1384 lines
56 KiB
Python
1384 lines
56 KiB
Python
import asyncio
|
||
from collections import defaultdict
|
||
from datetime import datetime, timedelta
|
||
from typing import Any
|
||
|
||
from src.common.database.sqlalchemy_database_api import db_get, db_query, db_save
|
||
from src.common.database.sqlalchemy_models import LLMUsage, Messages, OnlineTime
|
||
from src.common.logger import get_logger
|
||
from src.manager.async_task_manager import AsyncTask
|
||
from src.manager.local_store_manager import local_storage
|
||
|
||
logger = get_logger("maibot_statistic")
|
||
|
||
# 彻底异步化：删除原同步包装器 _sync_db_get，所有数据库访问统一使用 await db_get。
|
||
|
||
|
||
# 统计数据的键
|
||
TOTAL_REQ_CNT = "total_requests"
|
||
TOTAL_COST = "total_cost"
|
||
REQ_CNT_BY_TYPE = "requests_by_type"
|
||
REQ_CNT_BY_USER = "requests_by_user"
|
||
REQ_CNT_BY_MODEL = "requests_by_model"
|
||
REQ_CNT_BY_MODULE = "requests_by_module"
|
||
IN_TOK_BY_TYPE = "in_tokens_by_type"
|
||
IN_TOK_BY_USER = "in_tokens_by_user"
|
||
IN_TOK_BY_MODEL = "in_tokens_by_model"
|
||
IN_TOK_BY_MODULE = "in_tokens_by_module"
|
||
OUT_TOK_BY_TYPE = "out_tokens_by_type"
|
||
OUT_TOK_BY_USER = "out_tokens_by_user"
|
||
OUT_TOK_BY_MODEL = "out_tokens_by_model"
|
||
OUT_TOK_BY_MODULE = "out_tokens_by_module"
|
||
TOTAL_TOK_BY_TYPE = "tokens_by_type"
|
||
TOTAL_TOK_BY_USER = "tokens_by_user"
|
||
TOTAL_TOK_BY_MODEL = "tokens_by_model"
|
||
TOTAL_TOK_BY_MODULE = "tokens_by_module"
|
||
COST_BY_TYPE = "costs_by_type"
|
||
COST_BY_USER = "costs_by_user"
|
||
COST_BY_MODEL = "costs_by_model"
|
||
COST_BY_MODULE = "costs_by_module"
|
||
ONLINE_TIME = "online_time"
|
||
TOTAL_MSG_CNT = "total_messages"
|
||
MSG_CNT_BY_CHAT = "messages_by_chat"
|
||
TIME_COST_BY_TYPE = "time_costs_by_type"
|
||
TIME_COST_BY_USER = "time_costs_by_user"
|
||
TIME_COST_BY_MODEL = "time_costs_by_model"
|
||
TIME_COST_BY_MODULE = "time_costs_by_module"
|
||
AVG_TIME_COST_BY_TYPE = "avg_time_costs_by_type"
|
||
AVG_TIME_COST_BY_USER = "avg_time_costs_by_user"
|
||
AVG_TIME_COST_BY_MODEL = "avg_time_costs_by_model"
|
||
AVG_TIME_COST_BY_MODULE = "avg_time_costs_by_module"
|
||
STD_TIME_COST_BY_TYPE = "std_time_costs_by_type"
|
||
STD_TIME_COST_BY_USER = "std_time_costs_by_user"
|
||
STD_TIME_COST_BY_MODEL = "std_time_costs_by_model"
|
||
STD_TIME_COST_BY_MODULE = "std_time_costs_by_module"
|
||
|
||
|
||
class OnlineTimeRecordTask(AsyncTask):
|
||
"""在线时间记录任务"""
|
||
|
||
def __init__(self):
|
||
super().__init__(task_name="Online Time Record Task", run_interval=60)
|
||
|
||
self.record_id: int | None = None
|
||
"""记录ID"""
|
||
|
||
async def run(self): # sourcery skip: use-named-expression
|
||
try:
|
||
current_time = datetime.now()
|
||
extended_end_time = current_time + timedelta(minutes=1)
|
||
|
||
if self.record_id:
|
||
# 如果有记录，则更新结束时间
|
||
updated_rows = await db_query(
|
||
model_class=OnlineTime,
|
||
query_type="update",
|
||
filters={"id": self.record_id},
|
||
data={"end_timestamp": extended_end_time},
|
||
)
|
||
if updated_rows == 0:
|
||
# Record might have been deleted or ID is stale, try to find/create
|
||
self.record_id = None
|
||
|
||
if not self.record_id:
|
||
# 查找最近一分钟内的记录
|
||
recent_threshold = current_time - timedelta(minutes=1)
|
||
recent_records = await db_get(
|
||
model_class=OnlineTime,
|
||
filters={"end_timestamp": {"$gte": recent_threshold}},
|
||
order_by="-end_timestamp",
|
||
limit=1,
|
||
single_result=True,
|
||
)
|
||
|
||
if recent_records:
|
||
# 找到近期记录，更新它
|
||
self.record_id = recent_records["id"]
|
||
await db_query(
|
||
model_class=OnlineTime,
|
||
query_type="update",
|
||
filters={"id": self.record_id},
|
||
data={"end_timestamp": extended_end_time},
|
||
)
|
||
else:
|
||
# 创建新记录
|
||
new_record = await db_save(
|
||
model_class=OnlineTime,
|
||
data={
|
||
"timestamp": str(current_time),
|
||
"duration": 5, # 初始时长为5分钟
|
||
"start_timestamp": current_time,
|
||
"end_timestamp": extended_end_time,
|
||
},
|
||
)
|
||
if new_record:
|
||
self.record_id = new_record["id"]
|
||
|
||
except Exception as e:
|
||
logger.error(f"在线时间记录失败，错误信息：{e}")
|
||
|
||
|
||
def _format_online_time(online_seconds: int) -> str:
|
||
"""
|
||
格式化在线时间
|
||
:param online_seconds: 在线时间（秒）
|
||
:return: 格式化后的在线时间字符串
|
||
"""
|
||
total_online_time = timedelta(seconds=online_seconds)
|
||
|
||
days = total_online_time.days
|
||
hours = total_online_time.seconds // 3600
|
||
minutes = (total_online_time.seconds // 60) % 60
|
||
seconds = total_online_time.seconds % 60
|
||
if days > 0:
|
||
# 如果在线时间超过1天，则格式化为"X天X小时X分钟"
|
||
return f"{total_online_time.days}天{hours}小时{minutes}分钟{seconds}秒"
|
||
elif hours > 0:
|
||
# 如果在线时间超过1小时，则格式化为"X小时X分钟X秒"
|
||
return f"{hours}小时{minutes}分钟{seconds}秒"
|
||
else:
|
||
# 其他情况格式化为"X分钟X秒"
|
||
return f"{minutes}分钟{seconds}秒"
|
||
|
||
|
||
class StatisticOutputTask(AsyncTask):
|
||
"""统计输出任务"""
|
||
|
||
SEP_LINE = "-" * 84
|
||
|
||
def __init__(self, record_file_path: str = "maibot_statistics.html"):
|
||
# 延迟300秒启动，运行间隔300秒
|
||
super().__init__(task_name="Statistics Data Output Task", wait_before_start=0, run_interval=300)
|
||
|
||
self.name_mapping: dict[str, tuple[str, float]] = {}
|
||
"""
|
||
联系人/群聊名称映射 {聊天ID: (联系人/群聊名称, 记录时间（timestamp）)}
|
||
注：设计记录时间的目的是方便更新名称，使联系人/群聊名称保持最新
|
||
"""
|
||
|
||
self.record_file_path: str = record_file_path
|
||
"""
|
||
记录文件路径
|
||
"""
|
||
|
||
now = datetime.now()
|
||
if "deploy_time" in local_storage:
|
||
# 如果存在部署时间，则使用该时间作为全量统计的起始时间
|
||
deploy_time = datetime.fromtimestamp(local_storage["deploy_time"]) # type: ignore
|
||
else:
|
||
# 否则，使用最大时间范围，并记录部署时间为当前时间
|
||
deploy_time = datetime(2000, 1, 1)
|
||
local_storage["deploy_time"] = now.timestamp()
|
||
|
||
self.stat_period: list[tuple[str, timedelta, str]] = [
|
||
("all_time", now - deploy_time, "自部署以来"), # 必须保留"all_time"
|
||
("last_7_days", timedelta(days=7), "最近7天"),
|
||
("last_24_hours", timedelta(days=1), "最近24小时"),
|
||
("last_3_hours", timedelta(hours=3), "最近3小时"),
|
||
("last_hour", timedelta(hours=1), "最近1小时"),
|
||
]
|
||
"""
|
||
统计时间段 [(统计名称, 统计时间段, 统计描述), ...]
