Exemplo n.º 1
0
 @Override
 protected Boolean doInBackground(String... params) {
   try {
     Document doc = Jsoup.connect(params[0]).get();
     Element body = doc.body();
     Elements titleEs = body.select("td.title");
     Elements subTitleEs = body.select("td.subtext");
     int index = 1;
     if (!titleEs.isEmpty()) {
       if (mType == TYPE_REFRESH && mNews.size() > 0) {
         mNews.clear();
       }
       Iterator<Element> iterator = titleEs.iterator();
       Iterator<Element> subIt = subTitleEs.iterator();
       NewEntity entity = null;
       User user = null;
       while (iterator.hasNext()) {
         Element e = iterator.next();
         if (index % 2 == 0) {
           Element subE = subIt.next();
           Elements aTag = e.select("a");
           Elements spanTag = e.select("span.comhead");
           Elements subEa = subE.select("a");
           user = new User();
           user.setId(subEa.get(0).text());
           entity =
               new NewEntity(
                   aTag.get(0).attr("href"),
                   aTag.get(0).text(),
                   spanTag.isEmpty() ? null : spanTag.get(0).text(),
                   subE.html());
           entity.setDiscussUrl(subEa.get(1).attr("href"));
           // Log.i(LOG_TAG, entity.toString());
           mNews.add(entity);
         }
         index++;
       }
     }
     Elements more = doc.getElementsByAttributeValueStarting("href", "/x?fnid=");
     if (!more.isEmpty()) {
       mMoreURLPath = more.get(1).attr("href");
     }
     return true;
   } catch (IOException e) {
     Log.e(LOG_TAG, "", e);
     return false;
   }
 }
  @Scheduled(fixedDelay = 900000)
  public void loadBooksInfo() {
    int booksListPageNumber = 1;
    while (true) {
      Document document = null;
      try {
        URL url =
            new URL(
                "http://www.labirint.ru/genres/2308/?page="
                    + Integer.toString(booksListPageNumber++));
        document = Jsoup.parse(url, 5000);
      } catch (MalformedURLException ex) {

      } catch (IOException ex) {

      }
      Elements elements = document.getElementsByClass("product");
      if (elements.size() == 0) break;
      Iterator<Element> iterator = elements.iterator();
      while (iterator.hasNext()) {
        try {
          bookDao.saveBook(bookInfo(iterator.next()));
        } catch (Exception ex) {
          if (ex instanceof ConstraintViolationException) continue;
        }
      }
    }
  }
Exemplo n.º 3
0
  // Busca os endereços pelo número do CEP.
  public List<Address> getByCep(String cep) throws IOException {

    listEnderecos = new ArrayList<Address>();

    // mapeamento dos parametros que será passado na requisição
    Map<String, String> query = new HashMap<String, String>();
    query.put("CEP", cep);
    query.put("Metodo", "listaLogradouro");
    query.put("TipoConsulta", "cep");
    query.put("StartRow", "1");
    query.put("EndRow", "10");

    // Faz uma requisição no site do correios (www.buscacep.com.br) com Json, passando os parametros
    // mapeados,
    // requisição deverá ser do tipo post.
    // Armazena o retorno em uma variavel doc.
    Document doc =
        Jsoup.connect(Utils.adressCorreios)
            .data(query)
            .header("Origin", "http://www.buscacep.correios.com.br")
            .header("Referer", "http://www.buscacep.correios.com.br")
            .post();

    // Acessa o retorno do doc e percorre o resultado buscando as informações dos endereços
    // Armazena os resultados na lista de endereços criadas e retorna a mesma para que outras
    // classes possam acessar.
    Elements elements = doc.select("table").eq(2);
    Elements rows = elements.select("tr");

    Iterator<Element> rowIterator = rows.iterator();

    while (rowIterator.hasNext()) {
      Address enderecos = new Address();

      Element element = rowIterator.next();

