china_pubparser.java

本系统实现了从五个网站上搜索的图书进行整合后

			}
		}
		return price;
	}
	/**
	* Function:  getBookUrl
	* Description:  取出图书详细信息地址
	* Calls: no
	* Called By:  mainService
	* @param temStr as String
	* @return String
	* @throws no
	*/
	public String getBookUrl(String temStr) {
		String bookUrl = "";
		if(temStr.indexOf(":'") != -1&&temStr.length()>3)
		     bookUrl = "http://www.china-pub.com/" + temStr.substring(temStr.indexOf(":'")+2,temStr.length()-1);
		//System.out.println(bookUrl);
		return bookUrl;
	}
	/**
	* Function:  getBookDiscount
	* Description:  取出图书折扣
	* Calls: no
	* Called By:  mainService
	* @param temStr as String
	* @return String
	* @throws no
	*/
	public  String getBookDiscount(String temStr) {
		
		String bookDiscount = "";
		if(temStr.indexOf(":'") != -1&&temStr.length()>3)
		    bookDiscount = temStr.substring(temStr.indexOf(":'")+2,temStr.length()-1);
		return bookDiscount;

	}
	/**
	* Function:  getBookAuthor
	* Description:  取出图书作者
	* Calls: no
	* Called By:  mainService
	* @param temStr as String
	* @return String
	* @throws no
	*/
	public String getBookAuthor(String temStr) {
		String bookAuthor = "";
		if(temStr.indexOf(":'") != -1&&temStr.length()>3)
		   bookAuthor = temStr.substring(temStr.indexOf(":'")+2,temStr.length()-1);
		bookAuthor = bookAuthor.replace(" ", "");
        bookAuthor = bookAuthor.replace("，", " ");
        bookAuthor = bookAuthor.replace(",", " ");
        bookAuthor = bookAuthor.replace("；", " ");
		return bookAuthor;
	}
	/**
	* Function:  getBookImage
	* Description:  取出图书封面图片地址
	* Calls: no
	* Called By:  mainService
	* @param temStr as String
	* @return String
	* @throws no
	*/
	public String getBookImage(String temStr) {
		String bookImage = "";
		String tem = "";
		if(temStr.indexOf(":'") != -1&&temStr.length()>3)
		   tem = temStr.substring(temStr.indexOf(":'")+2,temStr.length()-1);
		 bookImage = "http://images.china-pub.com"+ tem + "/cover.jpg";
		return bookImage;
	}
	/**
	* Function:  getBookImage
	* Description:  取出图书封面图片地址
	* Calls: no
	* Called By:  mainService
	* @param temStr as String
	* @return String
	* @throws no
	*/
	public String getBookName(String temStr) {
		
		String bookName = "";
		if(temStr.indexOf(":'") != -1&&temStr.length()>3)
		  bookName = temStr.substring(temStr.indexOf(":'")+2,temStr.length()-1);
		return bookName;
	}
	/**
	* Function:  getBookFixPrice
	* Description:  取出图书市场定价
	* Calls: no
	* Called By:  mainService
	* @param temStr as String
	* @return String
	* @throws no
	*/
	public String getBookFixPrice(String temStr) {
		String bookPrice = "";
		if(temStr.indexOf(":'") != -1&&temStr.length()>3)
		   bookPrice = temStr.substring(temStr.indexOf(":'")+2,temStr.length()-1);
		return bookPrice;
	}
	/**
	* Function:  getBookPublishTime
	* Description:  取出图书出版时间
	* Calls: no
	* Called By:  mainService
	* @param temStr as String
	* @return String
	* @throws no
	*/
	public String getBookPublishTime(String temStr) {
		String bookPublishTime = null;
		if(temStr.indexOf(":'") != -1&&temStr.length()>10)
		 bookPublishTime = temStr.substring(temStr.indexOf(":'")+2,temStr.length()-8);
		return bookPublishTime;
	}
	/**
	* Function:  getBookISBN
	* Description:  取出图书ISBN
	* Calls: no
	* Called By:  mainService
	* @param temStr as String
	* @return String
	* @throws no
	*/
	public  String getBookISBN(String temStr) {
		
		String bookISBN = "";
		if(temStr.indexOf(":'") != -1&&temStr.length()>3)
		  bookISBN = temStr.substring(temStr.indexOf(":'")+2,temStr.length()-1);
		return bookISBN;
	}
	/**
	* Function:  getBookPublisher
	* Description:  取出图书出版社
	* Calls: no
	* Called By:  mainService
	* @param temStr as String
	* @return String
	* @throws no
	*/
	public String getBookPublisher(String temStr) {
		String bookPublisher = "";
		if(temStr.indexOf(":'") != -1&&temStr.length()>3)
			bookPublisher = temStr.substring(temStr.indexOf(":'")+2,temStr.length()-1);
		return bookPublisher;
		
