devbox
知新_RL
这个屌丝很懒，什么也没留下！
关注作者
热门标签
热门文章
当前位置:   article > 正文

python爬取B站视频_python爬b站电影

作者：知新_RL | 2024-04-18 02:07:03

赞

踩

python爬b站电影

参考：https://cloud.tencent.com/developer/article/1768680

参考的代码有点问题，请求头需要修改，上代码：

  1. import requests
  2. import re  # 正则表达式
  3. import pprint
  4. import json
  5. from moviepy.editor import AudioFileClip, VideoFileClip
  6. from bs4 import BeautifulSoup as bs
  7.  
  8. headers = {
  9.     # 防盗链 告诉服务器 我们请求的url网址是从哪里跳转过来的
  10.     'referer': 'https://www.bilibili.com/a',
  11.     'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/86.0.4240.198 Safari/537.36'
  12. }
  13.  
  14. def send_request(url):
  15.     response = requests.get(url=url, headers=headers)
  16.     return response
  17.  
  18. def get_video_data(html_data):
  19.     """解析视频数据"""
  20.  
  21.     # 提取视频的标题
  22.     soup = bs(html_data, 'lxml')
  23.     title = soup.find_all(name='h1',attrs={"class":"video-title special-text-indent"})[0].get_text()
  24.     # print(title)
  25.  
  26.     # 提取视频对应的json数据
  27.     json_data = re.findall('<script>window\.__playinfo__=(.*?)</script>', html_data)[0]
  28.     # print(json_data)  # json_data 字符串
  29.     json_data = json.loads(json_data)
  30.     pprint.pprint(json_data)
  31.  
  32.     # 提取音频的url地址
  33.     audio_url = json_data['data']['dash']['audio'][0]['backupUrl'][0]
  34.     print('解析到的音频地址:', audio_url)
  35.  
  36.     # 提取视频画面的url地址
  37.     video_url = json_data['data']['dash']['video'][0]['backupUrl'][0]
  38.     print('解析到的视频地址:', video_url)
  39.  
  40.     video_data = [title, audio_url, video_url]
  41.     return video_data
  42.  
  43. def save_data(file_name, audio_url, video_url):
  44.     # 请求数据
  45.     print('正在请求音频数据')
  46.     audio_data = send_request(audio_url).content
  47.     print('正在请求视频数据')
  48.     video_data = send_request(video_url).content
  49.     with open(file_name + '.mp3', mode='wb') as f:
  50.         f.write(audio_data)
  51.         print('正在保存音频数据')
  52.     with open(file_name + '.mp4', mode='wb') as f:
  53.         f.write(video_data)
  54.         print('正在保存视频数据')
  55.  
  56. def merge_data(video_name):
  57.     print('视频合成开始:', video_name)
  58.     audioclip = AudioFileClip(video_name+'.mp3')
  59.     videoclip = VideoFileClip(video_name+'.mp4')
  60.     # 3.获取视频和音频的时长
  61.     video_time = videoclip.duration
  62.     audio_time = audioclip.duration
  63.     # 4.对视频或者音频进行裁剪
  64.     if video_time > audio_time:
  65.         # 视频时长>音频时长，对视频进行截取
  66.         videoclip_new = videoclip.subclip(0, audio_time)
  67.         audioclip_new = audioclip
  68.     else:
  69.         # 音频时长>视频时长，对音频进行截取
  70.         videoclip_new = videoclip
  71.         audioclip_new = audioclip.subclip(0, video_time)
  72.     # 5.视频中加入音频
  73.     video_with_new_audio = videoclip_new.set_audio(audioclip_new)
  74.     # 6.写入到新的视频文件中
  75.     video_with_new_audio.write_videofile("output.mp4",
  76.                                          codec='libx264',
  77.                                          audio_codec='aac',
  78.                                          temp_audiofile='temp-video.m4a',
  79.                                          remove_temp=True
  80.                                          )
  81.     print('视频合成结束:', video_name)
  82.  
  83.  
  84. url = 'https://www.bilibili.com/video/BV1bK421a7qG/?spm_id_from=333.1007.tianma.6-4-22.click'
  85. response = send_request(url)
  86. response.encoding = requests.utils.get_encodings_from_content(response.text)[0]
  87. html_data = response.text
  88. video_data = get_video_data(html_data)
  89. save_data(video_data[0], video_data[1], video_data[2])
  90. merge_data(video_data[0])

