package?com.gx.reptile.util;

import?java.io.ByteArrayOutputStream;
import?java.io.File;
import?java.io.FileOutputStream;
import?java.io.IOException;
import?java.io.InputStream;
import?java.net.HttpURLConnection;
import?java.net.URL;
import?java.util.ArrayList;
import?java.util.List;

import?org.jsoup.Jsoup;
import?org.jsoup.nodes.Document;
import?org.jsoup.select.Elements;

/**
?*?
*?@ClassName:?ReptileUtil?
*?@Description:?圖片爬蟲?--?使用jsoup解析html頁面，獲取需要的路徑，進行循環下載
*?@author?zhoujie?
*?@date?2018年7月27日?下午8:26:01?
*?@version?V1.0
*?百度
?*/
public?class?ReptileUtil?{
	
	//定義路徑
	static?String?baseurl?=?“http://www.netbian.com/“;
	static?String?geturl?=?“http://www.netbian.com/desk/18321.htm“;
	static?String?filepath?=?“C:\\Users\\zj\\Desktop\\new\\“;

	public?static?void?main（String[]?args）?{
		System.out.println（“初始下載頁面：“+geturl）;
		String?html?=?getHtml（geturl）;?//html頁面內容
		List?srclists?=?getImgSrcListFromHtml（html）;?//圖片地址集合
		downloadImg（srclists?filepath）;?//下載圖片
		//獲取下一個頁面進行下載
		List?list?=?getNextPageUrl（html）;?
		System.out.println（list.size（））;
		for?（int?i?=?0;?i?			String?url?=?list.get（i）;
			System.out.println（“下一個下載頁面：“+url）;
			String?html2?=?getHtml（url）;?//html頁面內容
			List?srclists2?=?getImgSrcListFromHtml（html2）;?//圖片地址集合
			downloadImg（srclists2?filepath）;?//下載圖片
		}
		System.out.println（“下載完畢“）;
	}
	
	/**
	?*?
	*?@title:?getHtml?
	*?@Description:?獲取頁面內容
	*?@param?@param?url
	*?@param?@return??頁面內容
	*?@return?String??返回類型?
	*?@throws
	?*/
	public?static?String?getHtml（String?url）{
		String?html?=?““;
		try?{
			html?=?Jsoup.connect（url）.execute（）.body（）;
		}?catch?（IOException?e）?{
			e.printStackTrace（）;
		}
		return?html;
	}
	
	/**
	?*?
	*?@title:?getImgSrcListFromHtml?
	*?@Description:?獲取頁面內容圖片路徑
	*?@param?@param?html??頁面內容
	*?@param?@return????圖片路徑數組
	*?@return?ArrayList????返回類型?
	*?@throws
	?*/
	public?static?List?getImgSrcListFromHtml（String?html）{
		List?list?=?new?ArrayList<>（）;
		//解析成html頁面
		Document?document?=?Jsoup.parse（html）;
		//獲取目標
		Elements?elements?=?document.select（“div?[class=pic]“）.select（“img“）;
		int?len?=?elements.size（）;
		for?（int?i?=?0;?i?			list.add（elements.get（i）.attr（“src“））;
		}
		return?list;
	}
	
	/**
	?*?
	*?@title:?getNextPage?
	*?@Description:?從頁面內容中獲取下一個頁面路徑
	*?@param?????頁面內容
	*?@return?List??返回頁面url數組
	*?@throws
	?*/
	public?static?List?getNextPageUrl（String?html）{
		List?list?=?new?ArrayList<>（）;
		//解析成html頁面
		Document?document?=?Jsoup.parse（html）;
		//獲取目標
		Elements?elements?=?document.select（“div?[class=list]“）.select（“a“）;
		for?（int?i?=?0;i			String?url?=?baseurl?+?elements.get（i）.attr（“href“）;
			list.add（url）;
		}
		return?list;
	}
	
	/**
	?*?
	*?@title:?downloadImg?
	*?@Description:?下載圖片?--?通過獲取的流轉成

?屬性????????????大小?????日期????時間???名稱
-----------?---------??----------?-----??----
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\
?????文件?????????848??2018-07-27?20:19??ReptileDemo\.classpath
?????文件?????????911??2018-07-27?20:19??ReptileDemo\.project
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\.settings\
?????文件?????????564??2018-07-27?20:19??ReptileDemo\.settings\.jsdtscope
?????文件?????????364??2018-07-27?20:19??ReptileDemo\.settings\org.eclipse.jdt.core.prefs
?????文件?????????492??2018-07-27?20:19??ReptileDemo\.settings\org.eclipse.wst.common.component
?????文件?????????345??2018-07-27?20:19??ReptileDemo\.settings\org.eclipse.wst.common.project.facet.core.xml
?????文件??????????49??2018-07-27?20:19??ReptileDemo\.settings\org.eclipse.wst.jsdt.ui.superType.container
?????文件???????????6??2018-07-27?20:19??ReptileDemo\.settings\org.eclipse.wst.jsdt.ui.superType.name
?????文件???????????0??2018-07-27?20:22??ReptileDemo\readme
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\src\
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\src\com\
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\src\com\gx\
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\src\com\gx\reptile\
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\src\com\gx\reptile\util\
?????文件????????5263??2018-07-28?08:46??ReptileDemo\src\com\gx\reptile\util\ReptileUtil.java
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\WebRoot\
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\WebRoot\meta-INF\
?????文件??????????39??2018-07-27?20:19??ReptileDemo\WebRoot\meta-INF\MANIFEST.MF
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\WebRoot\WEB-INF\
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\WebRoot\WEB-INF\classes\
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\WebRoot\WEB-INF\classes\com\
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\WebRoot\WEB-INF\classes\com\gx\
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\WebRoot\WEB-INF\classes\com\gx\reptile\
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\WebRoot\WEB-INF\classes\com\gx\reptile\util\
?????文件????????5600??2018-07-28?08:46??ReptileDemo\WebRoot\WEB-INF\classes\com\gx\reptile\util\ReptileUtil.class
?????目錄???????????0??2018-07-28?09:05??ReptileDemo\WebRoot\WEB-INF\lib\
?????文件??????395748??2018-07-27?20:34??ReptileDemo\WebRoot\WEB-INF\lib\jsoup-1.11.3.jar
?????文件?????????654??2018-07-27?20:19??ReptileDemo\WebRoot\WEB-INF\web.xml