src/be/nikiroo/fanfix/supported/Fanfiction.java

   1 package be.nikiroo.fanfix.supported;
   2
   3 import java.awt.image.BufferedImage;
   4 import java.io.IOException;
   5 import java.io.InputStream;
   6 import java.net.MalformedURLException;
   7 import java.net.URL;
   8 import java.text.SimpleDateFormat;
   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.StringUtils;
  19
  20 /**
  21  * Support class for <a href="http://www.fanfiction.net/">Faniction.net</a>
  22  * stories, a website dedicated to fanfictions of many, many different
  23  * universes, from TV shows to novels to games.
  24  *
  25  * @author niki
  26  */
  27 class Fanfiction extends BasicSupport {
  28         @Override
  29         protected boolean isHtml() {
  30                 return true;
  31         }
  32
  33         @Override
  34         public String getSourceName() {
  35                 return "Fanfiction.net";
  36         }
  37
  38         @Override
  39         protected MetaData getMeta(URL source, InputStream in) throws IOException {
  40                 MetaData meta = new MetaData();
  41
  42                 meta.setTitle(getTitle(reset(in)));
  43                 meta.setAuthor(getAuthor(reset(in)));
  44                 meta.setDate(getDate(reset(in)));
  45                 meta.setTags(getTags(reset(in)));
  46                 meta.setSource(getSourceName());
  47                 meta.setUrl(source.toString());
  48                 meta.setPublisher(getSourceName());
  49                 meta.setUuid(source.toString());
  50                 meta.setLuid("");
  51                 meta.setLang("EN");
  52                 meta.setSubject(getSubject(reset(in)));
  53                 meta.setType(getType().toString());
  54                 meta.setImageDocument(false);
  55                 meta.setCover(getCover(source, reset(in)));
  56
  57                 return meta;
  58         }
  59
  60         private String getSubject(InputStream in) {
  61                 String line = getLine(in, "id=pre_story_links", 0);
  62                 if (line != null) {
  63                         int pos = line.lastIndexOf('"');
  64                         if (pos >= 1) {
  65                                 line = line.substring(pos + 1);
  66                                 pos = line.indexOf('<');
  67                                 if (pos >= 0) {
  68                                         return StringUtils.unhtml(line.substring(0, pos)).trim();
  69                                 }
  70                         }
  71                 }
  72
  73                 return null;
  74         }
  75
  76         private List<String> getTags(InputStream in) throws IOException {
  77                 List<String> tags = new ArrayList<String>();
  78
  79                 String key = "title=\"Send Private Message\"";
  80                 String line = getLine(in, key, 2);
  81                 if (line != null) {
  82                         key = "Rated:";
  83                         int pos = line.indexOf(key);
  84                         if (pos >= 0) {
  85                                 line = line.substring(pos + key.length());
  86                                 key = "Chapters:";
  87                                 pos = line.indexOf(key);
  88                                 if (pos >= 0) {
  89                                         line = line.substring(0, pos);
  90                                         line = StringUtils.unhtml(line).trim();
  91                                         if (line.endsWith("-")) {
  92                                                 line = line.substring(0, line.length() - 1);
  93                                         }
  94
  95                                         for (String tag : line.split("-")) {
  96                                                 tags.add(StringUtils.unhtml(tag).trim());
  97                                         }
  98                                 }
  99                         }
 100                 }
 101
 102                 return tags;
 103         }
 104
 105         private String getTitle(InputStream in) {
 106                 int i = 0;
 107                 @SuppressWarnings("resource")
 108                 Scanner scan = new Scanner(in, "UTF-8");
 109                 scan.useDelimiter("\\n");
 110                 while (scan.hasNext()) {
 111                         String line = scan.next();
 112                         if (line.contains("xcontrast_txt")) {
 113                                 if ((++i) == 2) {
 114                                         line = StringUtils.unhtml(line).trim();
 115                                         if (line.startsWith("Follow/Fav")) {
 116                                                 line = line.substring("Follow/Fav".length()).trim();
 117                                         }
 118
 119                                         return StringUtils.unhtml(line).trim();
 120                                 }
 121                         }
 122                 }
 123
 124                 return null;
 125         }
 126
 127         private String getAuthor(InputStream in) {
 128                 String author = null;
 129
 130                 int i = 0;
 131                 @SuppressWarnings("resource")
 132                 Scanner scan = new Scanner(in, "UTF-8");
 133                 scan.useDelimiter("\\n");
 134                 while (scan.hasNext()) {
 135                         String line = scan.next();
 136                         if (line.contains("xcontrast_txt")) {
 137                                 if ((++i) == 3) {
 138                                         author = StringUtils.unhtml(line).trim();
