优化关键词

This commit is contained in:
liuwei
2025-05-21 17:51:31 +08:00
parent 8f88e1bd9b
commit 06f84efbb5

View File

@@ -39,10 +39,17 @@ class InterventionBot:
self.messages_per_minute_threshold = reply_threshold.get("messages_per_minute_threshold", 3)
self.analysis_window_minutes = reply_threshold.get("analysis_window_minutes", 5)
# 冷却时间配置(秒)
self.cooldown_seconds = 20
self.last_intervention_time = None
# 最近话题记录
self.last_topic = None
self.last_topic_time = None
self.topic_cooldown = timedelta(seconds=60)
def is_morning_window(self, timestamp):
"""检查是否在早晨签到时间窗口"""
try:
# 处理不同类型的时间戳
if isinstance(timestamp, float):
message_datetime = datetime.fromtimestamp(timestamp)
message_time = message_datetime.time()
@@ -56,14 +63,12 @@ class InterventionBot:
return False
else:
return False
return self.morning_window[0] <= message_time <= self.morning_window[1]
except Exception as e:
print(f"[早晨窗口检测] 错误: {e}")
return False
def detect_topic(self, message):
"""检测消息所属话题类型"""
if not isinstance(message, str):
return None
message_lower = message.lower()
@@ -80,64 +85,47 @@ class InterventionBot:
return None
def rule_morning_signin(self, timestamp, messages):
"""规则1早晨签到"""
return self.is_morning_window(timestamp) and any("签到" in msg or "" in msg for msg in messages[-5:])
def rule_hot_topic(self, message, messages):
"""规则2热点话题参与"""
return self.detect_topic(message) == "hot_topic" and len(
[m for m in messages[-5:] if self.detect_topic(m) == "hot_topic"]) >= 3
def rule_tech_discussion(self, message, messages):
"""规则3技术性讨论"""
return self.detect_topic(message) == "tech"
def rule_fish_discussion(self, message, messages):
"""规则4养鱼话题"""
return self.detect_topic(message) == "fish"
def rule_mechanism_interaction(self, message, messages):
"""规则5群内机制互动"""
return self.detect_topic(message) == "mechanism"
def rule_humor_tease(self, message, messages):
"""规则6幽默与调侃"""
return any(emoji in message for emoji in self.emojis) or "哈哈" in message or len(
[m for m in messages[-5:] if any(e in m for e in self.emojis)]) >= 2
def rule_news_reaction(self, message, messages):
"""规则7猎奇或社会新闻反应"""
return self.detect_topic(message) == "news"
def rule_high_reply_rate(self, timestamp, chat_log):
"""规则8高回复频率每分钟消息数超过阈值"""
try:
# 检查timestamp类型并转换
if isinstance(timestamp, float):
# 如果是浮点数Unix时间戳转换为datetime对象
current_time = datetime.fromtimestamp(timestamp)
elif isinstance(timestamp, str):
# 如果是字符串,尝试解析
try:
current_time = datetime.strptime(timestamp, "%Y-%m-%d %H:%M:%S")
except ValueError:
# 如果字符串格式不匹配,尝试将其转换为浮点数再处理
try:
current_time = datetime.fromtimestamp(float(timestamp))
except:
# 如果转换失败,使用当前时间
current_time = datetime.now()
else:
# 其他类型,使用当前时间
current_time = datetime.now()
window_start = current_time - timedelta(minutes=self.analysis_window_minutes)
# 计算时间窗口内的消息数
recent_messages = []
for msg in chat_log:
try:
# 同样处理消息时间戳
msg_timestamp = msg.get("timestamp")
if isinstance(msg_timestamp, float):
msg_time = datetime.fromtimestamp(msg_timestamp)
@@ -157,14 +145,11 @@ class InterventionBot:
except (ValueError, KeyError, TypeError):
continue
# 如果消息太少,不触发
if len(recent_messages) < self.messages_per_minute_threshold:
return False
# 计算消息频率
messages_per_minute = len(recent_messages) / self.analysis_window_minutes
