 | * | |
659 | * @param source | |
660 | * the source URL of the story | |
661 | * @param content | |
662 | * the textual content | |
ed08c171 NR |
663 | * @param pg |
664 | * the optional progress reporter | |
68e370a4 NR |
665 | * |
666 | * @return the {@link Paragraph}s | |
667 | * | |
668 | * @throws IOException | |
669 | * in case of I/O error | |
670 | */ | |
ed08c171 NR |
671 | protected List<Paragraph> makeParagraphs(URL source, String content, |
672 | Progress pg) throws IOException { | |
673 | if (pg == null) { | |
674 | pg = new Progress(); | |
675 | } | |
676 | ||
08fe2e33 NR |
677 | if (isHtml()) { |
678 | // Special <HR> processing: | |
679 | content = content.replaceAll("(<hr [^>]*>)|(<hr/>)|(<hr>)", | |
d5a7153c | 680 | "<br/>* * *<br/>"); |
08fe2e33 NR |
681 | } |
682 | ||
68e370a4 | 683 | List<Paragraph> paras = new ArrayList<Paragraph>(); |
08fe2e33 | 684 | |
d5a7153c NR |
685 | if (content != null && !content.trim().isEmpty()) { |
686 | if (isHtml()) { | |
ed08c171 NR |
687 | String[] tab = content.split("(<p>|</p>|<br>|<br/>)"); |
688 | pg.setMinMax(0, tab.length); | |
689 | int i = 1; | |
690 | for (String line : tab) { | |
691 | if (line.startsWith("[") && line.endsWith("]")) { | |
692 | pg.setName("Extracting image " + i); | |
693 | } | |
d5a7153c | 694 | paras.add(makeParagraph(source, line.trim())); |
ed08c171 | 695 | pg.setProgress(i++); |
d5a7153c | 696 | } |
ed08c171 | 697 | pg.setName(null); |
d5a7153c | 698 | } else { |
ed08c171 | 699 | List<String> lines = new ArrayList<String>(); |
d5a7153c NR |
700 | BufferedReader buff = null; |
701 | try { | |
702 | buff = new BufferedReader( | |
703 | new InputStreamReader(new ByteArrayInputStream( | |
704 | content.getBytes("UTF-8")), "UTF-8")); | |
705 | for (String line = buff.readLine(); line != null; line = buff | |
706 | .readLine()) { | |
ed08c171 | 707 | lines.add(line.trim()); |
68e370a4 | 708 | } |
d5a7153c NR |
709 | } finally { |
710 | if (buff != null) { | |
711 | buff.close(); | |
68e370a4 | 712 | } |
08fe2e33 | 713 | } |
ed08c171 NR |
714 | |
715 | pg.setMinMax(0, lines.size()); | |
716 | int i = 0; | |
717 | for (String line : lines) { | |
718 | if (line.startsWith("[") && line.endsWith("]")) { | |
719 | pg.setName("Extracting image " + i); | |
720 | } | |
721 | paras.add(makeParagraph(source, line)); | |
722 | pg.setProgress(i++); | |
723 | } | |
724 | pg.setName(null); | |
08fe2e33 NR |
725 | } |
726 | ||
d5a7153c NR |
727 | // Check quotes for "bad" format |
728 | List<Paragraph> newParas = new ArrayList<Paragraph>(); | |
729 | for (Paragraph para : paras) { | |
730 | newParas.addAll(requotify(para)); | |
731 | } | |
732 | paras = newParas; | |
08fe2e33 | 733 | |
d5a7153c NR |
734 | // Remove double blanks/brks |
735 | fixBlanksBreaks(paras); | |
736 | } | |
08fe2e33 | 737 | |
68e370a4 NR |
738 | return paras; |
739 | } | |
08fe2e33 | 740 | |
d5a7153c NR |
741 | /** |
742 | * Convert the given line into a single {@link Paragraph}. | |
743 | * | |
744 | * @param source | |
745 | * the source URL of the story | |
746 | * @param line | |
747 | * the textual content of the paragraph | |
748 | * | |
749 | * @return the {@link Paragraph} | |
750 | */ | |
751 | private Paragraph makeParagraph(URL source, String line) { | |
16a81ef7 | 752 | Image image = null; |
d5a7153c | 753 | if (line.startsWith("[") && line.endsWith("]")) { |
2a25f781 NR |
754 | image = getImage(this, source, line.substring(1, line.length() - 1) |
755 | .trim()); | |
d5a7153c NR |
756 | } |
757 | ||
758 | if (image != null) { | |
759 | return new Paragraph(image); | |
d5a7153c | 760 | } |
211f7ddb NR |
761 | |
762 | return processPara(line); | |
d5a7153c NR |
763 | } |
764 | ||
68e370a4 NR |
765 | /** |
766 | * Fix the {@link ParagraphType#BLANK}s and {@link ParagraphType#BREAK}s of | |
767 | * those {@link Paragraph}s. | |
768 | * <p> | |
769 | * The resulting list will not contain a starting or trailing blank/break | |
770 | * nor 2 blanks or breaks following each other. | |
771 | * | |
772 | * @param paras | |
773 | * the list of {@link Paragraph}s to fix | |
774 | */ | |
775 | protected void fixBlanksBreaks(List<Paragraph> paras) { | |
