• python下载视频


    由于自己平时喜欢在b站看看番剧、学习视频之类的,恰好这段时间等了好几年的Re0 season2开始更新了,打算将视频全部保存到本地,于是打开pycharm开始干活…

    一、预期目标

    • 下载非大会员视频,最高1080p画质
    • 使用多线程下载加快下载速度
    • 鉴于本地网络环境不佳、考虑加入断点续传功能
    • 支持终端^c退出

    二、网址分析

    b站的视频信息全部保存在网页源码中的window.__playinfo__变量中,其实就是个json格式的字符串, 所以我们用正则表达式很容易提取出来,如下图:

    获得视频播放地址后如果直接在浏览器中请求服务器会返回403 Forbidden,打开开发者模式发现b站的请求方式为option,所以我们也照着这样去申请服务器资源。

    注意我们写脚本去爬取视频的时候需要填入自己的cookie,否则有些视频无法获取,当然如果你有大会员也可以下载大会员专属视频。
    补充一点:b站视频都是视频流和音频流分开存储的,所以将视频和音频下载完了之后,最后用 ffmpeg合并视频、音频。

    三、编码细节

    为了支持断点续传、我们需要记录每次请求头中的'Range': 'bytes=xx-yy'其中xx,yy为我们自定义的请求范围。所以我们再写个日志类用于记录请求信息,使用json格式保存到本地,方便操作。
    为了实现ctrl-c结束程序,使用signal模块捕获signal.SIGINT signal.SIGTERM即ctrl-c信号、终端信号,用全局变量is_exit控制线程是否退出,当程序接收到信号时is_exit设置为True线程退出。注意线程创建start之后不要jion方法阻塞,我们轮询的方式判断thread_list[i].is_alive(),如果所以线程运行完毕,程序就正常向下执行。
    如果程序因为网络故障某个线程退出之后,其他线程并未设置退出,继续进行下载,直到下载完成或报错为止。

    四、具体实现

    """
    ===================================
        -*- coding:utf-8 -*-
        Author     :GadyPu
        E_mail     :Gadypy@gmail.com
        Time       :2020/9/3 0003 上午 11:44
        FileName   :bili_downloader.py
    ===================================
    """
    import sys
    import signal
    import requests
    import time
    import math
    import os
    import json
    import re
    import threading
    import copy
    from lxml import etree
    from decimal import Decimal
    requests.packages.urllib3.disable_warnings()
    
    is_exit = False
    
    class Logger(object):
    
        def __init__(self, *args):
            if len(args) == 1:
                with open(os.path.join(os.getcwd(), args[0] + '.log'), 'r', encoding = 'utf-8') as rf:
                    self.log_file = json.loads(rf.read())
            else:
                self.log_file = {
                    'time': None,
                    'filename': args[0],
                    'filesize': args[1],
                    'downsize': 0,
                    'url': args[2],
                    'threads': []
                }
                for i in range(args[3]):
                    self.log_file['threads'].append({
                            'start': 0,
                            'end': 0,
                            'finshed': False
                        })
            self.lock = threading.Lock()
    
        def update_log(self, start, end, id, step, finshed = False):
            with self.lock:
                self.log_file['threads'][id]['start'] = start
                self.log_file['threads'][id]['end'] = end
                self.log_file['downsize'] += step
                if finshed:
                    self.log_file['threads'][id]['finshed'] = True
    
        def load_bound(self):
            with self.lock:
                return copy.deepcopy(self.log_file['threads'])
    
        def get_cur_size(self):
            with self.lock:
                return int(self.log_file['downsize'])
    
        def get_tot_size(self):
            with self.lock:
                return int(self.log_file['filesize'])
    
        def get_file_name(self):
            with self.lock:
                return self.log_file['filename']
    
        def get_req_url(self):
            with self.lock:
                return self.log_file['url']
    
        def save_log(self):
            with self.lock:
                self.log_file['time'] = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime())
                file_path = os.path.join(os.getcwd(), self.log_file['filename'] + '.log')
                with open(file_path, 'w+', encoding = 'utf-8') as wf:
                    wf.write(json.dumps(self.log_file))
    
        def failed(self):
            with self.lock:
                for i in self.log_file['threads']:
                    if not i['finshed']:
                        return True
                return False
    
        def show_log(self):
            print(self.log_file)
    
        def clear(self):
            file_path = os.path.join(os.getcwd(), self.log_file['filename'] + '.log')
            if os.path.exists(file_path):
                os.remove(file_path)
    
