更新数据截止日2026.01.08

This commit is contained in:
lychang
2026-01-08 01:06:27 +08:00
commit ce02565b09
37 changed files with 12587 additions and 0 deletions

735
chat_session.py Normal file
View File

@@ -0,0 +1,735 @@
import datetime
import json
import os.path
from typing import List, Tuple
import logging
import os
import threading
from concurrent.futures import ThreadPoolExecutor, as_completed
import requests
from requests.adapters import HTTPAdapter
from urllib3.util.retry import Retry
import pytz
from tools import DatabaseManager
from file import ObjectFileManager, LocalFileManager
TZ = pytz.timezone('Asia/Shanghai')
def convert_to_utc(time_str, time_format='%Y-%m-%d %H:%M:%S', source_timezone=None):
# 解析时间字符串为datetime对象
local_dt = datetime.datetime.strptime(time_str, time_format)
# 如果指定了源时区,则将本地时间转换为UTC时间
if source_timezone:
tz = pytz.timezone(source_timezone)
local_dt = tz.localize(local_dt)
utc_dt = local_dt.astimezone(pytz.UTC)
else:
# 如果没有指定源时区,则假设时间字符串已经是UTC时间
utc_dt = pytz.UTC.localize(local_dt)
return utc_dt
class AIAgent:
"""AI代理类,整合了各种AI功能"""
def __init__(self, deepseek_api_key: str = "sk-c2f04b4d635b4df0aace0432f3c6d6f4",
max_retries: int = 3, timeout: int = 30):
"""初始化AI助手
Args:
deepseek_api_key: DeepSeek API密钥
max_retries: 最大重试次数
timeout: 请求超时时间(秒)
"""
self.deepseek_api_key = deepseek_api_key
self.timeout = timeout
# 配置日志
self._setup_logging()
# 创建带重试机制的会话
self.session = self._create_session(max_retries)
# 配置常量
self.BASE_REPORT_URL = "https://a.cmdp.cn/basiceg/v1/json/tpldev"
self.CACHE_DIR = "cache"
def _setup_logging(self):
"""配置日志"""
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
)
self.logger = logging.getLogger(__name__)
@staticmethod
def _create_session(max_retries: int) -> requests.Session:
"""创建带重试机制的会话
Args:
max_retries: 最大重试次数
Returns:
配置好的会话对象
"""
session = requests.Session()
# 配置重试策略
retry_strategy = Retry(
total=max_retries,
status_forcelist=[429, 500, 502, 503, 504],
backoff_factor=1
)
# 为HTTP和HTTPS适配器添加重试机制
adapter = HTTPAdapter(max_retries=retry_strategy)
session.mount("http://", adapter)
session.mount("https://", adapter)
return session
def _make_request(self, method: str, url: str, **kwargs):
"""发送HTTP请求(带错误处理)
Args:
method: HTTP方法
url: 请求URL
**kwargs: 其他请求参数
Returns:
JSON响应或None
"""
try:
# 设置默认超时
if 'timeout' not in kwargs:
kwargs['timeout'] = self.timeout
response = self.session.request(method, url, **kwargs)
response.raise_for_status()
self.logger.info(f"请求成功: {url}")
return response.json()
except requests.exceptions.RequestException as e:
self.logger.error(f"请求失败: {url} - {str(e)}")
return {}
except json.JSONDecodeError as e:
self.logger.error(f"JSON解析失败: {url} - {str(e)}")
return {}
def chat_deepseek(self, message: str, role: str = "", model: str = "deepseek-chat") -> str:
"""与DeepSeek聊天
Args:
message: 用户消息
role: 系统角色
model: 使用的模型
Returns:
AI回复内容
"""
url = "https://api.deepseek.com/chat/completions"
headers = {
"Content-Type": "application/json",
"Authorization": f"Bearer {self.deepseek_api_key}"
}
messages = []
if role:
messages.append({"role": "system", "content": role})
messages.append({"role": "user", "content": message})
body = {
"model": model,
"messages": messages,
"stream": False
}
result = self._make_request("POST", url, json=body, headers=headers)
if result and "choices" in result:
return result.get('choices', [{}])[0]['message']['content']
self.logger.warning("DeepSeek API调用失败或返回空结果")
return ""
agent = AIAgent(max_retries=1, timeout=180)
class ChatSessionManager:
"""聊天会话管理器类"""
def __init__(self, db_manager: DatabaseManager):
"""
初始化聊天会话管理器
Args:
db_manager: 数据库管理工具
"""
self.files_dir = os.path.join("", "files")
