python 爬虫练习（一）

港控/mmm° 2021-09-25 06:20 501阅读 0赞

本文为熟悉基础爬虫语法的练习，不涉及高级功能。

开发环境为win10、python3.8.5、oracle12

在gbf（グランブルーファンタジー）的攻略网站爬取全部角色的基本信息，包括角色名、限定类别（ハロウィン、リミテッド、クリスマス等等）、稀有度、属性、种族、类型、武器、详情页面地址、综合评分、总体评价、角色头像图、角色全身图，技能详细数据并格式化存储在本地。

先看一下网页结构  
全角色索引页：[https://xn–bck3aza1a2if6kra4ee0hf.gamewith.jp/article/show/120415][https_xn_bck3aza1a2if6kra4ee0hf.gamewith.jp_article_show_120415]  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70]  
来看一下每一行的源代码（这个用f12看到的其实不是源代码，而是经过前端处理过的代码。不过这个网站好像大部分处理都是后台做的，因此用这个来做展示）  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70 1]  
注意第一行，每个角色的主要信息都在data-colX中。  
头像为data-original的值，注意不是img的src中的值，上图中src中的值应该是延迟加载后加上去的。  
详细信息的地址就是a标签的地址。

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70 2]  
id为score后的第一个加粗的数字接受角色的评分，目前发现加粗有两种方式，一个是b标签，还有一种是class样式，需要分别判断。

总体描述就是score下的第一个table的全部内容。

**注意：就在这篇文章编写过程中，网站对一部分内容进行了改版，导致最终有大概十几个角色的这两个信息没有获得到。本人因为懒没有对代码进行更新**

接下来是获取全身图和技能数据。

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70 3]  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70 4]