776 | boolean space = false; | |
777 | boolean brk = true; | |
778 | for (int i = 0; i < paras.size(); i++) { | |
779 | Paragraph para = paras.get(i); | |
780 | boolean thisSpace = para.getType() == ParagraphType.BLANK; | |
781 | boolean thisBrk = para.getType() == ParagraphType.BREAK; | |
782 | ||
783 | if (i > 0 && space && thisBrk) { | |
784 | paras.remove(i - 1); | |
785 | i--; | |
786 | } else if ((space || brk) && (thisSpace || thisBrk)) { | |
787 | paras.remove(i); | |
788 | i--; | |
08fe2e33 NR |
789 | } |
790 | ||
68e370a4 NR |
791 | space = thisSpace; |
792 | brk = thisBrk; | |
793 | } | |
08fe2e33 | 794 | |
68e370a4 NR |
795 | // Remove blank/brk at start |
796 | if (paras.size() > 0 | |
797 | && (paras.get(0).getType() == ParagraphType.BLANK || paras.get( | |
798 | 0).getType() == ParagraphType.BREAK)) { | |
799 | paras.remove(0); | |
800 | } | |
801 | ||
802 | // Remove blank/brk at end | |
803 | int last = paras.size() - 1; | |
804 | if (paras.size() > 0 | |
805 | && (paras.get(last).getType() == ParagraphType.BLANK || paras | |
806 | .get(last).getType() == ParagraphType.BREAK)) { | |
807 | paras.remove(last); | |
08fe2e33 NR |
808 | } |
809 | } | |
810 | ||
68e370a4 NR |
811 | /** |
812 | * Get the default cover related to this subject (see <tt>.info</tt> files). | |
813 | * | |
814 | * @param subject | |
815 | * the subject | |
816 | * | |
817 | * @return the cover if any, or NULL | |
818 | */ | |
16a81ef7 | 819 | static Image getDefaultCover(String subject) { |
68686a37 NR |
820 | if (subject != null && !subject.isEmpty() |
821 | && Instance.getCoverDir() != null) { | |
822 | try { | |
823 | File fileCover = new File(Instance.getCoverDir(), subject); | |
333f0e7b | 824 | return getImage(null, fileCover.toURI().toURL(), subject); |
68686a37 NR |
825 | } catch (MalformedURLException e) { |
826 | } | |
827 | } | |
828 | ||
829 | return null; | |
830 | } | |
831 | ||
08fe2e33 NR |
832 | /** |
833 | * Return the list of supported image extensions. | |
834 | * | |
a4143cd7 NR |
835 | * @param emptyAllowed |
836 | * TRUE to allow an empty extension on first place, which can be | |
837 | * used when you may already have an extension in your input but | |
838 | * are not sure about it | |
839 | * | |
08fe2e33 NR |
840 | * @return the extensions |
841 | */ | |
68686a37 | 842 | static String[] getImageExt(boolean emptyAllowed) { |
08fe2e33 NR |
843 | if (emptyAllowed) { |
844 | return new String[] { "", ".png", ".jpg", ".jpeg", ".gif", ".bmp" }; | |
08fe2e33 | 845 | } |
211f7ddb NR |
846 | |
847 | return new String[] { ".png", ".jpg", ".jpeg", ".gif", ".bmp" }; | |
08fe2e33 NR |
848 | } |
849 | ||
a4143cd7 NR |
850 | /** |
851 | * Check if the given resource can be a local image or a remote image, then | |
852 | * refresh the cache with it if it is. | |
853 | * | |
854 | * @param source | |
855 | * the story source | |
856 | * @param line | |
857 | * the resource to check | |
858 | * | |
859 | * @return the image if found, or NULL | |
860 | * | |
861 | */ | |
16a81ef7 | 862 | static Image getImage(BasicSupport support, URL source, String line) { |
333f0e7b | 863 | URL url = getImageUrl(support, source, line); |
68686a37 | 864 | if (url != null) { |
16a81ef7 NR |
865 | if ("file".equals(url.getProtocol())) { |
866 | if (new File(url.getPath()).isDirectory()) { | |
867 | return null; | |
868 | } | |
869 | } | |
68686a37 NR |
870 | InputStream in = null; |
871 | try { | |
872 | in = Instance.getCache().open(url, getSupport(url), true); | |
16a81ef7 | 873 | return new Image(in); |
68686a37 NR |
874 | } catch (IOException e) { |
875 | } finally { | |
876 | if (in != null) { | |
877 | try { | |
878 | in.close(); | |
879 | } catch (IOException e) { | |
880 | } | |
881 | } | |
882 | } | |
883 | } | |
884 | ||
885 | return null; | |
886 | } | |
887 | ||
08fe2e33 NR |
888 | /** |
889 | * Check if the given resource can be a local image or a remote image, then | |
890 | * refresh the cache with it if it is. | |
891 | * | |
892 | * @param source | |
893 | * the story source | |
894 | * @param line | |
895 | * the resource to check | |
896 | * | |
