supported/Fimfiction.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.util.ArrayList;
   9 import java.util.HashMap;
  10 import java.util.List;
  11 import java.util.Map;
  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.Progress;
  18 import be.nikiroo.utils.StringUtils;
  19
  20 /**
  21  * Support class for <a href="http://www.fimfiction.net/">FimFiction.net</a>
  22  * stories, a website dedicated to My Little Pony.
  23  *
  24  * @author niki
  25  */
  26 class Fimfiction extends BasicSupport {
  27         @Override
  28         protected boolean isHtml() {
  29                 return true;
  30         }
  31
  32         @Override
  33         public String getSourceName() {
  34                 return "FimFiction.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("MLP");
  52                 meta.setType(getType().toString());
  53                 meta.setImageDocument(false);
  54                 meta.setCover(getCover(reset(in)));
  55
  56                 return meta;
  57         }
  58
  59         @Override
  60         public Map<String, String> getCookies() {
  61                 Map<String, String> cookies = new HashMap<String, String>();
  62                 cookies.put("view_mature", "true");
  63                 return cookies;
  64         }
  65
  66         private List<String> getTags(InputStream in) {
  67                 List<String> tags = new ArrayList<String>();
  68                 tags.add("MLP");
  69
  70                 @SuppressWarnings("resource")
  71                 Scanner scan = new Scanner(in, "UTF-8");
  72                 scan.useDelimiter("\\n");
  73                 boolean started = false;
  74                 while (scan.hasNext()) {
  75                         String line = scan.next();
  76
  77                         if (!started) {
  78                                 started = line.contains("\"story_container\"");
  79                         }
  80
  81                         if (started && line.contains("class=\"tag-")) {
  82                                 if (line.contains("index.php")) {
  83                                         break; // end of *this story* tags
  84                                 }
  85
  86                                 String keyword = "title=\"";
  87                                 Scanner tagScanner = new Scanner(line);
  88                                 tagScanner.useDelimiter(keyword);
  89                                 if (tagScanner.hasNext()) {
  90                                         tagScanner.next();// Ignore first one
  91                                 }
  92                                 while (tagScanner.hasNext()) {
  93                                         String tag = tagScanner.next();
  94                                         if (tag.contains("\"")) {
  95                                                 tag = tag.split("\"")[0];
  96                                                 tag = StringUtils.unhtml(tag).trim();
  97                                                 if (!tag.isEmpty() && !tags.contains(tag)) {
  98                                                         tags.add(tag);
  99                                                 }
 100                                         }
 101                                 }
 102                                 tagScanner.close();
 103                         }
 104                 }
 105
 106                 return tags;
 107         }
 108
 109         private String getTitle(InputStream in) {
 110                 String line = getLine(in, " property=\"og:title\"", 0);
 111                 if (line != null) {
 112                         int pos = -1;
 113                         for (int i = 0; i < 3; i++) {
 114                                 pos = line.indexOf('"', pos + 1);
 115                         }
 116
 117                         if (pos >= 0) {
 118                                 line = line.substring(pos + 1);
 119                                 pos = line.indexOf('"');
 120                                 if (pos >= 0) {
 121                                         return StringUtils.unhtml(line.substring(0, pos)).trim();
 122                                 }
 123                         }
 124                 }
 125
 126                 return null;
 127         }
 128
 129         private String getAuthor(InputStream in) {
 130                 String line = getLine(in, " href=\"/user/", 0);
 131                 if (line != null) {
 132                         int pos = line.indexOf('"');
 133                         if (pos >= 0) {
 134                                 line = line.substring(pos + 1);
 135                                 pos = line.indexOf('"');
 136                                 if (pos >= 0) {
 137                                         line = line.substring(0, pos);
 138                                         pos = line.lastIndexOf('/');
 139                                         if (pos >= 0) {
 140                                                 line = line.substring(pos + 1);
 141                                                 return line.replace('+', ' ');
 142                                         }
 143                                 }
 144                         }
 145                 }
 146
 147                 return null;
 148         }
 149
 150         private String getDate(InputStream in) {
 151                 String line = getLine(in, "<span class=\"date\">", 0);
 152                 if (line != null) {
 153                         int pos = -1;
 154                         for (int i = 0; i < 3; i++) {
 155                                 pos = line.indexOf('>', pos + 1);
 156                         }
 157
 158                         if (pos >= 0) {
 159                                 line = line.substring(pos + 1);
 160                                 pos = line.indexOf('<');
 161                                 if (pos >= 0) {
