private boolean recuperaDadosAluno(String html) {
      HtmlCleaner cleaner = new HtmlCleaner();
      TagNode root = cleaner.clean(html);

      String table_name = context.getString(R.string.janusmob_table_ficha_aluno_id);

      TagNode table = null;
      for (TagNode node : root.getElementsByName("table", true)) {
        if (table_name.equals(node.getAttributeByName("id"))) {
          table = node;
          break;
        }
      }

      if (table == null) {
        return false;
      }

      String attribute_class_name = context.getString(R.string.janusmob_attribute_td_class_name);
      List<Campo> l = new LinkedList<Campo>();
      TagNode nodeValor = null;
      for (TagNode nodeCampo : table.getElementsByName("td", true)) {
        if (attribute_class_name.equals(nodeCampo.getAttributeByName("class"))) {
          nodeValor = nodeCampo.getParent().getElementsByName("td", false)[1];
          l.add(
              new Campo(
                  nodeCampo.getText().toString().split(":")[0],
                  nodeValor.getText().toString().trim().replaceAll("\"", "")));
        }
      }

      dbAdapter.setCampos(l);

      return true;
    }
    private boolean recuperaDadosPessoais(String html) {
      HtmlCleaner cleaner = new HtmlCleaner();
      TagNode root = cleaner.clean(html);

      String table_name = context.getString(R.string.janusmob_table_dados_aluno_id);

      TagNode table = null;
      for (TagNode node : root.getElementsByName("table", true)) {
        if (table_name.equals(node.getAttributeByName("id"))) {
          table = node;
          break;
        }
      }

      if (table == null) {
        return false;
      }

      String attribute_class_name = context.getString(R.string.janusmob_attribute_class_name);
      String campo_nusp = context.getString(R.string.janusmob_campo_nusp);
      String campo_nome = context.getString(R.string.janusmob_campo_nome);
      String campo_email = context.getString(R.string.janusmob_campo_email);
      String campo_cpf = context.getString(R.string.janusmob_campo_cpf);
      Usuario usuario = new Usuario();
      for (TagNode node : table.getElementsByName("span", true)) {
        if (attribute_class_name.equals(node.getAttributeByName("class"))) {
          TagNode parent = node.getParent();
          parent.removeChild(node);
          if (node.getText().toString().trim().equals(campo_nusp)) {
            usuario.setNusp(parent.getText().toString().trim());
          } else if (node.getText().toString().trim().equals(campo_nome)) {
            usuario.setNome(parent.getText().toString().trim());
          } else if (node.getText().toString().trim().equals(campo_email)) {
            usuario.setEmail(parent.getText().toString().trim());
          } else if (node.getText().toString().trim().equals(campo_cpf)) {
            usuario.setCpf(parent.getText().toString().trim());
          }
        }
      }

      if (lembrarSenha) {
        usuario.setSenha(senha);
      } else {
        usuario.setSenha(null);
      }

      dbAdapter.setUsuario(usuario);

      return true;
    }
    /**
     * @param html
     * @return <code>true</code> se a página possui formulário de login. <code>false</code> caso
     *     contrário.
     */
    private boolean autenticaUsuario(String html) {
      HtmlCleaner cleaner = new HtmlCleaner();
      TagNode root = cleaner.clean(html);

      String formulario_name = context.getString(R.string.janusmob_formulario);
      String campo_usuario_name = context.getString(R.string.janusmob_campo_usuario);
      String campo_senha_name = context.getString(R.string.janusmob_campo_senha);

      TagNode loginform = null;
      for (TagNode node : root.getElementsByName("form", true)) {
        if (node.getAttributeByName("name").equals(formulario_name)) {
          loginform = node;
          break;
        }
      }

      if (loginform == null) {
        return false;
      }

      // }
      // StringBuffer postData = new StringBuffer(String.format(
      // "%s=%s&%s=%s", campo_usuario_name, usuario,
      // campo_senha_name, senha));
      //
      // for (TagNode node : loginform.getElementsByName("input", true)) {
      // if (!(node.getAttributeByName("name")
      // .equals(campo_usuario_name) || node.getAttributeByName(
      // "name").equals(campo_senha_name))) {
      // postData.append("&")
      // .append(node.getAttributeByName("name"))
      // .append("=")
      // .append(node.getAttributeByName("value"));
      // }
      // }
      //
      // webView.postUrl(
      // context.getString(R.string.janusmob_pagina_login),
      // EncodingUtils.getBytes(postData.toString(), "base64"));

