Commit | Line | Data |
---|---|---|
158b372d NR |
1 | package be.nikiroo.fanfix.searchable; |
2 | ||
3 | import java.io.IOException; | |
4 | import java.net.URL; | |
5 | import java.util.ArrayList; | |
6 | import java.util.Arrays; | |
7 | import java.util.HashMap; | |
8 | import java.util.List; | |
9 | import java.util.Map; | |
10 | ||
11 | import org.jsoup.nodes.Document; | |
12 | import org.jsoup.nodes.Element; | |
13 | import org.jsoup.select.Elements; | |
14 | ||
15 | import be.nikiroo.fanfix.Instance; | |
16 | import be.nikiroo.fanfix.bundles.StringId; | |
17 | import be.nikiroo.fanfix.data.MetaData; | |
18 | import be.nikiroo.fanfix.supported.SupportType; | |
19 | ||
20 | /** | |
21 | * A {@link BasicSearchable} for Fanfiction.NET. | |
22 | * | |
23 | * @author niki | |
24 | */ | |
25 | class Fanfiction extends BasicSearchable { | |
76ec935e NR |
26 | static private String BASE_URL = "http://fanfiction.net/"; |
27 | ||
158b372d NR |
28 | /** |
29 | * Create a new {@link Fanfiction}. | |
30 | * | |
31 | * @param type | |
32 | * {@link SupportType#FANFICTION} | |
33 | */ | |
34 | public Fanfiction(SupportType type) { | |
35 | super(type); | |
36 | } | |
37 | ||
38 | @Override | |
39 | public List<SearchableTag> getTags() throws IOException { | |
40 | String storiesName = null; | |
41 | String crossoversName = null; | |
42 | Map<String, String> stories = new HashMap<String, String>(); | |
43 | Map<String, String> crossovers = new HashMap<String, String>(); | |
44 | ||
76ec935e | 45 | Document mainPage = load(BASE_URL, true); |
158b372d NR |
46 | Element menu = mainPage.getElementsByClass("dropdown").first(); |
47 | if (menu != null) { | |
48 | Element ul = menu.getElementsByClass("dropdown-menu").first(); | |
49 | if (ul != null) { | |
50 | Map<String, String> currentList = null; | |
51 | for (Element li : ul.getElementsByTag("li")) { | |
52 | if (li.hasClass("disabled")) { | |
53 | if (storiesName == null) { | |
54 | storiesName = li.text(); | |
55 | currentList = stories; | |
56 | } else { | |
57 | crossoversName = li.text(); | |
58 | currentList = crossovers; | |
59 | } | |
60 | } else if (currentList != null) { | |
61 | Element a = li.getElementsByTag("a").first(); | |
62 | if (a != null) { | |
63 | currentList.put(a.absUrl("href"), a.text()); | |
64 | } | |
65 | } | |
66 | } | |
67 | } | |
68 | } | |
69 | ||
70 | List<SearchableTag> tags = new ArrayList<SearchableTag>(); | |
71 | ||
72 | if (storiesName != null) { | |
76ec935e | 73 | SearchableTag tag = new SearchableTag(null, storiesName, false); |
158b372d | 74 | for (String id : stories.keySet()) { |
76ec935e | 75 | tag.add(new SearchableTag(id, stories.get(id), true, false)); |
158b372d NR |
76 | } |
77 | tags.add(tag); | |
78 | } | |
79 | ||
80 | if (crossoversName != null) { | |
76ec935e | 81 | SearchableTag tag = new SearchableTag(null, crossoversName, false); |
158b372d | 82 | for (String id : crossovers.keySet()) { |
76ec935e | 83 | tag.add(new SearchableTag(id, crossovers.get(id), false, false)); |
158b372d NR |
84 | } |
85 | tags.add(tag); | |
86 | } | |
87 | ||
88 | return tags; | |
89 | } | |
90 | ||
91 | @Override | |
92 | protected void fillTag(SearchableTag tag) throws IOException { | |
93 | if (tag.getId() == null || tag.isComplete()) { | |
94 | return; | |
95 | } | |
96 | ||
76ec935e | 97 | boolean subtagIsLeaf = !tag.getId().contains("/crossovers/"); |
158b372d | 98 | |
76ec935e | 99 | Document doc = load(tag.getId(), false); |
158b372d NR |
100 | Element list = doc.getElementById("list_output"); |
101 | if (list != null) { | |
102 | Element table = list.getElementsByTag("table").first(); | |
103 | if (table != null) { | |
104 | for (Element div : table.getElementsByTag("div")) { | |
105 | Element a = div.getElementsByTag("a").first(); | |
106 | Element span = div.getElementsByTag("span").first(); | |
107 | ||
108 | if (a != null) { | |
109 | SearchableTag subtag = new SearchableTag( | |
76ec935e | 110 | a.absUrl("href"), a.text(), subtagIsLeaf); |
158b372d NR |
111 | tag.add(subtag); |
112 | if (span != null) { | |
113 | String nr = span.text(); | |
114 | if (nr.startsWith("(")) { | |
115 | nr = nr.substring(1); | |
116 | } | |
117 | if (nr.endsWith(")")) { | |
118 | nr = nr.substring(0, nr.length() - 1); | |
119 | } | |
120 | nr = nr.trim(); | |
121 | ||
122 | long count = 0; | |
123 | try { | |
124 | if (nr.toLowerCase().endsWith("m")) { | |