 139                                         break;
 140                                 }
 141                         }
 142                 }
 143
 144                 return fixAuthor(author);
 145         }
 146
 147         private String getDate(InputStream in) {
 148                 String key = "Published: <span data-xutime='";
 149                 String line = getLine(in, key, 0);
 150                 if (line != null) {
 151                         int pos = line.indexOf(key);
 152                         if (pos >= 0) {
 153                                 line = line.substring(pos + key.length());
 154                                 pos = line.indexOf('\'');
 155                                 if (pos >= 0) {
 156                                         line = line.substring(0, pos).trim();
 157                                         try {
 158                                                 SimpleDateFormat sdf = new SimpleDateFormat(
 159                                                                 "YYYY-MM-dd");
 160                                                 return sdf
 161                                                                 .format(new Date(1000 * Long.parseLong(line)));
 162                                         } catch (NumberFormatException e) {
 163                                                 Instance.syserr(new IOException(
 164                                                                 "Cannot convert publication date: " + line, e));
 165                                         }
 166                                 }
 167                         }
 168                 }
 169
 170                 return null;
 171         }
 172
 173         @Override
 174         protected String getDesc(URL source, InputStream in) {
 175                 return getLine(in, "title=\"Send Private Message\"", 1);
 176         }
 177
 178         private BufferedImage getCover(URL url, InputStream in) {
 179                 String key = "class='cimage";
 180                 String line = getLine(in, key, 0);
 181                 if (line != null) {
 182                         int pos = line.indexOf(key);
 183                         if (pos >= 0) {
 184                                 line = line.substring(pos + key.length());
 185                                 key = "src='";
 186                                 pos = line.indexOf(key);
 187                                 if (pos >= 0) {
 188                                         line = line.substring(pos + key.length());
 189                                         pos = line.indexOf('\'');
 190                                         if (pos >= 0) {
 191                                                 line = line.substring(0, pos);
 192                                                 if (line.startsWith("//")) {
 193                                                         line = url.getProtocol() + "://"
 194                                                                         + line.substring(2);
 195                                                 } else if (line.startsWith("//")) {
 196                                                         line = url.getProtocol() + "://" + url.getHost()
 197                                                                         + "/" + line.substring(1);
 198                                                 } else {
 199                                                         line = url.getProtocol() + "://" + url.getHost()
 200                                                                         + "/" + url.getPath() + "/" + line;
 201                                                 }
 202
 203                                                 return getImage(this, null, line);
 204                                         }
 205                                 }
 206                         }
 207                 }
 208
 209                 return null;
 210         }
 211
 212         @Override
 213         protected List<Entry<String, URL>> getChapters(URL source, InputStream in) {
 214                 List<Entry<String, URL>> urls = new ArrayList<Entry<String, URL>>();
 215
 216                 String base = source.toString();
 217                 int pos = base.lastIndexOf('/');
 218                 String suffix = base.substring(pos); // including '/' at start
 219                 base = base.substring(0, pos);
 220                 if (base.endsWith("/1")) {
 221                         base = base.substring(0, base.length() - 1); // including '/' at end
 222                 }
 223
 224                 String line = getLine(in, "id=chap_select", 0);
 225                 String key = "<option  value=";
 226                 int i = 1;
 227
 228                 if (line != null) {
 229                         for (pos = line.indexOf(key); pos >= 0; pos = line
 230                                         .indexOf(key, pos), i++) {
 231                                 pos = line.indexOf('>', pos);
 232                                 if (pos >= 0) {
 233                                         int endOfName = line.indexOf('<', pos);
 234                                         if (endOfName >= 0) {
 235                                                 String name = line.substring(pos + 1, endOfName);
 236                                                 String chapNum = i + ".";
 237                                                 if (name.startsWith(chapNum)) {
 238                                                         name = name.substring(chapNum.length(),
 239                                                                         name.length());
 240                                                 }