      webView.loadUrl(
          String.format(
              "javascript:document.getElementById(\"%s\").value=%s;", campo_senha_name, senha));

      webView.loadUrl(
          String.format(
              "javascript:document.getElementById(\"%s\").value=%s;", campo_usuario_name, usuario));

      webView.loadUrl(
          String.format(
              "javascript:document.getElementById('%s').submit();",
              context.getString(R.string.janusmob_formulario)));

      return true;
    }
 private String getMensagemErro(String html) {
   TagNode root = new HtmlCleaner().clean(html);
   String idSpanMensagemErro = context.getString(R.string.janusmob_span_erro_id);
   TagNode[] tags = root.getElementsByName("span", true);
   for (TagNode node : tags) {
     if (idSpanMensagemErro.equals(node.getAttributeByName("id"))) {
       return node.getText().toString();
     }
   }
   return context.getString(R.string.janusmob_mensagem_erro_login);
 }
 private boolean usuarioLogado(String html) {
   TagNode root = new HtmlCleaner().clean(html);
   String idLinkLogout = context.getString(R.string.janusmob_link_logout_id);
   TagNode[] tags = root.getElementsByName("a", true);
   for (TagNode node : tags) {
     if (idLinkLogout.equals(node.getAttributeByName("id"))) {
       return true;
     }
   }
   return false;
 }
  public static ArrayList<ContentValues> parsePosts(
      TagNode aThread,
      int aThreadId,
      int unreadIndex,
      int opId,
      AwfulPreferences prefs,
      int startIndex) {
    ArrayList<ContentValues> result = new ArrayList<ContentValues>();
    boolean lastReadFound = false;
    int index = startIndex;
    String update_time = new Timestamp(System.currentTimeMillis()).toString();
    Log.v(TAG, "Update time: " + update_time);
    try {
      if (!Constants.isICS()
          || !prefs.inlineYoutube) { // skipping youtube support for now, it kinda sucks.
        aThread = convertVideos(aThread);
      }
      TagNode[] postNodes = aThread.getElementsByAttValue("class", "post", true, true);

      for (TagNode node : postNodes) {
        // fyad status, to prevent processing postbody twice if we are in fyad
        ContentValues post = new ContentValues();
        post.put(THREAD_ID, aThreadId);

        // We'll just reuse the array of objects rather than create
        // a ton of them
        int id = Integer.parseInt(node.getAttributeByName("id").replaceAll("post", ""));
        post.put(ID, id);
        post.put(AwfulProvider.UPDATED_TIMESTAMP, update_time);
        post.put(POST_INDEX, index);
        if (index > unreadIndex) {
          post.put(PREVIOUSLY_READ, 0);
          lastReadFound = true;
        } else {
          post.put(PREVIOUSLY_READ, 1);
        }
        index++;
        post.put(IS_MOD, 0);
        post.put(IS_ADMIN, 0);
        TagNode[] postContent = node.getElementsHavingAttribute("class", true);
        for (TagNode pc : postContent) {
          if (pc.getAttributeByName("class").contains("author")) {
            post.put(USERNAME, pc.getText().toString().trim());
          }

          if (pc.getAttributeByName("class").contains("role-mod")) {
            post.put(IS_MOD, 1);
          }

          if (pc.getAttributeByName("class").contains("role-admin")) {
            post.put(IS_ADMIN, 1);
          }

          if (pc.getAttributeByName("class").equalsIgnoreCase("title")
              && pc.getChildTags().length > 0) {
            TagNode[] avatar = pc.getElementsByName("img", true);

            if (avatar.length > 0) {
              post.put(AVATAR, avatar[0].getAttributeByName("src"));
            }
            post.put(AVATAR_TEXT, pc.getText().toString().trim());
          }

          if (pc.getAttributeByName("class").equalsIgnoreCase("postbody")
              || pc.getAttributeByName("class").contains("complete_shit")) {
            TagNode[] images = pc.getElementsByName("img", true);