897 | * @return the image URL if found, or NULL | |
898 | * | |
899 | */ | |
333f0e7b | 900 | static URL getImageUrl(BasicSupport support, URL source, String line) { |
08fe2e33 NR |
901 | URL url = null; |
902 | ||
68686a37 NR |
903 | if (line != null) { |
904 | // try for files | |
68686a37 | 905 | if (source != null) { |
68686a37 | 906 | try { |
2ab2e40a NR |
907 | |
908 | String relPath = null; | |
909 | String absPath = null; | |
910 | try { | |
911 | String path = new File(source.getFile()).getParent(); | |
912 | relPath = new File(new File(path), line.trim()) | |
913 | .getAbsolutePath(); | |
914 | } catch (Exception e) { | |
915 | // Cannot be converted to path (one possibility to take | |
916 | // into account: absolute path on Windows) | |
917 | } | |
918 | try { | |
919 | absPath = new File(line.trim()).getAbsolutePath(); | |
920 | } catch (Exception e) { | |
921 | // Cannot be converted to path (at all) | |
922 | } | |
923 | ||
68686a37 | 924 | for (String ext : getImageExt(true)) { |
16a81ef7 NR |
925 | File absFile = new File(absPath + ext); |
926 | File relFile = new File(relPath + ext); | |
927 | if (absPath != null && absFile.exists() | |
928 | && absFile.isFile()) { | |
929 | url = absFile.toURI().toURL(); | |
930 | } else if (relPath != null && relFile.exists() | |
931 | && relFile.isFile()) { | |
932 | url = relFile.toURI().toURL(); | |
68686a37 | 933 | } |
08fe2e33 | 934 | } |
68686a37 | 935 | } catch (Exception e) { |
2ab2e40a | 936 | // Should not happen since we control the correct arguments |
08fe2e33 | 937 | } |
68686a37 | 938 | } |
08fe2e33 | 939 | |
68686a37 NR |
940 | if (url == null) { |
941 | // try for URLs | |
942 | try { | |
08fe2e33 | 943 | for (String ext : getImageExt(true)) { |
f1fb834c NR |
944 | if (Instance.getCache() |
945 | .check(new URL(line + ext), true)) { | |
08fe2e33 | 946 | url = new URL(line + ext); |
333f0e7b | 947 | break; |
08fe2e33 NR |
948 | } |
949 | } | |
68686a37 NR |
950 | |
951 | // try out of cache | |
952 | if (url == null) { | |
953 | for (String ext : getImageExt(true)) { | |
954 | try { | |
955 | url = new URL(line + ext); | |
333f0e7b | 956 | Instance.getCache().refresh(url, support, true); |
68686a37 NR |
957 | break; |
958 | } catch (IOException e) { | |
959 | // no image with this ext | |
960 | url = null; | |
961 | } | |
962 | } | |
963 | } | |
964 | } catch (MalformedURLException e) { | |
965 | // Not an url | |
08fe2e33 | 966 | } |
08fe2e33 | 967 | } |
08fe2e33 | 968 | |
68686a37 NR |
969 | // refresh the cached file |
970 | if (url != null) { | |
971 | try { | |
333f0e7b | 972 | Instance.getCache().refresh(url, support, true); |
68686a37 NR |
973 | } catch (IOException e) { |
974 | // woops, broken image | |
975 | url = null; | |
976 | } | |
08fe2e33 NR |
977 | } |
978 | } | |
979 | ||
980 | return url; | |
981 | } | |
982 | ||
373da363 NR |
983 | /** |
984 | * Open the input file that will be used through the support. | |
315f14ae NR |
985 | * <p> |
986 | * Can return NULL, in which case you are supposed to work without an | |
987 | * {@link InputStream}. | |
373da363 NR |
988 | * |
989 | * @param source | |
990 | * the source {@link URL} | |
991 | * | |
992 | * @return the {@link InputStream} | |
993 | * | |
994 | * @throws IOException | |
995 | * in case of I/O error | |
996 | */ | |
997 | protected InputStream openInput(URL source) throws IOException { | |
998 | return Instance.getCache().open(source, this, false); | |
999 | } | |
1000 | ||
08fe2e33 NR |
1001 | /** |
1002 | * Reset then return {@link BasicSupport#in}. | |
1003 | * | |
1004 | * @return {@link BasicSupport#in} | |
08fe2e33 | 1005 | */ |
68686a37 NR |
1006 | protected InputStream getInput() { |
1007 | return reset(in); | |
08fe2e33 NR |
1008 | } |
1009 | ||
1010 | /** | |
1011 | * Fix the author name if it is prefixed with some "by" {@link String}. | |
1012 | * | |
1013 | * @param author | |
1014 | * the author with a possible prefix | |
1015 | * | |
1016 | * @return the author without prefixes | |
1017 | */ | |
68686a37 | 1018 | protected String fixAuthor(String author) { |
08fe2e33 NR |
1019 | if (author != null) { |
1020 | for (String suffix : new String[] { " ", ":" }) { | |
