1. # -*- coding: UTF-8 -*-
    3. import requests, json, re
    4. from selenium import webdriver
    5. from selenium.webdriver.common.by import By
    6. from selenium.webdriver.support.ui import WebDriverWait
    7. from selenium.webdriver.support import expected_conditions as EC
    8. from random import randint
    9. # 文件路径
    10. MUSIC_PATH='music/'
    11. # chrome driver路径
    12. DRIVERS_PATH='drivers/chromedriver-mac'
    14. class ParseQQmusic:
    16.     def __init__(self,ranking):
    17.         self.url =ranking
    18.         self.music_name =None
    19.         self.headers = {
    20.             'user-agent': 'Mozilla/5.0 (iPhone; CPU iPhone OS 11_0 like Mac OS X) AppleWebKit/604.1.38 (KHTML, like Gecko) Version/11.0 Mobile/15A372 Safari/604.1',
    21.             'Referer': 'https://u.y.qq.com'
    22.         }
    23.         self.session = requests.session()
    26.     def download_music(self,songmid):
    27.         data = {
    28.             "req": {
    29.                 "module": "CDN.SrfCdnDispatchServer",
    30.                 "method": "GetCdnDispatch",
    31.                 "param": {
    32.                     "guid": '',
    33.                     "calltype": 0,
    34.                     "userip": ""
    35.                 }
    36.             },
    37.             "req_0": {
    38.                 "module": "vkey.GetVkeyServer",
    39.                 "method": "CgiGetVkey",
    40.                 "param": {
    41.                     "guid": str(random_with_N_digits(9)),
    42.                     "songmid": [songmid],
    43.                     "songtype": [0],
    44.                     "uin": "0",
    45.                     "loginflag": 1,
    46.                     "platform": "20"
    47.                 }
    48.             },
    49.             "comm": {
    50.                 "uin": 0,
    51.                 "format": "json",
    52.                 "ct": 24,
    53.                 "cv": 0
    54.             }
    55.         }
    56.         api = 'http://u.y.qq.com/cgi-bin/musicu.fcg'
    57.         jsondata = json.dumps(data)
    58.         response = self.session.post(api, data=jsondata)
    59.         resJson = json.loads(response.content, encoding='utf-8')
    60.         sip = resJson.get('req_0').get('data').get('sip')[0]
    61.         purl = resJson.get('req_0').get('data').get('midurlinfo')[0].get('purl')
    62.         if purl != None and purl.strip():
    63.             print(self.music_name)
    64.             download_url= sip + purl
    65.             # 下载到本地
    66.             music = self.session.get(download_url,headers=self.headers)
    67.             # 文件名去除特殊符号
    68.             with open(MUSIC_PATH + "{}.m4a".format(re.sub(r'[\s+|@<>:\\"/]', '', self.music_name)), "wb") as m:
    69.                 m.write(music.content)
    70.     def view_html(self):
    71.         # qq音乐页面是js加载的，这里用chrome headless模式访问
    72.         option = webdriver.ChromeOptions()
    73.         option.add_argument('headless')
    74.         driver = webdriver.Chrome(DRIVERS_PATH, chrome_options=option)
    75.         # 排行榜页面
    76.         driver.get(self.url)
    77.         print(driver.title)
    78.         try:
    79.             # 等待播放列表加载完毕
    80.             WebDriverWait(driver, 10).until(EC.presence_of_element_located((By.CLASS_NAME, "songlist__songname_txt")))
    81.             lis = driver.find_elements_by_class_name('songlist__songname_txt')
    82.             pattern = re.compile(r'https://y.qq.com/n/yqq/song/(\S+).html')
    83.             for i in range(lis.__len__()):
    84.                 li = lis.__getitem__(i)
    85.                 a = li.find_element_by_class_name('js_song')
    86.                 # 获得songid
    87.                 href = a.get_attribute('href')
    88.                 music_name = a.get_attribute('title')
    89.                 self.music_name=music_name
    91.                 m = pattern.match(href)
    92.                 self.download_music(m.group(1))
    94.         finally:
    95.             driver.quit()
    97.     def start(self):
    98.         self.view_html()
    100. def random_with_N_digits(n):
    101.     range_start = 10**(n-1)
    102.     range_end = (10**n)-1
    103.     return randint(range_start, range_end)
    106. if __name__ == '__main__':
    107.     # 目标 抖音排行榜
    108.     qm = ParseQQmusic('https://y.qq.com/n/yqq/toplist/60.html#stat=y_new.toplist.menu.60')
    109.     qm.start()