全身图为abillity下第一个非广告非隐藏元素的图片，技能为在此之后直到广告的全部页面。

下面是全部代码  
建表语句

CREATE TABLE "C##TEST"."CRAWLER_GBF_BASE_SRC" 
       (	"ID" VARCHAR2(20), 
    	"PERSON_NAME" VARCHAR2(100), 
    	"RARE_DESC" VARCHAR2(100), 
    	"SHUXING" VARCHAR2(100), 
    	"ZHONGZU" VARCHAR2(100), 
    	"LEIXING" VARCHAR2(100), 
    	"WEAPON" VARCHAR2(100), 
    	"IMG_URL" VARCHAR2(1000), 
    	"DETAIL_URL" VARCHAR2(1000)
       ) SEGMENT CREATION IMMEDIATE 
      PCTFREE 10 PCTUSED 40 INITRANS 1 MAXTRANS 255 
     NOCOMPRESS LOGGING
      STORAGE(INITIAL 65536 NEXT 1048576 MINEXTENTS 1 MAXEXTENTS 2147483645
      PCTINCREASE 0 FREELISTS 1 FREELIST GROUPS 1
      BUFFER_POOL DEFAULT FLASH_CACHE DEFAULT CELL_FLASH_CACHE DEFAULT)
      TABLESPACE "USERS" ;
    CREATE TABLE "C##TEST"."CRAWLER_GBF_BASE_CLEAN1" 
       (	"ID" VARCHAR2(20), 
    	"PERSON_NAME" VARCHAR2(100), 
    	"EXT_DESC" VARCHAR2(100), 
    	"RARE_DESC" VARCHAR2(100), 
    	"SHUXING" VARCHAR2(100), 
    	"ZHONGZU" VARCHAR2(100), 
    	"LEIXING" VARCHAR2(100), 
    	"WEAPON" VARCHAR2(100), 
    	"IMG_URL" VARCHAR2(1000), 
    	"DETAIL_URL" VARCHAR2(1000), 
    	"OVERALL_SCORE" VARCHAR2(100), 
    	"DETAIL_TOP" NCLOB, 
    	"DETAIL_SKILL" NCLOB
       ) SEGMENT CREATION IMMEDIATE 
      PCTFREE 10 PCTUSED 40 INITRANS 1 MAXTRANS 255 
     NOCOMPRESS LOGGING
      STORAGE(INITIAL 65536 NEXT 8192 MINEXTENTS 1 MAXEXTENTS 2147483645
      PCTINCREASE 0 FREELISTS 1 FREELIST GROUPS 1
      BUFFER_POOL DEFAULT FLASH_CACHE DEFAULT CELL_FLASH_CACHE DEFAULT)
      TABLESPACE "USERS" 
     LOB ("DETAIL_TOP") STORE AS SECUREFILE (
      TABLESPACE "USERS" ENABLE STORAGE IN ROW CHUNK 8192
      NOCACHE LOGGING  NOCOMPRESS  KEEP_DUPLICATES 
      STORAGE(INITIAL 106496 NEXT 1048576 MINEXTENTS 1 MAXEXTENTS 2147483645
      PCTINCREASE 0
      BUFFER_POOL DEFAULT FLASH_CACHE DEFAULT CELL_FLASH_CACHE DEFAULT)) 
     LOB ("DETAIL_SKILL") STORE AS SECUREFILE (
      TABLESPACE "USERS" ENABLE STORAGE IN ROW CHUNK 8192
      NOCACHE LOGGING  NOCOMPRESS  KEEP_DUPLICATES 
      STORAGE(INITIAL 106496 NEXT 1048576 MINEXTENTS 1 MAXEXTENTS 2147483645
      PCTINCREASE 0
      BUFFER_POOL DEFAULT FLASH_CACHE DEFAULT CELL_FLASH_CACHE DEFAULT)) ;
      CREATE UNIQUE INDEX "C##TEST"."SYS_IL0000074601C00013$$" ON "C##TEST"."CRAWLER_GBF_BASE_CLEAN1" (
      PCTFREE 10 INITRANS 2 MAXTRANS 255 
      STORAGE(INITIAL 65536 NEXT 1048576 MINEXTENTS 1 MAXEXTENTS 2147483645
      PCTINCREASE 0 FREELISTS 1 FREELIST GROUPS 1
      BUFFER_POOL DEFAULT FLASH_CACHE DEFAULT CELL_FLASH_CACHE DEFAULT)
      TABLESPACE "USERS" 
      PARALLEL (DEGREE 0 INSTANCES 0) ;
      CREATE UNIQUE INDEX "C##TEST"."SYS_IL0000074601C00012$$" ON "C##TEST"."CRAWLER_GBF_BASE_CLEAN1" (
      PCTFREE 10 INITRANS 2 MAXTRANS 255 
      STORAGE(INITIAL 65536 NEXT 1048576 MINEXTENTS 1 MAXEXTENTS 2147483645
      PCTINCREASE 0 FREELISTS 1 FREELIST GROUPS 1
      BUFFER_POOL DEFAULT FLASH_CACHE DEFAULT CELL_FLASH_CACHE DEFAULT)
      TABLESPACE "USERS" 
      PARALLEL (DEGREE 0 INSTANCES 0) ;
       COMMENT ON COLUMN "C##TEST"."CRAWLER_GBF_BASE_CLEAN1"."PERSON_NAME" IS '名字';
       COMMENT ON COLUMN "C##TEST"."CRAWLER_GBF_BASE_CLEAN1"."EXT_DESC" IS '额外描述';
       COMMENT ON COLUMN "C##TEST"."CRAWLER_GBF_BASE_CLEAN1"."RARE_DESC" IS '稀有度';
       COMMENT ON COLUMN "C##TEST"."CRAWLER_GBF_BASE_CLEAN1"."SHUXING" IS '属性';
       COMMENT ON COLUMN "C##TEST"."CRAWLER_GBF_BASE_CLEAN1"."ZHONGZU" IS '种族';
       COMMENT ON COLUMN "C##TEST"."CRAWLER_GBF_BASE_CLEAN1"."LEIXING" IS '类型';
       COMMENT ON COLUMN "C##TEST"."CRAWLER_GBF_BASE_CLEAN1"."WEAPON" IS '武器';
       COMMENT ON COLUMN "C##TEST"."CRAWLER_GBF_BASE_CLEAN1"."OVERALL_SCORE" IS '评分';
    
