import org.htmlparser.tags.TableTag; //导入依赖的package包/类
private Assertion assertion() throws SyntaxError {
   NodeList headerColumns = new NodeList();
   headerColumns.add( new TableColumn() );
   TableHeader tableHeader = new TableHeader();
   tableHeader.setChildren( headerColumns );
   NodeList rowColumns = new NodeList();
   rowColumns.add( new TableColumn() );
   TableRow tableRow = new TableRow();
   tableRow.setChildren( rowColumns );
   NodeList rows = new NodeList();
   rows.add( tableHeader );
   rows.add( tableRow );
   TableTag tableTag = new TableTag();
   tableTag.setChildren( rows );
   TestPage testPage = new WikiTestPage( new WikiPageDummy());
   SlimTestContext testContext = new SlimTestContextImpl( testPage );
   Table table = new HtmlTable( tableTag );
   List<SlimAssertion> list = new QueryTable( table, "id", testContext ).getAssertions();
   return list.get( 0 );
}
 

import org.htmlparser.tags.TableTag; //导入依赖的package包/类
private void preencheMapaFeriadosEstaduais() throws IOException, ParserException,ParseException {
    String estadosPage =  recuperarDadosEstado();
    StringBuilder stringBuilder = new StringBuilder(estadosPage);
    stringBuilder.delete(0,estadosPage.indexOf("<h3"));
    NodeList nodeEstadoList = filterTable(stringBuilder.toString());
    String todosMeses[] = {"janeiro", "fevereiro", "março", "abril", "maio", "junho", "julho", "agosto", "setembro", "outubro", "novembro", "dezembro"};
    Map<String,String> mapaMeses = new HashMap<String,String>();
    int i = 1;
    for (String mes:todosMeses){
        String valor = String.valueOf(i++);
        if(valor.length()< 2){
            valor ="0"+valor;
        }
        mapaMeses.put(mes,valor);
    }

    String estado = null;
    for (Node node:nodeEstadoList.toNodeArray()){
        if(node instanceof TableTag){
            NodeList lista = ((TableTag) node).searchFor(TableColumn.class, true);
            SimpleNodeIterator iterator  = lista.elements();
            while (iterator.hasMoreNodes()){
                Feriado feriado = new Feriado();
                Node data = iterator.nextNode();
                String[] dataExtenso = data.toPlainTextString().split(" de ");
                feriado.setData(dataExtenso[0] + "/" + mapaMeses.get(dataExtenso[1]) + "/2015");
                Node nome = iterator.nextNode();
                feriado.setNome(nome.toPlainTextString());
                Node lei = iterator.nextNode();
                if(dataExtenso[0].length()==1){
                    dataExtenso[0] = "0"+dataExtenso[0];
                }
                System.out.println(dataExtenso[0] + "/" + mapaMeses.get(dataExtenso[1]) + "/2015,"+nome.toPlainTextString()+","+mapaEstados.get(estado));
                mapaFeriadosEstado.get(estado).add(feriado);
            }

        }
        if(node instanceof HeadingTag){
            estado =  node.getChildren().toHtml().trim();
            if(node.getChildren().elementAt(0).getChildren() != null){
                estado =  node.getChildren().elementAt(0).getChildren().toHtml().trim();
            }
            mapaFeriadosEstado.put(estado,new ArrayList<Feriado>());
        }

    }
}
 

import org.htmlparser.tags.TableTag; //导入依赖的package包/类
/**
	 * 更新最新的部门发文
	 * @param httpConn
	 * @throws Exception
	 */
	public List visit(boolean isByCookie) throws Exception{
		log4.info("======访问网站============cookie="+isByCookie);
		
		String newUrls = null;
		TableColumn[] arrColumns;
		LinkTag lt = null;
	
		String title, dates, codes, link;
		HtmlPage page = null;
		TableTag tableContent[] = null;
		int order_count = 0;
		List list = new LinkedList();
		
		if(this.getUrl().startsWith("http:")){
			newUrls = this.getUrl();
		}
		else{
			if(this.base_url.endsWith("/")){
				newUrls = this.base_url+"/"+this.getUrl();
			}
			else{
				newUrls = this.base_url+this.getUrl();
			}
			
		}
//		if(true){
//			page = move2Urls(this.getHttpURLConnection(), newUrls);
//			page.getBody().toHtml();
////			return page.getBody().toHtml();
//		}
		log4.info("newUrls="+newUrls);

		String str = this.move2UrlsHtml(this.getHttpURLConnection(), newUrls, isByCookie);
		
		String[] msgs = str.split("\n");
		List htmls = new LinkedList();
		for(String msg: msgs){
			msg = msg.trim();
			msg = msg.replaceAll(" ", "");
			msg = msg.replaceAll("<", "");
			msg = msg.replaceAll(">", "");
			msg = msg.replaceAll(""", "");
			msg = msg.replaceAll("td", "");
			msg = msg.replaceAll("tr", "");
//			msg = msg.replaceAll("&", "&");
//			msg = msg.replaceAll("<", "&lt;");
//			msg = msg.replaceAll(">", "&gt;");
//			msg = msg.replaceAll("\"", "&quot;");
//			msg = msg.replaceAll("'", "&apos;");
			if(!ErrorCode.isEmpty(msg)){
				htmls.add(msg);
			}
		}
	
		return htmls;
	}