	}
	public String getBookPrice(String bookElement) {
		
		return null;
	}
	public String getBookContent(String bookElement) {
		return null;
	}
	/**
	* Function:  getNextPageUrl
	* Description:  取出下一页链接
	* Calls: no
	* Called By:  processService.startMultithread()
	* @param doc as Document
	* @return String
	* @throws no
	*/
	public String getNextPageUrl(Document doc) {
		/* 初始化为no，表示没有下一页 */
		String nextPageUrl = "no";
		Node divNode = null;
		Element divElement = null;
		
		NodeList divList = doc.getElementsByTagName("div");
		/*把divNodeList中的节点保存在依次存入divNode中*/
		for (int i = 0; i < divList.getLength(); i++) {
			divNode = divList.item(i);
			/*将divNode类型转换成divElement类型*/
			if(divNode.getNodeType()==Node.ELEMENT_NODE){
				divElement = (Element)divNode; 
				/*选出 属性id是fanye的divElement*/
				if("fanye".equals(divElement.getAttribute("id"))){
					NodeList aList = divElement.getChildNodes();
					for(int j = 0;j<aList.getLength();j++){
						Node aNode = aList.item(j);
						if(aNode.getNodeType() == Node.ELEMENT_NODE){
							Element aElement = (Element)aNode;
							if("FONT".equals(aElement.getNodeName())){
								j++;
								if(j<aList.getLength()){
									Node tempNode = aList.item(j);
									if("A".equals(tempNode.getNodeName())){
										Element temElement = (Element)tempNode;
										nextPageUrl = "http://www.china-pub.com/s/"+temElement.getAttribute("href");
									}
								}
								break;
							}
						}
					}
					break;
				}
			}
		}
		return nextPageUrl;
	}
	public long getRecordNum(Document doc) {
		long num = 0;
		NodeList servers = doc.getElementsByTagName("div");
		for (int i = 0; i < servers.getLength(); i++) {
			
			Node node = servers.item(i);
			if(node.getNodeType() == Node.ELEMENT_NODE){
				Element element = (Element)node;
				if("break".equals(element.getAttribute("class"))){
					NodeList spanList = element.getChildNodes();
					for(int j = 0;j<spanList.getLength();j++){
						Node spanNode = spanList.item(j);
						if(spanNode.getNodeType() == Node.ELEMENT_NODE){
							Element spanElement = (Element)spanNode;
							if("red14".equals(spanElement.getAttribute("class"))){
								if(spanElement.hasChildNodes()&&null != spanElement.getFirstChild().getNodeValue()){
									String strNum = spanElement.getFirstChild().getNodeValue();
									if(strNum.indexOf("(") != -1&& strNum.indexOf("种")!= -1)
									    strNum = strNum.substring(strNum.indexOf("(")+1, strNum.indexOf("种")-2).trim();
									if(strNum.indexOf("+") != -1)
										strNum = strNum.substring(0, strNum.length()-1);
									num = Long.valueOf(strNum);
									//System.out.println(num);
									break;
								}
							}
						}
					}
					break;
				}
			}
		}
		return num;
	}

    public static void main(String args [])throws Exception{
    	China_pubparser ex = new China_pubparser();
    	Document doc = ex.nekohtmlParser("http://www.china-pub.com/s/?type=&ref=&tid=0&key1=java");
    	Price price = ex.getDetailInfo(doc);
    	System.out.println(price.getChina_pubDiscount() + ">>" + price.getChina_pubPrice() + price.getChina_pubUrl());
    	ArrayList<Book> list = ex.mainService(doc,false);
    	for(Iterator it = list.iterator();it.hasNext();){
    		Book book = (Book)it.next();
    		System.out.println(book.getBookAuthor() + ">>" + book.getBookFixPrice()
    				          + ">>" + book.getBookImage() + ">>" + book.getBookISBN()
    				          + ">>" + book.getBookName() + ">>" + book.getBookProspectus()
    				          + ">>" + book.getBookPublisher() + ">>" + book.getBookPublishTime()
    				          + ">>" + book.getPrice().getChina_pubDiscount() + ">>" + book.getPrice().getChina_pubPrice());
    	} 	
    	//System.out.println(ex.getNextPageUrl(doc));
//    	ex.getRecordNum(doc);
    	
    }

}