      Elements logradouro = element.children().select("td").eq(0);
      enderecos.setLogradouro(logradouro.text());
      Elements bairro = element.children().select("td").eq(1);
      enderecos.setBairro(bairro.text());
      Elements cidade = element.children().select("td").eq(2);
      Elements estado = element.children().select("td").eq(3);
      StringBuilder sbLocalidade = new StringBuilder();
      sbLocalidade.append(cidade.text());
      sbLocalidade.append("/");
      sbLocalidade.append(estado.text());
      enderecos.setLocalidade(sbLocalidade.toString());
      Elements codigopostal = element.children().select("td").eq(4);
      enderecos.setCEP(codigopostal.text());

      listEnderecos.add(enderecos);
    }

    return listEnderecos;
  }
  private List<Comment> parseComments(Element commentsElement, Book book) {
    List<Comment> comments = new ArrayList<Comment>();
    Elements elements = commentsElement.getElementsByClass("product-comment");
    Iterator<Element> iterator = elements.iterator();
    String str = "";
    while (iterator.hasNext()) {
      Comment comment = new Comment();
      Element element = iterator.next();
      try {
        str =
            element
                .getElementsByClass("comment-user-avatar")
                .get(0)
                .getElementsByTag("a")
                .get(0)
                .attr("title");
        comment.setAuthor((str.length() <= 100) ? str : str.substring(0, 99));
      } catch (Exception ex) {
        comment.setAuthor("");
      }

      try {
        SimpleDateFormat simpleDateFormat = new SimpleDateFormat("dd.mm.yyyy HH:mm:ss");
        str =
            element
                .getElementsByClass("comment-footer")
                .get(0)
                .getElementsByClass("date")
                .get(0)
                .text()
                .trim();
        comment.setDate(new Date(simpleDateFormat.parse(str).getTime()));
      } catch (Exception ex) {
        comment.setDate(null);
      }
      try {
        if (element.getElementsByAttributeValueContaining("id", "fullcomment").size() > 0) {
          str = element.getElementsByAttributeValueContaining("id", "fullcomment").get(0).text();
          comment.setComment((str.length() <= 5000) ? str : str.substring(0, 4999));
        } else if (element.getElementsByClass("comment-text").size() > 0) {
          str = element.getElementsByClass("comment-text").get(0).text();
          comment.setComment((str.length() <= 5000) ? str : str.substring(0, 4999));
        } else {
          str = element.getElementsByAttributeValueContaining("id", "shortcomment").get(0).text();
          comment.setComment((str.length() <= 5000) ? str : str.substring(0, 4999));
        }
      } catch (Exception ex) {
        comment.setComment("");
      }
      comment.setBook(book);
      comments.add(comment);
    }
    return comments;
  }
Exemplo n.º 5
0
  /**
   * Extract content with jsoup maybe later.
   *
   * @param doc
   * @return
   */
  public static List<Item> extractItem(Document doc) {
    List<Item> itemList = new ArrayList<Item>();
    Elements itemRows = doc.select("tr");
    Iterator iterator = itemRows.iterator();
    while (iterator.hasNext()) {
      Element element = (Element) iterator.next();
      Element titleElement = element.select(".title a").first();
      if (titleElement == null) {
        continue;
      }
      String titleStr = titleElement.text().trim();
      String urlStr = titleElement.attr("href").trim();

      Element comHeadElement = element.select(".comhead").first();
      if (comHeadElement == null) {
        continue;
      }

      String comheadStr = comHeadElement.text().trim();

      Element pointsElement = element.select("span[id^=score_]").first();
      if (pointsElement == null) {
        continue;
      }
      String pointsStr = pointsElement.text();
      if (pointsStr == null) {
        continue;
      }
      String[] pointsArr = pointsStr.split(" ");
      if (pointsArr.length != 2) {
        continue;
      }
      int points = -1;
      try {
        points = Integer.parseInt(pointsArr[0]);
      } catch (NumberFormatException e) {
      }

      if (points < 0) {
        continue;
      }
      Element userElement = element.select("a[href^=user]").first();
      if (userElement == null) {
        continue;
      }

      String user = userElement.text().trim();

      Element dateElement = element.select(".subtext").first();
    }

    return itemList;
  }
Exemplo n.º 6
0
  // Busca o Cep pelo logradouro.
  public List<String> getByAdress(String address) throws IOException {

    listAddress = new ArrayList<String>();