效果

小姐姐挺靓，就是左下角有水印，想办法去除水印，参考：python实战之去除视频水印&字幕_python 去除视频水印-CSDN博客

  1. import os
  2. import sys
  3. import cv2
  4. import numpy
  5. from moviepy import editor
  6.  
  7. TEMP_VIDEO = 'temp.mp4'
  8.  
  9.  
  10. class WatermarkRemover():
  11.  
  12.     def __init__(self, video_path, output, threshold: int, kernel_size: int):
  13.         self.threshold = threshold  # 阈值分割所用阈值
  14.         self.kernel_size = kernel_size  # 膨胀运算核尺寸
  15.         self.video_path = video_path
  16.         self.output = output
  17.  
  18.  
  19.     #根据用户手动选择的ROI（Region of Interest，感兴趣区域）框选水印或字幕位置。
  20.     def select_roi(self, img: numpy.ndarray, hint: str) -> list:
  21.         '''
  22.     框选水印或字幕位置，SPACE或ENTER键退出
  23.     :param img: 显示图片
  24.     :return: 框选区域坐标
  25.     '''
  26.         COFF = 0.7
  27.         w, h = int(COFF * img.shape[1]), int(COFF * img.shape[0])
  28.         resize_img = cv2.resize(img, (w, h))
  29.         roi = cv2.selectROI(hint, resize_img, False, False)
  30.         cv2.destroyAllWindows()
  31.         watermark_roi = [int(roi[0] / COFF), int(roi[1] / COFF), int(roi[2] / COFF), int(roi[3] / COFF)]
  32.         return watermark_roi
  33.  
  34.  
  35.     #对输入的蒙版进行膨胀运算，扩大蒙版的范围
  36.     def dilate_mask(self, mask: numpy.ndarray) -> numpy.ndarray:
  37.  
  38.         '''
  39.     对蒙版进行膨胀运算
  40.     :param mask: 蒙版图片
  41.     :return: 膨胀处理后蒙版
  42.     '''
  43.         kernel = numpy.ones((self.kernel_size, self.kernel_size), numpy.uint8)
  44.         mask = cv2.dilate(mask, kernel)
  45.         return mask
  46.     
  47.     #根据手动选择的ROI区域，在单帧图像中生成水印或字幕的蒙版。
  48.     def generate_single_mask(self, img: numpy.ndarray, roi: list, threshold: int) -> numpy.ndarray:
  49.         '''
  50.     通过手动选择的ROI区域生成单帧图像的水印蒙版
  51.     :param img: 单帧图像
  52.     :param roi: 手动选择区域坐标
  53.     :param threshold: 二值化阈值
  54.     :return: 水印蒙版
  55.     '''
  56.         # 区域无效，程序退出
  57.         if len(roi) != 4:
  58.             print('NULL ROI!')
  59.             sys.exit()
  60.  
  61.         # 复制单帧灰度图像ROI内像素点
  62.         roi_img = numpy.zeros((img.shape[0], img.shape[1]), numpy.uint8)
  63.         start_x, end_x = int(roi[1]), int(roi[1] + roi[3])
  64.         start_y, end_y = int(roi[0]), int(roi[0] + roi[2])
  65.         gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
  66.         roi_img[start_x:end_x, start_y:end_y] = gray[start_x:end_x, start_y:end_y]
  67.  
  68.         # 阈值分割
  69.         _, mask = cv2.threshold(roi_img, threshold, 255, cv2.THRESH_BINARY)
  70.         return mask
  71.  
  72.     #通过截取视频中多帧图像生成多张水印蒙版，并通过逻辑与计算生成最终的水印蒙版
  73.     def generate_watermark_mask(self, video_path: str) -> numpy.ndarray:
  74.         '''
  75.     截取视频中多帧图像生成多张水印蒙版，通过逻辑与计算生成最终水印蒙版
  76.     :param video_path: 视频文件路径
  77.     :return: 水印蒙版
  78.     '''
  79.         video = cv2.VideoCapture(video_path)
  80.         success, frame = video.read()
  81.         roi = self.select_roi(frame, 'select watermark ROI')
  82.         mask = numpy.ones((frame.shape[0], frame.shape[1]), numpy.uint8)
  83.         mask.fill(255)
  84.  
  85.         step = video.get(cv2.CAP_PROP_FRAME_COUNT) // 5
  86.         index = 0
  87.         while success:
  88.             if index % step == 0:
  89.                 mask = cv2.bitwise_and(mask, self.generate_single_mask(frame, roi, self.threshold))
  90.             success, frame = video.read()
  91.             index += 1
  92.         video.release()
  93.  
  94.         return self.dilate_mask(mask)
  95.  
  96.     #根据手动选择的ROI区域，在单帧图像中生成字幕的蒙版。
  97.     def generate_subtitle_mask(self, frame: numpy.ndarray, roi: list) -> numpy.ndarray:
  98.         '''
  99.     通过手动选择ROI区域生成单帧图像字幕蒙版
  100.     :param frame: 单帧图像
  101.     :param roi: 手动选择区域坐标
  102.     :return: 字幕蒙版
  103.     '''
  104.         mask = self.generate_single_mask(frame, [0, roi[1], frame.shape[1], roi[3]], self.threshold)  # 仅使用ROI横坐标区域
  105.         return self.dilate_mask(mask)
  106.  
  107.     def inpaint_image(self, img: numpy.ndarray, mask: numpy.ndarray) -> numpy.ndarray:
  108.         '''
  109.     修复图像
  110.     :param img: 单帧图像
  111.     :parma mask: 蒙版
  112.     :return: 修复后图像
  113.     '''
  114.         telea = cv2.inpaint(img, mask, 1, cv2.INPAINT_TELEA)
  115.         return telea
  116.  
  117.  
