教你如何使Python爬取酷我在线音乐

{
    "code": 200,
    "curTime": 1649482287185,
    "data": {
        "total": "741",
        "list": [
            {
                "musicrid": "MUSIC_11690555",
                "barrage": "0",
                "ad_type": "",
                "artist": "aiko",
                "mvpayinfo": {
                    "play": 0,
                    "vid": 8530326,
                    "down": 0
                },
                "nationid": "0",
                "pic": "http://img4.kuwo.cn/star/starheads/500/24/88/4146545084.jpg",
                "isstar": 0,
                "rid": 11690555,
                "duration": 362,
                "score100": "42",
                "ad_subtype": "0",
                "content_type": "0",
                "track": 1,
                "hasLossless": true,
                "hasmv": 1,
                "releaseDate": "1970-01-01",
                "album": "",
                "albumid": 0,
                "pay": "16515324",
                "artistid": 1907,
                "albumpic": "http://img4.kuwo.cn/star/starheads/500/24/88/4146545084.jpg",
                "originalsongtype": 0,
                "songTimeMinutes": "06:02",
                "isListenFee": false,
                "pic120": "http://img4.kuwo.cn/star/starheads/120/24/88/4146545084.jpg",
                "name": "恋をしたのは",
                "online": 1,
                "payInfo": {
                    "play": "1100",
                    "nplay": "00111",
                    "overseas_nplay": "11111",
                    "local_encrypt": "1",
                    "limitfree": 0,
                    "refrain_start": 89150,
                    "feeType": {
                        "song": "1",
                        "vip": "1"
                    },
                    "down": "1111",
                    "ndown": "11111",
                    "download": "1111",
                    "cannotDownload": 0,
                    "overseas_ndown": "11111",
                    "refrain_end": 126247,
                    "cannotOnlinePlay": 0
                },
                "tme_musician_adtype": "0"
            }
        ]
    },
    "msg": "success",
    "profileId": "site",
    "reqId": "4b55cf4b0171253c33ce1d71b999c42f",
    "tId": ""
}

# coding:utf-8
from copy import deepcopy
from dataclasses import dataclass


class Entity:
    """ Entity abstract class """

    def __setitem__(self, key, value):
        self.__dict__[key] = value

    def __getitem__(self, key):
        return self.__dict__[key]

    def get(self, key, default=None):
        return self.__dict__.get(key, default)

    def copy(self):
        return deepcopy(self)


@dataclass
class SongInfo(Entity):
    """ Song infORMation """
    file: str = None
    title: str = None
    singer: str = None
    album: str = None
    year: int = None
    genre: str = None
    duration: int = None
    track: int = None
    trackTotal: int = None
    disc: int = None
    discTotal: int = None
    createTime: int = None
    modifiedTime: int = None

# coding:utf-8
from copy import deepcopy


def exceptionHandler(*default):
    """ decorator for exception handling

    Parameters
    ----------
    *default:
        the default value returned when an exception occurs
    """

    def outer(func):

        def inner(*args, **kwargs):
            try:
                return func(*args, **kwargs)
            except BaseException as e:
                print(e)
                value = deepcopy(default)
                if len(value) == 0:
                    return None
                elif len(value) == 1:
                    return value[0]
                else:
                    return value

        return inner

    return outer

# coding:utf-8
import JSON
from urllib import parse
from typing import List, Tuple

import requests


class KuWoMusicCrawler:
    """ Crawler of KuWo Music """

    def __init__(self):
        super().__init__()
        self.headers = {
            'User-Agent': 'Mozilla/5.0 (windows NT 10.0; Win64; x64) '
                          'AppleWEBKit/537.36 (Khtml, like Gecko) Chrome/71.0.3578.98 Safari/537.36',
            'Cookie': 'kw_token=C713RK6IJ8J',
            'csrf': 'C713RK6IJ8J',
            'Host': 'www.kuwo.cn',
            'Referer': ''
        }

    @exceptionHandler([], 0)
    def getSongInfos(self, key_Word: str, page_num=1, page_size=10) -> Tuple[List[SongInfo], int]:
        key_word = parse.quote(key_word)

        # configure request header
        headers = self.headers.copy()
        headers["Referer"] = 'http://www.kuwo.cn/search/list?key='+key_word

        # send request for song information
        url = f'http://www.kuwo.cn/api/www/search/searchMusicBykeyWord?key={key_word}&pn={page_num}&rn={page_size}&reqId=c06e0e50-fe7c-11eb-9998-47e7e13a7206'
        response = requests.get(url, headers=headers)
        response.raise_for_status()

        # parse the response data
        song_infos = []
        data = json.loads(response.text)['data']
        for info in data['list']:
            song_info = SongInfo()
            song_info['rid'] = info['rid']
            song_info.title = info['name']
            song_info.singer = info['artist']
            song_info.album = info['album']
            song_info.year = info['releaseDate'].split('-')[0]
            song_info.track = info['track']
            song_info.trackTotal = info['track']
            song_info.duration = info["duration"]
            song_info.genre = 'Pop'
            song_info['coverPath'] = info.get('albumpic', '')
            song_infos.append(song_info)

        return song_infos, int(data['total'])

@exceptionHandler('')
def getSongUrl(self, song_info: SongInfo) -> str:
    # configure request header
    headers = self.headers.copy()
    headers.pop('Referer')
    headers.pop('csrf')

    # send request for play url
    url = f"http://www.kuwo.cn/api/v1/www/music/playUrl?mid={song_info['rid']}&type=convert_url3"
    response = requests.get(url, headers=headers)
    response.raise_for_status()
    play_url = json.loads(response.text)['data']['url']

    return play_url

@exceptionHandler('')
def downloadSong(self, song_info: SongInfo, save_dir: str) -> str:
    # get play url
    url = self.getSongUrl(song_info)
    if not url:
        return ''

    # send request for binary data of audio
    headers = self.headers.copy()
    headers.pop('Referer')
    headers.pop('csrf')
    headers.pop('Host')
    response = requests.get(url, headers=headers)
    response.raise_for_status()

    # save audio file
    song_path = os.path.join(
        save_dir, f"{song_info.singer} - {song_info.title}.mp3")
    with open(song_path, 'wb') as f:
        f.write(data)

    return song