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.data.MetaData;
  17 import be.nikiroo.utils.StringUtils;
  18
  19 /**
  20  * Support class for <a href="http://www.fanfiction.net/">Faniction.net</a>
  21  * stories, a website dedicated to fanfictions of many, many different
  22  * universes, from TV shows to novels to games.
  23  *
  24  * @author niki
  25  */
  26 class Fanfiction extends BasicSupport {
  27         @Override
  28         protected boolean isHtml() {
  29                 return true;
  30         }
  31
  32         @Override
  33         public String getSourceName() {
  34                 return "Fanfiction.net";
  35         }
  36
  37         @Override
  38         protected MetaData getMeta(URL source, InputStream in) throws IOException {
  39                 MetaData meta = new MetaData();
  40
  41                 meta.setTitle(getTitle(reset(in)));
  42                 meta.setAuthor(getAuthor(reset(in)));
  43                 meta.setDate(getDate(reset(in)));
  44                 meta.setTags(getTags(reset(in)));
  45                 meta.setSource(getSourceName());
  46                 meta.setUrl(source.toString());
  47                 meta.setPublisher(getSourceName());
  48                 meta.setUuid(source.toString());
  49                 meta.setLuid("");
  50                 meta.setLang("EN");
  51                 meta.setSubject(getSubject(reset(in)));
  52                 meta.setType(getType().toString());
  53                 meta.setImageDocument(false);
  54                 meta.setCover(getCover(source, reset(in)));
  55
  56                 return meta;
  57         }
  58
  59         private String getSubject(InputStream in) {
  60                 String line = getLine(in, "id=pre_story_links", 0);
  61                 if (line != null) {
  62                         int pos = line.lastIndexOf('"');
  63                         if (pos >= 1) {
  64                                 line = line.substring(pos + 1);
  65                                 pos = line.indexOf('<');
  66                                 if (pos >= 0) {
  67                                         return StringUtils.unhtml(line.substring(0, pos)).trim();
  68                                 }
  69                         }
  70                 }
  71
  72                 return null;
  73         }
  74
  75         private List<String> getTags(InputStream in) throws IOException {
  76                 List<String> tags = new ArrayList<String>();
  77
  78                 String key = "title=\"Send Private Message\"";
  79                 String line = getLine(in, key, 2);
  80                 if (line != null) {
  81                         key = "Rated:";
  82                         int pos = line.indexOf(key);
  83                         if (pos >= 0) {
  84                                 line = line.substring(pos + key.length());
  85                                 key = "Chapters:";
  86                                 pos = line.indexOf(key);
  87                                 if (pos >= 0) {
  88                                         line = line.substring(0, pos);
  89                                         line = StringUtils.unhtml(line).trim();
  90                                         if (line.endsWith("-")) {
  91                                                 line = line.substring(0, line.length() - 1);
  92                                         }
  93
  94                                         for (String tag : line.split("-")) {
  95                                                 tags.add(StringUtils.unhtml(tag).trim());
  96                                         }
  97                                 }
  98                         }
  99                 }
 100
 101                 return tags;
 102         }
 103
 104         private String getTitle(InputStream in) {
 105                 int i = 0;
 106                 @SuppressWarnings("resource")
 107                 Scanner scan = new Scanner(in, "UTF-8");
 108                 scan.useDelimiter("\\n");
 109                 while (scan.hasNext()) {
 110                         String line = scan.next();
 111                         if (line.contains("xcontrast_txt")) {
 112                                 if ((++i) == 2) {
 113                                         line = StringUtils.unhtml(line).trim();
 114                                         if (line.startsWith("Follow/Fav")) {
 115                                                 line = line.substring("Follow/Fav".length()).trim();
 116                                         }
 117
 118                                         return StringUtils.unhtml(line).trim();
 119                                 }
 120                         }
 121                 }
 122
 123                 return null;
 124         }
 125
 126         private String getAuthor(InputStream in) {
 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                                         return StringUtils.unhtml(line).trim();
 136                                 }
 137                         }
 138                 }
 139
 140                 return null;
 141         }
 142
 143         private String getDate(InputStream in) {
 144                 String key = "Published: <span data-xutime='";
 145                 String line = getLine(in, key, 0);
 146                 if (line != null) {
 147                         int pos = line.indexOf(key);
 148                         if (pos >= 0) {
 149                                 line = line.substring(pos + key.length());
 150                                 pos = line.indexOf('\'');
 151                                 if (pos >= 0) {
 152                                         line = line.substring(0, pos).trim();
 153                                         try {
 154                                                 SimpleDateFormat sdf = new SimpleDateFormat(
 155                                                                 "YYYY-MM-dd");
 156                                                 return sdf
 157                                                                 .format(new Date(1000 * Long.parseLong(line)));
 158                                         } catch (NumberFormatException e) {
 159                                                 Instance.syserr(new IOException(
 160                                                                 "Cannot convert publication date: " + line, e));
 161                                         }
 162                                 }
 163                         }
 164                 }
 165
 166                 return null;
 167         }
 168
 169         @Override
 170         protected String getDesc(URL source, InputStream in) {
 171                 return getLine(in, "title=\"Send Private Message\"", 1);
 172         }
 173
 174         private BufferedImage getCover(URL url, InputStream in) {
 175                 String key = "class='cimage";
