| 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 { |
| 26 | /** |
| 27 | * Create a new {@link Fanfiction}. |
| 28 | * |
| 29 | * @param type |
| 30 | * {@link SupportType#FANFICTION} |
| 31 | */ |
| 32 | public Fanfiction(SupportType type) { |
| 33 | super(type); |
| 34 | } |
| 35 | |
| 36 | @Override |
| 37 | public List<SearchableTag> getTags() throws IOException { |
| 38 | String storiesName = null; |
| 39 | String crossoversName = null; |
| 40 | Map<String, String> stories = new HashMap<String, String>(); |
| 41 | Map<String, String> crossovers = new HashMap<String, String>(); |
| 42 | |
| 43 | Document mainPage = load("http://fanfiction.net/"); |
| 44 | Element menu = mainPage.getElementsByClass("dropdown").first(); |
| 45 | if (menu != null) { |
| 46 | Element ul = menu.getElementsByClass("dropdown-menu").first(); |
| 47 | if (ul != null) { |
| 48 | Map<String, String> currentList = null; |
| 49 | for (Element li : ul.getElementsByTag("li")) { |
| 50 | if (li.hasClass("disabled")) { |
| 51 | if (storiesName == null) { |
| 52 | storiesName = li.text(); |
| 53 | currentList = stories; |
| 54 | } else { |
| 55 | crossoversName = li.text(); |
| 56 | currentList = crossovers; |
| 57 | } |
| 58 | } else if (currentList != null) { |
| 59 | Element a = li.getElementsByTag("a").first(); |
| 60 | if (a != null) { |
| 61 | currentList.put(a.absUrl("href"), a.text()); |
| 62 | } |
| 63 | } |
| 64 | } |
| 65 | } |
| 66 | } |
| 67 | |
| 68 | List<SearchableTag> tags = new ArrayList<SearchableTag>(); |
| 69 | |
| 70 | if (storiesName != null) { |
| 71 | SearchableTag tag = new SearchableTag(null, storiesName, true); |
| 72 | for (String id : stories.keySet()) { |
| 73 | tag.add(new SearchableTag(id, stories.get(id), false)); |
| 74 | } |
| 75 | tags.add(tag); |
| 76 | } |
| 77 | |
| 78 | if (crossoversName != null) { |
| 79 | SearchableTag tag = new SearchableTag(null, crossoversName, true); |
| 80 | for (String id : crossovers.keySet()) { |
| 81 | tag.add(new SearchableTag(id, crossovers.get(id), false)); |
| 82 | } |
| 83 | tags.add(tag); |
| 84 | } |
| 85 | |
| 86 | return tags; |
| 87 | } |
| 88 | |
| 89 | @Override |
| 90 | protected void fillTag(SearchableTag tag) throws IOException { |
| 91 | if (tag.getId() == null || tag.isComplete()) { |
| 92 | return; |
| 93 | } |
| 94 | |
| 95 | boolean subtagIsComplete = !tag.getId().contains("/crossovers/"); |
| 96 | |
| 97 | Document doc = load(tag.getId()); |
| 98 | Element list = doc.getElementById("list_output"); |
| 99 | if (list != null) { |
| 100 | Element table = list.getElementsByTag("table").first(); |
| 101 | if (table != null) { |
| 102 | for (Element div : table.getElementsByTag("div")) { |
| 103 | Element a = div.getElementsByTag("a").first(); |
| 104 | Element span = div.getElementsByTag("span").first(); |
| 105 | |
| 106 | if (a != null) { |
| 107 | SearchableTag subtag = new SearchableTag( |
| 108 | a.absUrl("href"), a.text(), subtagIsComplete); |
| 109 | tag.add(subtag); |
| 110 | if (span != null) { |
| 111 | String nr = span.text(); |
| 112 | if (nr.startsWith("(")) { |
| 113 | nr = nr.substring(1); |
| 114 | } |
| 115 | if (nr.endsWith(")")) { |
| 116 | nr = nr.substring(0, nr.length() - 1); |
| 117 | } |
| 118 | nr = nr.trim(); |
| 119 | |
| 120 | long count = 0; |
| 121 | try { |
| 122 | if (nr.toLowerCase().endsWith("m")) { |
| 123 | count = Long.parseLong(nr.substring(0, |
| 124 | nr.length() - 1).trim()); |