    // mapeamento dos parametros que será passado na requisição
    Map<String, String> query = new HashMap<String, String>();

    query.put("relaxation", address);
    query.put("TipoCep", "ALL");
    query.put("semelhante", "N");
    query.put("cfm", "1");
    query.put("Metodo", "listaLogradouro");
    query.put("TipoConsulta", "relaxation");
    query.put("StartRow", "1");
    query.put("EndRow", "10");

    // Faz uma requisição no site do correios (www.buscacep.com.br) com Json, passando os parametros
    // mapeados,
    // requisição deverá ser do tipo post.
    // Armazena o retorno em uma variavel doc.
    Document doc =
        Jsoup.connect(Utils.adressCorreios)
            .timeout(20000)
            .data(query)
            .header("Origin", "http://www.buscacep.correios.com.br")
            .header("Referer", "http://www.buscacep.correios.com.br")
            .post();

    // Acessa o retorno do doc e percorre o resultado buscando as informações de Cep de acordo com o
    // endereço passado.
    // Armazena os resultados na lista criada e retorna a mesma para que outras classes possam
    // acessar
    Elements elements = doc.select("table").eq(2);
    Elements rows = elements.select("tr");

    Iterator<Element> rowIterator = rows.iterator();

    while (rowIterator.hasNext()) {
      Address enderecos = new Address();

      Element element = rowIterator.next();

      Elements codigopostal = element.children().select("td").eq(4);

      enderecos.setCEP(codigopostal.text());

      listAddress.add(enderecos.getCEP());
    }

    return listAddress;
  }
Exemplo n.º 7
0
 static {
   try {
     locData = Jsoup.parse(new FileInputStream(LocLabel.locFile), "UTF-8", "", Parser.xmlParser());
     locStrList = new LinkedList<String>();
     Elements locElements = locData.select("[Name]");
     Iterator<Element> eleIterator = locElements.iterator();
     while (eleIterator.hasNext()) {
       Element curElement = eleIterator.next();
       locStrList.add(curElement.attr("Name"));
     }
   } catch (IOException e) {
     e.printStackTrace();
   }
 }
Exemplo n.º 8
0
  private static void alterElement(Element e) {
    org.jsoup.select.Elements s = e.children();
    Iterator<Element> ele = s.iterator();
    int i = 0;
    while (ele.hasNext()) {
      Element r = ele.next();

      if (!r.tag().getName().equals("p")) {
        r.tagName("p"); // plain replace
        //				Element rtemp = r.clone();
        //				Element ep = new Element(Tag.valueOf("p"), "");
        //				ep.appendChild(rtemp);
        //				r.replaceWith(ep);
        //				StringBuffer bf = new StringBuffer();
        //				bf.append("<k>").append(r.toString()).append("</k>");
        //				r.html(bf.toString());
        //				System.out.println(r.tagName());

      }
      i++;
    }
  }
Exemplo n.º 9
0
  public boolean scrapInternal(String url, String marker) throws IOException {

    Document document;
    try {
      document = Jsoup.connect(url).get();
    } catch (IOException e) {

      // try again
      document = Jsoup.connect(url).get();
    }

    Elements hrefElements = document.select("a");

    Iterator<Element> eltIterator = hrefElements.iterator();

    boolean postFound = false;
    while (eltIterator.hasNext()) {
      String href = eltIterator.next().attr("href");
      if (StringUtil.isBlank(href)) {
        continue;
      } else if (parser.isPostPage(href)) {
        try {
          BlogPost blogPost = parser.getRepository().findOneByUrl(href);
          if (blogPost == null) {
            scrapBlogPost(href, null, false, marker);
            postFound = true;
          } else if (!marker.equals(blogPost.getMarker())) {
            scrapBlogPost(href, blogPost, true, marker);
          }
        } catch (Exception exp) {
          exp.printStackTrace();
        }