1021 | for (String byString : Instance.getConfig() | |
1022 | .getString(Config.BYS).split(",")) { | |
1023 | byString += suffix; | |
1024 | if (author.toUpperCase().startsWith(byString.toUpperCase())) { | |
1025 | author = author.substring(byString.length()).trim(); | |
1026 | } | |
1027 | } | |
1028 | } | |
1029 | ||
1030 | // Special case (without suffix): | |
1031 | if (author.startsWith("©")) { | |
1032 | author = author.substring(1); | |
1033 | } | |
1034 | } | |
1035 | ||
1036 | return author; | |
1037 | } | |
1038 | ||
1039 | /** | |
1040 | * Check quotes for bad format (i.e., quotes with normal paragraphs inside) | |
1041 | * and requotify them (i.e., separate them into QUOTE paragraphs and other | |
1042 | * paragraphs (quotes or not)). | |
1043 | * | |
1044 | * @param para | |
a4143cd7 | 1045 | * the paragraph to requotify (not necessarily a quote) |
08fe2e33 NR |
1046 | * |
1047 | * @return the correctly (or so we hope) quotified paragraphs | |
1048 | */ | |
68e370a4 | 1049 | protected List<Paragraph> requotify(Paragraph para) { |
08fe2e33 NR |
1050 | List<Paragraph> newParas = new ArrayList<Paragraph>(); |
1051 | ||
68686a37 NR |
1052 | if (para.getType() == ParagraphType.QUOTE |
1053 | && para.getContent().length() > 2) { | |
08fe2e33 NR |
1054 | String line = para.getContent(); |
1055 | boolean singleQ = line.startsWith("" + openQuote); | |
1056 | boolean doubleQ = line.startsWith("" + openDoubleQuote); | |
1057 | ||
b4dc6ab5 NR |
1058 | // Do not try when more than one quote at a time |
1059 | // (some stories are not easily readable if we do) | |
1060 | if (singleQ | |
1061 | && line.indexOf(closeQuote, 1) < line | |
1062 | .lastIndexOf(closeQuote)) { | |
1063 | newParas.add(para); | |
1064 | return newParas; | |
1065 | } | |
1066 | if (doubleQ | |
1067 | && line.indexOf(closeDoubleQuote, 1) < line | |
1068 | .lastIndexOf(closeDoubleQuote)) { | |
1069 | newParas.add(para); | |
1070 | return newParas; | |
1071 | } | |
1072 | // | |
1073 | ||
08fe2e33 NR |
1074 | if (!singleQ && !doubleQ) { |
1075 | line = openDoubleQuote + line + closeDoubleQuote; | |
793f1071 NR |
1076 | newParas.add(new Paragraph(ParagraphType.QUOTE, line, para |
1077 | .getWords())); | |
08fe2e33 | 1078 | } else { |
a6395bef | 1079 | char open = singleQ ? openQuote : openDoubleQuote; |
08fe2e33 | 1080 | char close = singleQ ? closeQuote : closeDoubleQuote; |
a6395bef NR |
1081 | |
1082 | int posDot = -1; | |
1083 | boolean inQuote = false; | |
1084 | int i = 0; | |
1085 | for (char car : line.toCharArray()) { | |
1086 | if (car == open) { | |
1087 | inQuote = true; | |
1088 | } else if (car == close) { | |
1089 | inQuote = false; | |
1090 | } else if (car == '.' && !inQuote) { | |
1091 | posDot = i; | |
1092 | break; | |
1093 | } | |
1094 | i++; | |
08fe2e33 NR |
1095 | } |
1096 | ||
1097 | if (posDot >= 0) { | |
1098 | String rest = line.substring(posDot + 1).trim(); | |
1099 | line = line.substring(0, posDot + 1).trim(); | |
793f1071 NR |
1100 | long words = 1; |
1101 | for (char car : line.toCharArray()) { | |
1102 | if (car == ' ') { | |
1103 | words++; | |
1104 | } | |
1105 | } | |
1106 | newParas.add(new Paragraph(ParagraphType.QUOTE, line, words)); | |
68686a37 NR |
1107 | if (!rest.isEmpty()) { |
1108 | newParas.addAll(requotify(processPara(rest))); | |
1109 | } | |
08fe2e33 NR |
1110 | } else { |
1111 | newParas.add(para); | |
1112 | } | |
1113 | } | |
1114 | } else { | |
1115 | newParas.add(para); | |
1116 | } | |
1117 | ||
1118 | return newParas; | |
1119 | } | |
1120 | ||
1121 | /** | |
1122 | * Process a {@link Paragraph} from a raw line of text. | |
1123 | * <p> | |
1124 | * Will also fix quotes and HTML encoding if needed. | |
1125 | * | |
1126 | * @param line | |
1127 | * the raw line | |
1128 | * | |
1129 | * @return the processed {@link Paragraph} | |
1130 | */ | |
22848428 | 1131 | protected Paragraph processPara(String line) { |
08fe2e33 NR |
1132 | line = ifUnhtml(line).trim(); |
1133 | ||
1134 | boolean space = true; | |
1135 | boolean brk = true; | |
1136 | boolean quote = false; | |
1137 | boolean tentativeCloseQuote = false; | |
1138 | char prev = '\0'; | |