self.messages_dir = os.path.join("", "messages")
self.cache_path = None
# 初始化文件管理器
self.obj = ObjectFileManager()
self.lfm = LocalFileManager(self.files_dir)
self.dbm = db_manager
self.users = []
self.groups = []
self.members = {}
self.messages = []
self.files = {}
# 确保目录存在
os.makedirs(self.files_dir, exist_ok=True)
os.makedirs(self.messages_dir, exist_ok=True)
def execute_query(self, query: str, params=None, fetch_all=True):
result = []
if fetch_all:
rows = self.dbm.execute_query(query, params)
for row in rows:
result.append(list(row))
else:
rows = self.dbm.execute_query(query, params, False)
result = list(rows)
return result
def execute_event(self, query: str, params=None):
return self.dbm.execute_event(query, params)
def status(self):
row = self.dbm.execute_query("SELECT datetime FROM information order by datetime desc limit 1", fetch_all=False)
if row:
return row[0]
else:
return None
def _download_file(self, file_info, files_dir, history, lock):
"""
下载单个文件的线程函数
Args:
file_info: 文件信息
files_dir: 文件保存目录
history: 聊天历史字典
lock: 线程锁
"""
try:
path = file_info["path"].replace("s3://", "")
file_content = self.obj.get_file(path)
if path.endswith(".json"):
data = json.loads(file_content)
with lock:
history[data["msgid"]] = data
else:
# 保存非JSON文件
filename = path.split("/")[-1]
with open(os.path.join(files_dir, filename), "wb") as f:
f.write(file_content)
return path, True
except Exception as e:
return path, False
def download_chat_history(self, year: int, month: int, day: int, hour: int, max_workers: int = 5) -> bool:
"""
下载指定时间的聊天历史(多线程版本)
Args:
year: 年份
month: 月份
day: 日期
hour: 小时
max_workers: 最大线程数
Returns:
下载是否成功
"""
message_history_name = f"{year}_{month:02d}_{day:02d}_{hour:02d}.json"
print(f"Downloading chat history for {year}-{month:02d}-{day:02d}-{hour:02d} (多线程模式)")
try:
file_list = self.obj.get_file_list(f"zysl-log/wework/chat_msg/{year}/{month}/{day}/{hour}/")
print(f"共计{len(file_list)}个文件,使用{max_workers}个线程下载")
history = {}
failed_files = []
if file_list:
lock = threading.Lock()
# 使用线程池下载文件
with ThreadPoolExecutor(max_workers=max_workers) as executor:
# 提交所有下载任务
future_to_file = {
executor.submit(self._download_file, file_info, self.files_dir, history, lock): file_info
for file_info in file_list
}
# 处理完成的任务
completed_count = 0
total_count = len(file_list)
for future in as_completed(future_to_file):
file_info = future_to_file[future]
completed_count += 1
try:
path, success = future.result()
if success:
print(f"[{completed_count}/{total_count}] 下载成功: {path}")
else:
print(f"[{completed_count}/{total_count}] 下载失败: {path}")
failed_files.append(path)
except Exception as e:
print(f"[{completed_count}/{total_count}] 处理异常: {file_info['path']} - {str(e)}")
failed_files.append(file_info['path'])
# 检查是否有失败的文件
if failed_files:
print(f"警告: {len(failed_files)}个文件下载失败")
print("失败文件列表:", failed_files)
# 保存聊天历史JSON文件
output_path = os.path.join(self.messages_dir, message_history_name)
with open(output_path, "w", encoding="utf-8") as f:
json.dump(history, f, ensure_ascii=False, indent=2)
print(f"Downloading chat history for {year}-{month:02d}-{day:02d}-{hour:02d} complete")
print(f"成功下载: {len(file_list) - len(failed_files)}/{len(file_list)} 个文件")
# 如果有失败的文件,返回False
return len(failed_files) == 0
except Exception as e:
error_file = os.path.join(self.messages_dir, f"{message_history_name}.unfinished")
with open(error_file, "w", encoding="utf-8") as f:
f.write(str(e))
print(
f"Downloading chat history for {year}-{month:02d}-{day:02d}-{hour:02d} failed: please check the {message_history_name}.unfinished")
return False
def _parse_message(self, data, is_mixed: bool = False, idx: int = 0):
