From 1d990613f1c3c2de06a1886ffb91c9849e731dca Mon Sep 17 00:00:00 2001
From: p4ut7fwj5 <3346195219@qq.com>
Date: Sun, 15 Sep 2024 18:47:57 +0800
Subject: [PATCH] ADD file via upload

---
 word_cloud.py | 67 +++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 67 insertions(+)
 create mode 100644 word_cloud.py

diff --git a/word_cloud.py b/word_cloud.py
new file mode 100644
index 0000000..83b7f55
--- /dev/null
+++ b/word_cloud.py
@@ -0,0 +1,67 @@
+"""
+    说明：写入execl，生成词云图
+"""
+# import pandas as pd
+from wordcloud import WordCloud
+import matplotlib.pyplot as plt
+from collections import Counter
+import openpyxl
+import jieba
+import re
+from bullet_screen import get_bullet_screen
+from bv_maker import get_bv, BV_NUM
+jieba.setLogLevel(jieba.logging.INFO)   #去除warning
+
+def confirm(bullet):
+    """
+    确认是否与AI应用相关
+    :param bullet:
+    :return:True/False
+    """
+    keywords = ['人工智能', 'AI', '机器学习', '深度学习', '神经网络', '自动驾驶', '自然语言处理', '智能', 'ai']  # 设置AI应用关键词
+    bullet = bullet.lower()
+    obj = r'(?<![a-zA-Z])(?:' + '|'.join(re.escape(keyword) for keyword in keywords) + r')(?![a-zA-Z])'
+    if re.search(obj, bullet):
+        return True
+    else:
+        return False
+
+def make_word_cloud(bullet_screen_list):
+    """
+    将AI应用相关弹幕写入execcl，并完成词云图展示
+    :param bullet_screen_list:
+    :return:
+    """
+    ai_related_bullet = []
+    for bullet in bullet_screen_list:
+        if confirm(bullet):
+            ai_related_bullet.append(bullet)    #经确认后添加到ai相关弹幕列表
+    # print(ai_related_bullet)
+    top8 =Counter(ai_related_bullet).most_common(8) #找到数量前八的弹幕
+    print(top8) #展示数量前8弹幕
+    #使用openpyxl将前八的弹幕写入execl表
+    wb = openpyxl.Workbook()
+    sheet = wb.active
+    sheet.title = "AI Related Bullet-Screen"
+    sheet.append(["弹幕内容", "出现次数"])
+    for item in top8:
+        sheet.append([item[0], item[1]])
+    wb.save('ai_bullet_screen示例.xlsx')
+    text = ' '.join(ai_related_bullet)
+    cut_text = ' '.join(jieba.cut(text))  # 使用jieba分词
+    wordcloud = WordCloud(  #词云图设置
+        font_path='msyh.ttc',
+        width=800,
+        height=400,
+        background_color='white',
+        colormap='cool'
+    ).generate(cut_text)
+    plt.figure(figsize=(10, 5))
+    plt.imshow(wordcloud, interpolation='bilinear')
+    plt.axis('off')
+    plt.show()  #展示词云图
+
+if __name__ == '__main__':
+    bv_list = get_bv(BV_NUM)
+    bullet_screen_list = get_bullet_screen(bv_list)
+    make_word_cloud(bullet_screen_list)