|
||
"""
|
||
|
||
def _statistic_console_output(self, stats: dict[str, Any], now: datetime):
|
||
"""
|
||
输出统计数据到控制台
|
||
:param stats: 统计数据
|
||
:param now: 基准当前时间
|
||
"""
|
||
# 输出最近一小时的统计数据
|
||
|
||
output = [
|
||
self.SEP_LINE,
|
||
f" 最近1小时的统计数据 (自{now.strftime('%Y-%m-%d %H:%M:%S')}开始，详细信息见文件：{self.record_file_path})",
|
||
self.SEP_LINE,
|
||
self._format_total_stat(stats["last_hour"]),
|
||
"",
|
||
self._format_model_classified_stat(stats["last_hour"]),
|
||
"",
|
||
self._format_chat_stat(stats["last_hour"]),
|
||
self.SEP_LINE,
|
||
"",
|
||
]
|
||
|
||
logger.info("\n" + "\n".join(output))
|
||
|
||
async def run(self):
|
||
try:
|
||
now = datetime.now()
|
||
logger.info("正在收集统计数据(异步)...")
|
||
stats = await self._collect_all_statistics(now)
|
||
logger.info("统计数据收集完成")
|
||
self._statistic_console_output(stats, now)
|
||
await self._generate_html_report(stats, now)
|
||
logger.info("统计数据输出完成")
|
||
except Exception as e:
|
||
logger.exception(f"输出统计数据过程中发生异常，错误信息：{e}")
|
||
|
||
async def run_async_background(self):
|
||
"""
|
||
备选方案：完全异步后台运行统计输出
|
||
使用此方法可以让统计任务完全非阻塞
|
||
"""
|
||
|
||
async def _async_collect_and_output():
|
||
try:
|
||
now = datetime.now()
|
||
logger.info("(后台) 正在收集统计数据(异步)...")
|
||
stats = await self._collect_all_statistics(now)
|
||
self._statistic_console_output(stats, now)
|
||
await self._generate_html_report(stats, now)
|
||
logger.info("统计数据后台输出完成")
|
||
except Exception as e:
|
||
logger.exception(f"后台统计数据输出过程中发生异常：{e}")
|
||
|
||
# 创建后台任务，立即返回
|
||
asyncio.create_task(_async_collect_and_output())
|
||
|
||
# -- 以下为统计数据收集方法 --
|
||
|
||
@staticmethod
|
||
async def _collect_model_request_for_period(collect_period: list[tuple[str, datetime]]) -> dict[str, Any]:
|
||
"""
|
||
收集指定时间段的LLM请求统计数据
|
||
|
||
:param collect_period: 统计时间段
|
||
"""
|
||
if not collect_period:
|
||
return {}
|
||
|
||
# 排序-按照时间段开始时间降序排列（最晚的时间段在前）
|
||
collect_period.sort(key=lambda x: x[1], reverse=True)
|
||
|
||
stats = {
|
||
period_key: {
|
||
TOTAL_REQ_CNT: 0,
|
||
REQ_CNT_BY_TYPE: defaultdict(int),
|
||
REQ_CNT_BY_USER: defaultdict(int),
|
||
REQ_CNT_BY_MODEL: defaultdict(int),
|
||
REQ_CNT_BY_MODULE: defaultdict(int),
|
||
IN_TOK_BY_TYPE: defaultdict(int),
|
||
IN_TOK_BY_USER: defaultdict(int),
|
||
IN_TOK_BY_MODEL: defaultdict(int),
|
||
IN_TOK_BY_MODULE: defaultdict(int),
|
||
OUT_TOK_BY_TYPE: defaultdict(int),
|
||
OUT_TOK_BY_USER: defaultdict(int),
|
||
OUT_TOK_BY_MODEL: defaultdict(int),
|
||
OUT_TOK_BY_MODULE: defaultdict(int),
|
||
TOTAL_TOK_BY_TYPE: defaultdict(int),
|
||
TOTAL_TOK_BY_USER: defaultdict(int),
|
||
TOTAL_TOK_BY_MODEL: defaultdict(int),
|
||
TOTAL_TOK_BY_MODULE: defaultdict(int),
|
||
TOTAL_COST: 0.0,
|
||
COST_BY_TYPE: defaultdict(float),
|
||
COST_BY_USER: defaultdict(float),
|
||
COST_BY_MODEL: defaultdict(float),
|
||
COST_BY_MODULE: defaultdict(float),
|
||
TIME_COST_BY_TYPE: defaultdict(list),
|
||
TIME_COST_BY_USER: defaultdict(list),
|
||
TIME_COST_BY_MODEL: defaultdict(list),
|
||
TIME_COST_BY_MODULE: defaultdict(list),
|
||
AVG_TIME_COST_BY_TYPE: defaultdict(float),
|
||
AVG_TIME_COST_BY_USER: defaultdict(float),
|
||
AVG_TIME_COST_BY_MODEL: defaultdict(float),
|
||
AVG_TIME_COST_BY_MODULE: defaultdict(float),
|
||
STD_TIME_COST_BY_TYPE: defaultdict(float),
|
||
STD_TIME_COST_BY_USER: defaultdict(float),
|
||
STD_TIME_COST_BY_MODEL: defaultdict(float),
|
||
STD_TIME_COST_BY_MODULE: defaultdict(float),
|
||
}
|
||
for period_key, _ in collect_period
|
||
}
|
||
|
||
# 以最早的时间戳为起始时间获取记录
|
||
query_start_time = collect_period[-1][1]
|
||
records = (
|
||
await db_get(
|
||
model_class=LLMUsage,
|
||
filters={"timestamp": {"$gte": query_start_time}},
|
||
order_by="-timestamp",
|
||
)
|
||
or []
|
||
)
|
||
|
||
for record in records:
|
||
if not isinstance(record, dict):
|
||
continue
|
||
|
||
record_timestamp = record.get("timestamp")
|
||
if isinstance(record_timestamp, str):
|
||
record_timestamp = datetime.fromisoformat(record_timestamp)
|
||
|
||
if not record_timestamp:
|
||
continue
|
||
|
||
for idx, (_, period_start) in enumerate(collect_period):
|
||
if record_timestamp >= period_start:
|
||
for period_key, _ in collect_period[idx:]:
|
||
stats[period_key][TOTAL_REQ_CNT] += 1
|
||
|
||
request_type = record.get("request_type") or "unknown"
|
||
user_id = record.get("user_id") or "unknown"
|
||
model_name = record.get("model_name") or "unknown"
|
||
|
||
# 提取模块名：如果请求类型包含"."，取第一个"."之前的部分
|
||
module_name = request_type.split(".")[0] if "." in request_type else request_type
|
||
|
||
stats[period_key][REQ_CNT_BY_TYPE][request_type] += 1
|
||
stats[period_key][REQ_CNT_BY_USER][user_id] += 1
|
||
stats[period_key][REQ_CNT_BY_MODEL][model_name] += 1
|
||
stats[period_key][REQ_CNT_BY_MODULE][module_name] += 1
|
||
|
||
prompt_tokens = record.get("prompt_tokens") or 0
|
||
completion_tokens = record.get("completion_tokens") or 0
|
||
total_tokens = prompt_tokens + completion_tokens
|
||
|
||
stats[period_key][IN_TOK_BY_TYPE][request_type] += prompt_tokens
|
||
stats[period_key][IN_TOK_BY_USER][user_id] += prompt_tokens
|
||
stats[period_key][IN_TOK_BY_MODEL][model_name] += prompt_tokens
|
||
stats[period_key][IN_TOK_BY_MODULE][module_name] += prompt_tokens
|
||
|
||
stats[period_key][OUT_TOK_BY_TYPE][request_type] += completion_tokens
|
||
stats[period_key][OUT_TOK_BY_USER][user_id] += completion_tokens
|
||
stats[period_key][OUT_TOK_BY_MODEL][model_name] += completion_tokens
|
||
stats[period_key][OUT_TOK_BY_MODULE][module_name] += completion_tokens
|
||
|
||
stats[period_key][TOTAL_TOK_BY_TYPE][request_type] += total_tokens
|
||
stats[period_key][TOTAL_TOK_BY_USER][user_id] += total_tokens
|
||