1139 | int dashCount = 0; | |
793f1071 | 1140 | long words = 1; |
08fe2e33 NR |
1141 | |
1142 | StringBuilder builder = new StringBuilder(); | |
1143 | for (char car : line.toCharArray()) { | |
1144 | if (car != '-') { | |
1145 | if (dashCount > 0) { | |
1146 | // dash, ndash and mdash: - – — | |
1147 | // currently: always use mdash | |
1148 | builder.append(dashCount == 1 ? '-' : '—'); | |
1149 | } | |
1150 | dashCount = 0; | |
1151 | } | |
1152 | ||
1153 | if (tentativeCloseQuote) { | |
1154 | tentativeCloseQuote = false; | |
22848428 | 1155 | if (Character.isLetterOrDigit(car)) { |
08fe2e33 NR |
1156 | builder.append("'"); |
1157 | } else { | |
22848428 NR |
1158 | // handle double-single quotes as double quotes |
1159 | if (prev == car) { | |
1160 | builder.append(closeDoubleQuote); | |
1161 | continue; | |
22848428 | 1162 | } |
211f7ddb NR |
1163 | |
1164 | builder.append(closeQuote); | |
08fe2e33 NR |
1165 | } |
1166 | } | |
1167 | ||
1168 | switch (car) { | |
1169 | case ' ': // note: unbreakable space | |
1170 | case ' ': | |
1171 | case '\t': | |
1172 | case '\n': // just in case | |
1173 | case '\r': // just in case | |
793f1071 NR |
1174 | if (builder.length() > 0 |
1175 | && builder.charAt(builder.length() - 1) != ' ') { | |
1176 | words++; | |
1177 | } | |
08fe2e33 NR |
1178 | builder.append(' '); |
1179 | break; | |
1180 | ||
1181 | case '\'': | |
1182 | if (space || (brk && quote)) { | |
1183 | quote = true; | |
22848428 NR |
1184 | // handle double-single quotes as double quotes |
1185 | if (prev == car) { | |
1186 | builder.deleteCharAt(builder.length() - 1); | |
1187 | builder.append(openDoubleQuote); | |
1188 | } else { | |
1189 | builder.append(openQuote); | |
1190 | } | |
1191 | } else if (prev == ' ' || prev == car) { | |
1192 | // handle double-single quotes as double quotes | |
1193 | if (prev == car) { | |
1194 | builder.deleteCharAt(builder.length() - 1); | |
1195 | builder.append(openDoubleQuote); | |
1196 | } else { | |
1197 | builder.append(openQuote); | |
1198 | } | |
08fe2e33 NR |
1199 | } else { |
1200 | // it is a quote ("I'm off") or a 'quote' ("This | |
1201 | // 'good' restaurant"...) | |
1202 | tentativeCloseQuote = true; | |
1203 | } | |
1204 | break; | |
1205 | ||
1206 | case '"': | |
1207 | if (space || (brk && quote)) { | |
1208 | quote = true; | |
1209 | builder.append(openDoubleQuote); | |
1210 | } else if (prev == ' ') { | |
1211 | builder.append(openDoubleQuote); | |
1212 | } else { | |
1213 | builder.append(closeDoubleQuote); | |
1214 | } | |
1215 | break; | |
1216 | ||
1217 | case '-': | |
1218 | if (space) { | |
1219 | quote = true; | |
1220 | } else { | |
1221 | dashCount++; | |
1222 | } | |
1223 | space = false; | |
1224 | break; | |
1225 | ||
1226 | case '*': | |
1227 | case '~': | |
1228 | case '/': | |
1229 | case '\\': | |
1230 | case '<': | |
1231 | case '>': | |
1232 | case '=': | |
1233 | case '+': | |
1234 | case '_': | |
1235 | case '–': | |
1236 | case '—': | |
1237 | space = false; | |
1238 | builder.append(car); | |
1239 | break; | |
1240 | ||
1241 | case '‘': | |
1242 | case '`': | |
1243 | case '‹': | |
1244 | case '﹁': | |
1245 | case '〈': | |
1246 | case '「': | |
1247 | if (space || (brk && quote)) { | |
1248 | quote = true; | |
1249 | builder.append(openQuote); | |
1250 | } else { | |
22848428 NR |
1251 | // handle double-single quotes as double quotes |
1252 | if (prev == car) { | |
1253 | builder.deleteCharAt(builder.length() - 1); | |
1254 | builder.append(openDoubleQuote); | |
1255 | } else { | |
1256 | builder.append(openQuote); | |
1257 | } | |
08fe2e33 NR |
1258 | } |
1259 | space = false; | |
1260 | brk = false; | |
1261 | break; | |
1262 | ||
1263 | case '’': | |
1264 | case '›': | |
1265 | case '﹂': | |
1266 | case '〉': | |
1267 | case '」': | |
1268 | space = false; | |
1269 | brk = false; | |
22848428 NR |
1270 | // handle double-single quotes as double quotes |
1271 | if (prev == car) { | |
1272 | builder.deleteCharAt(builder.length() - 1); | |
1273 | builder.append(closeDoubleQuote); | |
1274 | } else { | |
1275 | builder.append(closeQuote); | |
1276 | } | |
08fe2e33 NR |
1277 | break; |
