Commit | Line | Data |
---|---|---|
08fe2e33 NR |
1 | package be.nikiroo.fanfix.supported; |
2 | ||
3 | import java.io.IOException; | |
4 | import java.io.InputStream; | |
5 | import java.net.MalformedURLException; | |
6 | import java.net.URL; | |
7 | import java.text.SimpleDateFormat; | |
ce297a79 | 8 | import java.util.AbstractMap; |
08fe2e33 NR |
9 | import java.util.ArrayList; |
10 | import java.util.Date; | |
11 | import java.util.List; | |
12 | import java.util.Map.Entry; | |
13 | import java.util.Scanner; | |
14 | ||
15 | import be.nikiroo.fanfix.Instance; | |
22848428 | 16 | import be.nikiroo.fanfix.bundles.Config; |
68686a37 | 17 | import be.nikiroo.fanfix.data.MetaData; |
16a81ef7 | 18 | import be.nikiroo.utils.Image; |
ed08c171 | 19 | import be.nikiroo.utils.Progress; |
08fe2e33 NR |
20 | import be.nikiroo.utils.StringUtils; |
21 | ||
22 | /** | |
23 | * Support class for <a href="http://www.fanfiction.net/">Faniction.net</a> | |
24 | * stories, a website dedicated to fanfictions of many, many different | |
25 | * universes, from TV shows to novels to games. | |
26 | * | |
27 | * @author niki | |
28 | */ | |
0ffa4754 | 29 | class Fanfiction extends BasicSupport_Deprecated { |
08fe2e33 NR |
30 | @Override |
31 | protected boolean isHtml() { | |
32 | return true; | |
33 | } | |
34 | ||
35 | @Override | |
36 | public String getSourceName() { | |
37 | return "Fanfiction.net"; | |
38 | } | |
39 | ||
40 | @Override | |
68686a37 NR |
41 | protected MetaData getMeta(URL source, InputStream in) throws IOException { |
42 | MetaData meta = new MetaData(); | |
43 | ||
44 | meta.setTitle(getTitle(reset(in))); | |
45 | meta.setAuthor(getAuthor(reset(in))); | |
46 | meta.setDate(getDate(reset(in))); | |
47 | meta.setTags(getTags(reset(in))); | |
48 | meta.setSource(getSourceName()); | |
2206ef66 | 49 | meta.setUrl(source.toString()); |
68686a37 NR |
50 | meta.setPublisher(getSourceName()); |
51 | meta.setUuid(source.toString()); | |
52 | meta.setLuid(""); | |
ce297a79 | 53 | meta.setLang("en"); // TODO! |
68686a37 NR |
54 | meta.setSubject(getSubject(reset(in))); |
55 | meta.setType(getType().toString()); | |
56 | meta.setImageDocument(false); | |
57 | meta.setCover(getCover(source, reset(in))); | |
58 | ||
59 | return meta; | |
60 | } | |
61 | ||
62 | private String getSubject(InputStream in) { | |
08fe2e33 NR |
63 | String line = getLine(in, "id=pre_story_links", 0); |
64 | if (line != null) { | |
65 | int pos = line.lastIndexOf('"'); | |
66 | if (pos >= 1) { | |
67 | line = line.substring(pos + 1); | |
68 | pos = line.indexOf('<'); | |
69 | if (pos >= 0) { | |
68686a37 | 70 | return StringUtils.unhtml(line.substring(0, pos)).trim(); |
08fe2e33 NR |
71 | } |
72 | } | |
73 | } | |
74 | ||
75 | return null; | |
76 | } | |
77 | ||
211f7ddb | 78 | private List<String> getTags(InputStream in) { |
68686a37 | 79 | List<String> tags = new ArrayList<String>(); |
08fe2e33 NR |
80 | |
81 | String key = "title=\"Send Private Message\""; | |
82 | String line = getLine(in, key, 2); | |
83 | if (line != null) { | |
84 | key = "Rated:"; | |
85 | int pos = line.indexOf(key); | |
86 | if (pos >= 0) { | |
87 | line = line.substring(pos + key.length()); | |
88 | key = "Chapters:"; | |
89 | pos = line.indexOf(key); | |
90 | if (pos >= 0) { | |
91 | line = line.substring(0, pos); | |
92 | line = StringUtils.unhtml(line).trim(); | |
93 | if (line.endsWith("-")) { | |
94 | line = line.substring(0, line.length() - 1); | |
95 | } | |
96 | ||
97 | for (String tag : line.split("-")) { | |
68686a37 | 98 | tags.add(StringUtils.unhtml(tag).trim()); |
08fe2e33 NR |
99 | } |
100 | } | |
101 | } | |
102 | } | |
103 | ||
104 | return tags; | |
105 | } | |
106 | ||
68686a37 | 107 | private String getTitle(InputStream in) { |
08fe2e33 NR |
108 | int i = 0; |
109 | @SuppressWarnings("resource") | |