    CREATE SEQUENCE  "C##TEST"."SEQ_CRAWLER"  MINVALUE 1 MAXVALUE 9999999999999999999999999999 INCREMENT BY 1 START WITH 4681 CACHE 20 NOORDER  NOCYCLE  NOKEEP  NOSCALE  GLOBAL ;

python代码  
工具类

''' Created on 2021年3月16日 @author: song '''
    def saveFile(filename, contents): 
      #fh = open(filename, 'w') 
        fh = open(filename, 'a+', encoding='utf-8')
        fh.write(contents)
        fh.write("\n")
        fh.close() 
    def cleanFile(filename): 
      #fh = open(filename, 'w') 
        fh = open(filename, 'w', encoding='utf-8')
        fh.write("")
        fh.close() 
    def resultSetFormat(cur):
        from collections import namedtuple
        fieldName = [i[0].lower() for i in cur.description]
        Record = namedtuple('Record', fieldName)
        rows = cur.fetchall()
        return list(map(Record._make, rows))
    def saveImgFromHttp(fileName, imgUrl):
        print("准备下载文件：", fileName)
        
        import urllib.request
        import os.path
        
        
        
        if os.path.isfile(fileName):
            print("warning: 文件下载--文件【", fileName, "】已存在")
            return
        
        if not fileName or not imgUrl:
            print("warning: 文件下载--参数为空")
            return
        
        conn = urllib.request.urlopen(imgUrl)
        f = open(fileName, 'wb')
        f.write(conn.read())
        f.close()

主类

''' Created on 2021年3月16日 @author: song '''
    from urllib import request
    from pyquery import PyQuery as pq
    from crawler import MyUtil
    import json
    import cx_Oracle
    import re
    import html
    from bs4 import BeautifulSoup
    import time
    import logging
    
    P_URL = "https://xn--bck3aza1a2if6kra4ee0hf.gamewith.jp/article/show/120415"
    P_FILE_SRC = "E:\\python_test\\gbfSrc_allCharacter.txt"
    P_FILE_JSON = "E:\\python_test\\gbfJson_allCharacter.txt"
    P_DIR_IMG = "E:/python_test/img/"
    #大图
    P_DIR_BIG_IMG = "E:/python_test/bigImg/"
    
    P_DB = cx_Oracle.connect('c##test', 'test', 'localhost:1521/orcl')
    P_DB.autocommit = True
    P_CURSOR = P_DB.cursor()
    
    def doCrawler():
        resp = request.urlopen(P_URL)
        print("crawler start...")
        htmlSrc = resp.read().decode("utf-8");
        print("crawler done")
        MyUtil.cleanFile(P_FILE_SRC)
        MyUtil.saveFile(P_FILE_SRC, htmlSrc)
        
    def doParseSrcToJson_1():
        
        print("parse src html start...")
        
        with open(P_FILE_SRC, "r", encoding='utf-8') as f:
            temp = f.read()
        obj = pq(temp)
        
    # obj = pq(filename='E:\\python_test\\v0316.txt', encoding='utf-8')
        trs = obj(".gbf-hyouka_table").eq(0)("tr")
        
        jsonObj = { "data": []};
        
        for i in range(trs.length):
            
            thisJsonObj = { };
            
            thisTr = trs.eq(i)
            #角色全名
            person_name = thisTr.attr("data-col1")
            #稀有度
            rare_desc = thisTr.attr("data-col2")
            #属性
            shuxing = thisTr.attr("data-col3")
            #种族
            zhongzu = thisTr.attr("data-col4")
            #评分（未处理）
            leixing = thisTr.attr("data-col5")
            #武器
            weapon = thisTr.attr("data-col6")
            
