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