1278 | ||
1279 | case '«': | |
1280 | case '“': | |
1281 | case '﹃': | |
1282 | case '《': | |
1283 | case '『': | |
1284 | if (space || (brk && quote)) { | |
1285 | quote = true; | |
1286 | builder.append(openDoubleQuote); | |
1287 | } else { | |
1288 | builder.append(openDoubleQuote); | |
1289 | } | |
1290 | space = false; | |
1291 | brk = false; | |
1292 | break; | |
1293 | ||
1294 | case '»': | |
1295 | case '”': | |
1296 | case '﹄': | |
1297 | case '》': | |
1298 | case '』': | |
1299 | space = false; | |
1300 | brk = false; | |
1301 | builder.append(closeDoubleQuote); | |
1302 | break; | |
1303 | ||
1304 | default: | |
1305 | space = false; | |
1306 | brk = false; | |
1307 | builder.append(car); | |
1308 | break; | |
1309 | } | |
1310 | ||
1311 | prev = car; | |
1312 | } | |
1313 | ||
1314 | if (tentativeCloseQuote) { | |
1315 | tentativeCloseQuote = false; | |
1316 | builder.append(closeQuote); | |
1317 | } | |
1318 | ||
1319 | line = builder.toString().trim(); | |
1320 | ||
1321 | ParagraphType type = ParagraphType.NORMAL; | |
1322 | if (space) { | |
1323 | type = ParagraphType.BLANK; | |
1324 | } else if (brk) { | |
1325 | type = ParagraphType.BREAK; | |
1326 | } else if (quote) { | |
1327 | type = ParagraphType.QUOTE; | |
1328 | } | |
1329 | ||
793f1071 | 1330 | return new Paragraph(type, line, words); |
08fe2e33 NR |
1331 | } |
1332 | ||
1333 | /** | |
a4143cd7 | 1334 | * Remove the HTML from the input <b>if</b> {@link BasicSupport#isHtml()} is |
08fe2e33 NR |
1335 | * true. |
1336 | * | |
1337 | * @param input | |
1338 | * the input | |
1339 | * | |
1340 | * @return the no html version if needed | |
1341 | */ | |
1342 | private String ifUnhtml(String input) { | |
1343 | if (isHtml() && input != null) { | |
1344 | return StringUtils.unhtml(input); | |
1345 | } | |
1346 | ||
1347 | return input; | |
1348 | } | |
1349 | ||
1350 | /** | |
1351 | * Return a {@link BasicSupport} implementation supporting the given | |
1352 | * resource if possible. | |
1353 | * | |
1354 | * @param url | |
1355 | * the story resource | |
1356 | * | |
1357 | * @return an implementation that supports it, or NULL | |
1358 | */ | |
1359 | public static BasicSupport getSupport(URL url) { | |
1360 | if (url == null) { | |
1361 | return null; | |
1362 | } | |
1363 | ||
1364 | // TEXT and INFO_TEXT always support files (not URLs though) | |
1365 | for (SupportType type : SupportType.values()) { | |
1366 | if (type != SupportType.TEXT && type != SupportType.INFO_TEXT) { | |
1367 | BasicSupport support = getSupport(type); | |
1368 | if (support != null && support.supports(url)) { | |
1369 | return support; | |
1370 | } | |
1371 | } | |
1372 | } | |
1373 | ||
373da363 NR |
1374 | for (SupportType type : new SupportType[] { SupportType.INFO_TEXT, |
1375 | SupportType.TEXT }) { | |
08fe2e33 NR |
1376 | BasicSupport support = getSupport(type); |
1377 | if (support != null && support.supports(url)) { | |
1378 | return support; | |
1379 | } | |
1380 | } | |
1381 | ||
1382 | return null; | |
1383 | } | |
1384 | ||
1385 | /** | |
1386 | * Return a {@link BasicSupport} implementation supporting the given type. | |
1387 | * | |
1388 | * @param type | |
1389 | * the type | |
1390 | * | |
1391 | * @return an implementation that supports it, or NULL | |
1392 | */ | |
1393 | public static BasicSupport getSupport(SupportType type) { | |
1394 | switch (type) { | |
1395 | case EPUB: | |
1396 | return new Epub().setType(type); | |
1397 | case INFO_TEXT: | |
1398 | return new InfoText().setType(type); | |
1399 | case FIMFICTION: | |
315f14ae NR |
1400 | try { |
1401 | // Can fail if no client key or NO in options | |
1402 | return new FimfictionApi().setType(type); | |
1403 | } catch (IOException e) { | |
1404 | return new Fimfiction().setType(type); | |
1405 | } | |
08fe2e33 NR |
1406 | case FANFICTION: |
1407 | return new Fanfiction().setType(type); | |
1408 | case TEXT: | |
1409 | return new Text().setType(type); | |
1410 | case MANGAFOX: | |
1411 | return new MangaFox().setType(type); | |
1412 | case E621: | |
1413 | return new E621().setType(type); | |
a4143cd7 NR |
1414 | case YIFFSTAR: |
1415 | return new YiffStar().setType(type); | |