110 | Scanner scan = new Scanner(in, "UTF-8"); | |
111 | scan.useDelimiter("\\n"); | |
112 | while (scan.hasNext()) { | |
113 | String line = scan.next(); | |
114 | if (line.contains("xcontrast_txt")) { | |
115 | if ((++i) == 2) { | |
116 | line = StringUtils.unhtml(line).trim(); | |
117 | if (line.startsWith("Follow/Fav")) { | |
118 | line = line.substring("Follow/Fav".length()).trim(); | |
119 | } | |
120 | ||
68686a37 | 121 | return StringUtils.unhtml(line).trim(); |
08fe2e33 NR |
122 | } |
123 | } | |
124 | } | |
125 | ||
126 | return null; | |
127 | } | |
128 | ||
68686a37 | 129 | private String getAuthor(InputStream in) { |
b4dc6ab5 NR |
130 | String author = null; |
131 | ||
08fe2e33 NR |
132 | int i = 0; |
133 | @SuppressWarnings("resource") | |
134 | Scanner scan = new Scanner(in, "UTF-8"); | |
135 | scan.useDelimiter("\\n"); | |
136 | while (scan.hasNext()) { | |
137 | String line = scan.next(); | |
138 | if (line.contains("xcontrast_txt")) { | |
139 | if ((++i) == 3) { | |
b4dc6ab5 NR |
140 | author = StringUtils.unhtml(line).trim(); |
141 | break; | |
08fe2e33 NR |
142 | } |
143 | } | |
144 | } | |
145 | ||
0ffa4754 | 146 | return BasicSupportHelper.fixAuthor(author); |
08fe2e33 NR |
147 | } |
148 | ||
68686a37 | 149 | private String getDate(InputStream in) { |
08fe2e33 NR |
150 | String key = "Published: <span data-xutime='"; |
151 | String line = getLine(in, key, 0); | |
152 | if (line != null) { | |
153 | int pos = line.indexOf(key); | |
154 | if (pos >= 0) { | |
155 | line = line.substring(pos + key.length()); | |
156 | pos = line.indexOf('\''); | |
157 | if (pos >= 0) { | |
158 | line = line.substring(0, pos).trim(); | |
159 | try { | |
160 | SimpleDateFormat sdf = new SimpleDateFormat( | |
161 | "YYYY-MM-dd"); | |
162 | return sdf | |
163 | .format(new Date(1000 * Long.parseLong(line))); | |
164 | } catch (NumberFormatException e) { | |
16a81ef7 NR |
165 | Instance.getTraceHandler().error( |
166 | new IOException( | |
167 | "Cannot convert publication date: " | |
168 | + line, e)); | |
08fe2e33 NR |
169 | } |
170 | } | |
171 | } | |
172 | } | |
173 | ||
174 | return null; | |
175 | } | |
176 | ||
177 | @Override | |
178 | protected String getDesc(URL source, InputStream in) { | |
179 | return getLine(in, "title=\"Send Private Message\"", 1); | |
180 | } | |
181 | ||
16a81ef7 | 182 | private Image getCover(URL url, InputStream in) { |
08fe2e33 NR |
183 | String key = "class='cimage"; |
184 | String line = getLine(in, key, 0); | |
185 | if (line != null) { | |
186 | int pos = line.indexOf(key); | |
187 | if (pos >= 0) { | |
188 | line = line.substring(pos + key.length()); | |
189 | key = "src='"; | |
190 | pos = line.indexOf(key); | |
191 | if (pos >= 0) { | |
192 | line = line.substring(pos + key.length()); | |
193 | pos = line.indexOf('\''); | |
194 | if (pos >= 0) { | |
195 | line = line.substring(0, pos); | |
196 | if (line.startsWith("//")) { | |
197 | line = url.getProtocol() + "://" | |
198 | + line.substring(2); | |
199 | } else if (line.startsWith("//")) { | |
200 | line = url.getProtocol() + "://" + url.getHost() | |
201 | + "/" + line.substring(1); | |
202 | } else { | |
203 | line = url.getProtocol() + "://" + url.getHost() | |
204 | + "/" + url.getPath() + "/" + line; | |
205 | } | |
206 | ||
333f0e7b | 207 | return getImage(this, null, line); |
08fe2e33 NR |
208 | } |
209 | } | |
210 | } | |
211 | } | |
212 | ||
213 | return null; | |
214 | } | |
215 | ||
216 | @Override | |
ed08c171 NR |
217 | protected List<Entry<String, URL>> getChapters(URL source, InputStream in, |
218 | Progress pg) { | |
08fe2e33 NR |
219 | List<Entry<String, URL>> urls = new ArrayList<Entry<String, URL>>(); |
220 | ||
221 | String base = source.toString(); | |
222 | int pos = base.lastIndexOf('/'); | |
223 | String suffix = base.substring(pos); // including '/' at start | |