            for (TagNode img : images) {
              // don't alter video mock buttons
              if ((img.hasAttribute("class")
                  && img.getAttributeByName("class").contains("videoPlayButton"))) {
                continue;
              }
              boolean dontLink = false;
              TagNode parent = img.getParent();
              String src = img.getAttributeByName("src");

              if ((parent != null && parent.getName().equals("a"))
                  || (img.hasAttribute("class")
                      && img.getAttributeByName("class")
                          .contains("nolink"))) { // image is linked, don't override
                dontLink = true;
              }
              if (src.contains(".gif")) {
                img.setAttribute(
                    "class",
                    (img.hasAttribute("class") ? img.getAttributeByName("class") + " " : "")
                        + "gif");
              }

              if (img.hasAttribute("title")) {
                if (!prefs.showSmilies) { // kill all emotes
                  String name = img.getAttributeByName("title");
                  img.setName("p");
                  img.addChild(new ContentNode(name));
                }
              } else {
                if (!lastReadFound && prefs.hideOldImages || !prefs.imagesEnabled) {
                  if (!dontLink) {
                    img.setName("a");
                    img.setAttribute("href", src);
                    img.addChild(new ContentNode(src));
                  } else {
                    img.setName("p");
                    img.addChild(new ContentNode(src));
                  }
                } else {
                  if (!dontLink) {
                    img.setName("a");
                    img.setAttribute("href", src);
                    TagNode newimg = new TagNode("img");
                    if (!prefs.imgurThumbnails.equals("d") && src.contains("i.imgur.com")) {
                      int lastSlash = src.lastIndexOf('/');
                      if (src.length() - lastSlash <= 9) {
                        int pos = src.length() - 4;
                        src = src.substring(0, pos) + prefs.imgurThumbnails + src.substring(pos);
                      }
                    }
                    newimg.setAttribute("src", src);
                    img.addChild(newimg);
                  }
                }
              }
            }

            StringBuffer fixedContent = new StringBuffer();
            Matcher fixCharMatch = fixCharacters_regex.matcher(NetworkUtils.getAsString(pc));

            while (fixCharMatch.find()) {
              fixCharMatch.appendReplacement(fixedContent, "");
            }

            fixCharMatch.appendTail(fixedContent);
            post.put(CONTENT, fixedContent.toString());
          }

          if (pc.getAttributeByName("class").equalsIgnoreCase("postdate")) {
            post.put(
                DATE,
                NetworkUtils.unencodeHtml(pc.getText().toString())
                    .replaceAll("[^\\w\\s:,]", "")
                    .trim());
          }

          if (pc.getAttributeByName("class").equalsIgnoreCase("profilelinks")) {
            TagNode[] links = pc.getElementsHavingAttribute("href", true);

            if (links.length > 0) {
              String href = links[0].getAttributeByName("href").trim();
              String userId = href.substring(href.lastIndexOf("rid=") + 4);

              post.put(USER_ID, userId);

              if (Integer.toString(opId).equals(userId)) { // ugh
                post.put(IS_OP, 1);
              } else {
                post.put(IS_OP, 0);
              }
            }
          }
          if (pc.getAttributeByName("class").equalsIgnoreCase("editedby")
              && pc.getChildTags().length > 0) {
            post.put(EDITED, "<i>" + pc.getChildTags()[0].getText().toString() + "</i>");
          }
        }
        TagNode[] editImgs = node.getElementsByAttValue("alt", "Edit", true, true);

        if (editImgs.length > 0) {
          post.put(EDITABLE, 1);
        } else {
          post.put(EDITABLE, 0);
        }
        result.add(post);
      }

      Log.i(
          TAG,
          Integer.toString(postNodes.length) + " posts found, " + result.size() + " posts parsed.");
    } catch (Exception e) {
      e.printStackTrace();
    }