    class DownThread(threading.Thread):
        def __init__(self, headers: dict, log: Logger, id: int, start, end, url, session):
            super().__init__()
            self.headers = copy.deepcopy(headers)
            self.file_name = log.get_file_name()
            self.thread_id = id
            self.log = log
            self.__start = start
            self.__end = end
            self.url = url
            self.session = session
    
        def run(self):
            fp = open(self.file_name, 'rb+')
            __start, __end = self.__start, self.__end
            __step = 1024 * 512
            finshed = True
            b_read = 0
            print(f'thread[{self.thread_id}] 开始下载 Range:{__start}-{__end}')
            t1 = time.time()
            i, tm = 0, (__end - __start + 1) // __step + 1
            while __start <= __end:
                __to = __end if __start + __step > __end else __start + __step
                self.headers.update({'Range': 'bytes=' + str(__start) + '-' + str(__to)})
                try:
                    res = self.session.get(url = self.url, headers = self.headers, verify = False, timeout = 5)
                    if res.status_code in [200, 206]:
                        fp.seek(__start)
                        fp.tell()
                        fp.write(res.content)
                        #print(f'thread[{self.thread_id}] 正在下载 Range:{__start}-{__to}')
                        print('
    thread[{}] 正在下载{:^3.0f}%'.format(self.thread_id, (i / tm) * 100), end = '')
                        i += 1
                        b_read += len(res.content)
                        #self.log.update_log(__start, __end, self.thread_id, len(res.content))
    
                        if is_exit:
                            print('
    ', f'接收到^c信号 thread[{self.thread_id}]退出')
                            self.log.update_log(__start, __end, self.thread_id, b_read)
                            finshed = False
                            break
    
                except Exception as e:
                    print(e)
                    print('
    ', f'thread[{self.thread_id}]:网络错误请重新下载')
                    self.log.update_log(__start, __end, self.thread_id, b_read)
                    finshed = False
                    break
                __start = __to + 1
            if finshed:
                print('
    ', f'thread[{self.thread_id}]->下载完成 花费:{round(time.time() - t1, 2)}s')
                self.log.update_log(__start - 1, __end, self.thread_id, b_read, True)
            fp.close()
    
    class DownBilibili(object):
    
        def __init__(self, url, thread_num, path, sesdata):
            self.headers = {
                'Cookie': "SESSDATA=96b2608a%2C1617615923%2C7e662*a1;",
                'origin': 'https://www.bilibili.com',
                'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/84.0.4147.125 Safari/537.36'
            }
            self.url = url
            self.pat_src = r'window.__playinfo__=(.*?)</script>'
            self.pat_tit = r'<title.*>(.*?)</title>'
            self.pat_inf = r'window.__INITIAL_STATE__=(.*?)</script>'
            self.video_url = ''
            self.audio_url = ''
            self.titles = ''
            self.sesdata = sesdata
            self.thread_num = thread_num
            self.log = None
            self.out_put = []
            if path != '':
                os.chdir(path)
    
        def get_file_size(self, e: int) -> str:
            if e <= 0:
                return ''
            t = ["B", "KB", "MB", "GB", "TB", "PB", "EB", "ZB", "YB"]
            n = math.floor(math.log2(e) / math.log2(1024))
            return str(Decimal(e / math.pow(1024, n)).quantize(Decimal("0.00"))) + t[n]
    
        def get_video_info(self):
    
            try:
                self.headers.update({'Referer': self.url})
                if self.sesdata:
                    self.headers.update({'Cookie': self.sesdata})
                resp = requests.get(url = self.url, headers = self.headers, verify = False)
                self.titles = re.findall(self.pat_tit, resp.text)[0].split('_')[0]
            except Exception as err:
                print(err, '
    无法获取视频信息请重试')
                return False
            try:
                info_json = json.loads(re.findall(self.pat_src, resp.text)[0])
                self.video_url = info_json['data']['dash']['video'][0]['baseUrl']
                self.audio_url = info_json['data']['dash']['audio'][0]['baseUrl']
            except:
                bv_info = json.loads(re.findall(self.pat_inf, resp.text)[0].split(';')[0])
                ep_id = self.url.split('/')[-1]
                ep_id = ep_id[2:] if ep_id.startswith('ep') else ''
                vip_api = f"https://api.bilibili.com/pgc/player/web/playurl?cid={bv_info['epList'][0]['cid']}&bvid={bv_info['epList'][0]['bvid']}&ep_id={ep_id}&fnval=80"
                info_json = requests.get(url = vip_api, headers = self.headers, verify = False).json()
                if info_json['message'] != 'success':
                    print('大会员专属视频,请填入SESSDATA后重试!')
                    return False
                self.video_url = info_json['result']['dash']['video'][0]['baseUrl']
                self.audio_url = info_json['result']['dash']['audio'][0]['baseUrl']
            return True
    