# 记录日志,便于调试
if messages_per_minute >= self.messages_per_minute_threshold:
print(f"[高频率检测] 当前消息频率: {messages_per_minute:.2f}/分钟,阈值: {self.messages_per_minute_threshold}/分钟")
@@ -174,7 +159,6 @@ class InterventionBot:
return False
def should_intervene(self, timestamp, message, messages, chat_log):
"""判断是否需要介入"""
rules = [
self.rule_morning_signin,
self.rule_hot_topic,
@@ -186,28 +170,50 @@ class InterventionBot:
self.rule_high_reply_rate
]
current_time = datetime.now() if not isinstance(timestamp, datetime) else timestamp
if isinstance(timestamp, str):
try:
current_time = datetime.strptime(timestamp, "%Y-%m-%d %H:%M:%S")
except:
current_time = datetime.now()
elif isinstance(timestamp, float):
current_time = datetime.fromtimestamp(timestamp)
if self.last_intervention_time:
if (current_time - self.last_intervention_time).total_seconds() < self.cooldown_seconds:
return False
current_topic = self.detect_topic(message)
if self.last_topic == current_topic and self.last_topic_time:
if (current_time - self.last_topic_time) < self.topic_cooldown:
return False
for rule in rules:
if rule == self.rule_morning_signin:
if rule(timestamp, messages):
self.last_intervention_time = current_time
self.last_topic = current_topic
self.last_topic_time = current_time
return True
elif rule == self.rule_high_reply_rate:
if rule(timestamp, chat_log):
self.last_intervention_time = current_time
self.last_topic = current_topic
self.last_topic_time = current_time
return True
elif rule(message, messages):
self.last_intervention_time = current_time
self.last_topic = current_topic
self.last_topic_time = current_time
return True
return False
def process_message(self, timestamp, message, messages, chat_log):
"""处理单条消息,返回介入状态"""
if self.should_intervene(timestamp, message, messages, chat_log):
return True
return False
return self.should_intervene(timestamp, message, messages, chat_log)
def process_chat_log(self, chat_log):
"""处理聊天记录,返回每条消息的介入状态"""
messages = [line["message"] for line in chat_log]
results = []
for i, line in enumerate(chat_log):
timestamp = line["timestamp"]
message = line["message"]
@@ -217,13 +223,10 @@ class InterventionBot:
"message": message,
"intervention": intervention
})
return results
# 示例用法
if __name__ == "__main__":
# 模拟聊天记录
sample_chat_log = [
{"timestamp": "2025-03-14 08:06:38", "user_id": "Jyunere", "message": "白嫖马斯克每个月150刀的额度应该能玩很久了。"},
{"timestamp": "2025-03-14 08:06:54", "user_id": "Jyunere", "message": "啥情况?卷了?"},
@@ -231,8 +234,7 @@ if __name__ == "__main__":
{"timestamp": "2025-03-14 09:12:28", "user_id": "Jyunere", "message": "我同事的鸿蒙确实流畅。"},
{"timestamp": "2025-03-14 09:35:21", "user_id": "Jyunere", "message": "垃圾MIUI"},
{"timestamp": "2025-05-21 14:31:57", "user_id": "wxid_4re8ddo26dxb52", "message": "年轻人随随便便就能深蹲200"},
{"timestamp": "2025-05-21 14:32:20", "user_id": "liu79830956",
"message": "@水牛 过分了啊,报错还扣积分 赔我200"},
{"timestamp": "2025-05-21 14:32:20", "user_id": "liu79830956", "message": "@水牛 过分了啊,报错还扣积分 赔我200"},
{"timestamp": "2025-05-21 14:32:39", "user_id": "Jyunere", "message": "哈哈,识别到指令了。"},
{"timestamp": "2025-05-21 14:32:42", "user_id": "wxid_z8uo70zywfpn12", "message": "检测到天 气了"},
{"timestamp": "2025-05-21 14:35:08", "user_id": "liu79830956", "message": "这螺蛳粉估计要明天也吃不上了[旺柴]"}