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.getTraceHandler().error(
 161                                                                 new IOException(
 162                                                                                 "Cannot convert publication date: "
 163                                                                                                 + line, e));
 164                                         }
 165                                 }
 166                         }
 167                 }
 168
 169                 return null;
 170         }
 171
 172         @Override
 173         protected String getDesc(URL source, InputStream in) {
 174                 return getLine(in, "title=\"Send Private Message\"", 1);
 175         }
 176
 177         private Image getCover(URL url, InputStream in) {
 178                 String key = "class='cimage";
 179                 String line = getLine(in, key, 0);
 180                 if (line != null) {
 181                         int pos = line.indexOf(key);
 182                         if (pos >= 0) {
 183                                 line = line.substring(pos + key.length());
 184                                 key = "src='";
 185                                 pos = line.indexOf(key);
 186                                 if (pos >= 0) {
 187                                         line = line.substring(pos + key.length());
 188                                         pos = line.indexOf('\'');
 189                                         if (pos >= 0) {
 190                                                 line = line.substring(0, pos);
 191                                                 if (line.startsWith("//")) {
 192                                                         line = url.getProtocol() + "://"
 193                                                                         + line.substring(2);
 194                                                 } else if (line.startsWith("//")) {
 195                                                         line = url.getProtocol() + "://" + url.getHost()
 196                                                                         + "/" + line.substring(1);
 197                                                 } else {
 198                                                         line = url.getProtocol() + "://" + url.getHost()
 199                                                                         + "/" + url.getPath() + "/" + line;
 200                                                 }
 201
 202                                                 return getImage(this, null, line);
 203                                         }
 204                                 }
 205                         }
 206                 }
 207
 208                 return null;
 209         }
 210
 211         @Override
 212         protected List<Entry<String, URL>> getChapters(URL source, InputStream in,
 213                         Progress pg) {
 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                                                         urls.add(new AbstractMap.SimpleEntry<String, URL>(
 244                                                                         name.trim(), new URL(base + i + suffix)));
 245                                                 } catch (MalformedURLException e) {
 246                                                         Instance.getTraceHandler()
 247                                                                         .error(new IOException(
 248                                                                                         "Cannot parse chapter " + i
 249                                                                                                         + " url: "
 250                                                                                                         + (base + i + suffix), e));
 251                                                 }
 252                                         }
 253                                 }
 254                         }
 255                 } else {
 256                         // only one chapter:
 257                         final String chapName = getTitle(reset(in));
 258                         final URL chapURL = source;
 259                         urls.add(new Entry<String, URL>() {
 260                                 @Override
 261                                 public URL setValue(URL value) {
 262                                         return null;
 263                                 }
 264
 265                                 @Override
 266                                 public URL getValue() {
 267                                         return chapURL;
 268                                 }
 269
 270                                 @Override
 271                                 public String getKey() {
 272                                         return chapName;
 273                                 }
 274                         });
 275                 }
 276
 277                 return urls;
 278         }
 279
 280         @Override
 281         protected String getChapterContent(URL source, InputStream in, int number,
 282                         Progress pg) {
 283                 StringBuilder builder = new StringBuilder();
 284                 String startAt = "class='storytext ";
 285                 String endAt1 = "function review_init";
 286                 String endAt2 = "id=chap_select";
 287                 boolean ok = false;
 288
 289                 @SuppressWarnings("resource")
 290                 Scanner scan = new Scanner(in, "UTF-8");
 291                 scan.useDelimiter("\\n");
 292                 while (scan.hasNext()) {
 293                         String line = scan.next();
 294                         if (!ok && line.contains(startAt)) {
 295                                 ok = true;
 296                         } else if (ok && (line.contains(endAt1) || line.contains(endAt2))) {
 297                                 ok = false;
 298                                 break;
 299                         }
 300
 301                         if (ok) {
 302                                 // First line may contain the title and chap name again
 303                                 if (builder.length() == 0) {
 304                                         int pos = line.indexOf("<hr");
 305                                         if (pos >= 0) {
 306                                                 boolean chaptered = false;
 307                                                 for (String lang : Instance.getConfig().getList(
 308                                                                 Config.CONF_CHAPTER)) {
 309                                                         String chapterWord = Instance.getConfig()
 310                                                                         .getStringX(Config.CONF_CHAPTER, lang);
 311                                                         int posChap = line.indexOf(chapterWord + " ");
 312                                                         if (posChap < pos) {
 313                                                                 chaptered = true;
 314                                                                 break;
 315                                                         }
 316                                                 }
 317
 318                                                 if (chaptered) {
 319                                                         line = line.substring(pos);
 320                                                 }
 321                                         }
 322                                 }
 323
 324                                 builder.append(line);
 325                                 builder.append(' ');
 326                         }
 327                 }
 328
 329                 return builder.toString();
 330         }
 331
 332         @Override
 333         protected boolean supports(URL url) {
 334                 return "fanfiction.net".equals(url.getHost())
 335                                 || "www.fanfiction.net".equals(url.getHost());
 336         }
 337 }