        def get_download_size(self, src_url):
            try:
                session = requests.session()
                self.headers.update({'referer': self.url})
                session.options(url = src_url, headers = self.headers, verify = False)
                self.headers.update({'Range': 'bytes=' + str(0) + '-' + str(1024)})
                res = session.get(url = src_url, headers = self.headers, verify = False)
                size = int(res.headers['Content-Range'].split('/')[1])
                return size, session
            except Exception:
                print('无法获取服务器资源请重试')
            return -1, None
    
        def handler(self, signum, frame):
            global is_exit
            is_exit = True
    
        def assigned_tasks(self, size: int, log_exit: bool, src_url: str, session):
            step = size // self.thread_num
            thread_list = []
            if log_exit:
                bound = self.log.load_bound()
            for i in range(self.thread_num):
                __session = copy.copy(session)
                if log_exit:
                    __x, __y = bound[i]['start'], bound[i]['end']
                    # if __x == __y:
                    #     continue
                    t = DownThread(self.headers, self.log, i, __x, __y, src_url, __session)
                else:
                    start = 0 if not i else end + 1
                    end = start + step if i != self.thread_num - 1 else size - 1
                    t = DownThread(self.headers, self.log, i, start, end, src_url, __session)
                t.setDaemon(True)
                t.start()
                thread_list.append(t)
    
            #[_.join() for _ in thread_list]
    
            signal.signal(signal.SIGINT, self.handler)
            signal.signal(signal.SIGTERM, self.handler)
            while True:
                alive = False
                for i in range(self.thread_num):
                    alive |= thread_list[i].is_alive()
                if not alive:
                    break
                time.sleep(0.5)
    
        def download(self, src_url, file_name):
            global is_exit
            is_exit = False
            size, session = self.get_download_size(src_url)
            if size == -1:
                print('无法获取服务器资源请重试')
                return False
            log_exit = os.path.exists(os.path.join(os.getcwd(), file_name + '.log'))
            if log_exit:
                self.log = Logger(file_name)
                if not self.log.failed():
                    return True
                cur = self.log.get_cur_size()
                url = self.log.get_req_url()
                print(f'继续上次下载进度[已下载:{self.get_file_size(cur)} {round(cur / size * 100, 2)}%]', url)
                print('开始重新下载')
            else:
                print(f'开始下载【{file_name}】、文件大小为{self.get_file_size(size)}')
                self.log = Logger(file_name, size, src_url, self.thread_num)
                with open(file_name, 'wb') as fp:
                    fp.truncate(size)
    
            self.assigned_tasks(size, log_exit, src_url, session)
            self.log.save_log()
            if self.log.failed():
                return False
            return True
    
        def work(self):
            res = self.get_video_info()
            if not res:
                print('无法获取视频下载链接请重试')
                return None
            self.titles = re.sub('[/:*?"<>|]', '-' , self.titles)
            self.out_put = [self.titles + '-video_out.m4s', self.titles + '-audio_out.m4s']
            for i in range(2):
                if not i:
                    finshed = self.download(self.video_url, self.out_put[i])
                else:
                    finshed = self.download(self.audio_url, self.out_put[i])
                if not finshed:
                    print(f'【{self.out_put[i]}】' + '下载失败请重试')
                    return None
            self.merge()
            print('Done!')
    
        def merge(self):
            print('开始合并视频音频文件')
            os.popen('ffmpeg -i ' + f'"{self.out_put[0]}"' + ' -i ' + f'"{self.out_put[1]}"' + ' -codec copy ' + f'"{self.titles}.mp4"').read()
            print('合并完成开始删除缓存文件')
            [os.remove(os.path.join(os.getcwd(), i)) for i in self.out_put]
            [os.remove(os.path.join(os.getcwd(), i + '.log')) for i in self.out_put]
    
    def print_usage():
        print('''usage: bili_downloader [OPTION]...
    A simple tools for download bilibili videos.
    Support ctrl-c cancel task and restart task. 
    You can run it by command line arguments.
    download options:
      -b                     bilibili videos id. eg: 'BV1J5411b7XQ'
      -t                     download threads number, default 4
      -d                     the path that you want to save videos, default current dir path
      -s                     input your cookies, eg: 'SESSDATA=xx' to get the vip videos
      -h                     show this help message and exit
    ps: If you need to download a member video please enter your 'SESSDATA', otherwise you'd better not do that.   
    Refer: <https://www.cnblogs.com/GadyPu/>
    ''')
    
    if __name__ == '__main__':
    