stats[period_key][TOTAL_TOK_BY_MODEL][model_name] += total_tokens
|
||
stats[period_key][TOTAL_TOK_BY_MODULE][module_name] += total_tokens
|
||
|
||
cost = record.get("cost") or 0.0
|
||
stats[period_key][TOTAL_COST] += cost
|
||
stats[period_key][COST_BY_TYPE][request_type] += cost
|
||
stats[period_key][COST_BY_USER][user_id] += cost
|
||
stats[period_key][COST_BY_MODEL][model_name] += cost
|
||
stats[period_key][COST_BY_MODULE][module_name] += cost
|
||
|
||
# 收集time_cost数据
|
||
time_cost = record.get("time_cost") or 0.0
|
||
if time_cost > 0: # 只记录有效的time_cost
|
||
stats[period_key][TIME_COST_BY_TYPE][request_type].append(time_cost)
|
||
stats[period_key][TIME_COST_BY_USER][user_id].append(time_cost)
|
||
stats[period_key][TIME_COST_BY_MODEL][model_name].append(time_cost)
|
||
stats[period_key][TIME_COST_BY_MODULE][module_name].append(time_cost)
|
||
break
|
||
|
||
# 计算平均耗时和标准差
|
||
for period_key in stats:
|
||
for category in [REQ_CNT_BY_TYPE, REQ_CNT_BY_USER, REQ_CNT_BY_MODEL, REQ_CNT_BY_MODULE]:
|
||
time_cost_key = f"time_costs_by_{category.split('_')[-1]}"
|
||
avg_key = f"avg_time_costs_by_{category.split('_')[-1]}"
|
||
std_key = f"std_time_costs_by_{category.split('_')[-1]}"
|
||
|
||
for item_name in stats[period_key][category]:
|
||
time_costs = stats[period_key][time_cost_key].get(item_name, [])
|
||
if time_costs:
|
||
# 计算平均耗时
|
||
avg_time_cost = sum(time_costs) / len(time_costs)
|
||
stats[period_key][avg_key][item_name] = round(avg_time_cost, 3)
|
||
|
||
# 计算标准差
|
||
if len(time_costs) > 1:
|
||
variance = sum((x - avg_time_cost) ** 2 for x in time_costs) / len(time_costs)
|
||
std_time_cost = variance**0.5
|
||
stats[period_key][std_key][item_name] = round(std_time_cost, 3)
|
||
else:
|
||
stats[period_key][std_key][item_name] = 0.0
|
||
else:
|
||
stats[period_key][avg_key][item_name] = 0.0
|
||
stats[period_key][std_key][item_name] = 0.0
|
||
return stats
|
||
|
||
@staticmethod
|
||
async def _collect_online_time_for_period(
|
||
collect_period: list[tuple[str, datetime]], now: datetime
|
||
) -> dict[str, Any]:
|
||
"""
|
||
收集指定时间段的在线时间统计数据
|
||
|
||
:param collect_period: 统计时间段
|
||
"""
|
||
if not collect_period:
|
||
return {}
|
||
|
||
collect_period.sort(key=lambda x: x[1], reverse=True)
|
||
|
||
stats = {
|
||
period_key: {
|
||
ONLINE_TIME: 0.0,
|
||
}
|
||
for period_key, _ in collect_period
|
||
}
|
||
|
||
query_start_time = collect_period[-1][1]
|
||
records = (
|
||
await db_get(
|
||
model_class=OnlineTime,
|
||
filters={"end_timestamp": {"$gte": query_start_time}},
|
||
order_by="-end_timestamp",
|
||
)
|
||
or []
|
||
)
|
||
|
||
for record in records:
|
||
if not isinstance(record, dict):
|
||
continue
|
||
|
||
record_end_timestamp = record.get("end_timestamp")
|
||
if isinstance(record_end_timestamp, str):
|
||
record_end_timestamp = datetime.fromisoformat(record_end_timestamp)
|
||
|
||
record_start_timestamp = record.get("start_timestamp")
|
||
if isinstance(record_start_timestamp, str):
|
||
record_start_timestamp = datetime.fromisoformat(record_start_timestamp)
|
||
|
||
if not record_end_timestamp or not record_start_timestamp:
|
||
continue
|
||
|
||
for idx, (_, period_boundary_start) in enumerate(collect_period):
|
||
if record_end_timestamp >= period_boundary_start:
|
||
# Calculate effective end time for this record in relation to 'now'
|
||
effective_end_time = min(record_end_timestamp, now)
|
||
|
||
for period_key, current_period_start_time in collect_period[idx:]:
|
||
# Determine the portion of the record that falls within this specific statistical period
|
||
overlap_start = max(record_start_timestamp, current_period_start_time)
|
||
overlap_end = effective_end_time # Already capped by 'now' and record's own end
|
||
|
||
if overlap_end > overlap_start:
|
||
stats[period_key][ONLINE_TIME] += (overlap_end - overlap_start).total_seconds()
|
||
break
|
||
return stats
|
||
|
||
async def _collect_message_count_for_period(self, collect_period: list[tuple[str, datetime]]) -> dict[str, Any]:
|
||
"""
|
||
收集指定时间段的消息统计数据
|
||
|
||
:param collect_period: 统计时间段
|
||
"""
|
||
if not collect_period:
|
||
return {}
|
||
|
||
collect_period.sort(key=lambda x: x[1], reverse=True)
|
||
|
||
stats = {
|
||
period_key: {
|
||
TOTAL_MSG_CNT: 0,
|
||
MSG_CNT_BY_CHAT: defaultdict(int),
|
||
}
|
||
for period_key, _ in collect_period
|
||
}
|
||
|
||
query_start_timestamp = collect_period[-1][1].timestamp() # Messages.time is a DoubleField (timestamp)
|
||
records = (
|
||
await db_get(
|
||
model_class=Messages,
|
||
filters={"time": {"$gte": query_start_timestamp}},
|
||
order_by="-time",
|
||
)
|
||
or []
|
||
)
|
||
|
||
for message in records:
|
||
if not isinstance(message, dict):
|
||
continue
|
||
message_time_ts = message.get("time") # This is a float timestamp
|
||
|
||
if not message_time_ts:
|
||
continue
|
||
|
||
chat_id = None
|
||
chat_name = None
|
||
|
||
# Logic based on SQLAlchemy model structure, aiming to replicate original intent
|
||
if message.get("chat_info_group_id"):
|
||
chat_id = f"g{message['chat_info_group_id']}"
|
||
chat_name = message.get("chat_info_group_name") or f"群{message['chat_info_group_id']}"
|
||
elif message.get("user_id"): # Fallback to sender's info for chat_id if not a group_info based chat
|
||
# This uses the message SENDER's ID as per original logic's fallback
|
||
chat_id = f"u{message['user_id']}" # SENDER's user_id
|
||
chat_name = message.get("user_nickname") # SENDER's nickname
|
||
else:
|
||
# If neither group_id nor sender_id is available for chat identification
|
||
logger.warning(f"Message (PK: {message.get('id', 'N/A')}) lacks group_id and user_id for chat stats.")
