今日影評動態爬蟲

代碼：

import requests
import sqlite3
import json
import random

def find():
    url = 'http://www.1905.com/api/content/index.php'
    headers = {
        'User-Agent':'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/80.0.3987.163 Safari/537.36'
        }

    params = {
        'callback':'reloadList',
        'm':'converged',
        'a': 'info',
        'type':'jryp',
        'year': '2018',
        'month': '2'        
        }
    response = requests.get(url,params=params,headers=headers)
    result = response.text
    result = result.replace('reloadList(','').replace(')','')
    result = json.loads(result)
##    print(result)
    for i in result['info']:
##        print(i['url'],i['title'],i['thumb'])
        img = requests.get(i['thumb']).content
        img_name = random.randint(10000,99999)
        with open('img/%s.jpg'%img_name,'wb') as w:
            w.write(img)
##        createDB()
        save_data(content=i['title'],link=i['url'],img=img_name)
##        break

def createDB():
    conn = sqlite3.connect('film.db')
    c = conn.cursor()
    c.execute('CREATE TABLE filmdata(id INTEGER PRIMARY KEY AUTOINCREMENT,content text,link text,img text)')
    conn.commit()
    conn.close()

def save_data(content,link,img):
    conn = sqlite3.connect('film.db')
    c = conn.cursor()
    c.execute("INSERT into filmdata(content,link,img) VALUES('{0}','{1}','{2}')".format(content,link,img))
    conn.commit()
    conn.close()

def showdata():
    conn = sqlite3.connect('film.db')
    c = conn.cursor()
    res = c.execute('SELECT * from filmdata')
    print(res)
    for i in res:
        print(i[1])
    conn.close()    

find()
showdata()

結果：