        url = 'https://www.bilibili.com/video/'
        args = sys.argv
        length = len(args)
        if length == 1:
            print_usage()
        else:
            kwargs = {
                'bv': '',
                'num': 4,
                'path': '',
                'sesdata': ''
            }
            if '-b' in args:
                id = args.index('-b')
                if id != length - 1:
                    bv = args[id + 1]
                    kwargs['bv'] = bv
                    if bv.startswith('ss') or bv.startswith('ep'):
                        url  = 'https://www.bilibili.com/bangumi/play/'
            if '-h' in args:
                print_usage()
                sys.exit()
            if '-t' in args:
                id = args.index('-t')
                if id != length - 1:
                    num = int(args[id + 1])
                    kwargs['num'] = num
            if '-d' in args:
                id = args.index('-d')
                if id != length - 1 and os.path.exists(args[id + 1]):
                    kwargs['path'] = args[id + 1]
            if '-s' in args:
                id = args.index('-s')
                if id != length - 1:
                    sesdata = args[id + 1]
                    kwargs['sesdata'] = sesdata
            if kwargs['bv'] == '':
                print_usage()
                sys.exit()
            d = DownBilibili(url + kwargs['bv'], kwargs['num'], kwargs['path'], kwargs['sesdata'])
            d.work()

    如果出现:“[WinError 10048] 通常每个套接字地址(协议/网络地址/端口)只允许使用一次“,关闭终端重新打开终端再运行程序即可。

    本人没有大会员所以某些番剧不能第一时间观看,所以又在互联网上找了其它网站,简单分析了一下他的视频存储方式保存在m3u8文件里,具体实现如下,支持断点续传,有兴趣的小伙伴可以看一下(此网站需要自备梯子)

    """
    ===================================
        -*- coding:utf-8 -*-
        Author     :GadyPu
        E_mail     :Gadypy@gmail.com
        Time       :2020/8/22 0022 下午 01:41
        FileName   :fetch_video.py
    ===================================
    """
    import os
    import re
    import sys
    import time
    import requests
    import urllib.parse
    import threading
    from queue import Queue
    import warnings
    warnings.filterwarnings("ignore")
    
    class FetchVideos(object):
    
        def __init__(self, url):
            self.url = url
            self.m3u8_url = None
            self.que = Queue()
            self.headers = {
                'Referer': url,
                'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/84.0.4147.125 Safari/537.36',
                'Connection': 'close'
            }
            self.is_write = False
            self.lock = threading.Lock()
            self.m4s_headers_url = None
            self.m4s_headers_bytes = b''
            self.pat = r'mac_url=unescape((.*))'
            self.download_failed_lists = []#'faild.txt'
    
        def unescape(self, _str):
            _str = urllib.parse.unquote(_str)
            return re.sub(r'%u([a-fA-F0-9]{4}|[a-fA-F0-9]{2})', lambda m: chr(int(m.group(1), 16)), _str)
    
        def get_m3u8_url(self):
            try:
                response = requests.get(url = self.url, headers = self.headers, verify = False)
                res = re.findall(self.pat, response.text)[0]
                res = self.unescape(res).split(r"'")[1].split('#')[-1]
                self.m3u8_url = res[res.find('$') + 1:]
                print('the m3u8 url is:' + self.m3u8_url)
            except:
                print('cannot get the m3u8_url...')
                sys.exit()
    
        def get_m4s_headers(self):
            try:
                response = requests.get(url = self.m4s_headers_url, headers = self.headers, verify = False)
                self.m4s_headers_bytes = response.content
            except:
                print('cannot get m4s headers...')
                sys.exit()
    
        def init_ts_url(self):
            page = 1
            try:
                response = requests.get(url = self.m3u8_url, headers = self.headers, verify = False)
                api_url = self.m3u8_url[: self.m3u8_url.rfind('/') + 1]
                time.sleep(0.5)
                resp = requests.get(url = api_url + response.text.split('
    ')[2], headers = self.headers, verify = False)
                print(resp.url)
                for i in resp.text.split('
    '):
                    if i.endswith('.m4s'):
                        self.que.put((page, api_url + i))
                        page += 1
                    elif i.startswith('#EXT-X-MAP:URI='):
                        self.m4s_headers_url = api_url + i.split('#EXT-X-MAP:URI=')[1].split('"')[1]
                        print(self.m4s_headers_url)
                print(f'total {self.que.qsize()} files to be downloading...')
            except:
                print('cannot get m4s_url_lists...')
                sys.exit()
    