| 125 | count *= 1000000; |
| 126 | } else if (nr.toLowerCase().endsWith("k")) { |
| 127 | count = Long.parseLong(nr.substring(0, |
| 128 | nr.length() - 1).trim()); |
| 129 | count *= 1000; |
| 130 | } else { |
| 131 | count = Long.parseLong(nr); |
| 132 | } |
| 133 | } catch (NumberFormatException pe) { |
| 134 | } |
| 135 | |
| 136 | subtag.setCount(count); |
| 137 | } |
| 138 | } |
| 139 | } |
| 140 | } |
| 141 | } |
| 142 | |
| 143 | tag.setComplete(true); |
| 144 | } |
| 145 | |
| 146 | @Override |
| 147 | public List<MetaData> search(String search) throws IOException { |
| 148 | // TODO Auto-generated method stub |
| 149 | return null; |
| 150 | } |
| 151 | |
| 152 | @Override |
| 153 | public List<MetaData> search(SearchableTag tag) throws IOException { |
| 154 | List<MetaData> metas = new ArrayList<MetaData>(); |
| 155 | |
| 156 | if (tag.getId() != null) { |
| 157 | Document doc = load(tag.getId()); |
| 158 | for (Element story : doc.getElementsByClass("z-list")) { |
| 159 | String title = ""; |
| 160 | String url = ""; |
| 161 | String coverUrl = ""; |
| 162 | |
| 163 | Element stitle = story.getElementsByClass("stitle").first(); |
| 164 | if (stitle != null) { |
| 165 | title = stitle.text(); |
| 166 | url = stitle.absUrl("href"); |
| 167 | Element cover = stitle.getElementsByTag("img").first(); |
| 168 | if (cover != null) { |
| 169 | // note: see data-original if needed? |
| 170 | coverUrl = cover.absUrl("src"); |
| 171 | } |
| 172 | } |
| 173 | |
| 174 | String author = ""; |
| 175 | |
| 176 | Elements as = story.getElementsByTag("a"); |
| 177 | if (as.size() > 1) { |
| 178 | author = as.get(1).text(); |
| 179 | } |
| 180 | |
| 181 | String resume = ""; |
| 182 | String tags = ""; |
| 183 | |
| 184 | Elements divs = story.getElementsByTag("div"); |
| 185 | if (divs.size() > 1 && divs.get(1).childNodeSize() > 0) { |
| 186 | resume = divs.get(1).text(); |
| 187 | if (divs.size() > 2) { |
| 188 | tags = divs.get(2).text(); |
| 189 | resume = resume.substring(0, |
| 190 | resume.length() - tags.length()).trim(); |
| 191 | } |
| 192 | } |
| 193 | |
| 194 | MetaData meta = new MetaData(); |
| 195 | meta.setAuthor(author); |
| 196 | // meta.setCover(cover); //TODO ? |
| 197 | meta.setImageDocument(false); |
| 198 | meta.setResume(getSupport().makeChapter(new URL(tag.getId()), |
| 199 | 0, Instance.getTrans().getString(StringId.DESCRIPTION), |
| 200 | resume)); |
| 201 | meta.setSource(getType().getSourceName()); |
| 202 | meta.setTags(Arrays.asList(tags.split(" *- *"))); |
| 203 | meta.setTitle(title); |
| 204 | meta.setUrl(url); |
| 205 | |
| 206 | metas.add(meta); |
| 207 | } |
| 208 | } |
| 209 | |
| 210 | return metas; |
| 211 | } |
| 212 | |
| 213 | public static void main(String[] args) throws IOException { |
| 214 | Fanfiction f = new Fanfiction(SupportType.FANFICTION); |
| 215 | |
| 216 | SearchableTag cartoons = f.getTags().get(0).getChildren().get(2); |
| 217 | f.fillTag(cartoons); |
| 218 | SearchableTag mlp = cartoons.getChildren().get(2); |
| 219 | System.out.println(mlp); |
| 220 | |
| 221 | SearchableTag ccartoons = f.getTags().get(1).getChildren().get(0); |
| 222 | f.fillTag(ccartoons); |
| 223 | SearchableTag cmlp = ccartoons.getChildren().get(0); |
| 224 | System.out.println(cmlp); |
| 225 | |
| 226 | f.fillTag(cmlp); |
| 227 | System.out.println(cmlp); |
| 228 | |
| 229 | List<MetaData> metas = f.search(mlp); |
| 230 | System.out.println(metas); |
| 231 | } |
| 232 | } |