            #缩略图url
            imgUrl = thisTr("img").eq(0).attr("data-original");
            #详细页面url
            detailUrl = thisTr("a").eq(0).attr("href");
            
            
            thisJsonObj["person_name"] = person_name;
            thisJsonObj["rare_desc"] = rare_desc;
            thisJsonObj["shuxing"] = shuxing;
            thisJsonObj["zhongzu"] = zhongzu;
            
            thisJsonObj["leixing"] = leixing;
            thisJsonObj["weapon"] = weapon;
            thisJsonObj["img_url"] = imgUrl;
            thisJsonObj["detail_url"] = detailUrl;
            
            #print(thisJsonObj)
            if person_name != None:
                jsonObj["data"].append(thisJsonObj)
        
        MyUtil.cleanFile(P_FILE_JSON)
        MyUtil.saveFile(P_FILE_JSON, json.dumps(jsonObj, ensure_ascii=False))
        
        print("parse src html done, count [" + str(len(jsonObj["data"])) + "]")
        1
        
    #存入数据库
    def doSaveDatabase():
        cursor = P_DB.cursor()
        
        cursor.execute('''truncate table crawler_gbf_base_src''')
        
        with open(P_FILE_JSON, 'r', encoding = 'utf8') as fp:
            jsonData = json.load(fp)["data"]
            for i in range(len(jsonData)):
                cursor.execute(''' insert into crawler_gbf_base_src( id, person_name, rare_desc, shuxing, zhongzu, leixing, weapon, img_url, detail_url ) values( seq_crawler.nextval, :person_name, :rare_desc, :shuxing, :zhongzu, :leixing, :weapon, :img_url, :detail_url ) ''', jsonData[i])
        
        'ﾊﾞﾗﾝｽ'
        
        1
    
    #第一轮数据清洗，主要处理原始基本数据
    def dateClean_v1():
        P_CURSOR.execute('''truncate table crawler_gbf_base_clean1''')
        P_CURSOR.execute('''update crawler_gbf_base_src t1 set t1.person_name = replace(t1.person_name, '<br>')''')
        P_CURSOR.execute(''' insert into crawler_gbf_base_clean1(id, person_name, ext_desc, rare_desc, shuxing, zhongzu, leixing, weapon, img_url, detail_url) select regexp_replace(t1.detail_url, '([^/]*/)*(.+)', '\\2') id, regexp_substr(t1.person_name, '[^(<\\?？＆]+') person_name, regexp_replace(regexp_substr(t1.person_name, '\$(.+)\$'), '\$|\$') ext_desc, t1.rare_desc, t1.shuxing, t1.zhongzu, t1.leixing, replace(t1.weapon, '？', '/') weapon, t1.img_url, t1.detail_url from crawler_gbf_base_src t1 ''')
        1
    
    #爬取缩略图
    def doCrawlerThumbnail():
        P_CURSOR.execute('''select t1.* from crawler_gbf_base_clean1 t1 order by to_number(t1.id) desc''');
        resultList = MyUtil.resultSetFormat(P_CURSOR)
        #缩略图
        for i in range(len(resultList)):
            tempUrl = P_DIR_IMG + "thumbnail_" + re.match('([^/]*/)*([^/]*)', 
                                                         resultList[i].img_url, re.S).group(2)
            print(tempUrl)
            MyUtil.saveImgFromHttp(tempUrl, resultList[i].img_url)
        1
    
    def innerDoDetailParse_before(detail_url, pid):
        try:
            time.sleep(2)
            innerDoDetailParse(detail_url, pid)
        except Exception as err:
            logging.exception(err)
            exit()
    # finally:
    # print("f")
    
    def innerDoDetailParse(detail_url, pid):
    # with open("E:\\python_test\\test.txt", "r", encoding='utf-8') as f:
    # temp = f.read()
    # obj = pq(temp)
        import threading
        print("%s threading is printed %s"%(threading.current_thread().name, pid))
    # time.sleep(1)
    # return;
    