msg_type = data["msgtype"]
if msg_type == "text":
content = data['text']["content"]
elif msg_type in ["file", "image", "video", "voice", "emotion", "meeting_voice_call"]:
file_data = data[msg_type]
suffixs = [f"_{idx}", ".png", ".jpg", ".gif", ".mp4", "_.png", "_.jpg", "_.gif", "_.mp4",
("_" + file_data["filename"]) if msg_type == "file" else ""]
file_path = ""
for suffix in suffixs:
temp_path = f"{data['msgid']}_mixed{suffix}" if is_mixed else f"{data['msgid']}_{msg_type}{suffix}"
if self.lfm.exists(temp_path):
file_path = temp_path
break
if file_path != "":
ext_mapping = {
"file": (file_data.get('fileext', ''), "filesize"),
"image": ("png", file_data.get("filesize", 0)),
"video": ("mp4", file_data.get("filesize", 0)),
"voice": ("wav", file_data.get("voice_size", 0)),
"emotion": ("gif" if file_data.get("type", 0) == 1 else "png", file_data.get("imagesize", 0)),
"meeting_voice_call": ("wav", self.lfm.get_size(file_path)),
}
file_info = {
"id": file_data.get("md5sum", self.lfm.get_md5(file_path)),
"size": ext_mapping[msg_type][1],
"extension": ext_mapping[msg_type][0],
"content": file_path,
"original": self.cache_path + file_path,
"type": msg_type
}
self.files[file_info["id"]] = file_info
content = file_info["id"]
else:
file_path = f"{data['msgid']}_{msg_type}_0"
file_info = {
"id": file_data.get("md5sum", file_path),
"size": file_data.get('filesize', 0),
"extension": file_data.get('fileext', "unknown"),
"content": file_path,
"original": self.cache_path + file_path,
"type": msg_type
}
self.files[file_info["id"]] = file_info
content = file_info["id"]
elif msg_type == "link":
data = data[msg_type]
content = f'{data["description"]}{data["title"]}({data["link_url"]})'
else:
print(f"未知消息类型: {msg_type}")
content = json.dumps(data, ensure_ascii=False)
return content
def _parse_mixed_message(self, param, idx: int):
tp = param["msgtype"]
content = self._parse_message(param, True, idx)
if tp in ["file", "image", "video", "voice", "emotion", "meeting_voice_call"]:
self.files[content]["content"] = f'{param["msgid"]}_mixed_{idx}'
idx += 1
return idx, content
def parse_chat_history(self, year: int, month: int, day: int, hour: int):
"""
解析聊天历史数据
Args:
year: 年份
month: 月份
day: 日期
hour: 小时
Returns:
(用户列表, 群组列表, 成员映射, 消息列表, 文件映射)
"""
message_history_name = f"{year}_{month:02d}_{day:02d}_{hour:02d}.json"
history_path = os.path.join(self.messages_dir, message_history_name)
# 如果文件不存在,先下载
if not os.path.exists(history_path):
self.download_chat_history(year, month, day, hour,10)
with open(history_path, "r", encoding="utf-8") as f:
message_history = json.load(f)
try:
for message_data in message_history.values():
# 只处理发送的消息
if message_data.get("action") != "send":
continue
# 提取基本信息
from_user = message_data["from"]
to_users = message_data["tolist"]
room_id = message_data.get("roomid", "")
# 更新用户列表
self.users.extend([from_user] + to_users)
self.users = list(set(self.users))
# 确定接收者类型
to_type = 1 if room_id else 0
receiver = room_id if to_type == 1 else to_users[0]
# 处理群组信息
if room_id:
if room_id not in self.members:
self.members[room_id] = to_users + [from_user]
else:
self.members[room_id].extend(to_users + [from_user])
self.members[room_id] = list(set(self.members[room_id]))
if room_id not in self.groups:
self.groups.append(room_id)
# 构建消息对象
message = {
"id": message_data["msgid"],
"type": message_data["msgtype"],
"time": message_data["msgtime"],
"toType": to_type,
"from": from_user,
"to": receiver
}
# 根据消息类型处理内容
msg_type = message_data["msgtype"]
if msg_type == "mixed":
temp = []
idx = 0
for t in message_data["mixed"]["item"]:
idx, content = self._parse_mixed_message({"msgid": message_data["msgid"],