    return result;
  }
  private static TagNode convertVideos(TagNode contentNode) {
    TagNode[] videoNodes = contentNode.getElementsByAttValue("class", "bbcode_video", true, true);
    TagNode[] youtubeNodes =
        contentNode.getElementsByAttValue("class", "youtube-player", true, true);
    for (TagNode youTube : youtubeNodes) {
      String src = youTube.getAttributeByName("src");
      int height = Integer.parseInt(youTube.getAttributeByName("height"));
      int width = Integer.parseInt(youTube.getAttributeByName("width"));
      Matcher youtube = youtubeHDId_regex.matcher(src);
      if (youtube.find()) {
        String videoId = youtube.group(1);
        String link = "http://www.youtube.com/watch?v=" + videoId;
        String image = "http://img.youtube.com/vi/" + videoId + "/0.jpg";
        youTube.setName("a");
        youTube.setAttribute("href", link);
        youTube.removeAttribute("type");
        youTube.removeAttribute("frameborder");
        youTube.removeAttribute("src");
        youTube.removeAttribute("height");
        youTube.removeAttribute("width");
        youTube.setAttribute(
            "style",
            "background-image:url("
                + image
                + ");background-size:cover;background-repeat:no-repeat;background-position:center; position:relative;display:block;text-align:center; width:"
                + width
                + "; height:"
                + height);
        TagNode img = new TagNode("img");
        img.setAttribute("class", "nolink videoPlayButton");
        img.setAttribute("src", "file:///android_res/drawable/ic_menu_video.png");
        img.setAttribute(
            "style", "position:absolute;top:50%;left:50%;margin-top:-16px;margin-left:-16px;");
        youTube.addChild(img);
      }
    }
    for (TagNode node : videoNodes) {
      try {
        String src = null;
        int height = 0;
        int width = 0;
        TagNode[] object = node.getElementsByName("object", false);
        if (object.length > 0) {
          height = Integer.parseInt(object[0].getAttributeByName("height"));
          width = Integer.parseInt(object[0].getAttributeByName("width"));
          TagNode[] emb = object[0].getElementsByName("embed", true);
          if (emb.length > 0) {
            src = emb[0].getAttributeByName("src");
          }
        }
        if (src != null && height != 0 && width != 0) {
          String link = null, image = null;
          Matcher youtube = youtubeId_regex.matcher(src);
          Matcher vimeo = vimeoId_regex.matcher(src);
          if (youtube
              .find()) { // we'll leave in the old youtube code in case something gets reverted
            String videoId = youtube.group(1);
            link = "http://www.youtube.com/watch?v=" + videoId;
            image = "http://img.youtube.com/vi/" + videoId + "/0.jpg";
          } else if (vimeo.find()) {
            String videoId = vimeo.group(1);
            TagNode vimeoXML;
            try {
              vimeoXML = NetworkUtils.get("http://vimeo.com/api/v2/video/" + videoId + ".xml");
            } catch (Exception e) {
              e.printStackTrace();
              continue;
            }
            if (vimeoXML.findElementByName("mobile_url", true) != null) {
              link = vimeoXML.findElementByName("mobile_url", true).getText().toString();
            } else {
              link = vimeoXML.findElementByName("url", true).getText().toString();
            }
            image = vimeoXML.findElementByName("thumbnail_large", true).getText().toString();
          } else {
            node.removeAllChildren();
            TagNode ln = new TagNode("a");
            ln.setAttribute("href", src);
            ln.addChild(new ContentNode(src));
            node.addChild(ln);
            continue;
          }
          node.removeAllChildren();
          node.setAttribute(
              "style",
              "background-image:url("
                  + image
                  + ");background-size:cover;background-repeat:no-repeat;background-position:center; position:relative;text-align:center; width:"
                  + width
                  + "; height:"
                  + height);
          node.setAttribute("onclick", "location.href=\"" + link + "\"");
          TagNode img = new TagNode("img");
          img.setAttribute("class", "nolink videoPlayButton");
          img.setAttribute("src", "file:///android_res/drawable/ic_menu_video.png");
          img.setAttribute(
              "style", "position:absolute;top:50%;left:50%;margin-top:-23px;margin-left:-32px;");
          node.addChild(img);
        }
      } catch (Exception e) {
        continue; // if we fail to convert the video tag, we can still display the rest.
      }
    }

    return contentNode;
  }
Esempio n. 8
0
  String getAdditionInfo(String url) {
    String s = "";
    if (url.contains("bongda")) {
      try {
        HtmlCleaner cleaner = new HtmlCleaner();
        String id = url.substring(url.lastIndexOf("/"));
        id = id.substring(1, id.indexOf("."));
        Uri ss = Uri.parse("http://m.bongdaplus.vn/Story.aspx?sid=" + id);