|
||
continue
|
||
|
||
if not chat_id: # Should not happen if above logic is correct
|
||
continue
|
||
|
||
# Update name_mapping
|
||
if chat_id in self.name_mapping:
|
||
if chat_name != self.name_mapping[chat_id][0] and message_time_ts > self.name_mapping[chat_id][1]:
|
||
self.name_mapping[chat_id] = (chat_name, message_time_ts)
|
||
else:
|
||
self.name_mapping[chat_id] = (chat_name, message_time_ts)
|
||
|
||
for idx, (_, period_start_dt) in enumerate(collect_period):
|
||
if message_time_ts >= period_start_dt.timestamp():
|
||
for period_key, _ in collect_period[idx:]:
|
||
stats[period_key][TOTAL_MSG_CNT] += 1
|
||
stats[period_key][MSG_CNT_BY_CHAT][chat_id] += 1
|
||
break
|
||
return stats
|
||
|
||
async def _collect_all_statistics(self, now: datetime) -> dict[str, dict[str, Any]]:
|
||
"""
|
||
收集各时间段的统计数据
|
||
:param now: 基准当前时间
|
||
"""
|
||
|
||
last_all_time_stat = None
|
||
|
||
if "last_full_statistics" in local_storage:
|
||
# 如果存在上次完整统计数据，则使用该数据进行增量统计
|
||
last_stat: dict[str, Any] = local_storage["last_full_statistics"] # 上次完整统计数据 # type: ignore
|
||
|
||
self.name_mapping = last_stat["name_mapping"] # 上次完整统计数据的名称映射
|
||
last_all_time_stat = last_stat["stat_data"] # 上次完整统计的统计数据
|
||
last_stat_timestamp = datetime.fromtimestamp(last_stat["timestamp"]) # 上次完整统计数据的时间戳
|
||
self.stat_period = [item for item in self.stat_period if item[0] != "all_time"] # 删除"所有时间"的统计时段
|
||
self.stat_period.append(("all_time", now - last_stat_timestamp, "自部署以来的"))
|
||
|
||
stat_start_timestamp = [(period[0], now - period[1]) for period in self.stat_period]
|
||
|
||
stat = {item[0]: {} for item in self.stat_period}
|
||
|
||
model_req_stat, online_time_stat, message_count_stat = await asyncio.gather(
|
||
self._collect_model_request_for_period(stat_start_timestamp),
|
||
self._collect_online_time_for_period(stat_start_timestamp, now),
|
||
self._collect_message_count_for_period(stat_start_timestamp),
|
||
)
|
||
|
||
# 统计数据合并
|
||
# 合并三类统计数据
|
||
for period_key, _ in stat_start_timestamp:
|
||
stat[period_key].update(model_req_stat[period_key])
|
||
stat[period_key].update(online_time_stat[period_key])
|
||
stat[period_key].update(message_count_stat[period_key])
|
||
|
||
if last_all_time_stat:
|
||
# 若存在上次完整统计数据，则将其与当前统计数据合并
|
||
for key, val in last_all_time_stat.items():
|
||
# 确保当前统计数据中存在该key
|
||
if key not in stat["all_time"]:
|
||
continue
|
||
|
||
if isinstance(val, dict):
|
||
# 是字典类型，则进行合并
|
||
for sub_key, sub_val in val.items():
|
||
# 普通的数值或字典合并
|
||
if sub_key in stat["all_time"][key]:
|
||
# 检查是否为嵌套的字典类型（如版本统计）
|
||
if isinstance(sub_val, dict) and isinstance(stat["all_time"][key][sub_key], dict):
|
||
# 合并嵌套字典
|
||
for nested_key, nested_val in sub_val.items():
|
||
if nested_key in stat["all_time"][key][sub_key]:
|
||
stat["all_time"][key][sub_key][nested_key] += nested_val
|
||
else:
|
||
stat["all_time"][key][sub_key][nested_key] = nested_val
|
||
else:
|
||
# 普通数值累加
|
||
stat["all_time"][key][sub_key] += sub_val
|
||
else:
|
||
stat["all_time"][key][sub_key] = sub_val
|
||
else:
|
||
# 直接合并
|
||
stat["all_time"][key] += val
|
||
|
||
# 更新上次完整统计数据的时间戳
|
||
# 将所有defaultdict转换为普通dict以避免类型冲突
|
||
clean_stat_data = self._convert_defaultdict_to_dict(stat["all_time"])
|
||
local_storage["last_full_statistics"] = {
|
||
"name_mapping": self.name_mapping,
|
||
"stat_data": clean_stat_data,
|
||
"timestamp": now.timestamp(),
|
||
}
|
||
|
||
return stat
|
||
|
||
def _convert_defaultdict_to_dict(self, data):
|
||
# sourcery skip: dict-comprehension, extract-duplicate-method, inline-immediately-returned-variable, merge-duplicate-blocks
|
||
"""递归转换defaultdict为普通dict"""
|
||
if isinstance(data, defaultdict):
|
||
# 转换defaultdict为普通dict
|
||
result = {}
|
||
for key, value in data.items():
|
||
result[key] = self._convert_defaultdict_to_dict(value)
|
||
return result
|
||
elif isinstance(data, dict):
|
||
# 递归处理普通dict
|
||
result = {}
|
||
for key, value in data.items():
|
||
result[key] = self._convert_defaultdict_to_dict(value)
|
||
return result
|
||
else:
|
||
# 其他类型直接返回
|
||
return data
|
||
|
||
# -- 以下为统计数据格式化方法 --
|
||
|
||
@staticmethod
|
||
def _format_total_stat(stats: dict[str, Any]) -> str:
|
||
"""
|
||
格式化总统计数据
|
||
"""
|
||
|
||
output = [
|
||
f"总在线时间: {_format_online_time(stats[ONLINE_TIME])}",
|
||
f"总消息数: {stats[TOTAL_MSG_CNT]}",
|
||
f"总请求数: {stats[TOTAL_REQ_CNT]}",
|
||
f"总花费: {stats[TOTAL_COST]:.4f}¥",
|
||
"",
|
||
]
|
||
|
||
return "\n".join(output)
|
||
|
||
@staticmethod
|
||
def _format_model_classified_stat(stats: dict[str, Any]) -> str:
|
||
"""
|
||
格式化按模型分类的统计数据
|
||
"""
|
||
if stats[TOTAL_REQ_CNT] <= 0:
|
||
return ""
|
||
data_fmt = "{:<32} {:>10} {:>12} {:>12} {:>12} {:>9.4f}¥ {:>10} {:>10}"
|
||
|
||
output = [
|
||
" 模型名称 调用次数 输入Token 输出Token Token总量 累计花费 平均耗时(秒) 标准差(秒)",
|
||
]
|
||
for model_name, count in sorted(stats[REQ_CNT_BY_MODEL].items()):
|
||
name = f"{model_name[:29]}..." if len(model_name) > 32 else model_name
|
||
in_tokens = stats[IN_TOK_BY_MODEL][model_name]
|
||
out_tokens = stats[OUT_TOK_BY_MODEL][model_name]
|
||
tokens = stats[TOTAL_TOK_BY_MODEL][model_name]
|
||