"msgtype": t["type"],
t["type"]: json.loads(t["content"])}, idx)
temp.append([t["type"], content])
message["content"] = json.dumps(temp, ensure_ascii=False)
else:
message["content"] = self._parse_message(message_data)
self.messages.append(message)
except Exception as e:
print(f"解析聊天记录失败: {e}")
raise e
def save_to_sqlite(self, sql: str, data: List[Tuple]) -> None:
"""
批量保存数据到SQLite数据库
Args:
sql: SQL语句
data: 数据列表
"""
if not data:
return
print(f"开始插入 {len(data)} 条记录")
self.dbm.execute_event(sql, data)
print(f"成功插入 {len(data)} 条记录")
def save_information_to_sqlite(self, date_time: str, message_num: int, file_num: int, user_num: int,
group_num: int) -> None:
"""
保存会话信息到SQLite数据库
Args:
date_time: 会话时间
message_num: 消息数量
file_num: 文件数量
user_num: 用户数量
group_num: 群组数量
"""
self.save_to_sqlite('''
INSERT OR IGNORE INTO "information"
(datetime, message, file, user, "group")
VALUES (?, ?, ?, ?, ?)
''', [(date_time, message_num, file_num, user_num, group_num)])
def save_all_to_sqlite(self) -> None:
"""
保存所有聊天数据到SQLite数据库
"""
# 保存文件信息
file_data = []
for file_info in self.files.values():
file_data.append((
file_info.get('id'),
file_info.get('type'),
file_info.get('content'),
file_info.get('original', ''),
file_info.get('extension', ''),
file_info.get('size', 0)
))
self.save_to_sqlite('''
INSERT OR IGNORE INTO file
(id, type, content, original, extension, size)
VALUES (?, ?, ?, ?, ?, ?)
''', file_data)
# 保存用户信息
user_data = [(user, user) for user in self.users]
self.save_to_sqlite('''
INSERT OR IGNORE INTO user
(id, name)
VALUES (?, ?)
''', user_data)
# 保存消息信息
message_data = []
for message in self.messages:
message_data.append((
message.get('id'),
message.get('type'),
message.get('time'),
message.get('toType'),
message.get('from'),
message.get('to'),
message.get('content')
))
self.save_to_sqlite('''
INSERT OR IGNORE INTO message
(id, type, time, toType, sender, receiver, content)
VALUES (?, ?, ?, ?, ?, ?, ?)
''', message_data)
# 保存群组信息
group_data = [(group, group) for group in self.groups]
self.save_to_sqlite('''
INSERT OR IGNORE INTO "group"
(id, name)
VALUES (?, ?)
''', group_data)
# 保存成员信息
member_data = []
for group_id, group_users in self.members.items():
for user in group_users:
member_data.append((group_id, user))
self.save_to_sqlite('''
INSERT OR IGNORE INTO "member"
(groupId, userId)
VALUES (?, ?)
''', member_data)
def process_chat_session(self, year: int, month: int, day: int, hour: int) -> bool:
"""
处理完整的聊天会话流程
Args:
year: 年份
month: 月份
day: 日期
hour: 小时
Returns:
处理是否成功
"""
self.users = []
self.groups = []
self.members = {}
self.messages = []
self.files = {}
try:
self.cache_path = f"zysl-log/wework/chat_msg/{year}/{month}/{day}/{hour}/"
# 解析聊天历史
self.parse_chat_history(year, month, day, hour)
# 保存到数据库
self.save_all_to_sqlite()
self.save_information_to_sqlite(f"{year}-{month:02d}-{day:02d}-{hour:02d}", len(self.messages),
len(self.files), len(self.users), len(self.groups))
print(f"成功处理 {year}-{month:02d}-{day:02d}-{hour:02d} 的聊天会话")
return True
except Exception as e:
print(f"处理聊天会话失败: {e}")
return False
def get_chat_history(self, start_time_str: str, end_time_str: str) -> list:
"""
获取指定时间的聊天历史记录
Args:
start_time_str: 开始时间,格式为 "YYYY-MM-DD HH:MM:SS"
end_time_str: 结束时间,格式为 "YYYY-MM-DD HH:MM:SS"
Returns:
聊天历史记录列表
"""
start_time = convert_to_utc(start_time_str, source_timezone="Asia/Shanghai").timestamp()
end_time = convert_to_utc(end_time_str, source_timezone="Asia/Shanghai").timestamp()
res = self.dbm.execute_query('''
SELECT message.time, message.content
FROM message
WHERE (message.type = 'text'
or message.type = 'mixed')
AND message.time >= ?
