上文介紹了爬取知乎問題資訊的整個過程,這裡介紹下爬取問題下所有答案的內容和圖片,大致過程相同,部分核心程式碼不同.

爬取一個問題的所有內容流程大致如下:

• 一個問題url
• 請求url,獲取問題下的答案個數(我不需要,因為之前獲取問題資訊的時候儲存了問題的回答個數)
• 通過答案的介面去獲取答案(如果一次獲取5個答案,總計100個答案,需要計算的出訪問20次答案介面)[答案的介面地址如下圖所示]
• 答案介面返回的內容儲存到mysql
• 提取內容中的圖片地址,儲存到本地



爬取程式碼:

從mysql庫中查到question的id, 然後直接訪問 答案介面 去獲取資料.

answer_template="https://www.zhihu.com/api/v4/questions/%s/answers?include=data[*].is_normal,admin_closed_comment,reward_info,is_collapsed,annotation_action,annotation_detail,collapse_reason,is_sticky,collapsed_by,suggest_edit,comment_count,can_
comment,content,editable_content,voteup_count,reshipment_settings,comment_permission,created_time,updated_time,review_info,relevant_info,question,excerpt,relationship.is_authorized,is_author,voting,is_thanked,is_nothelp;data[*].mark_infos[*].url;dat
a[*].author.follower_count,badge[?(type=best_answerer)].topics&limit=5&offset=%s&sort_by=default"
    def check_login(self, response):
         #從mysql中讀取question的資訊,來進行爬取
         db = MySQLdb.connect("localhost", "root", "", "crawl", charset='utf8' )
         cursor = db.cursor()
         selectsql="select questionid,answer_num from  zhihu_question where id in ( 251,138,93,233,96,293,47,24,288,151,120,311,214,33) ;"
         try:
             cursor.execute(selectsql)    
             results = cursor.fetchall()
             for row in results:
                 questionid = row[0]
                 answer_num = row[1]
                 fornum = answer_num/5 #計算需要訪問答案介面的次數
                 print("questionid : "+ str(questionid)+"   answer_Num: "+str(answer_num))
                 for i in range(fornum+1):
                     answer_url = self.answer_template % (str(questionid), str(i*5))
                     yield scrapy.Request(answer_url,callback=self.parse_answer, headers=self.headers) 
         except Exception as e:
             print(e)
         db.close()
 

解析response

parser_anser解析接口裡的內容,這裡就比較方便了, 因為是json格式的 程式碼如下:

def parse_answer(self,response):
        #測試時把返回結果寫到本地, 然後寫pythonmain方法測試,測試方法都在test_code目錄下
        #temfn= str(random.randint(0,100))
        #f = open("/var/www/html/scrapy/answer/"+temfn,'wb')
        #f.write(response.body)
        #f.write("------")
        #f.close() 
        res=json.loads(response.text)
        #print (res)
        data=res['data']
        # 一次返回多個(預設5個)答案, 需要遍歷
        for od in data:
            #print(od)
            item = AnswerItem()
            item['answer_id']=str(od['id'])  #  answer id
            item['question_id']=str(od['question']['id'])
            item['question_title']=od['question']['title']
            item['author_url_token']=od['author']['url_token']
            item['author_name']=od['author']['name']
            item['voteup_count']=str(od['voteup_count'])
            item['comment_count']=str(od["comment_count"])
            item['content']=od['content']
            yield item
            testh = etree.HTML(od['content'])
            itemimg = MyImageItem()
            itemimg['question_answer_id'] = str(od['question']['id'])+"/"+str(od['id'])
            itemimg['image_urls']=testh.xpath("//img/@data-original")
            yield itemimg
 

成果展示

爬取了4w+個答案和12G圖片(個人伺服器只有12G空間了~) 



爬取收藏夾下的答案內容和圖片:

爬取收藏夾下的回答的流程和爬取問題下回答基本流程一樣,區別在於:

1. 問題的start_urls為多個,收藏夾是一個一個爬取
2. 問題頁面上找到了內容介面,返回json.方便. 收藏夾頁面沒有找到介面(我沒有找到),我是訪問每頁,然後解析的html.

構造每頁的起始地址:



解析html核心程式碼: