Heroku/heroku/log.py

549 lines
18 KiB
Python

"""Main logging part"""
# ©️ Dan Gazizullin, 2021-2023
# This file is a part of Hikka Userbot
# 🌐 https://github.com/hikariatama/Hikka
# You can redistribute it and/or modify it under the terms of the GNU AGPLv3
# 🔑 https://www.gnu.org/licenses/agpl-3.0.html
# ©️ Codrago, 2024-2025
# This file is a part of Heroku Userbot
# 🌐 https://github.com/coddrago/Heroku
# You can redistribute it and/or modify it under the terms of the GNU AGPLv3
# 🔑 https://www.gnu.org/licenses/agpl-3.0.html
import asyncio
import contextlib
import inspect
import io
import linecache
import logging
import re
import sys
import traceback
import typing
from logging.handlers import RotatingFileHandler
import herokutl
from aiogram.exceptions import TelegramNetworkError
from herokutl.errors import PersistentTimestampOutdatedError
from . import utils
from .tl_cache import CustomTelegramClient
from .types import BotInlineCall, Module
from .web.debugger import WebDebugger
# Monkeypatch linecache to make interactive line debugger available
# in werkzeug web debugger
# This is weird, but the only adequate approach
# https://github.com/pallets/werkzeug/blob/3115aa6a6276939f5fd6efa46282e0256ff21f1a/src/werkzeug/debug/tbtools.py#L382-L416
old = linecache.getlines
def getlines(filename: str, module_globals=None) -> str:
"""
Get the lines for a Python source file from the cache.
Update the cache if it doesn't contain an entry for this file already.
Modified version of original `linecache.getlines`, which returns the
source code of Heroku modules properly. This is needed for
interactive line debugger in werkzeug web debugger.
"""
try:
if filename.startswith("<") and filename.endswith(">"):
module = filename[1:-1].split(maxsplit=1)[-1]
if (module.startswith("heroku.modules")) and module in sys.modules:
return list(
map(
lambda x: f"{x}\n",
sys.modules[module].__loader__.get_source().splitlines(),
)
)
except Exception:
logging.debug("Can't get lines for %s", filename, exc_info=True)
return old(filename, module_globals)
linecache.getlines = getlines
def override_text(exception: Exception) -> typing.Optional[str]:
"""Returns error-specific description if available, else `None`"""
if isinstance(exception, (TelegramNetworkError, asyncio.exceptions.TimeoutError)):
return "✈️ <b>You have problems with internet connection on your server.</b>"
elif isinstance(exception, PersistentTimestampOutdatedError):
return "✈️ <b>Telegram has problems with their datacenters.</b>"
return None
class HerokuException:
def __init__(
self,
message: str,
full_stack: str,
sysinfo: typing.Optional[
typing.Tuple[object, Exception, traceback.TracebackException]
] = None,
):
self.message = message
self.full_stack = full_stack
self.sysinfo = sysinfo
self.debug_url = None
@classmethod
def from_exc_info(
cls,
exc_type: object,
exc_value: Exception,
tb: traceback.TracebackException,
stack: typing.Optional[typing.List[inspect.FrameInfo]] = None,
comment: typing.Optional[typing.Any] = None,
) -> "HerokuException":
def to_hashable(dictionary: dict) -> dict:
dictionary = dictionary.copy()
for key, value in dictionary.items():
if isinstance(value, dict):
dictionary[key] = to_hashable(value)
else:
try:
if (
getattr(getattr(value, "__class__", None), "__name__", None)
== "Database"
):
dictionary[key] = "<Database>"
elif isinstance(
value,
(herokutl.TelegramClient, CustomTelegramClient),
):
dictionary[key] = f"<{value.__class__.__name__}>"
elif len(str(value)) > 512:
dictionary[key] = f"{str(value)[:512]}..."