        URI fine =
            new URI(
                ss.getScheme(),
                ss.getUserInfo(),
                ss.getHost(),
                ss.getPort(),
                ss.getPath(),
                ss.getQuery(),
                ss.getFragment());
        HttpClient httpclient = new DefaultHttpClient();
        httpclient.getParams().setParameter(ClientPNames.COOKIE_POLICY, CookiePolicy.BEST_MATCH);
        HttpGet httpget = new HttpGet(fine);
        httpget.addHeader(
            "user-agent", "Mozilla/5.0 (Linux; U; Android 2.3.3) Gecko/20100101 Firefox/8.0");
        httpget.addHeader("accept-language", "en-us,en;q=0.5");
        HttpResponse response = httpclient.execute(httpget);
        InputStream binaryreader = new BufferedInputStream(response.getEntity().getContent());
        BufferedReader buf = new BufferedReader(new InputStreamReader(binaryreader));
        String sss;
        StringBuilder sb = new StringBuilder();
        while ((sss = buf.readLine()) != null) {
          sb.append(sss);
        }

        //	    		URL u = new URL(ss.getScheme(), ss.getHost(), ss.getPort(), ss.getPath());
        TagNode root = cleaner.clean(sb.toString());
        TagNode div = root.findElementByAttValue("class", "story-body", true, false);
        TagNode[] child = div.getElementsByName("strong", true);
        for (int i = 0; i < child.length; i++) child[i].removeFromTree();
        child = div.getElementsByAttValue("class", "listing", true, false);
        for (int i = 0; i < child.length; i++) child[i].removeFromTree();
        child = div.getElementsByName("ul", true);
        for (int i = 0; i < child.length; i++) child[i].removeFromTree();
        child = div.getElementsByName("img", true);
        for (int i = 0; i < child.length; i++)
          child[i].setAttribute("style", "width: 100%; height: auto;");
        s = cleaner.getInnerHtml(div);
        if (s.indexOf("VideoPlaying(") > 0) {
          int x = s.indexOf("VideoPlaying(");
          int st = s.indexOf("'", x);
          int en = s.indexOf("'", st + 1);
          String urlvideo = "http://bongdaplus.vn" + s.substring(st + 1, en);
          s = s + "<a href=\"" + urlvideo + "\"> Video </a></br></br>";
        }
      } catch (Exception e) {
        e.printStackTrace();
      }
    } else if (url.contains("teamtalk")) {
      try {
        HtmlCleaner cleaner = new HtmlCleaner();
        Uri ss = Uri.parse(url);
        URL u = new URL(ss.getScheme(), ss.getHost(), ss.getPort(), ss.getPath());
        TagNode root = cleaner.clean(u);
        TagNode div = root.findElementByAttValue("class", "tt-article-text", true, false);
        TagNode[] child = div.getElementsByName("p", true);
        for (int i = 0; i < child.length; i++)
          if (child[i].hasAttribute("class") || child[i].hasAttribute("style"))
            child[i].removeFromTree();
        child = div.getElementsByName("img", true);
        for (int i = 0; i < child.length; i++)
          child[i].setAttribute("style", "max-width: 80%; height: auto;");
        s = cleaner.getInnerHtml(div);
      } catch (Exception e) {
        e.printStackTrace();
      }
    } else if (url.contains("licheuro")) {
      try {
        HtmlCleaner cleaner = new HtmlCleaner();
        Uri ss = Uri.parse(url);
        URL u = new URL(ss.getScheme(), ss.getHost(), ss.getPort(), ss.getPath());
        TagNode root = cleaner.clean(u);
        TagNode div = root.findElementByAttValue("class", "entry", true, false);

        TagNode[] child = div.getElementsByName("span", true);
        for (int i = 0; i < child.length; i++) child[i].removeFromTree();
        child = div.getElementsByName("strong", true);
        for (int i = 0; i < child.length; i++) {
          child[i].getParent().addChildren(child[i].getChildren());
          child[i].getParent().addChild(child[i].getText());
          child[i].removeFromTree();
        }
        child = div.getElementsByName("h1", true);
        for (int i = 0; i < child.length; i++) {
          child[i].getParent().addChildren(child[i].getChildren());
          child[i].getParent().addChild(child[i].getText());
          child[i].removeFromTree();
        }
        child = div.getElementsByAttValue("class", "boxpost", true, true);
        for (int i = 0; i < child.length; i++) child[i].removeFromTree();
        child = div.getElementsByAttValue("class", "ratingblock ", true, true);
        for (int i = 0; i < child.length; i++) child[i].removeFromTree();
        child = div.getElementsByName("iframe", true);
        for (int i = 0; i < child.length; i++) child[i].removeFromTree();
        child = div.getElementsByName("blockqoute", true);
        for (int i = 0; i < child.length; i++) {
          child[i].getParent().addChildren(child[i].getChildren());
          child[i].removeFromTree();
        }

