supported/YiffStar.java

   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.util.AbstractMap;
   8 import java.util.ArrayList;
   9 import java.util.HashMap;
  10 import java.util.List;
  11 import java.util.Map;
  12 import java.util.Map.Entry;
  13 import java.util.Scanner;
  14
  15 import be.nikiroo.fanfix.Instance;
  16 import be.nikiroo.fanfix.bundles.Config;
  17 import be.nikiroo.fanfix.data.MetaData;
  18 import be.nikiroo.utils.Image;
  19 import be.nikiroo.utils.Progress;
  20 import be.nikiroo.utils.StringUtils;
  21
  22 /**
  23  * Support class for <a href="https://sofurry.com/">SoFurry.com</a>, a Furry
  24  * website supporting images and stories (we only retrieve the stories).
  25  *
  26  * @author niki
  27  */
  28 class YiffStar extends BasicSupport_Deprecated {
  29         @Override
  30         protected MetaData getMeta(URL source, InputStream in) throws IOException {
  31                 MetaData meta = new MetaData();
  32
  33                 meta.setTitle(getTitle(reset(in)));
  34                 meta.setAuthor(getAuthor(reset(in)));
  35                 meta.setDate("");
  36                 meta.setTags(getTags(reset(in)));
  37                 meta.setUrl(source.toString());
  38                 meta.setUuid(source.toString());
  39                 meta.setLuid("");
  40                 meta.setLang("en");
  41                 meta.setSubject("Furry");
  42                 meta.setImageDocument(false);
  43                 meta.setCover(getCover(source, reset(in)));
  44
  45                 return meta;
  46         }
  47
  48         @Override
  49         protected boolean supports(URL url) {
  50                 String host = url.getHost();
  51                 if (host.startsWith("www.")) {
  52                         host = host.substring("www.".length());
  53                 }
  54
  55                 return "sofurry.com".equals(host);
  56         }
  57
  58         @Override
  59         protected boolean isHtml() {
  60                 return true;
  61         }
  62
  63         @Override
  64         public void login() throws IOException {
  65                 // Note: this should not be necessary anymore
  66                 // (the "/guest" trick is enough)
  67                 String login = Instance.getInstance().getConfig().getString(Config.LOGIN_YIFFSTAR_USER);
  68                 String password = Instance.getInstance().getConfig().getString(Config.LOGIN_YIFFSTAR_PASS);
  69
  70                 if (login != null && !login.isEmpty() && password != null
  71                                 && !password.isEmpty()) {
  72
  73                         Map<String, String> post = new HashMap<String, String>();
  74                         post.put("LoginForm[sfLoginUsername]", login);
  75                         post.put("LoginForm[sfLoginPassword]", password);
  76                         post.put("YII_CSRF_TOKEN", "");
  77                         post.put("yt1", "Login");
  78                         post.put("returnUrl", "/");
  79
  80                         // Cookies will actually be retained by the cache manager once
  81                         // logged in
  82                         Instance.getInstance().getCache()
  83                                         .openNoCache(new URL("https://www.sofurry.com/user/login"), this, post, null, null).close();
  84                 }
  85         }
  86
  87         @Override
  88         public URL getCanonicalUrl(URL source) {
  89                 try {
  90                         if (source.getPath().startsWith("/view")) {
  91                                 source = guest(source.toString());
  92                                 // NO CACHE because we don't want the NotLoggedIn message later
  93                                 InputStream in = Instance.getInstance().getCache().openNoCache(source, this, null, null, null);
  94                                 String line = getLine(in, "/browse/folder/", 0);
  95                                 if (line != null) {
  96                                         String[] tab = line.split("\"");
  97                                         if (tab.length > 1) {
  98                                                 String groupUrl = source.getProtocol() + "://"
  99                                                                 + source.getHost() + tab[1];
 100                                                 return guest(groupUrl);
 101                                         }
 102                                 }
 103                         }
 104                 } catch (Exception e) {
 105                         Instance.getInstance().getTraceHandler().error(e);
 106                 }
 107
 108                 return super.getCanonicalUrl(source);
 109         }
 110
 111         private List<String> getTags(InputStream in) {
 112                 List<String> tags = new ArrayList<String>();
 113
 114                 String line = getLine(in, "class=\"sf-story-big-tags", 0);
 115                 if (line != null) {
 116                         String[] tab = StringUtils.unhtml(line).split(",");
 117                         for (String possibleTag : tab) {
 118                                 String tag = possibleTag.trim();
 119                                 if (!tag.isEmpty() && !tag.equals("...") && !tags.contains(tag)) {
 120                                         tags.add(tag);
 121                                 }
 122                         }
 123                 }
 124
 125                 return tags;
 126         }
 127
 128         private Image getCover(URL source, InputStream in) throws IOException {
 129
 130                 List<Entry<String, URL>> chaps = getChapters(source, in, null);
 131                 if (!chaps.isEmpty()) {
 132                         in = Instance.getInstance().getCache().open(chaps.get(0).getValue(), this, true);
 133                         String line = getLine(in, " name=\"og:image\"", 0);