cost = stats[COST_BY_MODEL][model_name]
|
||
avg_time_cost = stats[AVG_TIME_COST_BY_MODEL][model_name]
|
||
std_time_cost = stats[STD_TIME_COST_BY_MODEL][model_name]
|
||
output.append(
|
||
data_fmt.format(name, count, in_tokens, out_tokens, tokens, cost, avg_time_cost, std_time_cost)
|
||
)
|
||
|
||
output.append("")
|
||
return "\n".join(output)
|
||
|
||
def _format_chat_stat(self, stats: dict[str, Any]) -> str:
|
||
"""
|
||
格式化聊天统计数据
|
||
"""
|
||
if stats[TOTAL_MSG_CNT] <= 0:
|
||
return ""
|
||
output = ["聊天消息统计:", " 联系人/群组名称 消息数量"]
|
||
output.extend(
|
||
f"{self.name_mapping[chat_id][0][:32]:<32} {count:>10}"
|
||
for chat_id, count in sorted(stats[MSG_CNT_BY_CHAT].items())
|
||
)
|
||
output.append("")
|
||
return "\n".join(output)
|
||
|
||
@staticmethod
|
||
def _get_chat_display_name_from_id(chat_id: str) -> str:
|
||
"""从chat_id获取显示名称"""
|
||
try:
|
||
# 首先尝试从chat_stream获取真实群组名称
|
||
from src.chat.message_receive.chat_stream import get_chat_manager
|
||
|
||
chat_manager = get_chat_manager()
|
||
|
||
if chat_id in chat_manager.streams:
|
||
stream = chat_manager.streams[chat_id]
|
||
if stream.group_info and hasattr(stream.group_info, "group_name"):
|
||
group_name = stream.group_info.group_name
|
||
if group_name and group_name.strip():
|
||
return group_name.strip()
|
||
elif stream.user_info and hasattr(stream.user_info, "user_nickname"):
|
||
user_name = stream.user_info.user_nickname
|
||
if user_name and user_name.strip():
|
||
return user_name.strip()
|
||
|
||
# 如果从chat_stream获取失败，尝试解析chat_id格式
|
||
if chat_id.startswith("g"):
|
||
return f"群聊{chat_id[1:]}"
|
||
elif chat_id.startswith("u"):
|
||
return f"用户{chat_id[1:]}"
|
||
else:
|
||
return chat_id
|
||
except Exception as e:
|
||
logger.warning(f"获取聊天显示名称失败: {e}")
|
||
return chat_id
|
||
|
||
# 移除_generate_versions_tab方法
|
||
|
||
async def _generate_html_report(self, stat: dict[str, Any], now: datetime):
|
||
"""
|
||
生成HTML格式的统计报告
|
||
:param stat: 统计数据
|
||
:param now: 基准当前时间
|
||
:return: HTML格式的统计报告
|
||
"""
|
||
|
||
# 移除版本对比内容相关tab和内容
|
||
tab_list = [
|
||
f'<button class="tab-link" onclick="showTab(event, \'{period[0]}\')">{period[2]}</button>'
|
||
for period in self.stat_period
|
||
]
|
||
tab_list.append('<button class="tab-link" onclick="showTab(event, \'charts\')">数据图表</button>')
|
||
|
||
def _format_stat_data(stat_data: dict[str, Any], div_id: str, start_time: datetime) -> str:
|
||
"""
|
||
格式化一个时间段的统计数据到html div块
|
||
:param stat_data: 统计数据
|
||
:param div_id: div的ID
|
||
:param start_time: 统计时间段开始时间
|
||
"""
|
||
# format总在线时间
|
||
|
||
# 按模型分类统计
|
||
model_rows = "\n".join(
|
||
[
|
||
f"<tr>"
|
||
f"<td>{model_name}</td>"
|
||
f"<td>{count}</td>"
|
||
f"<td>{stat_data[IN_TOK_BY_MODEL][model_name]}</td>"
|
||
f"<td>{stat_data[OUT_TOK_BY_MODEL][model_name]}</td>"
|
||
f"<td>{stat_data[TOTAL_TOK_BY_MODEL][model_name]}</td>"
|
||
f"<td>{stat_data[COST_BY_MODEL][model_name]:.4f} ¥</td>"
|
||
f"<td>{stat_data[AVG_TIME_COST_BY_MODEL][model_name]:.3f} 秒</td>"
|
||
f"<td>{stat_data[STD_TIME_COST_BY_MODEL][model_name]:.3f} 秒</td>"
|
||
f"</tr>"
|
||
for model_name, count in sorted(stat_data[REQ_CNT_BY_MODEL].items())
|
||
]
|
||
)
|
||
# 按请求类型分类统计
|
||
type_rows = "\n".join(
|
||
[
|
||
f"<tr>"
|
||
f"<td>{req_type}</td>"
|
||
f"<td>{count}</td>"
|
||
f"<td>{stat_data[IN_TOK_BY_TYPE][req_type]}</td>"
|
||
f"<td>{stat_data[OUT_TOK_BY_TYPE][req_type]}</td>"
|
||
f"<td>{stat_data[TOTAL_TOK_BY_TYPE][req_type]}</td>"
|
||
f"<td>{stat_data[COST_BY_TYPE][req_type]:.4f} ¥</td>"
|
||
f"<td>{stat_data[AVG_TIME_COST_BY_TYPE][req_type]:.3f} 秒</td>"
|
||
f"<td>{stat_data[STD_TIME_COST_BY_TYPE][req_type]:.3f} 秒</td>"
|
||
f"</tr>"
|
||
for req_type, count in sorted(stat_data[REQ_CNT_BY_TYPE].items())
|
||
]
|
||
)
|
||
# 按模块分类统计
|
||
module_rows = "\n".join(
|
||
[
|
||
f"<tr>"
|
||
f"<td>{module_name}</td>"
|
||
f"<td>{count}</td>"
|
||
f"<td>{stat_data[IN_TOK_BY_MODULE][module_name]}</td>"
|
||
f"<td>{stat_data[OUT_TOK_BY_MODULE][module_name]}</td>"
|
||
f"<td>{stat_data[TOTAL_TOK_BY_MODULE][module_name]}</td>"
|
||
f"<td>{stat_data[COST_BY_MODULE][module_name]:.4f} ¥</td>"
|
||
f"<td>{stat_data[AVG_TIME_COST_BY_MODULE][module_name]:.3f} 秒</td>"
|
||
f"<td>{stat_data[STD_TIME_COST_BY_MODULE][module_name]:.3f} 秒</td>"
|
||
f"</tr>"
|
||
for module_name, count in sorted(stat_data[REQ_CNT_BY_MODULE].items())
|
||
]
|
||
)
|
||
|
||
# 聊天消息统计
|
||
chat_rows = "\n".join(
|
||
[
|
||
f"<tr><td>{self.name_mapping[chat_id][0]}</td><td>{count}</td></tr>"
|
||
for chat_id, count in sorted(stat_data[MSG_CNT_BY_CHAT].items())
|
||
]
|
||
)
|
||
# 生成HTML
|
||
return f"""
|
||
<div id=\"{div_id}\" class=\"tab-content\">
|
||
<p class=\"info-item\">
|
||
<strong>统计时段: </strong>
|
||
{start_time.strftime("%Y-%m-%d %H:%M:%S")} ~ {now.strftime("%Y-%m-%d %H:%M:%S")}
|
||
</p>
|
||
<p class=\"info-item\"><strong>总在线时间: </strong>{_format_online_time(stat_data[ONLINE_TIME])}</p>
|
||
<p class=\"info-item\"><strong>总消息数: </strong>{stat_data[TOTAL_MSG_CNT]}</p>
|
||
<p class=\"info-item\"><strong>总请求数: </strong>{stat_data[TOTAL_REQ_CNT]}</p>
|
||
<p class=\"info-item\"><strong>总花费: </strong>{stat_data[TOTAL_COST]:.4f} ¥</p>
|
||
|
||
<h2>按模型分类统计</h2>
|
||
<table>
|
||
<tr><th>模块名称</th><th>调用次数</th><th>输入Token</th><th>输出Token</th><th>Token总量</th><th>累计花费</th><th>平均耗时(秒)</th><th>标准差(秒)</th></tr>
|
||
<tbody>
|
||
{model_rows}
|
||
</tbody>
|
||
</table>
|
||
|
||
<h2>按模块分类统计</h2>
|
||
<table>
|
||
<thead>
|
||
<tr><th>模块名称</th><th>调用次数</th><th>输入Token</th><th>输出Token</th><th>Token总量</th><th>累计花费</th><th>平均耗时(秒)</th><th>标准差(秒)</th></tr>
|
||
</thead>
|
||
<tbody>
|
||
{module_rows}
|
||
</tbody>
|
||
</table>
|
||
|
||
<h2>按请求类型分类统计</h2>
|
||
<table>
|
||
<thead>
|
||
<tr><th>请求类型</th><th>调用次数</th><th>输入Token</th><th>输出Token</th><th>Token总量</th><th>累计花费</th><th>平均耗时(秒)</th><th>标准差(秒)</th></tr>
|
||
</thead>
|
||
<tbody>
|
||
{type_rows}
|
||
</tbody>
|
||
</table>
|
||
|
||
<h2>聊天消息统计</h2>
|
||
<table>
|
||
<thead>
|
||
<tr><th>联系人/群组名称</th><th>消息数量</th></tr>
|
||
</thead>
|
||
<tbody>
|
||
{chat_rows}
|
||
</tbody>
|
||
</table>
|
||
|
||
|
||
</div>
|
||
"""
|
||
|
||
tab_content_list = [
|
||
_format_stat_data(stat[period[0]], period[0], now - period[1])
|
||
for period in self.stat_period
|
||
if period[0] != "all_time"
|
||
]
|
||
|
||
tab_content_list.append(
|
||
_format_stat_data(stat["all_time"], "all_time", datetime.fromtimestamp(local_storage["deploy_time"])) # type: ignore
|
||
)
|
||
|
||
# 不再添加版本对比内容
|
||
# 添加图表内容 (修正缩进)
|
||
chart_data = await self._generate_chart_data(stat)
|
||
tab_content_list.append(self._generate_chart_tab(chart_data))
|
||
|
||
joined_tab_list = "\n".join(tab_list)
|
||
joined_tab_content = "\n".join(tab_content_list)
|
||
|
||
html_template = (
|
||
"""
|
||
<!DOCTYPE html>
|
||
<html lang="zh-CN">
|
||
<head>
|
||
<meta charset="UTF-8">
|
||
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
||
<title>MaiBot运行统计报告</title>
|
||
<script src="https://cdn.jsdelivr.net/npm/chart.js"></script>
|
||
<style>
|
||
body {
|
||
font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "Helvetica Neue", Arial, sans-serif;
|
||
margin: 0;
|
||
padding: 20px;
|
||
background-color: #f4f7f6;
|
||
color: #333;
|
||
line-height: 1.6;
|
||
}
|
||
.container {
|
||
max-width: 900px;
|
||
margin: 20px auto;
|
||
background-color: #fff;
|
||
padding: 25px;
|
||
border-radius: 8px;
|
||
box-shadow: 0 2px 10px rgba(0,0,0,0.1);
|
||
}
|
||
h1, h2 {
|
||
color: #2c3e50;
|
||
border-bottom: 2px solid #3498db;
|
||
padding-bottom: 10px;
|
||
margin-top: 0;
|
||
}
|
||
h1 {
|
||
text-align: center;
|
||
font-size: 2em;
|
||
}
|
||
h2 {
|
||
font-size: 1.5em;
|
||
margin-top: 30px;
|
||
}
|
||
p {
|
||
margin-bottom: 10px;
|
||
}
|
||
.info-item {
|
||
background-color: #ecf0f1;
|
||
padding: 8px 12px;
|
||
border-radius: 4px;
|
||
margin-bottom: 8px;
|
||
font-size: 0.95em;
|
||
}
|
||
.info-item strong {
|
||
color: #2980b9;
|
||
}
|
||
table {
|
||
width: 100%;
|
||
border-collapse: collapse;
|
||
margin-top: 15px;
|
||
font-size: 0.9em;
|
||
}
|
||
th, td {
|
||
border: 1px solid #ddd;
|
||
padding: 10px;
|
||
text-align: left;
|
||
}
|
||
th {
|
||
background-color: #3498db;
|
||
color: white;
|
||
font-weight: bold;
|
||
}
|
||
tr:nth-child(even) {
|
||
background-color: #f9f9f9;
|
||
}
|
||
.footer {
|
||
text-align: center;
|
||
margin-top: 30px;
|
||
font-size: 0.8em;
|
||
color: #7f8c8d;
|
||
}
|
||
.tabs {
|
||
overflow: hidden;
|
||
background: #ecf0f1;
|
||
display: flex;
|
||
}
|
||
.tabs button {
|
||
background: inherit; border: none; outline: none;
|
||
padding: 14px 16px; cursor: pointer;
|
||
transition: 0.3s; font-size: 16px;
|
||
}
|
||
.tabs button:hover {
|
||
background-color: #d4dbdc;
|
||
}
|
||
.tabs button.active {
|
||
background-color: #b3bbbd;
|
||
}
|
||
.tab-content {
|
||
display: none;
|
||
padding: 20px;
|
||
background-color: #fff;
|
||
border: 1px solid #ccc;
|
||
}
|
||
.tab-content.active {
|
||
display: block;
|
||
}
|
||
</style>
|
||
</head>
|
||
<body>
|
||
"""
|
||
+ f"""
|
||
<div class="container">
|
||
<h1>MaiBot运行统计报告</h1>
|
||
<p class="info-item"><strong>统计截止时间:</strong> {now.strftime("%Y-%m-%d %H:%M:%S")}</p>
|
||
|
||
<div class="tabs">
|
||
{joined_tab_list}
|
||
</div>
|
||
|
||
{joined_tab_content}
|
||
</div>
|
||
"""
|
||
+ """
|
||
<script>
|
||
let i, tab_content, tab_links;
|
||
tab_content = document.getElementsByClassName("tab-content");
|
||
tab_links = document.getElementsByClassName("tab-link");
|
||
|
||
tab_content[0].classList.add("active");
|
||
tab_links[0].classList.add("active");
|
||
|
||
function showTab(evt, tabName) {{
|
||
for (i = 0; i < tab_content.length; i++) tab_content[i].classList.remove("active");
|
||
for (i = 0; i < tab_links.length; i++) tab_links[i].classList.remove("active");
|
||
document.getElementById(tabName).classList.add("active");
|
||
evt.currentTarget.classList.add("active");
|
||
}}
|
||
</script>
|
||
</body>
|
||
</html>
|
||
"""
|
||
)
|
||
|
||
with open(self.record_file_path, "w", encoding="utf-8") as f:
|
||
f.write(html_template)
|
||
|
||
async def _generate_chart_data(self, stat: dict[str, Any]) -> dict:
|
||
"""生成图表数据 (异步)"""
|
||
now = datetime.now()
|
||
chart_data: dict[str, Any] = {}
|
||
|
||
time_ranges = [
|
||
