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