    # if pid == "166845":
    # print(pid)
    
        resp = request.urlopen(detail_url)
        htmlSrc = resp.read().decode("utf-8");
        
        obj = pq(htmlSrc)
        j1 = obj("#score")
        j1.next()
        
        detail_score = ""
        #大图
        detail_img_top = ""
        detail_top = ""
        detail_skill = ""
        #简介1
        topFlag = 0#简介下只取第一个img和一个table
        imgFlag = 0
        skillFlag = 0
        while (j1.next() and (imgFlag + skillFlag) < 2):
            j1 = j1.next()
            
            
            
            
            
            if str(j1).find("<noscript") == 0:
                continue
            
            if str(j1).find("<img") == 0 and imgFlag == 0:
                imgFlag = 1
                detail_img_top = html.unescape(str(j1.attr("data-original") if j1.attr("data-original") else j1.attr("src")))
            
            if j1.find("img").length > 0 and imgFlag == 0:
                imgFlag = 1
                detail_img_top = html.unescape(str(j1.find("img").attr("data-original")))
            
            if str(j1).find("<table") == 0 and skillFlag == 0:
                skillFlag = 1
                detail_score = j1.find("b").html() if j1.find("b").html() else j1.find(".bolder").html() 
                detail_top = BeautifulSoup(html.unescape(str(j1)), features="lxml").prettify()
            
            
            1
        
        #c-rectangle-ad-wrap noscript script
        j2 = obj("#abillity")#技能部分
        while (j2.next()):
            
            if str(j2).find("<script") > -1:
                j2 = j2.next()
                continue
            
            if j2.has_class("c-rectangle-ad-wrap"):
    # print("end: ", html.unescape(str(j2)))
                break
            
            detail_skill += html.unescape(str(j2))
            j2 = j2.next()
            1
        detail_skill = BeautifulSoup(detail_skill, features="lxml").prettify()
        
        print("综合评分：", detail_score)
        if not detail_top or not detail_skill:
            print("------------- ", pid, " 描述为空")
    # print("detail_img_top：", detail_img_top)
    # print("技能：", detail_skill)
    # print("综合评价", detail_top)
        
        MyUtil.saveImgFromHttp(P_DIR_BIG_IMG + "bigImg_" + pid + ".jpg", detail_img_top)
        print("do update... pid: ", pid)
        
    # myCursor = P_DB.cursor()
        
        p_db = cx_Oracle.connect('c##test', 'test', 'localhost:1521/orcl')
        p_db.autocommit = True
        p_cursor = p_db.cursor()
        
        p_cursor.execute(''' update crawler_gbf_base_clean1 t1 set t1.detail_skill = :detail_skill, t1.detail_top = :detail_top, t1.overall_score = :overall_score where t1.id = :id ''', 
        { "detail_skill": detail_skill.replace("・", "·"),
         "detail_top": detail_top.replace("・", "·"),
         "overall_score": detail_score,
         "id": pid}
        )
        
    # myCursor.close()
        
        1
    #爬取详细信息
    def doCrawlerDetail():
        from concurrent.futures.thread import ThreadPoolExecutor
        threadPool = ThreadPoolExecutor(max_workers = 5, thread_name_prefix = "gbf_")
        