      } else if (parser.isListPage(href) && !summaryQueue.contains(href)) {
        summaryQueue.add(href);
      }
    }
    return postFound;
  }
Exemplo n.º 10
0
 Hashtable<String, String> getFilters(String nodeId) {
   Hashtable<String, String> sdata = new Hashtable<String, String>();
   Elements ttFilterNodes =
       baseNode.select("#doc #bd-wrap #bd #sidebar .bd #" + nodeId + " .default_show");
   for (Iterator<Element> iterator = ttFilterNodes.iterator(); iterator.hasNext(); ) {
     String key = "", value = "";
     Element liElement = (Element) iterator.next();
     if (liElement.getElementsByTag("a") != null && liElement.getElementsByTag("a").size() > 0) {
       Element link = liElement.getElementsByTag("a").get(0);
       key = link.text().toString();
     } else {
       Element activeNode = liElement.getElementsByClass("active").get(0);
       key = activeNode.text().toString();
     }
     if (liElement.getElementsByClass("refiner_count") != null
         && liElement.getElementsByClass("refiner_count").size() > 0) {
       Element countNode = liElement.getElementsByClass("refiner_count").get(0);
       value = countNode.text().toString().replace("(", "").replace(")", "");
     }
     sdata.put(key, value);
   }
   return sdata;
 }
Exemplo n.º 11
0
  public List<Standing> parse() throws IOException {
    List<Standing> ret = new ArrayList<Standing>();

    Database db = Database.getInstance();

    Document doc;
    try {
      doc = Jsoup.parse(new URL(url), 5000);
    } catch (Exception ex) {
      logger.warn("Failed fetching standings at " + url, ex);
      return ret;
    }

    Elements rows = doc.select("div.Section1 table");

    Iterator<Element> tableIter = rows.iterator();
    tableIter.next(); // skip 2, selector won't work for some reason
    tableIter.next();

    Element table = tableIter.next();

    for (Element row : table.select("tr:gt(1)")) { // skip first 2 headers
      AscentStanding s = (AscentStanding) game.createStanding();

      // 0: rank
      // 1: team #
      // 2: qualification score
      // 3: AP (?)
      // 4: coopertition points
      // 5: teleop points
      // 6: record (wins-losses-ties)
      // 7 disqualifications
      // 8: matches played

      s.setRank(Integer.parseInt(row.child(0).text()));

      Team team = db.getTeam(Integer.parseInt(row.child(1).text()));
      if (team == null) {
        logger.warn("Unknown team " + row.child(1).text() + ", skipping");
        continue;
      }
      s.setTeam(game.getEntry(team));

      s.setQualificationScore(Float.parseFloat(row.child(2).text()));
      // todo: old/invalid!
      // s.setHybridPoints(Float.parseFloat(row.child(3).text()));
      // s.setBridgePoints(Float.parseFloat(row.child(4).text()));
      s.setTeleopPoints(Float.parseFloat(row.child(5).text()));

      s.setCoopertitionPoints((int) Float.parseFloat(row.child(4).text()));

      String[] wlt = StringUtils.split(row.child(6).text(), '-');
      s.setWins(Integer.parseInt(wlt[0]));
      s.setLosses(Integer.parseInt(wlt[1]));
      s.setTies(Integer.parseInt(wlt[2]));

      s.setDisqualifications(Integer.parseInt(row.child(7).text()));
      s.setMatchesPlayed(Integer.parseInt(row.child(8).text()));

      ret.add(s);
    }

    return ret;
  }
Exemplo n.º 12
0
  public void onPostExecute(AnswerObject result) {
    Document doc = Jsoup.parse(result.getHTML());
    sendHTMLatBug(doc.html());
    if (doc.select("span.notLoggedText").text().length() > 0) {
      Intent BackToLoginIntent = new Intent(this, TuCanMobileActivity.class);
      BackToLoginIntent.putExtra("lostSession", true);
      startActivity(BackToLoginIntent);
    } else {
      if (PREPCall == false) {
        String Title = doc.select("h1").text();

        TextView SingleEventTitle = (TextView) findViewById(R.id.singleevent_title);
        SingleEventTitle.setText(Title);
        Elements Deltarows = doc.select("table[courseid]").first().select("tr");
        Element rows;
        if (Deltarows.size() == 1) {
          rows = Deltarows.get(0).select("td").first();
        } else {
          rows = Deltarows.get(1).select("td").first();
        }

        Elements Paragraphs = rows.select("p");
        Iterator<Element> PaIt = Paragraphs.iterator();
        ArrayList<String> titles = new ArrayList<String>();
        ArrayList<String> values = new ArrayList<String>();

        while (PaIt.hasNext()) {

          Element next = PaIt.next();
          String[] information = crop(next.html());
          titles.add(information[0]);
          values.add(information[1]);
        }