else:
dictionary[key] = str(value)
except Exception:
dictionary[key] = f"<{value.__class__.__name__}>"
return dictionary
full_traceback = traceback.format_exc().replace(
"Traceback (most recent call last):\n",
"",
)
line_regex = re.compile(r' File "(.*?)", line ([0-9]+), in (.+)')
def format_line(line: str) -> str:
filename_, lineno_, name_ = line_regex.search(line).groups()
return (
f"👉 <code>{utils.escape_html(filename_)}:{lineno_}</code> <b>in</b>"
f" <code>{utils.escape_html(name_)}</code>"
)
filename, lineno, name = next(
(
line_regex.search(line).groups()
for line in reversed(full_traceback.splitlines())
if line_regex.search(line)
),
(None, None, None),
)
full_traceback = "\n".join(
[
(
format_line(line)
if line_regex.search(line)
else f"<code>{utils.escape_html(line)}</code>"
)
for line in full_traceback.splitlines()
]
)
caller = utils.find_caller(stack or inspect.stack())
return cls(
message=override_text(exc_value)
or (
"{}<b>🎯 Source:</b> <code>{}:{}</code><b> in"
" </b><code>{}</code>\n<b>❓ Error:</b> <code>{}</code>{}"
).format(
(
(
"🔮 <b>Cause: method </b><code>{}</code><b> of"
" </b><code>{}</code>\n\n"
).format(
utils.escape_html(caller.__name__),
utils.escape_html(caller.__self__.__class__.__name__),
)
if (
caller
and hasattr(caller, "__self__")
and hasattr(caller, "__name__")
)
else ""
),
utils.escape_html(filename),
lineno,
utils.escape_html(name),
utils.escape_html(
"".join(
traceback.format_exception_only(exc_type, exc_value)
).strip()
),
(
"\n💭 <b>Message:</b>"
f" <code>{utils.escape_html(str(comment))}</code>"
if comment
else ""
),
),
full_stack=full_traceback,
sysinfo=(exc_type, exc_value, tb),
)
class TelegramLogsHandler(logging.Handler):
"""
Keeps 2 buffers.
One for dispatched messages.
One for unused messages.
When the length of the 2 together is 100
truncate to make them 100 together,
first trimming handled then unused.
"""
def __init__(self, targets: list, capacity: int):
super().__init__(0)
self.buffer = []
self.handledbuffer = []
self._queue = []
self._mods = {}
self.tg_buff = []
self.force_send_all = False
self.tg_level = 20
self.ignore_common = False
self.web_debugger = None
self.targets = targets
self.capacity = capacity
self.lvl = logging.NOTSET
self._send_lock = asyncio.Lock()
def install_tg_log(self, mod: Module):
if getattr(self, "_task", False):
self._task.cancel()
self._mods[mod.tg_id] = mod
if mod.db.get(__name__, "debugger", False):
self.web_debugger = WebDebugger()
self._task = asyncio.ensure_future(self.queue_poller())
async def queue_poller(self):
while True:
with contextlib.suppress(Exception):
await self.sender()
await asyncio.sleep(3)
def setLevel(self, level: int):
self.lvl = level
def dump(self):
"""Return a list of logging entries"""
return self.handledbuffer + self.buffer
def dumps(
self,
lvl: int = 0,
client_id: typing.Optional[int] = None,
) -> typing.List[str]:
"""Return all entries of minimum level as list of strings"""
return [
self.targets[0].format(record)
for record in (self.buffer + self.handledbuffer)
if record.levelno >= lvl
and (not record.heroku_caller or client_id == record.heroku_caller)
]
async def _show_full_trace(
self,
call: BotInlineCall,
bot: "aiogram.Bot", # type: ignore # noqa: F821
item: HerokuException,
):
chunks = item.message + "\n\n<b>🪐 Full traceback:</b>\n" + item.full_stack
chunks = list(utils.smart_split(*herokutl.extensions.html.parse(chunks), 4096))
await call.edit(
chunks[0],
reply_markup=self._gen_web_debug_button(item),
)
for chunk in chunks[1:]:
await bot.send_message(chat_id=call.chat_id, text=chunk)
def _gen_web_debug_button(self, item: HerokuException) -> list:
if not item.sysinfo:
return []
if not (url := item.debug_url):
try:
url = self.web_debugger.feed(*item.sysinfo)
except Exception:
url = None
item.debug_url = url
return [
(
{
"text": "🐞 Web debugger",
"url": url,
}
if self.web_debugger
else {
"text": "🪲 Start debugger",
"callback": self._start_debugger,
"args": (item,),
}
)
]
async def _start_debugger(
self,
call: "InlineCall", # type: ignore # noqa: F821
item: HerokuException,
):
if not self.web_debugger:
self.web_debugger = WebDebugger()
await self.web_debugger.proxy_ready.wait()
url = self.web_debugger.feed(*item.sysinfo)
item.debug_url = url
await call.edit(
item.message,
reply_markup=self._gen_web_debug_button(item),
)
await call.answer(
(
"Web debugger started. You can get PIN using .debugger command. \n⚠️"
" !DO NOT GIVE IT TO ANYONE! ⚠️"
),
show_alert=True,
)
def get_logid_by_client(self, client_id: int) -> int:
return self._mods[client_id].logchat
async def sender(self):
async with self._send_lock:
self._queue = {
client_id: utils.chunks(
utils.escape_html(
"".join(
[
item[0]
for item in self.tg_buff
if isinstance(item[0], str)
and (
not item[1]
or item[1] == client_id
or self.force_send_all
)
]
)
),
4096,
)
for client_id in self._mods
}
self._exc_queue = {
client_id: [
self._mods[client_id].inline.bot.send_message(
self._mods[client_id].logchat,
item[0].message,
reply_markup=self._mods[client_id].inline.generate_markup(
[
{
"text": "🪐 Full traceback",
"callback": self._show_full_trace,
"args": (
self._mods[client_id].inline.bot,
item[0],
),
"disable_security": True,
},
*self._gen_web_debug_button(item[0]),
],
),
)
for item in self.tg_buff
if isinstance(item[0], HerokuException)
and (not item[1] or item[1] == client_id or self.force_send_all)
]
for client_id in self._mods
}
for exceptions in self._exc_queue.values():
for exc in exceptions:
asyncio.create_task(exc)
self.tg_buff = []
for client_id in self._mods:
if client_id not in self._queue:
continue
if len(self._queue[client_id]) > 5:
logfile = io.BytesIO(
"".join(self._queue[client_id]).encode("utf-8")
)
logfile.name = "heroku-logs.txt"
logfile.seek(0)
await self._mods[client_id].inline.bot.send_document(
self._mods[client_id].logchat,
logfile,
caption=(
"<b>🧳 Journals are too big to be sent as separate"
" messages</b>"
),
)
self._queue[client_id] = []
continue
while self._queue[client_id]:
if chunk := self._queue[client_id].pop(0):
asyncio.ensure_future(
self._mods[client_id].inline.bot.send_message(
self._mods[client_id].logchat,
f"<code>{chunk}</code>",
disable_notification=True,
)
)
def emit(self, record: logging.LogRecord):
try:
caller = next(
(
frame_info.frame.f_locals["_heroku_client_id_logging_tag"]
for frame_info in inspect.stack()
if isinstance(
getattr(getattr(frame_info, "frame", None), "f_locals", {}).get(
"_heroku_client_id_logging_tag"
),
int,
)
),
False,
)
if not isinstance(caller, int):
caller = None
except Exception:
caller = None
record.heroku_caller = caller
if record.levelno >= self.tg_level:
if record.exc_info:
exc = HerokuException.from_exc_info(
*record.exc_info,
stack=record.__dict__.get("stack", None),
comment=record.msg % record.args,
)
if not self.ignore_common or all(
field not in exc.message
for field in [
"InputPeerEmpty() does not have any entity type",
"https://docs.telethon.dev/en/stable/concepts/entities.html",
]
):
self.tg_buff += [(exc, caller)]
else:
self.tg_buff += [
(
_tg_formatter.format(record),
caller,
)
]
if len(self.buffer) + len(self.handledbuffer) >= self.capacity:
if self.handledbuffer:
del self.handledbuffer[0]
else:
del self.buffer[0]
self.buffer.append(record)
if record.levelno >= self.lvl >= 0:
self.acquire()
try:
for precord in self.buffer:
for target in self.targets:
if record.levelno >= target.level:
target.handle(precord)
self.handledbuffer = (
self.handledbuffer[-(self.capacity - len(self.buffer)) :]
+ self.buffer
)
self.buffer = []
finally:
self.release()
_main_formatter = logging.Formatter(
fmt="%(asctime)s [%(levelname)s] %(name)s: %(message)s",
datefmt="%Y-%m-%d %H:%M:%S",
style="%",
)
_tg_formatter = logging.Formatter(
fmt="[%(levelname)s] %(name)s: %(message)s\n",
datefmt=None,
style="%",
)
rotating_handler = RotatingFileHandler(
filename="heroku.log",
mode="a",
maxBytes=10 * 1024 * 1024,
backupCount=1,
encoding="utf-8",
delay=0,
)
rotating_handler.setFormatter(_main_formatter)
def init():
class NoFetchUpdatesFilter(logging.Filter):
def filter(self, record: logging.LogRecord) -> bool:
msg = record.getMessage()
return "Failed to fetch updates" not in msg
logging.getLogger("aiogram.dispatcher").addFilter(NoFetchUpdatesFilter())
handler = logging.StreamHandler()
handler.setLevel(logging.INFO)
handler.setFormatter(_main_formatter)
logging.getLogger().handlers = []
logging.getLogger().addHandler(
TelegramLogsHandler((handler, rotating_handler), 7000)
)
logging.getLogger().setLevel(logging.NOTSET)
logging.getLogger("herokutl").setLevel(logging.WARNING)
logging.getLogger("matplotlib").setLevel(logging.WARNING)
logging.getLogger("aiohttp").setLevel(logging.WARNING)
logging.getLogger("aiogram").setLevel(logging.WARNING)
logging.captureWarnings(True)