        def download_ts_thread(self, path):
            while not self.que.empty():
                d = self.que.get()
                try:
                    resp = requests.get(url = d[1], headers = self.headers,
                                            verify = False, stream = True, timeout = 6)
                    if resp.status_code == 200:
                        file_path = os.path.join(path, f'{d[0]}.m4s')
                        content = b''
                        for data in resp.iter_content(1024):
                            if data:
                                content += data
                        with open(file_path, 'wb') as wf:
                            wf.write(self.m4s_headers_bytes + content)
                        print(f'{d[0]}.m4s is download complished...')
                        time.sleep(0.2)
                    else:
                        print(f'the {d[0]}.m4s download failed...')
                        with self.lock:
                            self.download_failed_lists.append(d)
                except:
                    print(f'the {d[0]}.m4s download failed...')
                    if not self.is_write:
                        with open('header.bin', 'wb') as wf:
                            wf.write(self.m4s_headers_bytes)
                            self.is_write = True
                    with self.lock:
                        self.download_failed_lists.append(d)
    
        def merge_ts_files(self, filename):
            print('start merging...')
            dir = os.getcwd() + r'downloadvideos'
            merge_txt = [_ for _ in os.listdir(dir)]
            merge_txt.sort(key = lambda x: int(x.split('.')[0]))
            src = ''
            first = True
            for i in merge_txt:
                if first:
                    src = i
                    first = False
                else:
                    src = src + '+' + i
            os.chdir(dir)
            os.system('copy /b ' + src + f' {filename}.mp4')
            print('video merge completed...')
    
        def down_faild(self, lists: list):
            for i in lists:
                if i:
                    d = i.split(' ')
                    self.que.put((d[0], d[1]))
                    #print(d)
            with open('header.bin', 'rb') as rf:
                self.m4s_headers_bytes = rf.read()
    
        def remove_temp_files(self):
            path_text = os.path.join(os.getcwd(), 'failed.txt')
            path_bin = os.path.join(os.getcwd(), 'header.bin')
            if os.path.exists(path_text):
                os.remove(path_text)
            if os.path.exists(path_bin):
                os.remove(path_bin)
    
        def run(self):
            failed = False
            self.get_m3u8_url()
            with open('failed.txt', 'a+') as rf:
                rf.seek(0)
                lists = rf.read()
                if len(lists) > 1 and lists.split('
    ')[0][:86] == self.m3u8_url[:86]:
                    lists = lists.split('
    ')
                    self.down_faild(lists[1:])
                    failed = True
                else:
                    self.init_ts_url()
                    self.get_m4s_headers()
            down_path = os.getcwd() + r'downloadvideos'
            if not os.path.exists(down_path):
                os.makedirs(down_path)
            if not failed:
                [os.remove(os.path.join(down_path, i)) for i in os.listdir(down_path)]
            thred_list = []
            print('start downloading...')
            for i in range(3):
                t = threading.Thread(target = self.download_ts_thread, args = (down_path, ))
                t.setDaemon(True)
                thred_list.append(t)
                t.start()
            [t.join() for t in thred_list]
            if not len(self.download_failed_lists):
                self.remove_temp_files()
                self.merge_ts_files(self.m3u8_url.split('/')[4])
            else:
                with open('failed.txt', 'w+', encoding = 'utf-8') as wf:
                    wf.write(self.m3u8_url + '
    ')
                    [wf.write(str(i[0]) + ' ' + i[1] + '
    ') for i in self.download_failed_lists]
                print(f'total {len(self.download_failed_lists)} files download failed...')
    
    
    if __name__ == '__main__':
        #Re0 season2 ep9
        url = 'https://www.olevod.com/?m=vod-play-id-19657-src-1-num-9.html'
        d = FetchVideos(url = url)
        d.run()

    五、运行效果

    六、参考链接

    1、https://www.cnblogs.com/W-Y-C/p/13406392.html
    2、https://www.cnblogs.com/yuanyb/p/12296815.html
    3、https://blog.csdn.net/weixin_43107613/article/details/104754426

    七、更新日志

    • 2020.10.13 如果有大会员可以下载仅大会员可见的视频,默认最高画质,填入相应'SESSDATA'即可
  • 相关阅读:
    原型模板分享——国外高阶版“大众点评”TripAdvisor
    插画手绘入门教程
    十大用户体验设计失败案例,你知道吗?
    用摹客,轻松管理设计规范!
    摹客“独门绝技”之评论审阅篇
    政务管理系统原型模板分享
    产品经理如何做需求分析?
    python基础知socket编程
    异常处理
    元类,__call__方法和单例模式
  • 原文地址:https://www.cnblogs.com/GadyPu/p/13607071.html
Copyright © 2020-2023  润新知