 176                 String line = getLine(in, key, 0);
 177                 if (line != null) {
 178                         int pos = line.indexOf(key);
 179                         if (pos >= 0) {
 180                                 line = line.substring(pos + key.length());
 181                                 key = "src='";
 182                                 pos = line.indexOf(key);
 183                                 if (pos >= 0) {
 184                                         line = line.substring(pos + key.length());
 185                                         pos = line.indexOf('\'');
 186                                         if (pos >= 0) {
 187                                                 line = line.substring(0, pos);
 188                                                 if (line.startsWith("//")) {
 189                                                         line = url.getProtocol() + "://"
 190                                                                         + line.substring(2);
 191                                                 } else if (line.startsWith("//")) {
 192                                                         line = url.getProtocol() + "://" + url.getHost()
 193                                                                         + "/" + line.substring(1);
 194                                                 } else {
 195                                                         line = url.getProtocol() + "://" + url.getHost()
 196                                                                         + "/" + url.getPath() + "/" + line;
 197                                                 }
 198
 199                                                 return getImage(null, line);
 200                                         }
 201                                 }
 202                         }
 203                 }
 204
 205                 return null;
 206         }
 207
 208         @Override
 209         protected List<Entry<String, URL>> getChapters(URL source, InputStream in) {
 210                 List<Entry<String, URL>> urls = new ArrayList<Entry<String, URL>>();
 211
 212                 String base = source.toString();
 213                 int pos = base.lastIndexOf('/');
 214                 String suffix = base.substring(pos); // including '/' at start
 215                 base = base.substring(0, pos);
 216                 if (base.endsWith("/1")) {
 217                         base = base.substring(0, base.length() - 1); // including '/' at end
 218                 }
 219
 220                 String line = getLine(in, "id=chap_select", 0);
 221                 String key = "<option  value=";
 222                 int i = 1;
 223                 for (pos = line.indexOf(key); pos >= 0; pos = line.indexOf(key, pos), i++) {
 224                         pos = line.indexOf('>', pos);
 225                         if (pos >= 0) {
 226                                 int endOfName = line.indexOf('<', pos);
 227                                 if (endOfName >= 0) {
 228                                         String name = line.substring(pos + 1, endOfName);
 229                                         String chapNum = i + ".";
 230                                         if (name.startsWith(chapNum)) {
 231                                                 name = name.substring(chapNum.length(), name.length());
 232                                         }
 233
 234                                         try {
 235                                                 final String chapName = name.trim();
 236                                                 final URL chapURL = new URL(base + i + suffix);
 237                                                 urls.add(new Entry<String, URL>() {
 238                                                         public URL setValue(URL value) {
 239                                                                 return null;
 240                                                         }
 241
 242                                                         public URL getValue() {
 243                                                                 return chapURL;
 244                                                         }
 245
 246                                                         public String getKey() {
 247                                                                 return chapName;
 248                                                         }
 249                                                 });
 250                                         } catch (MalformedURLException e) {
 251                                                 Instance.syserr(new IOException("Cannot parse chapter "
 252                                                                 + i + " url: " + (base + i + suffix), e));
 253                                         }
 254                                 }
 255                         }
 256                 }
 257
 258                 return urls;
 259         }
 260
 261         @Override
 262         protected String getChapterContent(URL source, InputStream in, int number) {
 263                 StringBuilder builder = new StringBuilder();
 264                 String startAt = "class='storytext ";
 265                 String endAt1 = "function review_init";
 266                 String endAt2 = "id=chap_select";
 267                 boolean ok = false;
 268
 269                 @SuppressWarnings("resource")
 270                 Scanner scan = new Scanner(in, "UTF-8");
 271                 scan.useDelimiter("\\n");
 272                 while (scan.hasNext()) {
 273                         String line = scan.next();
 274                         if (!ok && line.contains(startAt)) {
 275                                 ok = true;
 276                         } else if (ok && (line.contains(endAt1) || line.contains(endAt2))) {
 277                                 ok = false;
 278                                 break;
 279                         }
 280
 281                         if (ok) {
 282                                 // First line may contain the title and chap name again
 283                                 if (builder.length() == 0) {
 284                                         int pos = line.indexOf("<hr");
 285                                         if (pos >= 0) {
 286                                                 line = line.substring(pos);
 287                                         }
 288                                 }
 289
 290                                 builder.append(line);
 291                         }
 292                 }
 293
 294                 return builder.toString();
 295         }
 296
 297         @Override
 298         protected boolean supports(URL url) {
 299                 return "fanfiction.net".equals(url.getHost())
 300                                 || "www.fanfiction.net".equals(url.getHost());
 301         }
 302 }