 241
 242                                                 try {
 243                                                         final String chapName = name.trim();
 244                                                         final URL chapURL = new URL(base + i + suffix);
 245                                                         urls.add(new Entry<String, URL>() {
 246                                                                 public URL setValue(URL value) {
 247                                                                         return null;
 248                                                                 }
 249
 250                                                                 public URL getValue() {
 251                                                                         return chapURL;
 252                                                                 }
 253
 254                                                                 public String getKey() {
 255                                                                         return chapName;
 256                                                                 }
 257                                                         });
 258                                                 } catch (MalformedURLException e) {
 259                                                         Instance.syserr(new IOException(
 260                                                                         "Cannot parse chapter " + i + " url: "
 261                                                                                         + (base + i + suffix), e));
 262                                                 }
 263                                         }
 264                                 }
 265                         }
 266                 } else {
 267                         // only one chapter:
 268                         final String chapName = getTitle(reset(in));
 269                         final URL chapURL = source;
 270                         urls.add(new Entry<String, URL>() {
 271                                 public URL setValue(URL value) {
 272                                         return null;
 273                                 }
 274
 275                                 public URL getValue() {
 276                                         return chapURL;
 277                                 }
 278
 279                                 public String getKey() {
 280                                         return chapName;
 281                                 }
 282                         });
 283                 }
 284
 285                 return urls;
 286         }
 287
 288         @Override
 289         protected String getChapterContent(URL source, InputStream in, int number) {
 290                 StringBuilder builder = new StringBuilder();
 291                 String startAt = "class='storytext ";
 292                 String endAt1 = "function review_init";
 293                 String endAt2 = "id=chap_select";
 294                 boolean ok = false;
 295
 296                 @SuppressWarnings("resource")
 297                 Scanner scan = new Scanner(in, "UTF-8");
 298                 scan.useDelimiter("\\n");
 299                 while (scan.hasNext()) {
 300                         String line = scan.next();
 301                         if (!ok && line.contains(startAt)) {
 302                                 ok = true;
 303                         } else if (ok && (line.contains(endAt1) || line.contains(endAt2))) {
 304                                 ok = false;
 305                                 break;
 306                         }
 307
 308                         if (ok) {
 309                                 // First line may contain the title and chap name again
 310                                 if (builder.length() == 0) {
 311                                         int pos = line.indexOf("<hr");
 312                                         if (pos >= 0) {
 313                                                 boolean chaptered = false;
 314                                                 for (String lang : Instance.getConfig()
 315                                                                 .getString(Config.CHAPTER).split(",")) {
 316                                                         String chapterWord = Instance.getConfig()
 317                                                                         .getStringX(Config.CHAPTER, lang);
 318                                                         int posChap = line.indexOf(chapterWord + " ");
 319                                                         if (posChap < pos) {
 320                                                                 chaptered = true;
 321                                                                 break;
 322                                                         }
 323                                                 }
 324
 325                                                 if (chaptered) {
 326                                                         line = line.substring(pos);
 327                                                 }
 328                                         }
 329                                 }
 330
 331                                 builder.append(line);
 332                         }
 333                 }
 334
 335                 return builder.toString();
 336         }
 337
 338         @Override
 339         protected boolean supports(URL url) {
 340                 return "fanfiction.net".equals(url.getHost())
 341                                 || "www.fanfiction.net".equals(url.getHost());
 342         }
 343 }