125 | count = Long.parseLong(nr.substring(0, | |
126 | nr.length() - 1).trim()); | |
127 | count *= 1000000; | |
128 | } else if (nr.toLowerCase().endsWith("k")) { | |
129 | count = Long.parseLong(nr.substring(0, | |
130 | nr.length() - 1).trim()); | |
131 | count *= 1000; | |
132 | } else { | |
133 | count = Long.parseLong(nr); | |
134 | } | |
135 | } catch (NumberFormatException pe) { | |
136 | } | |
137 | ||
138 | subtag.setCount(count); | |
139 | } | |
140 | } | |
141 | } | |
142 | } | |
143 | } | |
144 | ||
145 | tag.setComplete(true); | |
146 | } | |
147 | ||
148 | @Override | |
149 | public List<MetaData> search(String search) throws IOException { | |
76ec935e | 150 | // TODO /search/?reader=1&type=story&keywords=blablablab |
158b372d NR |
151 | return null; |
152 | } | |
153 | ||
154 | @Override | |
155 | public List<MetaData> search(SearchableTag tag) throws IOException { | |
156 | List<MetaData> metas = new ArrayList<MetaData>(); | |
157 | ||
158 | if (tag.getId() != null) { | |
76ec935e NR |
159 | Document doc = load(tag.getId(), false); |
160 | ||
161 | Element center = doc.getElementsByTag("center").first(); | |
162 | if (center != null) { | |
163 | int pages = -1; | |
164 | for (Element a : center.getElementsByTag("a")) { | |
165 | if (a.absUrl("href").contains("&p=")) { | |
166 | int thisLinkPages = -1; | |
167 | try { | |
168 | String[] tab = a.absUrl("href").split("="); | |
169 | tab = tab[tab.length - 1].split("&"); | |
170 | thisLinkPages = Integer | |
171 | .parseInt(tab[tab.length - 1]); | |
172 | } catch (Exception e) { | |
173 | } | |
174 | ||
175 | pages = Math.max(pages, thisLinkPages); | |
176 | } | |
177 | } | |
178 | ||
179 | tag.setPages(pages); | |
180 | } | |
181 | ||
158b372d NR |
182 | for (Element story : doc.getElementsByClass("z-list")) { |
183 | String title = ""; | |
184 | String url = ""; | |
185 | String coverUrl = ""; | |
186 | ||
187 | Element stitle = story.getElementsByClass("stitle").first(); | |
188 | if (stitle != null) { | |
189 | title = stitle.text(); | |
190 | url = stitle.absUrl("href"); | |
191 | Element cover = stitle.getElementsByTag("img").first(); | |
192 | if (cover != null) { | |
193 | // note: see data-original if needed? | |
194 | coverUrl = cover.absUrl("src"); | |
195 | } | |
196 | } | |
197 | ||
198 | String author = ""; | |
199 | ||
200 | Elements as = story.getElementsByTag("a"); | |
201 | if (as.size() > 1) { | |
202 | author = as.get(1).text(); | |
203 | } | |
204 | ||
205 | String resume = ""; | |
206 | String tags = ""; | |
207 | ||
208 | Elements divs = story.getElementsByTag("div"); | |
209 | if (divs.size() > 1 && divs.get(1).childNodeSize() > 0) { | |
210 | resume = divs.get(1).text(); | |
211 | if (divs.size() > 2) { | |
212 | tags = divs.get(2).text(); | |
213 | resume = resume.substring(0, | |
214 | resume.length() - tags.length()).trim(); | |
215 | } | |
216 | } | |
217 | ||
218 | MetaData meta = new MetaData(); | |
219 | meta.setAuthor(author); | |
220 | // meta.setCover(cover); //TODO ? | |
221 | meta.setImageDocument(false); | |
222 | meta.setResume(getSupport().makeChapter(new URL(tag.getId()), | |
223 | 0, Instance.getTrans().getString(StringId.DESCRIPTION), | |
224 | resume)); | |
225 | meta.setSource(getType().getSourceName()); | |
76ec935e | 226 | // TODO: remove tags to interpret them instead (lang, words..) |
158b372d NR |
227 | meta.setTags(Arrays.asList(tags.split(" *- *"))); |
228 | meta.setTitle(title); | |
229 | meta.setUrl(url); | |
230 | ||
231 | metas.add(meta); | |
232 | } | |
233 | } | |
234 | ||
235 | return metas; | |
236 | } | |
237 | ||
238 | public static void main(String[] args) throws IOException { | |
239 | Fanfiction f = new Fanfiction(SupportType.FANFICTION); | |
240 | ||
241 | SearchableTag cartoons = f.getTags().get(0).getChildren().get(2); | |
242 | f.fillTag(cartoons); | |
243 | SearchableTag mlp = cartoons.getChildren().get(2); | |
244 | System.out.println(mlp); | |
245 | ||
246 | SearchableTag ccartoons = f.getTags().get(1).getChildren().get(0); | |
247 | f.fillTag(ccartoons); | |
248 | SearchableTag cmlp = ccartoons.getChildren().get(0); | |
249 | System.out.println(cmlp); | |
250 | ||
251 | f.fillTag(cmlp); | |
252 | System.out.println(cmlp); | |
253 | ||
254 | List<MetaData> metas = f.search(mlp); | |
76ec935e NR |
255 | System.out.println(mlp.getPages()); |
256 | //System.out.println(metas); | |
158b372d NR |
257 | } |
258 | } |