import re import requests from bs4 import BeautifulSoup from collections import Counter import pandas as pd import time # 定义需要爬取的视频数量和搜索内容 BV_NUM = 300 # 需要获取的视频数量 SEARCH_CONTENT = "2024巴黎奥运会" # 搜索关键词 # 请求头，防止反爬 HEADERS = { 'User-Agent':'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36 Edg/126.0.0.0', "Referer": "https://search.bilibili.com/all?" } # 根据搜索关键词，爬取指定数量的视频BVID def get_bv(num): """ 参数： num: 需要获取的BVID数量 返回值： bv_list, 一个包含视频BVID的集合 """ bv_list = set() # 使用集合存储获取的BVID，以去重 page = 1 # 初始化页码 while len(bv_list) < num: # 构造搜索页面的URL search_url = f"https://search.bilibili.com/all?keyword={SEARCH_CONTENT}&page={page}" response = requests.get(search_url, headers=HEADERS) # 使用正则表达式提取BVID pattern = re.compile(r'aid:.*?bvid:"(?P.*?)",') matches = pattern.finditer(response.text) # 将BVID加入集合 for match in matches: bv_list.add(match.group("bvs")) # 如果达到了指定数量，直接返回结果 if len(bv_list) >= num: return bv_list # 增加页码，继续爬取下一页 page += 1 return bv_list # 通过bv号获取视频cid，进一步获取弹幕内容 def fetch_bullet_screen(bv_list): """ 参数 bv_list: 包含BV号的列表 返回值: 弹幕内容列表 """ my_bullet = [] # 存放所有弹幕的列表 for bv in bv_list: cid_url = f"https://api.bilibili.com/x/player/pagelist?bvid={bv}&jsonp=jsonp" response_cid = requests.get(cid_url, headers=HEADERS) response_cid_json = response_cid.json() # 获取视频的cid cid = response_cid_json.get("data", [{}])[0].get("cid") if not cid: print(f"无法获取 {bv} 的cid") continue # 获取对应cid的弹幕内容 response_bullet = requests.get(f"https://comment.bilibili.com/{cid}.xml", headers=HEADERS) response_bullet.encoding = "utf-8" # 解析XML格式的弹幕数据 soup = BeautifulSoup(response_bullet.text, "xml") danmus = soup.find_all("d") # 将弹幕内容添加到列表中 my_bullet.extend([danmu.text for danmu in danmus]) print(f"已成功爬取 {bv} 的弹幕") return my_bullet # 分析弹幕内容，提取包含AI相关词语的完整句子 def analyze_bullet_screen_with_ai_sentences(my_bullet , ai_keywords): """ 参数1： my_bullet: 弹幕内容列表 参数2： ai_keywords:关键词 返回值: 包含AI关键词的完整句子列表 返回值: 包含AI关键词的列表 返回值: 关键词统计 """ ai_sentences = [] # 包含关键词的完整句子 keyword_only_sentences = [] # 只包含关键词的句子 keyword_counts = Counter() # 计数器，用于统计每个关键词的频次 for bullet in my_bullet: contains_keyword = False for keyword in ai_keywords: if keyword in bullet: contains_keyword = True ai_sentences.append(bullet) keyword_only_sentences.append(keyword) keyword_counts[keyword] += 1 # 更新关键词计数 if contains_keyword: # 只保留包含的关键词 keywords_in_bullet = [keyword for keyword in ai_keywords if keyword in bullet] return ai_sentences, keyword_only_sentences, keyword_counts # 将关键词统计信息保存到Excel文件 def save_keyword_counts_to_excel(keyword_counts, path='keyword_counts.xlsx'): """ 参数1: keyword_counts: Counter对象，包含每个关键词及其出现次数 参数2: path: Excel文件名 """ counts_df = pd.DataFrame(keyword_counts.items(), columns=["关键词", "出现次数"]) # 按出现次数排序，从高到低 counts_df.sort_values(by='出现次数', ascending=False, inplace=True) counts_df.to_excel(path, index=False) print(f"关键词统计已保存到 {path}") # 将包含AI关键词的句子保存到Excel文件 def save_sentences_to_excel(ai_sentences, keyword_only_sentences,path1='ai_sentences.xlsx',path2='keyword_only_sentences.xlsx',choose1=True ,choose2=True): """ 参数1: ai_sentences: 包含AI关键词的句子列表 参数2: keyword_only_sentences: 只包含AI关键字的列表 参数3: path1: Excel文件名,保存AI关键词句子列表 参数4: path2: Excel文件名,保存AI关键词 参数5: choose1: 是否保存path1 参数6: choose2: 是否保存path2 """ if(choose1): ai_df = pd.DataFrame(ai_sentences, columns=["包含关键词的弹幕"]) ai_df.to_excel(path1, index=False) if(choose2): keyword_only_df = pd.DataFrame(keyword_only_sentences, columns=["关键词"]) keyword_only_df.to_excel(path2, index=False) print(f"包含关键词的弹幕句子已保存到 {path1}") print(f"只包含关键词的句子已保存到 {path2}") def main_bullet(): bv_list = get_bv(BV_NUM) # print(bv_list) bullet_screens = fetch_bullet_screen(bv_list) print(f"获取了 {len(bullet_screens)} 条弹幕") ai_keywords = [ "AI", "人工智能", "机器学习", "深度学习", "神经网络", "算法", "智能", "大数据", "自动化", "机器人", "计算机视觉", "自然语言处理", "NLP", "语音识别", "自动驾驶", "边缘计算", "强化学习", "生成对抗网络", "GAN", "迁移学习", "数据挖掘", "语义分析", "图像识别", "深度神经网络", "DNN", "决策树", "随机森林", "集成学习", "模糊逻辑", "专家系统", "计算智能", "大规模并行处理", "分布式系统", "物联网", "IoT", "云计算", "区块链", "量子计算", "图神经网络", "GNN", "人机交互", "HCI", "情感分析", "机器人过程自动化", "RPA", "无人机", "UAV", "智能城市", "云原生", "分布式学习", "元学习", "数字孪生", "自动化运维", "AIOps" ] ai_sentences, keyword_only_sentences, keyword_counts = analyze_bullet_screen_with_ai_sentences(bullet_screens, ai_keywords) print(f"提取了 {len(ai_sentences)} 条包含AI相关关键词的弹幕") save_sentences_to_excel(ai_sentences, keyword_only_sentences, choose2 = False) save_keyword_counts_to_excel(keyword_counts) if __name__ == '__main__': main_bullet() # 获取相关函数运行时间 def main_bullet2(): start_time = time.time() # 开始计时 bv_list = get_bv(BV_NUM) end_time = time.time() # 结束计时 print(f"get_bv函数耗时: {end_time - start_time:.4f} 秒") # print(bv_list) start_time = time.time() # 开始计时 bullet_screens = fetch_bullet_screen(bv_list) # print(f"获取了 {len(bullet_screens)} 条弹幕") end_time = time.time() # 结束计时 print(f"fetch_bullet_screen函数耗时: {end_time - start_time:.4f} 秒") ai_keywords = [ "AI", "人工智能", "机器学习", "深度学习", "神经网络", "算法", "智能", "大数据", "自动化", "机器人", "计算机视觉", "自然语言处理", "NLP", "语音识别", "自动驾驶", "边缘计算", "强化学习", "生成对抗网络", "GAN", "迁移学习", "数据挖掘", "语义分析", "图像识别", "深度神经网络", "DNN", "决策树", "随机森林", "集成学习", "模糊逻辑", "专家系统", "计算智能", "大规模并行处理", "分布式系统", "物联网", "IoT", "云计算", "区块链", "量子计算", "图神经网络", "GNN", "人机交互", "HCI", "情感分析", "机器人过程自动化", "RPA", "无人机", "UAV", "智能城市", "云原生", "分布式学习", "元学习", "数字孪生", "自动化运维", "AIOps" ] start_time = time.time() # 开始计时 ai_sentences, keyword_only_sentences, keyword_counts = analyze_bullet_screen_with_ai_sentences(bullet_screens, ai_keywords) # print(f"提取了 {len(ai_sentences)} 条包含AI相关关键词的弹幕") end_time = time.time() # 结束计时 print(f"analyze_bullet_screen_with_ai_sentences函数耗时: {end_time - start_time:.4f} 秒") start_time = time.time() # 开始计时 save_sentences_to_excel(ai_sentences, keyword_only_sentences, choose2 = False) end_time = time.time() # 结束计时 print(f"save_sentences_to_excel函数耗时: {end_time - start_time:.4f} 秒") start_time = time.time() # 开始计时 save_keyword_counts_to_excel(keyword_counts) end_time = time.time() # 结束计时 print(f"save_keyword_counts_to_excel函数耗时: {end_time - start_time:.4f} 秒")