f0608ab1 NR |
1416 | case E_HENTAI: |
1417 | return new EHentai().setType(type); | |
08fe2e33 NR |
1418 | case CBZ: |
1419 | return new Cbz().setType(type); | |
373da363 NR |
1420 | case HTML: |
1421 | return new Html().setType(type); | |
08fe2e33 NR |
1422 | } |
1423 | ||
1424 | return null; | |
1425 | } | |
68686a37 | 1426 | |
315f14ae NR |
1427 | /** |
1428 | * Reset the given {@link InputStream} and return it. | |
1429 | * | |
1430 | * @param in | |
1431 | * the {@link InputStream} to reset | |
1432 | * | |
1433 | * @return the same {@link InputStream} after reset | |
1434 | */ | |
1435 | static protected InputStream reset(InputStream in) { | |
1436 | try { | |
1437 | if (in != null) { | |
1438 | in.reset(); | |
1439 | } | |
1440 | } catch (IOException e) { | |
1441 | } | |
1442 | ||
1443 | return in; | |
1444 | } | |
1445 | ||
68686a37 NR |
1446 | /** |
1447 | * Return the first line from the given input which correspond to the given | |
1448 | * selectors. | |
1449 | * | |
1450 | * @param in | |
1451 | * the input | |
1452 | * @param needle | |
1453 | * a string that must be found inside the target line (also | |
1454 | * supports "^" at start to say "only if it starts with" the | |
1455 | * needle) | |
1456 | * @param relativeLine | |
1457 | * the line to return based upon the target line position (-1 = | |
1458 | * the line before, 0 = the target line...) | |
1459 | * | |
1460 | * @return the line | |
1461 | */ | |
315f14ae NR |
1462 | static protected String getLine(InputStream in, String needle, |
1463 | int relativeLine) { | |
68686a37 NR |
1464 | return getLine(in, needle, relativeLine, true); |
1465 | } | |
1466 | ||
1467 | /** | |
1468 | * Return a line from the given input which correspond to the given | |
1469 | * selectors. | |
1470 | * | |
1471 | * @param in | |
1472 | * the input | |
1473 | * @param needle | |
1474 | * a string that must be found inside the target line (also | |
1475 | * supports "^" at start to say "only if it starts with" the | |
1476 | * needle) | |
1477 | * @param relativeLine | |
1478 | * the line to return based upon the target line position (-1 = | |
1479 | * the line before, 0 = the target line...) | |
1480 | * @param first | |
1481 | * takes the first result (as opposed to the last one, which will | |
1482 | * also always spend the input) | |
1483 | * | |
1484 | * @return the line | |
1485 | */ | |
315f14ae NR |
1486 | static protected String getLine(InputStream in, String needle, |
1487 | int relativeLine, boolean first) { | |
68686a37 NR |
1488 | String rep = null; |
1489 | ||
315f14ae | 1490 | reset(in); |
68686a37 NR |
1491 | |
1492 | List<String> lines = new ArrayList<String>(); | |
1493 | @SuppressWarnings("resource") | |
1494 | Scanner scan = new Scanner(in, "UTF-8"); | |
1495 | int index = -1; | |
1496 | scan.useDelimiter("\\n"); | |
1497 | while (scan.hasNext()) { | |
1498 | lines.add(scan.next()); | |
1499 | ||
1500 | if (index == -1) { | |
1501 | if (needle.startsWith("^")) { | |
1502 | if (lines.get(lines.size() - 1).startsWith( | |
1503 | needle.substring(1))) { | |
1504 | index = lines.size() - 1; | |
1505 | } | |
1506 | ||
1507 | } else { | |
1508 | if (lines.get(lines.size() - 1).contains(needle)) { | |
1509 | index = lines.size() - 1; | |
1510 | } | |
1511 | } | |
1512 | } | |
1513 | ||
1514 | if (index >= 0 && index + relativeLine < lines.size()) { | |
1515 | rep = lines.get(index + relativeLine); | |
1516 | if (first) { | |
1517 | break; | |
1518 | } | |
1519 | } | |
1520 | } | |
1521 | ||
1522 | return rep; | |
1523 | } | |
f0608ab1 NR |
1524 | |
1525 | /** | |
1526 | * Return the text between the key and the endKey (and optional subKey can | |
1527 | * be passed, in this case we will look for the key first, then take the | |
1528 | * text between the subKey and the endKey). | |
1529 | * <p> | |
1530 | * Will only match the first line with the given key if more than one are | |
1531 | * possible. Which also means that if the subKey or endKey is not found on | |
1532 | * that line, NULL will be returned. | |
1533 | * | |
1534 | * @param in | |
1535 | * the input | |
1536 | * @param key | |