        P_CURSOR.execute('''select t1.* from crawler_gbf_base_clean1 t1 where t1.overall_score is null order by to_number(t1.id)''');
        resultList = MyUtil.resultSetFormat(P_CURSOR)
        
        for i in range(len(resultList)):
            if i > -1:
                if not resultList[i].overall_score :
                    future = threadPool.submit(innerDoDetailParse_before, resultList[i].detail_url, resultList[i].id)
    # future.result()
                
        threadPool.shutdown(wait=True)
    # print(resultList[i].detail_url)
        1
        
    def test0():
    # with open("E:\\python_test\\test.txt", "r", encoding='utf-8') as f:
    # temp = f.read()
    # obj = pq(temp)
        resp = request.urlopen("https://xn--bck3aza1a2if6kra4ee0hf.gamewith.jp/article/show/260553")
        print("crawler start...")
        htmlSrc = resp.read().decode("utf-8");
        import threading
        print("%s threading is printed "%(threading.current_thread().name))
    
        obj = pq(htmlSrc)
        j1 = obj("#score")
    # print(j1)
        j1.next()
        
        detail_score = ""
        #大图
        detail_img_top = ""
        detail_top = ""
        detail_skill = ""
        #简介1
        topFlag = 0#简介下只取第一个img和一个table
        imgFlag = 0
        skillFlag = 0
        while (j1.next() and (imgFlag + skillFlag) < 2):
            j1 = j1.next()
            
            print(j1)
            
            if str(j1).find("<noscript") == 0:
                continue
            
            if str(j1).find("<img") == 0 and imgFlag == 0:
                imgFlag = 1
                detail_img_top = html.unescape(str(j1.attr("data-original") if j1.attr("data-original") else j1.attr("src")))
            
            if j1.find("img").length > 0 and imgFlag == 0:
                imgFlag = 1
                detail_img_top = html.unescape(str(j1.find("img").attr("data-original")))
            
            if str(j1).find("<table") == 0 and skillFlag == 0:
                skillFlag = 1
                detail_score = j1.find("b").html() if j1.find("b").html() else j1.find(".bolder").html() 
                detail_top = BeautifulSoup(html.unescape(str(j1)), features="lxml").prettify()
            
            1
        
        #c-rectangle-ad-wrap noscript script
        j2 = obj("#abillity")#技能部分
        while (j2.next()):
            
            if str(j2).find("<script") > -1:
                j2 = j2.next()
                continue
            
            if j2.has_class("c-rectangle-ad-wrap"):
    # print("end: ", html.unescape(str(j2)))
                break
            
            detail_skill += html.unescape(str(j2))
            j2 = j2.next()
            1
        detail_skill = BeautifulSoup(detail_skill, features="lxml").prettify()
        
        print("点数：", detail_score)
        print("detail_img_top：", detail_img_top)
    # print("技能：", detail_skill)
    # print("综合评价", detail_top)
        
        return
        MyUtil.saveImgFromHttp(P_DIR_BIG_IMG + "bigImg_" + "20995" + ".jpg", detail_img_top)
        print("do update... pid: ", 20995)
        
        return
        1
        
    if __name__ == '__main__':
         doCrawler()
         doParseSrcToJson_1()
         doSaveDatabase()
         dateClean_v1()
         doCrawlerThumbnail()
    
         doCrawlerDetail()
    # test0()
        1
    
    P_DB.close()

成果  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70 5]  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70 6]  
注：缩略图后面的数字对应img\_url后面的数字  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70 7]  
注：全身图的数字对应数据库主键

[https_xn_bck3aza1a2if6kra4ee0hf.gamewith.jp_article_show_120415]: https://xn--bck3aza1a2if6kra4ee0hf.gamewith.jp/article/show/120415
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70]: /images/20210923/14989330437f4384bbbcc4796c40f747.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70 1]: /images/20210923/34da7203964b40a1805b5538f665d346.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70 2]: /images/20210923/0d5ead501a48401daf30cd7f472a1df2.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70 3]: /images/20210923/7c73b6aeca9f4c8cb6400d69d97ca9a2.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70 4]: /images/20210923/c81a2a6a3fb54ea8a88e488eb8fa9470.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70 5]: /images/20210923/55aea5cc079448068690f6f29e72b118.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70 6]: /images/20210923/540831b7f5a348eeb71af06681936895.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25heWlfMjI0_size_16_color_FFFFFF_t_70 7]: /images/20210923/faced18576ed4010aac528188ce8f803.png