("6h", 6, 10),
|
||
("12h", 12, 15),
|
||
("24h", 24, 15),
|
||
("48h", 48, 30),
|
||
]
|
||
|
||
# 依次处理（数据量不大，避免复杂度；如需可改 gather）
|
||
for range_key, hours, interval_minutes in time_ranges:
|
||
chart_data[range_key] = await self._collect_interval_data(now, hours, interval_minutes)
|
||
return chart_data
|
||
|
||
async def _collect_interval_data(self, now: datetime, hours: int, interval_minutes: int) -> dict:
|
||
start_time = now - timedelta(hours=hours)
|
||
time_points: list[datetime] = []
|
||
current_time = start_time
|
||
while current_time <= now:
|
||
time_points.append(current_time)
|
||
current_time += timedelta(minutes=interval_minutes)
|
||
|
||
total_cost_data = [0.0] * len(time_points)
|
||
cost_by_model: dict[str, list[float]] = {}
|
||
cost_by_module: dict[str, list[float]] = {}
|
||
message_by_chat: dict[str, list[int]] = {}
|
||
time_labels = [t.strftime("%H:%M") for t in time_points]
|
||
interval_seconds = interval_minutes * 60
|
||
|
||
# 单次查询 LLMUsage
|
||
llm_records = (
|
||
await db_get(
|
||
model_class=LLMUsage,
|
||
filters={"timestamp": {"$gte": start_time}},
|
||
order_by="-timestamp",
|
||
)
|
||
or []
|
||
)
|
||
for record in llm_records:
|
||
if not isinstance(record, dict) or not record.get("timestamp"):
|
||
continue
|
||
record_time = record["timestamp"]
|
||
if isinstance(record_time, str):
|
||
try:
|
||
record_time = datetime.fromisoformat(record_time)
|
||
except Exception:
|
||
continue
|
||
time_diff = (record_time - start_time).total_seconds()
|
||
idx = int(time_diff // interval_seconds)
|
||
if 0 <= idx < len(time_points):
|
||
cost = record.get("cost") or 0.0
|
||
total_cost_data[idx] += cost
|
||
model_name = record.get("model_name") or "unknown"
|
||
if model_name not in cost_by_model:
|
||
cost_by_model[model_name] = [0.0] * len(time_points)
|
||
cost_by_model[model_name][idx] += cost
|
||
request_type = record.get("request_type") or "unknown"
|
||
module_name = request_type.split(".")[0] if "." in request_type else request_type
|
||
if module_name not in cost_by_module:
|
||
cost_by_module[module_name] = [0.0] * len(time_points)
|
||
cost_by_module[module_name][idx] += cost
|
||
|
||
# 单次查询 Messages
|
||
msg_records = (
|
||
await db_get(
|
||
model_class=Messages,
|
||
filters={"time": {"$gte": start_time.timestamp()}},
|
||
order_by="-time",
|
||
)
|
||
or []
|
||
)
|
||
for msg in msg_records:
|
||
if not isinstance(msg, dict) or not msg.get("time"):
|
||
continue
|
||
msg_ts = msg["time"]
|
||
time_diff = msg_ts - start_time.timestamp()
|
||
idx = int(time_diff // interval_seconds)
|
||
if 0 <= idx < len(time_points):
|
||
if msg.get("chat_info_group_id"):
|
||
chat_name = msg.get("chat_info_group_name") or f"群{msg['chat_info_group_id']}"
|
||
elif msg.get("user_id"):
|
||
chat_name = msg.get("user_nickname") or f"用户{msg['user_id']}"
|
||
else:
|
||
continue
|
||
if chat_name not in message_by_chat:
|
||
message_by_chat[chat_name] = [0] * len(time_points)
|
||
message_by_chat[chat_name][idx] += 1
|
||
|
||
return {
|
||
"time_labels": time_labels,
|
||
"total_cost_data": total_cost_data,
|
||
"cost_by_model": cost_by_model,
|
||
"cost_by_module": cost_by_module,
|
||
"message_by_chat": message_by_chat,
|
||
}
|
||
|
||
@staticmethod
|
||
def _generate_chart_tab(chart_data: dict) -> str:
|
||
# sourcery skip: extract-duplicate-method, move-assign-in-block
|
||
"""生成图表选项卡HTML内容"""
|
||
|
||
# 生成不同颜色的调色板
|
||
colors = [
|
||
"#3498db",
|
||
"#e74c3c",
|
||
"#2ecc71",
|
||
"#f39c12",
|
||
"#9b59b6",
|
||
"#1abc9c",
|
||
"#34495e",
|
||
"#e67e22",
|
||
"#95a5a6",
|
||
"#f1c40f",
|
||
]
|
||
|
||
# 默认使用24小时数据生成数据集
|
||
default_data = chart_data["24h"]
|
||
|
||
# 为每个模型生成数据集
|
||
model_datasets = []
|
||
for i, (model_name, cost_data) in enumerate(default_data["cost_by_model"].items()):
|
||
color = colors[i % len(colors)]
|
||
model_datasets.append(f"""{{
|
||
label: '{model_name}',
|
||
data: {cost_data},
|
||
borderColor: '{color}',
|
||
backgroundColor: '{color}20',
|
||
tension: 0.4,
|
||
fill: false
|
||
}}""")
|
||
|
||
",\n ".join(model_datasets)
|
||
|
||
# 为每个模块生成数据集
|
||
module_datasets = []
|
||
for i, (module_name, cost_data) in enumerate(default_data["cost_by_module"].items()):
|
||
color = colors[i % len(colors)]
|
||
module_datasets.append(f"""{{
|
||
label: '{module_name}',
|
||
data: {cost_data},
|
||
borderColor: '{color}',
|
||
backgroundColor: '{color}20',
|
||
tension: 0.4,
|
||
fill: false
|
||
}}""")
|
||
|
||
",\n ".join(module_datasets)
|
||
|
||
# 为每个聊天流生成消息数据集
|
||
message_datasets = []
|
||
for i, (chat_name, message_data) in enumerate(default_data["message_by_chat"].items()):
|
||
color = colors[i % len(colors)]
|
||
message_datasets.append(f"""{{
|
||
label: '{chat_name}',
|
||
data: {message_data},
|
||
borderColor: '{color}',
|
||
backgroundColor: '{color}20',
|
||
tension: 0.4,
|
||
fill: false
|
||
}}""")
|
||
|
||
",\n ".join(message_datasets)
|
||
|
||
return f"""
|
||
<div id="charts" class="tab-content">
|
||
<h2>数据图表</h2>
|
||
|
||
<!-- 时间范围选择按钮 -->
|
||
<div style="margin: 20px 0; text-align: center;">
|
||
<label style="margin-right: 10px; font-weight: bold;">时间范围:</label>