        PropertyValueAdapter = new SingleEventAdapter(titles, values);
        setListAdapter(PropertyValueAdapter);

        // Termin-Selektor:
        // Terminselektor

        Iterator<Element> captionIt = doc.select("caption").iterator();
        Iterator<Element> DateTable = null;
        Iterator<Element> materialTable = null;
        while (captionIt.hasNext()) {
          Element next = captionIt.next();
          if (next.text().equals("Termine")) {
            System.out.println(next.parent().html());
            DateTable = next.parent().select("tr").iterator();
          } else if (next.text().contains("Material")) {

            materialTable = next.parent().select("tr").iterator();
          }
        }
        ArrayList<String> eventNumber = new ArrayList<String>();
        ArrayList<String> eventDate = new ArrayList<String>();
        ArrayList<String> eventTime = new ArrayList<String>();

        ArrayList<String> eventRoom = new ArrayList<String>();
        ArrayList<String> eventInstructor = new ArrayList<String>();

        while (DateTable.hasNext()) {
          Element next = DateTable.next();
          Elements cols = next.select("td");
          eventNumber.add(cols.get(0).text());
          eventDate.add(cols.get(1).text());
          eventTime.add(cols.get(2).text() + "-" + cols.get(3).text());
          eventRoom.add(cols.get(4).text());
          eventInstructor.add(cols.get(5).text());
        }

        DateAppointmentAdapter =
            new AppointmentAdapter(eventDate, eventTime, eventNumber, eventRoom, eventInstructor);

        int ct = 0;
        ArrayList<String> materialNumber = new ArrayList<String>();
        ArrayList<String> materialName = new ArrayList<String>();
        ArrayList<String> materialDesc = new ArrayList<String>();
        materialLink = new ArrayList<String>();
        ArrayList<String> materialFile = new ArrayList<String>();
        if (materialTable != null) {
          while (materialTable.hasNext()) {
            Element next = materialTable.next();

            if (next.select("td").size() > 1) {
              ct++;
              System.out.println(ct + "  " + (ct % 3));
              int mod = (ct % 3);
              switch (mod) {
                case 1:
                  materialNumber.add(next.select("td").get(0).text());
                  materialName.add(next.select("td").get(1).text());

                  break;
                case 2:
                  materialDesc.add(next.select("td").get(1).text());
                  if (next.attr("class").equals("tbdata_nob")) {
                    ct++;
                    materialLink.add("");
                    materialFile.add("");
                  }
                  break;
                case 0:
                  materialLink.add(next.select("td").get(1).select("a").attr("href"));
                  materialFile.add(next.select("td").get(1).select("a").text());
                  break;
              }
            }
          }
        }

        if (ct > 2) {
          FileAdapter =
              new AppointmentAdapter(
                  materialNumber, materialFile, null, materialName, materialDesc);
          thereAreFiles = true;
        } else
          FileAdapter =
              new ArrayAdapter<String>(
                  this, android.R.layout.simple_list_item_1, new String[] {"Kein Material"});

      } else {
        String nextlink =
            TucanMobile.TUCAN_PROT
                + TucanMobile.TUCAN_HOST
                + doc.select("div.detailout").select("a").attr("href");
        SimpleSecureBrowser callOverviewBrowser = new SimpleSecureBrowser(this);
        RequestObject thisRequest =
            new RequestObject(nextlink, localCookieManager, RequestObject.METHOD_GET, "");
        PREPCall = false;
        callOverviewBrowser.execute(thisRequest);
      }
    }
  }
Exemplo n.º 13
0
  public DataRecipe getRecipeData() {
    DataRecipe output = new DataRecipe();
    Hashtable<String, String> sdata;
    int i = 0;
    // Element baseNodeE = baseNode.select("body").first();
    String img;
    Elements baseNodes =
        baseNode.select("#doc #bd-wrap #bd #results #cols #left #main #web ol .recipe-article");
    for (Iterator<Element> iterator = baseNodes.iterator(); iterator.hasNext(); ) {
      sdata = new Hashtable<String, String>();
      Element divElement = (Element) iterator.next();
      Elements imgElements = divElement.select("img");
      if (imgElements != null && imgElements.size() > 0) {
        img = imgElements.get(0).attr("data-src");
        int ind = img.indexOf("http", 2);
        if (-1 != ind) {
          img = img.substring(ind);
        }
      } else {
        img = "http://illpop.com/img_illust/food/t_rice_m28.jpg";
      }
      sdata.put("image", img);