AND message.time <= ?
''', (str(start_time).replace(".0", "000"), str(end_time).replace(".0", "000")))
return res
@staticmethod
def analyze_qa_content(group_messages) -> dict:
group_qa = {}
for group_id in group_messages:
print(f"群ID: {group_id}")
print(f"消息数量: {len(group_messages[group_id])}")
content = ""
if len(group_messages[group_id]) <= 2:
continue
else:
message_mapping = {}
for msg in sorted(group_messages[group_id], key=lambda i: i[1]):
message_mapping[msg[4]] = msg[2]
content += json.dumps({
"user": msg[1],
"time": msg[2],
"content": msg[3],
"message_id": msg[4]
}, ensure_ascii=False) + ",\n"
template = f"""
以下是群中的消息记录
```json
[
{content}
]
```
""" + """
找到其中的问题和回答,如果没有回答就只整理问题,问题是与产品填报系统,上市公司,证券政策相关的问题
问题可能不只由一句话构成,回答也可能不只由一句话构成
多个问题需要分开放置
不需要任何解释说明只返回json格式
输出格式
```json
[
{
"question": ["message_id",...],
"answer": ["message_id",...]
}
]
```
"""
answer = agent.chat_deepseek(template)
answer = answer.replace("```json", "").replace("```", "") if "```json" in answer else answer
try:
answer = json.loads(answer)
except:
print(answer)
answer = []
for ans in answer:
time_list = [message_mapping.get(a, "") for a in ans["answer"]] + [message_mapping.get(q, "") for q
in ans["question"]]
start_time = min(time_list)
end_time = max(time_list)
ans["start_time"] = start_time
ans["end_time"] = end_time
group_qa[group_id] = answer
return group_qa
def save_qa_to_sqlite(self, group_qa: dict):
"""
将问答数据保存到 SQLite 数据库
Args:
group_qa: 问答数据字典,键为群ID,值为问答列表
"""
data = []
index_id = self.execute_query(f"SELECT id FROM qa order by id desc limit 1", fetch_all=False)[0]
for r in group_qa:
qas = group_qa.get(r, {})
for qa in qas:
Q = qa["question"]
A = qa["answer"]
index_id += 1
data.append((index_id, r, json.dumps(Q, ensure_ascii=False), json.dumps(A, ensure_ascii=False),
qa["start_time"],
qa["end_time"]))
self.execute_event('''
insert into qa (id, groupId, question, answer, start_time, end_time)
values (?, ?, ?, ?, ?, ?)
''', data)
def update_qa_data(self, start_time_str: str, end_time_str: str):
"""
更新指定时间范围内的问答内容
Args:
start_time_str: 开始时间,格式为 "YYYY-MM-DD HH:MM:SS"
end_time_str: 结束时间,格式为 "YYYY-MM-DD HH:MM:SS"
"""
start_time = convert_to_utc(start_time_str, source_timezone="Asia/Shanghai").timestamp()
end_time = convert_to_utc(end_time_str, source_timezone="Asia/Shanghai").timestamp()
data = self.execute_query('''
SELECT receiver, sender, time, content, id
FROM message
WHERE message.time >= ?
AND message.time <= ?
AND toType = 1
AND type in ('text', 'mixed', 'chatrecord')
''', (str(start_time).replace(".0", "000"), str(end_time).replace(".0", "000")))
group_messages = {}
for i in data:
group_id = i[0]
if group_id not in group_messages:
group_messages[group_id] = []
group_messages[group_id].append(i)
group_qa = self.analyze_qa_content(group_messages)
self.save_qa_to_sqlite(group_qa)