src/be/nikiroo/fanfix/supported/EHentai.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.List;
  10 import java.util.Map.Entry;
  11 import java.util.Scanner;
  12
  13 import be.nikiroo.fanfix.Instance;
  14 import be.nikiroo.fanfix.data.Chapter;
  15 import be.nikiroo.fanfix.data.MetaData;
  16 import be.nikiroo.fanfix.data.Story;
  17 import be.nikiroo.utils.Progress;
  18 import be.nikiroo.utils.StringUtils;
  19
  20 /**
  21  * Support class for <a href="https://e-hentai.org/">e-hentai.org</a>, a website
  22  * supporting mostly but not always NSFW comics, including some of MLP.
  23  *
  24  * @author niki
  25  */
  26 class EHentai extends BasicSupport {
  27         @Override
  28         public String getSourceName() {
  29                 return "e-hentai.org";
  30         }
  31
  32         @Override
  33         protected MetaData getMeta(URL source, InputStream in) throws IOException {
  34                 MetaData meta = new MetaData();
  35
  36                 meta.setTitle(getTitle(reset(in)));
  37                 meta.setAuthor(getAuthor(reset(in)));
  38                 meta.setDate("");
  39                 meta.setTags(getTags(reset(in)));
  40                 meta.setSource(getSourceName());
  41                 meta.setUrl(source.toString());
  42                 meta.setPublisher(getSourceName());
  43                 meta.setUuid(source.toString());
  44                 meta.setLuid("");
  45                 meta.setLang("EN");
  46                 meta.setSubject("Furry");
  47                 meta.setType(getType().toString());
  48                 meta.setImageDocument(true);
  49                 meta.setCover(getCover(source, reset(in)));
  50                 meta.setFakeCover(true);
  51
  52                 return meta;
  53         }
  54
  55         @Override
  56         public Story process(URL url, Progress pg) throws IOException {
  57                 // There is no chapters on e621, just pagination...
  58                 Story story = super.process(url, pg);
  59
  60                 Chapter only = new Chapter(1, null);
  61                 for (Chapter chap : story) {
  62                         only.getParagraphs().addAll(chap.getParagraphs());
  63                 }
  64
  65                 story.getChapters().clear();
  66                 story.getChapters().add(only);
  67
  68                 return story;
  69         }
  70
  71         @Override
  72         protected boolean supports(URL url) {
  73                 return "e-hentai.org".equals(url.getHost());
  74         }
  75
  76         @Override
  77         protected boolean isHtml() {
  78                 return true;
  79         }
  80
  81         private BufferedImage getCover(URL source, InputStream in)
  82                         throws IOException {
  83                 BufferedImage author = null;
  84                 String coverLine = getKeyLine(in, "<div id=\"gd1\"", " url(", ")");
  85                 if (coverLine != null) {
  86                         coverLine = StringUtils.unhtml(coverLine).trim();
  87                         author = getImage(this, source, coverLine);
  88                 }
  89
  90                 return author;
  91         }
  92
  93         private String getAuthor(InputStream in) {
  94                 String author = null;
  95
  96                 List<String> tagsAuthor = getTagsAuthor(in);
  97                 if (!tagsAuthor.isEmpty()) {
  98                         author = tagsAuthor.get(0);
  99                 }
 100
 101                 return author;
 102         }
 103
 104         private List<String> getTags(InputStream in) {
 105                 List<String> tags = new ArrayList<String>();
 106                 List<String> tagsAuthor = getTagsAuthor(in);
 107
 108                 for (int i = 1; i < tagsAuthor.size(); i++) {
 109                         tags.add(tagsAuthor.get(i));
 110                 }
 111
 112                 return tags;
 113         }
 114
 115         private List<String> getTagsAuthor(InputStream in) {
 116                 List<String> tags = new ArrayList<String>();
 117                 String tagLine = getKeyLine(in, "<meta name=\"description\"", "Tags: ",
 118                                 null);
 119                 if (tagLine != null) {
 120                         for (String tag : tagLine.split(",")) {
 121                                 String candi = tag.trim();
 122                                 if (!candi.isEmpty() && !tags.contains(candi)) {
 123                                         tags.add(candi);
 124                                 }
 125                         }
 126                 }
 127
 128                 return tags;
 129         }
 130
 131         private String getTitle(InputStream in) throws IOException {
 132                 String siteName = " - E-Hentai Galleries";
 133
 134                 String title = getLine(in, "<title>", 0);
 135                 if (title != null) {
 136                         title = StringUtils.unhtml(title).trim();
 137                         if (title.endsWith(siteName)) {
 138                                 title = title.substring(0, title.length() - siteName.length())
 139                                                 .trim();
 140                         }
 141                 }
 142
 143                 return title;
 144         }
 145
 146         @Override
 147         protected String getDesc(URL source, InputStream in) throws IOException {
 148                 String desc = null;
 149
 150                 String descLine = getKeyLine(in, "Uploader Comment", null,