        child = div.getElementsByAttValue("class", "dd_outer", true, true);
        for (int i = 0; i < child.length; i++) child[i].removeFromTree();
        child = div.getElementsByAttValue("id", "fb-root", true, true);
        for (int i = 0; i < child.length; i++) child[i].removeFromTree();
        child = div.getElementsByAttValue("id", "fbSEOComments", true, true);
        for (int i = 0; i < child.length; i++) child[i].removeFromTree();
        child = div.getElementsByAttValue("id", "ajax_comments_wrapper", true, true);
        for (int i = 0; i < child.length; i++) child[i].removeFromTree();
        child = div.getElementsByName("table", true);
        for (int i = 0; i < child.length; i++) {
          child[i].setAttribute("style", "max-width: 100%; height: auto;");
          child[i].setAttribute("width", "100%");
        }
        child = div.getElementsByName("script", true);
        for (int i = 0; i < child.length; i++) child[i].removeFromTree();
        child = div.getElementsByName("img", true);
        for (int i = 0; i < child.length; i++)
          child[i].setAttribute("style", "max-width: 80%; height: auto;");
        s = cleaner.getInnerHtml(div);
        if (s.indexOf("\"file\":") > 0) {
          int x = s.indexOf("\"file\":");
          int st = s.indexOf("http", x);
          int en = s.indexOf("\"", st + 1);
          String urlvideo = s.substring(st, en);
          s = s + "<a href=\"" + urlvideo + "\"> Video </a></br></br>";
        }
      } catch (Exception e) {
        e.printStackTrace();
      }
    }
    System.out.println(s);
    return s;
  }
Esempio n. 9
0
  @SuppressWarnings("unchecked")
  protected void handleFile(File file, int depth, Collection results) {
    File f = new File(FilenameUtils.normalize(file.getAbsolutePath()));
    logger.debug(f.getAbsoluteFile());
    try {
      HtmlCleaner cleaner = new HtmlCleaner();
      cleaner.setTransformations(ct);

      CleanerProperties props = cleaner.getProperties();
      //				props.setAdvancedXmlEscape(false);
      props.setUseEmptyElementTags(false);
      //				props.setTranslateSpecialEntities(false);
      //				props.setRecognizeUnicodeChars(false);

      TagNode node = cleaner.clean(f);

      TagNode tnBody = node.getAllElements(false)[1];
      List l = tnBody.getChildren();
      if (l != null
          && l.size() > 0) { // This is a hack to remove the <?xml in the beginning of body
        tnBody.removeChild(l.get(0));
      }

      for (int i = 1; i <= anzElements; i++) {
        String tag = config.getString("substitute[" + i + "]/@tag");
        String att = config.getString("substitute[" + i + "]/@att");
        String from = config.getString("substitute[" + i + "]/from");
        String to = config.getString("substitute[" + i + "]/to");
        to = subSpecial(to);

        TagNode[] imgs = node.getElementsByName(tag, true);

        for (TagNode tn : imgs) {
          String srcAtt = tn.getAttributeByName(att);
          int index = srcAtt.indexOf(from);
          if (index >= 0) {
            tn.addAttribute(att, to);
          }
        }
      }

      BrowserCompactXmlSerializer serializer = new BrowserCompactXmlSerializer(props);
      //			PrettyXmlSerializer serializer = new PrettyXmlSerializer(props);

      String s = serializer.getXmlAsString(node, "ISO-8859-1");

      Writer fw = null;
      try {
        fw = new FileWriter(f);
        fw.write(s);
      } catch (IOException e) {
        logger.error("", e);
      } finally {
        if (fw != null)
          try {
            fw.close();
          } catch (IOException e) {
          }
      }

      results.add(f.getAbsoluteFile());
    } catch (IOException e) {
      logger.error("", e);
    }
  }