def __init__(self, mode=0):
self.mode = mode # (0: training, 1: testing, 2: real world implementation)
self.MLP = 0 # multi-layer-perceptron
self.RNN = 1 # recurrent-neural-network
self.CNN = 2 # convolution-neural-network
self.training = 0 # training phase
self.testing = 1 # testing phase
self.real_world = 2 # real world implementation
self.paths = None
self.number_of_samples = None
self.number_of_classes = None
self.image_width = None
self.image_height = None
self.image_type = None
self.video_path = None
self.video_scale = None
self.video_overlap_size = None
self.video_patch_size = None
self.Initial()
self.Data = Dataset(self.paths, self.number_of_samples,
self.number_of_classes, self.image_width,
self.image_height, self.image_type)
self.VidData = VideoData(self.video_path, self.video_patch_size,
self.video_overlap_size)
def get_video_list(self, limit=0):
max_loops = 50
max_results = 50
if limit:
max_loops = math.floor(limit / 50)
max_results = limit % 50
end_of_videos = False
videos = []
timestamp = str(datetime.utcnow().replace(microsecond=0).isoformat()) \
+ "Z"
while not end_of_videos:
result_limit = max_results if not max_loops else 50
result = self.service.search().list(
part="snippet",
channelId=self.channel_id,
maxResults=result_limit,
order="date",
type='video',
publishedBefore=timestamp).execute()
for video in result.get("items", []):
videos.append(VideoData())
videos[-1].id = video.get("id").get("videoId")
snippet = video.get("snippet", [])
videos[-1].title = snippet.get("title")
videos[-1].published = snippet.get("publishedAt")
# Loop End
max_loops = max_loops - 1
if len(result.get("items", [])) < max_results or max_loops < 0:
end_of_videos = True
else:
timestamp = videos[-1].published
datestamp = datetime.strptime(timestamp,
"%Y-%m-%dT%H:%M:%S.%fZ")
datestamp = datestamp - timedelta(0, 1)
timestamp = str(datestamp.replace(microsecond=0).isoformat()) \
+ "Z"
return videos
def characterArt_demo():
video_url = 'https://www.bilibili.com/video/av79839682'
video = VideoData(video_url)
print('正在获取弹幕')
video.getDanmaku()
print('正在获取评论')
video.getVideoComment()
video_title = video.video_info['title']
# video_title = '「鬼灭之刃」温柔到让人哭泣的心——心碎者的柔歌'
video_path = 'C:\\Users\\William\\Desktop\\bili\\videos_data\\{}'.format(
video_title)
os.chdir(video_path)
img_file = 'cover.jpg'
danmaku_file = 'danmaku.txt'
replies_file = 'replies.txt'
print('正在生成字符画')
characterart.createCharacterArt(img_file, danmaku_file, replies_file)
print('已成功保存字符画')
from videodata import VideoData video_url = 'https://www.bilibili.com/video/av80565276' # 指定需要爬取的视频url video = VideoData(video_url) # 初始化视频对象，生成对应的文件夹并获取基本信息与封面 video.getDanmaku() # 获取视频的弹幕 video.getVideoComment() # 获取视频的评论 video.getVideoTags() # 获取视频标签 video.getVideoData() # 获取弹幕评论和标签 import bilibilidata # getRank(url=None, rank_type='all') bilibilidata.getRank() # 默认获取全站排行榜 # 指定rank_type参数：all、origin、bangumi、cinema、rookie bilibilidata.getRank(rank_type='bangumi') # 指定排行榜url rank_url = 'https://www.bilibili.com/ranking/cinema/23/0/3' bilibilidata.getRank(url=rank_url)
def tagsWordcloud_demo():
bilibilidata.getRank()
os.chdir(r'C:\Users\William\Desktop\bili\bilibili_data')
rank_data = pd.read_csv('all.csv')
print('成功获取排行榜数据')
rank_url = rank_data.url
rank_title = rank_data.title
url_head = r'https:'
for i in range(len(rank_url)):
rank_url[i] = url_head + rank_url[i]
# 爬取排行榜每一个视频的基本信息
print('正在爬取排行榜每一个视频的基本信息')
process_num = 1
for each in rank_url:
try:
os.chdir(r'C:\Users\William\Desktop\bili')
video = VideoData(each)
video.getVideoTags()
print(process_num)
process_num += 1
except:
process_num += 1
# 读取每一个视频的标签
tags = list()
os.chdir(r'C:\Users\William\Desktop\bili')
for each in rank_title:
try:
os.chdir(
r'C:\Users\William\Desktop\bili\videos_data\{}'.format(each))
with open('tags.txt', encoding='utf-8') as file:
tags.append(file.readlines())
except Exception as exc:
pass
tags_txt = str()
for each_video_tags in tags:
for each_tag in each_video_tags:
tags_txt += each_tag
tags_txt = tags_txt.replace('\n', ' ')
# 加载停用词
path = 'C:\\Users\\William\\Desktop\\bili\\百度停用词表.txt'
with open(path, encoding='utf-8') as file:
file_content = file.readlines()
stopwords = set()
for i in range(len(file_content)):
file_content[i] = file_content[i].replace("\n", "")
stopwords.add(file_content[i])
# 生成视频标签词云
wc = WordCloud(background_color='white',
font_path=font_path,
stopwords=stopwords,
width=1920,
height=1080,
scale=2)
wc.generate(tags_txt)
plt.imshow(wc)
plt.axis('off')
os.chdir('C:\\Users\\William\\Desktop\\bili')
plt.savefig('./排行榜标签词云.png')
def getDanmaku_demo():
video_url = 'https://www.bilibili.com/video/av78273046'
video3 = VideoData(video_url)
video3.getDanmaku()
def getVideoComment_demo():
video_url = 'https://www.bilibili.com/bangumi/play/ep285946'
video2 = VideoData(video_url)
video2.getVideoComment()
def getVideoInfo_demo():
video_url = 'https://www.bilibili.com/video/av68733672'
video1 = VideoData(video_url)