27dc7179 NR |
1537 | * the key to match (also supports "^" at start to say |
1538 | * "only if it starts with" the key) | |
f0608ab1 NR |
1539 | * @param subKey |
1540 | * the sub key or NULL if none | |
1541 | * @param endKey | |
1542 | * the end key or NULL for "up to the end" | |
1543 | * @return the text or NULL if not found | |
1544 | */ | |
315f14ae NR |
1545 | static protected String getKeyLine(InputStream in, String key, |
1546 | String subKey, String endKey) { | |
1547 | return getKeyText(getLine(in, key, 0), key, subKey, endKey); | |
1548 | } | |
1549 | ||
1550 | /** | |
1551 | * Return the text between the key and the endKey (and optional subKey can | |
1552 | * be passed, in this case we will look for the key first, then take the | |
1553 | * text between the subKey and the endKey). | |
1554 | * | |
1555 | * @param in | |
1556 | * the input | |
1557 | * @param key | |
1558 | * the key to match (also supports "^" at start to say | |
1559 | * "only if it starts with" the key) | |
1560 | * @param subKey | |
1561 | * the sub key or NULL if none | |
1562 | * @param endKey | |
1563 | * the end key or NULL for "up to the end" | |
1564 | * @return the text or NULL if not found | |
1565 | */ | |
1566 | static protected String getKeyText(String in, String key, String subKey, | |
f0608ab1 NR |
1567 | String endKey) { |
1568 | String result = null; | |
1569 | ||
315f14ae | 1570 | String line = in; |
f0608ab1 NR |
1571 | if (line != null && line.contains(key)) { |
1572 | line = line.substring(line.indexOf(key) + key.length()); | |
1573 | if (subKey == null || subKey.isEmpty() || line.contains(subKey)) { | |
1574 | if (subKey != null) { | |
1575 | line = line.substring(line.indexOf(subKey) | |
1576 | + subKey.length()); | |
1577 | } | |
1578 | if (endKey == null || line.contains(endKey)) { | |
1579 | if (endKey != null) { | |
1580 | line = line.substring(0, line.indexOf(endKey)); | |
1581 | result = line; | |
1582 | } | |
1583 | } | |
1584 | } | |
1585 | } | |
1586 | ||
1587 | return result; | |
1588 | } | |
315f14ae NR |
1589 | |
1590 | /** | |
1591 | * Return the text between the key and the endKey (optional subKeys can be | |
1592 | * passed, in this case we will look for the subKeys first, then take the | |
1593 | * text between the key and the endKey). | |
1594 | * | |
1595 | * @param in | |
1596 | * the input | |
1597 | * @param key | |
1598 | * the key to match | |
1599 | * @param endKey | |
1600 | * the end key or NULL for "up to the end" | |
1601 | * @param afters | |
1602 | * the sub-keys to find before checking for key/endKey | |
1603 | * | |
1604 | * @return the text or NULL if not found | |
1605 | */ | |
1606 | static protected String getKeyTextAfter(String in, String key, | |
1607 | String endKey, String... afters) { | |
1608 | ||
1609 | if (in != null && !in.isEmpty()) { | |
1610 | int pos = indexOfAfter(in, 0, afters); | |
1611 | if (pos < 0) { | |
1612 | return null; | |
1613 | } | |
1614 | ||
1615 | in = in.substring(pos); | |
1616 | } | |
1617 | ||
1618 | return getKeyText(in, key, null, endKey); | |
1619 | } | |
1620 | ||
1621 | /** | |
1622 | * Return the first index after all the given "afters" have been found in | |
1623 | * the {@link String}, or -1 if it was not possible. | |
1624 | * | |
1625 | * @param in | |
1626 | * the input | |
1627 | * @param startAt | |
1628 | * start at this position in the string | |
1629 | * @param afters | |
1630 | * the sub-keys to find before checking for key/endKey | |
1631 | * | |
1632 | * @return the text or NULL if not found | |
1633 | */ | |
1634 | static protected int indexOfAfter(String in, int startAt, String... afters) { | |
1635 | int pos = -1; | |
1636 | if (in != null && !in.isEmpty()) { | |
1637 | pos = startAt; | |
1638 | if (afters != null) { | |
1639 | for (int i = 0; pos >= 0 && i < afters.length; i++) { | |
1640 | String subKey = afters[i]; | |
1641 | if (!subKey.isEmpty()) { | |
1642 | pos = in.indexOf(subKey, pos); | |
1643 | if (pos >= 0) { | |
1644 | pos += subKey.length(); | |
1645 | } | |
1646 | } | |
1647 | } | |
1648 | } | |
1649 | } | |
1650 | ||
1651 | return pos; | |
1652 | } | |
08fe2e33 | 1653 | } |