224 | base = base.substring(0, pos); | |
225 | if (base.endsWith("/1")) { | |
226 | base = base.substring(0, base.length() - 1); // including '/' at end | |
227 | } | |
228 | ||
229 | String line = getLine(in, "id=chap_select", 0); | |
230 | String key = "<option value="; | |
231 | int i = 1; | |
08fe2e33 | 232 | |
333f0e7b NR |
233 | if (line != null) { |
234 | for (pos = line.indexOf(key); pos >= 0; pos = line | |
235 | .indexOf(key, pos), i++) { | |
236 | pos = line.indexOf('>', pos); | |
237 | if (pos >= 0) { | |
238 | int endOfName = line.indexOf('<', pos); | |
239 | if (endOfName >= 0) { | |
240 | String name = line.substring(pos + 1, endOfName); | |
241 | String chapNum = i + "."; | |
242 | if (name.startsWith(chapNum)) { | |
243 | name = name.substring(chapNum.length(), | |
244 | name.length()); | |
245 | } | |
246 | ||
247 | try { | |
ce297a79 NR |
248 | urls.add(new AbstractMap.SimpleEntry<String, URL>( |
249 | name.trim(), new URL(base + i + suffix))); | |
333f0e7b | 250 | } catch (MalformedURLException e) { |
16a81ef7 NR |
251 | Instance.getTraceHandler() |
252 | .error(new IOException( | |
253 | "Cannot parse chapter " + i | |
254 | + " url: " | |
255 | + (base + i + suffix), e)); | |
333f0e7b | 256 | } |
08fe2e33 NR |
257 | } |
258 | } | |
259 | } | |
333f0e7b NR |
260 | } else { |
261 | // only one chapter: | |
262 | final String chapName = getTitle(reset(in)); | |
263 | final URL chapURL = source; | |
264 | urls.add(new Entry<String, URL>() { | |
211f7ddb | 265 | @Override |
333f0e7b NR |
266 | public URL setValue(URL value) { |
267 | return null; | |
268 | } | |
269 | ||
211f7ddb | 270 | @Override |
333f0e7b NR |
271 | public URL getValue() { |
272 | return chapURL; | |
273 | } | |
274 | ||
211f7ddb | 275 | @Override |
333f0e7b NR |
276 | public String getKey() { |
277 | return chapName; | |
278 | } | |
279 | }); | |
08fe2e33 NR |
280 | } |
281 | ||
282 | return urls; | |
283 | } | |
284 | ||
285 | @Override | |
ed08c171 NR |
286 | protected String getChapterContent(URL source, InputStream in, int number, |
287 | Progress pg) { | |
08fe2e33 NR |
288 | StringBuilder builder = new StringBuilder(); |
289 | String startAt = "class='storytext "; | |
290 | String endAt1 = "function review_init"; | |
291 | String endAt2 = "id=chap_select"; | |
292 | boolean ok = false; | |
293 | ||
294 | @SuppressWarnings("resource") | |
295 | Scanner scan = new Scanner(in, "UTF-8"); | |
296 | scan.useDelimiter("\\n"); | |
297 | while (scan.hasNext()) { | |
298 | String line = scan.next(); | |
299 | if (!ok && line.contains(startAt)) { | |
300 | ok = true; | |
301 | } else if (ok && (line.contains(endAt1) || line.contains(endAt2))) { | |
302 | ok = false; | |
303 | break; | |
304 | } | |
305 | ||
306 | if (ok) { | |
307 | // First line may contain the title and chap name again | |
308 | if (builder.length() == 0) { | |
309 | int pos = line.indexOf("<hr"); | |
310 | if (pos >= 0) { | |
22848428 NR |
311 | boolean chaptered = false; |
312 | for (String lang : Instance.getConfig() | |
313 | .getString(Config.CHAPTER).split(",")) { | |
314 | String chapterWord = Instance.getConfig() | |
315 | .getStringX(Config.CHAPTER, lang); | |
316 | int posChap = line.indexOf(chapterWord + " "); | |
317 | if (posChap < pos) { | |
318 | chaptered = true; | |
319 | break; | |
320 | } | |
321 | } | |
322 | ||
323 | if (chaptered) { | |
324 | line = line.substring(pos); | |
325 | } | |
08fe2e33 NR |
326 | } |
327 | } | |
328 | ||
329 | builder.append(line); | |
406447a4 | 330 | builder.append(' '); |
08fe2e33 NR |
331 | } |
332 | } | |
333 | ||
334 | return builder.toString(); | |
335 | } | |
336 | ||
337 | @Override | |
338 | protected boolean supports(URL url) { | |
339 | return "fanfiction.net".equals(url.getHost()) | |
340 | || "www.fanfiction.net".equals(url.getHost()); | |
341 | } | |
342 | } |