  118.     def merge_audio(self, input_path: str, output_path: str, temp_path: str):
  119.         '''
  120.     合并音频与处理后视频
  121.     :param input_path: 原视频文件路径
  122.     :param output_path: 封装音视频后文件路径
  123.     :param temp_path: 无声视频文件路径
  124.     '''
  125.         with editor.VideoFileClip(input_path) as video:
  126.             audio = video.audio
  127.             with editor.VideoFileClip(temp_path) as opencv_video:
  128.                 clip = opencv_video.set_audio(audio)
  129.                 clip.to_videofile(output_path)
  130.  
  131.     def remove_video_watermark(self):
  132.         '''
  133.     去除视频水印
  134.     '''
  135.         if not os.path.exists(self.output):
  136.             os.makedirs(self.output)
  137.  
  138.         filenames = [os.path.join(self.video_path, i) for i in os.listdir(self.video_path)]
  139.         mask = None
  140.  
  141.         for i, name in enumerate(filenames):
  142.             if i == 0:
  143.                 # 生成水印蒙版
  144.                 mask = self.generate_watermark_mask(name)
  145.  
  146.             # 创建待写入文件对象
  147.             video = cv2.VideoCapture(name)
  148.             fps = video.get(cv2.CAP_PROP_FPS)
  149.             size = (int(video.get(cv2.CAP_PROP_FRAME_WIDTH)), int(video.get(cv2.CAP_PROP_FRAME_HEIGHT)))
  150.             video_writer = cv2.VideoWriter(TEMP_VIDEO, cv2.VideoWriter_fourcc(*'mp4v'), fps, size)
  151.  
  152.             # 逐帧处理图像
  153.             success, frame = video.read()
  154.  
  155.             while success:
  156.                 frame = self.inpaint_image(frame, mask)
  157.                 video_writer.write(frame)
  158.                 success, frame = video.read()
  159.  
  160.             video.release()
  161.             video_writer.release()
  162.  
  163.             # 封装视频
  164.             (_, filename) = os.path.split(name)
  165.             output_path = os.path.join(self.output, filename.split('.')[0] + '_no_watermark.mp4')  # 输出文件路径
  166.             self.merge_audio(name, output_path, TEMP_VIDEO)
  167.  
  168.     if os.path.exists(TEMP_VIDEO):
  169.         os.remove(TEMP_VIDEO)
  170.  
  171.     def remove_video_subtitle(self):
  172.         '''去除视频字幕'''
  173.         if not os.path.exists(self.output):
  174.             os.makedirs(self.output)
  175.  
  176.         filenames = [os.path.join(self.video_path, i) for i in os.listdir(self.video_path)]
  177.         roi = []
  178.  
  179.         for i, name in enumerate(filenames):
  180.             # 创建待写入文件对象
  181.             video = cv2.VideoCapture(name)
  182.             fps = video.get(cv2.CAP_PROP_FPS)
  183.             size = (int(video.get(cv2.CAP_PROP_FRAME_WIDTH)), int(video.get(cv2.CAP_PROP_FRAME_HEIGHT)))
  184.             video_writer = cv2.VideoWriter(TEMP_VIDEO, cv2.VideoWriter_fourcc(*'mp4v'), fps, size)
  185.  
  186.             # 逐帧处理图像
  187.             success, frame = video.read()
  188.             if i == 0:
  189.                 roi = self.select_roi(frame, 'select subtitle ROI')
  190.  
  191.             while success:
  192.                 mask = self.generate_subtitle_mask(frame, roi)
  193.                 frame = self.inpaint_image(frame, mask)
  194.                 video_writer.write(frame)
  195.                 success, frame = video.read()
  196.  
  197.             video.release()
  198.             video_writer.release()
  199.  
  200.             # 封装视频
  201.             (_, filename) = os.path.split(name)
  202.             output_path = os.path.join(OUTPUT_PATH, filename.split('.')[0] + '_no_sub.mp4')  # 输出文件路径
  203.             self.merge_audio(name, output_path, TEMP_VIDEO)
  204.  
  205.         if os.path.exists(TEMP_VIDEO):
  206.             os.remove(TEMP_VIDEO)
  207.  
  208.  # 去水印
  209. video_path = 'video'
  210. output_path = 'output'
  211. remover = WatermarkRemover(video_path,output_path,threshold=80, kernel_size=5)
  212. remover.remove_video_watermark()   
  213. #去字幕
  214. # remover = WatermarkRemover(video_path,output_path,threshold=80, kernel_size=5)
  215. # remover.remove_video_subtitle()

效果一般吧：

声明：本文内容由网友自发贡献，不代表【wpsshop博客】立场，版权归原作者所有，本站不承担相应法律责任。如您发现有侵权的内容，请联系我们。转载请注明出处：https://www.wpsshop.cn/w/知新_RL/article/detail/443447
推荐阅读
相关标签

Copyright © 2003-2013 www.wpsshop.cn 版权所有，并保留所有权利。

  

闽ICP备14008679号