      Elements titleLink = divElement.select(".yschttl");
      String title =
          titleLink
              .get(0)
              .text()
              .toString()
              .replace("<b>", "")
              .replace("</b>", "")
              .replace("\n", "");
      String url = titleLink.get(0).attr("href");
      sdata.put("title", title);
      int ind = url.indexOf("http", 2);
      if (-1 != ind) {
        url = URLDecoder.decode(url.substring(ind));
      }
      sdata.put("URL", url);

      Elements totalTimeNode = divElement.select(".totaltime .attr-content");
      String totaltime = "";
      if (totalTimeNode != null && totalTimeNode.size() > 0) {
        totaltime =
            totalTimeNode
                .get(0)
                .text()
                .toString()
                .replace("</span>", "")
                .replace("<span class=\"num\">", "")
                .replace("\n", "");
      }
      sdata.put("duration", totaltime);
      sdata.put("rating", "");
      Elements ingredientsNode = divElement.select(".ingredients .attr-content");
      String ingredients = "";
      if (ingredientsNode != null && ingredientsNode.size() > 0) {
        ingredients = ingredientsNode.get(0).text().toString();
      }
      sdata.put("ingredients", ingredients);
      output.put(i++, sdata);
    }
    return output;
  }
  protected Book bookInfo(Element bookIcon) {
    Document document = null;
    String productId = bookIcon.attr("data-product-id");
    try {
      URL url = new URL("http://www.labirint.ru/books/" + productId);
      document = Jsoup.parse(url, 5000);
    } catch (MalformedURLException ex) {

    } catch (IOException ex) {

    }
    Book book = new Book();
    Element bookElement = document.getElementById("product");

    String[] strs;
    String str;

    book.setProductId(Integer.parseInt(productId));

    try {
      String name;
      name = bookElement.getElementById("product-title").getElementsByTag("h1").text().trim();
      book.setName((name.length() <= 150) ? name : name.substring(0, 149));
    } catch (Exception ex) {
      book.setName("");
    }

    try {
      String author = "";
      String editor = "";
      Elements authorElements =
          bookElement.getElementById("product-specs").getElementsByClass("authors");
      Iterator<Element> iterator = authorElements.iterator();
      while (iterator.hasNext()) {
        str = iterator.next().text();
        if (str.contains("Автор")) {
          strs = str.split(":");
          if (strs.length == 2) author = strs[1].replaceAll("\"", "").trim();
          else author = strs[0].replaceAll("\"", "").trim();
        }
        if (str.contains("Редактор")) {
          strs = str.split(":");
          if (strs.length == 2) editor = strs[1].replaceAll("\"", "").trim();
          else editor = strs[0].replaceAll("\"", "").trim();
        }
      }
      book.setAuthor((author.length() <= 100) ? author : author.substring(0, 99));
      book.setEditor((editor.length() <= 100) ? editor : editor.substring(0, 99));
    } catch (Exception ex) {
      book.setAuthor("");
      book.setEditor("");
    }

    try {
      String publisherAndYear = "";
      str =
          bookElement.getElementById("product-specs").getElementsByClass("publisher").get(0).text();
      strs = str.split(":");
      if (strs.length == 2) publisherAndYear = strs[1].replaceAll("\"", "").trim();
      else publisherAndYear = strs[0].replaceAll("\"", "").trim();
      book.setPublisherAndYear(
          (publisherAndYear.length() <= 200)
              ? publisherAndYear
              : publisherAndYear.substring(0, 199));
    } catch (Exception ex) {
      book.setPublisherAndYear("");
    }