 162                                         return line.substring(0, pos).trim();
 163                                 }
 164                         }
 165                 }
 166
 167                 return null;
 168         }
 169
 170         @Override
 171         protected String getDesc(URL source, InputStream in) {
 172                 // the og: meta version is the SHORT resume, this is the LONG resume
 173                 return getLine(in, "class=\"description-text bbcode\"", 1);
 174         }
 175
 176         private BufferedImage getCover(InputStream in) {
 177                 // Note: the 'og:image' is the SMALL cover, not the full version
 178                 String cover = getLine(in, "class=\"story_container__story_image\"", 1);
 179                 if (cover != null) {
 180                         int pos = cover.indexOf('"');
 181                         if (pos >= 0) {
 182                                 cover = cover.substring(pos + 1);
 183                                 pos = cover.indexOf('"');
 184                                 if (pos >= 0) {
 185                                         cover = cover.substring(0, pos);
 186                                 }
 187                         }
 188                 }
 189
 190                 return getImage(this, null, cover);
 191         }
 192
 193         @Override
 194         protected List<Entry<String, URL>> getChapters(URL source, InputStream in,
 195                         Progress pg) {
 196                 List<Entry<String, URL>> urls = new ArrayList<Entry<String, URL>>();
 197                 @SuppressWarnings("resource")
 198                 Scanner scan = new Scanner(in, "UTF-8");
 199                 scan.useDelimiter("\\n");
 200                 boolean started = false;
 201                 while (scan.hasNext()) {
 202                         String line = scan.next().trim();
 203
 204                         if (!started) {
 205                                 started = line.equals("<!--Chapters-->");
 206                         } else {
 207                                 if (line.equals("</form>")) {
 208                                         break;
 209                                 }
 210
 211                                 if (line.startsWith("<a href=")
 212                                                 || line.contains("class=\"chapter-title\"")) {
 213                                         // Chapter name
 214                                         String name = line;
 215                                         int pos = name.indexOf('>');
 216                                         if (pos >= 0) {
 217                                                 name = name.substring(pos + 1);
 218                                                 pos = name.indexOf('<');
 219                                                 if (pos >= 0) {
 220                                                         name = name.substring(0, pos);
 221                                                 }
 222                                         }
 223                                         // Chapter content
 224                                         pos = line.indexOf('/');
 225                                         if (pos >= 0) {
 226                                                 line = line.substring(pos); // we take the /, not +1
 227                                                 pos = line.indexOf('"');
 228                                                 if (pos >= 0) {
 229                                                         line = line.substring(0, pos);
 230                                                 }
 231                                         }
 232
 233                                         try {
 234                                                 final String key = name;
 235                                                 final URL value = new URL("http://www.fimfiction.net"
 236                                                                 + line);
 237                                                 urls.add(new Entry<String, URL>() {
 238                                                         @Override
 239                                                         public URL setValue(URL value) {
 240                                                                 return null;
 241                                                         }
 242
 243                                                         @Override
 244                                                         public String getKey() {
 245                                                                 return key;
 246                                                         }
 247
 248                                                         @Override
 249                                                         public URL getValue() {
 250                                                                 return value;
 251                                                         }
 252                                                 });
 253                                         } catch (MalformedURLException e) {
 254                                                 Instance.syserr(e);
 255                                         }
 256                                 }
 257                         }
 258                 }
 259
 260                 return urls;
 261         }
 262
 263         @Override
 264         protected String getChapterContent(URL source, InputStream in, int number,
 265                         Progress pg) {
 266                 return getLine(in, "<div class=\"bbcode\">", 1);
 267         }
 268
 269         @Override
 270         protected boolean supports(URL url) {
 271                 return "fimfiction.net".equals(url.getHost())
 272                                 || "www.fimfiction.net".equals(url.getHost());
 273         }
 274 }