 151                                 "<div class=\"c7\"");
 152                 if (descLine != null) {
 153                         desc = StringUtils.unhtml(descLine);
 154                 }
 155
 156                 return desc;
 157         }
 158
 159         @Override
 160         protected List<Entry<String, URL>> getChapters(URL source, InputStream in,
 161                         Progress pg) throws IOException {
 162                 List<Entry<String, URL>> urls = new ArrayList<Entry<String, URL>>();
 163                 int last = 0; // no pool/show when only one page, first page == page 0
 164
 165                 @SuppressWarnings("resource")
 166                 Scanner scan = new Scanner(in, "UTF-8");
 167                 scan.useDelimiter(">");
 168                 while (scan.hasNext()) {
 169                         String line = scan.next();
 170                         if (line.contains(source.toString())) {
 171                                 String page = line.substring(line.indexOf(source.toString()));
 172                                 String pkey = "?p=";
 173                                 if (page.contains(pkey)) {
 174                                         page = page.substring(page.indexOf(pkey) + pkey.length());
 175                                         String number = "";
 176                                         while (!page.isEmpty() && page.charAt(0) >= '0'
 177                                                         && page.charAt(0) <= '9') {
 178                                                 number += page.charAt(0);
 179                                                 page = page.substring(1);
 180                                         }
 181                                         if (number.isEmpty()) {
 182                                                 number = "0";
 183                                         }
 184
 185                                         int current = Integer.parseInt(number);
 186                                         if (last < current) {
 187                                                 last = current;
 188                                         }
 189                                 }
 190                         }
 191                 }
 192
 193                 for (int i = 0; i <= last; i++) {
 194                         final String key = Integer.toString(i + 1);
 195                         final URL value = new URL(source.toString() + "?p=" + i);
 196                         urls.add(new Entry<String, URL>() {
 197                                 public URL setValue(URL value) {
 198                                         return null;
 199                                 }
 200
 201                                 public URL getValue() {
 202                                         return value;
 203                                 }
 204
 205                                 public String getKey() {
 206                                         return key;
 207                                 }
 208                         });
 209                 }
 210
 211                 return urls;
 212         }
 213
 214         @Override
 215         protected String getChapterContent(URL source, InputStream in, int number,
 216                         Progress pg) throws IOException {
 217                 String staticSite = "https://e-hentai.org/s/";
 218                 List<URL> pages = new ArrayList<URL>();
 219
 220                 @SuppressWarnings("resource")
 221                 Scanner scan = new Scanner(in, "UTF-8");
 222                 scan.useDelimiter("\"");
 223                 while (scan.hasNext()) {
 224                         String line = scan.next();
 225                         if (line.startsWith(staticSite)) {
 226                                 try {
 227                                         pages.add(new URL(line));
 228                                 } catch (MalformedURLException e) {
 229                                         Instance.syserr(new IOException(
 230                                                         "Parsing error, a link is not correctly parsed: "
 231                                                                         + line, e));
 232                                 }
 233                         }
 234                 }
 235
 236                 if (pg == null) {
 237                         pg = new Progress();
 238                 }
 239                 pg.setMinMax(0, pages.size());
 240                 pg.setProgress(0);
 241
 242                 StringBuilder builder = new StringBuilder();
 243
 244                 for (URL page : pages) {
 245                         InputStream pageIn = Instance.getCache().open(page, this, false);
 246                         try {
 247                                 String link = getKeyLine(pageIn, "id=\"img\"", "src=\"", "\"");
 248                                 if (link != null && !link.isEmpty()) {
 249                                         builder.append("[");
 250                                         builder.append(link);
 251                                         builder.append("]<br/>");
 252                                 }
 253                                 pg.add(1);
 254                         } finally {
 255                                 if (pageIn != null) {
 256                                         pageIn.close();
 257                                 }
 258                         }
 259                 }
 260
 261                 pg.done();
 262                 return builder.toString();
 263         }
 264 }