 134                         if (line != null) {
 135                                 int pos = -1;
 136                                 for (int i = 0; i < 3; i++) {
 137                                         pos = line.indexOf('"', pos + 1);
 138                                 }
 139
 140                                 if (pos >= 0) {
 141                                         line = line.substring(pos + 1);
 142                                         pos = line.indexOf('"');
 143                                         if (pos >= 0) {
 144                                                 line = line.substring(0, pos);
 145                                                 if (line.contains("/thumb?")) {
 146                                                         line = line.replace("/thumb?",
 147                                                                         "/auxiliaryContent?type=25&");
 148                                                         return getImage(this, null, line);
 149                                                 }
 150                                         }
 151                                 }
 152                         }
 153                 }
 154
 155                 return null;
 156         }
 157
 158         private String getAuthor(InputStream in) {
 159                 String author = getLine(in, "class=\"onlinestatus", 0);
 160                 if (author != null) {
 161                         return StringUtils.unhtml(author).trim();
 162                 }
 163
 164                 return null;
 165         }
 166
 167         private String getTitle(InputStream in) {
 168                 String title = getLine(in, "class=\"sflabel pagetitle", 0);
 169                 if (title != null) {
 170                         if (title.contains("(series)")) {
 171                                 title = title.replace("(series)", "");
 172                         }
 173                         return StringUtils.unhtml(title).trim();
 174                 }
 175
 176                 return null;
 177         }
 178
 179         @Override
 180         protected String getDesc(URL source, InputStream in) throws IOException {
 181                 return null; // TODO: no description at all? Cannot find one...
 182         }
 183
 184         @Override
 185         protected List<Entry<String, URL>> getChapters(URL source, InputStream in,
 186                         Progress pg) throws IOException {
 187                 List<Entry<String, URL>> urls = new ArrayList<Entry<String, URL>>();
 188
 189                 @SuppressWarnings("resource")
 190                 Scanner scan = new Scanner(in, "UTF-8");
 191                 scan.useDelimiter("\\n");
 192                 while (scan.hasNext()) {
 193                         String line = scan.next();
 194                         if (line.contains("\"/view/") && line.contains("title=")) {
 195                                 String[] tab = line.split("\"");
 196                                 if (tab.length > 5) {
 197                                         String link = tab[5];
 198                                         if (link.startsWith("/")) {
 199                                                 link = source.getProtocol() + "://" + source.getHost()
 200                                                                 + link;
 201                                         }
 202                                         urls.add(new AbstractMap.SimpleEntry<String, URL>(
 203                                                         StringUtils.unhtml(line).trim(), guest(link)));
 204                                 }
 205                         }
 206                 }
 207
 208                 return urls;
 209         }
 210
 211         @Override
 212         protected String getChapterContent(URL source, InputStream in, int number,
 213                         Progress pg) throws IOException {
 214                 StringBuilder builder = new StringBuilder();
 215
 216                 String startAt = "id=\"sfContentBody";
 217                 String endAt = "id=\"recommendationArea";
 218                 boolean ok = false;
 219
 220                 @SuppressWarnings("resource")
 221                 Scanner scan = new Scanner(in, "UTF-8");
 222                 scan.useDelimiter("\\n");
 223                 while (scan.hasNext()) {
 224                         String line = scan.next();
 225                         if (!ok && line.contains(startAt)) {
 226                                 ok = true;
 227                         } else if (ok && line.contains(endAt)) {
 228                                 ok = false;
 229                                 break;
 230                         }
 231
 232                         if (ok) {
 233                                 builder.append(line);
 234                                 builder.append(' ');
 235                         }
 236                 }
 237
 238                 return builder.toString();
 239         }
 240
 241         /**
 242          * Return a {@link URL} from the given link, but add the "/guest" part to it
 243          * to make sure we don't need to be logged-in to see it.
 244          *
 245          * @param link
 246          *            the link
 247          *
 248          * @return the {@link URL}
 249          *
 250          * @throws MalformedURLException
 251          *             in case of data error
 252          */
 253         private URL guest(String link) throws MalformedURLException {
 254                 if (link.contains("?")) {
 255                         if (link.contains("/?")) {
 256                                 return new URL(link.replace("?", "guest?"));
 257                         }
 258
 259                         return new URL(link.replace("?", "/guest?"));
 260                 }
 261
 262                 return new URL(link + "/guest");
 263         }
 264 }