    try {
      if (bookElement
              .getElementById("product-specs")
              .getElementsByClass("buying-pricenew-val-number")
              .size()
          > 0)
        str =
            bookElement
                .getElementById("product-specs")
                .getElementsByClass("buying-pricenew-val-number")
                .get(0)
                .text();
      else
        str =
            bookElement
                .getElementById("product-specs")
                .getElementsByClass("buying-price-val-number")
                .get(0)
                .text();
      book.setPrice(Double.parseDouble(str.length() <= 10 ? str : str.substring(0, 9)));
    } catch (Exception ex) {
      book.setPrice(null);
    }

    try {
      String isbn = "";
      str = bookElement.getElementById("product-specs").getElementsByClass("isbn").get(0).text();
      strs = str.split(":");
      if (strs.length == 2) isbn = strs[1].replaceAll("\"", "").trim();
      else isbn = strs[0].replaceAll("\"", "").trim();
      book.setIsbn((isbn.length() <= 20) ? isbn : isbn.substring(0, 19));
    } catch (Exception ex) {
      book.setIsbn("");
    }

    try {
      String pagesCount = "";
      str = bookElement.getElementById("product-specs").getElementsByClass("pages2").get(0).text();
      strs = str.split(":");
      if (strs.length == 2) pagesCount = strs[1].replaceAll("\"", "").trim();
      else pagesCount = strs[0].replaceAll("\"", "").trim();
      book.setPagesCount((pagesCount.length() <= 30) ? pagesCount : pagesCount.substring(0, 29));
    } catch (Exception ex) {
      book.setPagesCount("");
    }

    try {
      String decor = "";
      Document decorHtml =
          Jsoup.parse(new URL("http://www.labirint.ru/ajax/design/" + productId), 5000);
      book.setDecor(
          (decorHtml.text().length() <= 400)
              ? decorHtml.text()
              : decorHtml.text().substring(0, 399));
    } catch (Exception ex) {
      book.setDecor("");
    }

    try {
      String weight = "";
      str = bookElement.getElementById("product-specs").getElementsByClass("weight").get(0).text();
      strs = str.split(":");
      if (strs.length == 2) weight = strs[1].replaceAll("\"", "").trim();
      else weight = strs[0].replaceAll("\"", "").trim();
      book.setWeight((weight.length() <= 20) ? weight : weight.substring(0, 19));
    } catch (Exception ex) {
      book.setWeight("");
    }

    try {
      String dimensions = "";
      str =
          bookElement
              .getElementById("product-specs")
              .getElementsByClass("dimensions")
              .get(0)
              .text();
      strs = str.split(":");
      if (strs.length == 2) dimensions = strs[1].replaceAll("\"", "").trim();
      else dimensions = strs[0].replaceAll("\"", "").trim();
      book.setDimensions((dimensions.length() <= 30) ? dimensions : dimensions.substring(0, 29));
    } catch (Exception ex) {
      book.setDimensions("");
    }

    try {
      if (bookElement.getElementById("fullannotation") != null)
        book.setAnnotation(
            (bookElement.getElementById("fullannotation").text().length() <= 2000)
                ? bookElement.getElementById("fullannotation").text()
                : bookElement.getElementById("fullannotation").text().substring(0, 1999));
      else if (bookElement.getElementById("product-about") != null)
        book.setAnnotation(
            (bookElement.getElementById("product-about").text().length() <= 2000)
                ? bookElement.getElementById("product-about").text()
                : bookElement.getElementById("product-about").text().substring(0, 1999));
      else
        book.setAnnotation(
            (bookElement.getElementById("smallannotation").text().length() <= 2000)
                ? bookElement.getElementById("smallannotation").text()
                : bookElement.getElementById("smallannotation").text().substring(0, 1999));

    } catch (Exception ex) {
      book.setAnnotation("");
    }

    try {
      String imageUrl = "";
      if (document.getElementsByAttributeValue("property", "og:image") != null)
        imageUrl =
            document.getElementsByAttributeValue("property", "og:image").get(0).attr("content");
      else imageUrl = "http://img.labirint.ru/design/emptycover.png";
      book.setCoverImgUrl(imageUrl);
    } catch (Exception ex) {
      book.setCoverImgUrl("");
    }

    if (document.getElementById("product-comments") != null)
      book.setComments(parseComments(document.getElementById("product-comments"), book));

    return book;
  }