|
||
<button class="time-range-btn" onclick="switchTimeRange('6h')">6小时</button>
|
||
<button class="time-range-btn" onclick="switchTimeRange('12h')">12小时</button>
|
||
<button class="time-range-btn active" onclick="switchTimeRange('24h')">24小时</button>
|
||
<button class="time-range-btn" onclick="switchTimeRange('48h')">48小时</button>
|
||
</div>
|
||
|
||
<div style="margin-top: 20px;">
|
||
<div style="margin-bottom: 40px;">
|
||
<canvas id="totalCostChart" width="800" height="400"></canvas>
|
||
</div>
|
||
<div style="margin-bottom: 40px;">
|
||
<canvas id="costByModuleChart" width="800" height="400"></canvas>
|
||
</div>
|
||
<div style="margin-bottom: 40px;">
|
||
<canvas id="costByModelChart" width="800" height="400"></canvas>
|
||
</div>
|
||
<div>
|
||
<canvas id="messageByChatChart" width="800" height="400"></canvas>
|
||
</div>
|
||
</div>
|
||
|
||
<style>
|
||
.time-range-btn {{
|
||
background-color: #ecf0f1;
|
||
border: 1px solid #bdc3c7;
|
||
color: #2c3e50;
|
||
padding: 8px 16px;
|
||
margin: 0 5px;
|
||
border-radius: 4px;
|
||
cursor: pointer;
|
||
font-size: 14px;
|
||
transition: all 0.3s ease;
|
||
}}
|
||
|
||
.time-range-btn:hover {{
|
||
background-color: #d5dbdb;
|
||
}}
|
||
|
||
.time-range-btn.active {{
|
||
background-color: #3498db;
|
||
color: white;
|
||
border-color: #2980b9;
|
||
}}
|
||
</style>
|
||
|
||
<script>
|
||
const allChartData = {chart_data};
|
||
let currentCharts = {{}};
|
||
|
||
// 图表配置模板
|
||
const chartConfigs = {{
|
||
totalCost: {{
|
||
id: 'totalCostChart',
|
||
title: '总花费',
|
||
yAxisLabel: '花费 (¥)',
|
||
dataKey: 'total_cost_data',
|
||
fill: true
|
||
}},
|
||
costByModule: {{
|
||
id: 'costByModuleChart',
|
||
title: '各模块花费',
|
||
yAxisLabel: '花费 (¥)',
|
||
dataKey: 'cost_by_module',
|
||
fill: false
|
||
}},
|
||
costByModel: {{
|
||
id: 'costByModelChart',
|
||
title: '各模型花费',
|
||
yAxisLabel: '花费 (¥)',
|
||
dataKey: 'cost_by_model',
|
||
fill: false
|
||
}},
|
||
messageByChat: {{
|
||
id: 'messageByChatChart',
|
||
title: '各聊天流消息数',
|
||
yAxisLabel: '消息数',
|
||
dataKey: 'message_by_chat',
|
||
fill: false
|
||
}},
|
||
focusCyclesByAction: {{
|
||
id: 'focusCyclesByActionChart',
|
||
title: 'Focus循环按Action类型',
|
||
yAxisLabel: '循环数',
|
||
dataKey: 'focus_cycles_by_action',
|
||
fill: false
|
||
}},
|
||
focusTimeByStage: {{
|
||
id: 'focusTimeByStageChart',
|
||
title: 'Focus各阶段累计时间',
|
||
yAxisLabel: '时间 (秒)',
|
||
dataKey: 'focus_time_by_stage',
|
||
fill: false
|
||
}}
|
||
}};
|
||
|
||
function switchTimeRange(timeRange) {{
|
||
// 更新按钮状态
|
||
document.querySelectorAll('.time-range-btn').forEach(btn => {{
|
||
btn.classList.remove('active');
|
||
}});
|
||
event.target.classList.add('active');
|
||
|
||
// 更新图表数据
|
||
const data = allChartData[timeRange];
|
||
updateAllCharts(data, timeRange);
|
||
}}
|
||
|
||
function updateAllCharts(data, timeRange) {{
|
||
// 销毁现有图表
|
||
Object.values(currentCharts).forEach(chart => {{
|
||
if (chart) chart.destroy();
|
||
}});
|
||
|
||
currentCharts = {{}};
|
||
|
||
// 重新创建图表
|
||
createChart('totalCost', data, timeRange);
|
||
createChart('costByModule', data, timeRange);
|
||
createChart('costByModel', data, timeRange);
|
||
createChart('messageByChat', data, timeRange);
|
||
}}
|
||
|
||
function createChart(chartType, data, timeRange) {{
|
||
const config = chartConfigs[chartType];
|
||
const colors = ['#3498db', '#e74c3c', '#2ecc71', '#f39c12', '#9b59b6', '#1abc9c', '#34495e', '#e67e22', '#95a5a6', '#f1c40f'];
|
||
|
||
let datasets = [];
|
||
|
||
if (chartType === 'totalCost') {{
|
||
datasets = [{{
|
||
label: config.title,
|
||
data: data[config.dataKey],
|
||
borderColor: colors[0],
|
||
backgroundColor: 'rgba(52, 152, 219, 0.1)',
|
||
tension: 0.4,
|
||
fill: config.fill
|
||
}}];
|
||
}} else {{
|
||
let i = 0;
|
||
Object.entries(data[config.dataKey]).forEach(([name, chartData]) => {{
|
||
datasets.push({{
|
||
label: name,
|
||
data: chartData,
|
||
borderColor: colors[i % colors.length],
|
||
backgroundColor: colors[i % colors.length] + '20',
|
||
tension: 0.4,
|
||
fill: config.fill
|
||
}});
|
||
i++;
|
||
}});
|
||
}}
|
||
|
||
currentCharts[chartType] = new Chart(document.getElementById(config.id), {{
|
||
type: 'line',
|
||
data: {{
|
||
labels: data.time_labels,
|
||
datasets: datasets
|
||
}},
|
||
options: {{
|
||
responsive: true,
|
||
plugins: {{
|
||
title: {{
|
||
display: true,
|
||
text: timeRange + '内' + config.title + '趋势',
|
||
font: {{ size: 16 }}
|
||
}},
|
||
legend: {{
|
||
display: chartType !== 'totalCost',
|
||
position: 'top'
|
||
}}
|
||
}},
|
||
scales: {{
|
||
x: {{
|
||
title: {{
|
||
display: true,
|
||
text: '时间'
|
||
}},
|
||
ticks: {{
|
||
maxTicksLimit: 12
|
||
}}
|
||
}},
|
||
y: {{
|
||
title: {{
|
||
display: true,
|
||
text: config.yAxisLabel
|
||
}},
|
||
beginAtZero: true
|
||
}}
|
||
}},
|
||
interaction: {{
|
||
intersect: false,
|
||
mode: 'index'
|
||
}}
|
||
}}
|
||
}});
|
||
}}
|
||
|
||
// 初始化图表（默认24小时）
|
||
document.addEventListener('DOMContentLoaded', function() {{
|
||
updateAllCharts(allChartData['24h'], '24h');
|
||
}});
|
||
</script>
|
||
</div>
|
||
"""
|