diff --git a/app/bot/menu.py b/app/bot/menu.py index c13f9b0..9339737 100644 --- a/app/bot/menu.py +++ b/app/bot/menu.py @@ -2,9 +2,11 @@ from enum import StrEnum +from aiogram import F from aiogram.filters.callback_data import CallbackData from aiogram.types import InlineKeyboardMarkup, ReplyKeyboardMarkup from aiogram.utils.keyboard import InlineKeyboardBuilder, ReplyKeyboardBuilder +from magic_filter import MagicFilter from app.texts import Texts @@ -38,6 +40,12 @@ def section_by_button(texts: Texts) -> dict[str, Section]: } +def free_input(texts: Texts) -> MagicFilter: + """Input for a dialog step — anything but a main-menu button or a bot command: those must work from any step, + otherwise «🩺 Статус» pressed while the bot waits for a file name becomes the file name.""" + return ~F.text.in_(set(section_by_button(texts))) & ~F.text.regexp(r"^/(start|menu|help)(@\w+)?(\s|$)") + + def main_menu(texts: Texts, *, is_admin: bool) -> ReplyKeyboardMarkup: m = texts.menu kb = ReplyKeyboardBuilder() diff --git a/app/bot/routers/commit.py b/app/bot/routers/commit.py index 3466f33..3b720d0 100644 --- a/app/bot/routers/commit.py +++ b/app/bot/routers/commit.py @@ -16,7 +16,7 @@ from aiogram.fsm.state import State, StatesGroup from aiogram.types import BufferedInputFile, CallbackQuery, InlineKeyboardMarkup, Message from aiogram.utils.keyboard import InlineKeyboardBuilder -from app.bot.menu import Nav, NavAction +from app.bot.menu import Nav, NavAction, free_input from app.db.models import ChangeOp, User from app.repo.drafts import DraftFull, DraftService, DraftView from app.repo.git import Entry, GitError, UploadsRepo @@ -520,7 +520,7 @@ def build_router( # ---- text input -------------------------------------------------------------------------------------------- - @router.message(Flow.file_name, F.text) + @router.message(Flow.file_name, F.text, free_input(texts)) async def got_file_name(message: Message, state: FSMContext, member: User) -> None: data = await state.get_data() directory = data.get("commit_path", "") @@ -541,7 +541,7 @@ def build_router( await state.update_data(commit_target=path, commit_op=ChangeOp.ADD.value) await message.answer(t.ask_content.format(path=path), reply_markup=to_dir_keyboard(texts)) - @router.message(Flow.dir_name, F.text) + @router.message(Flow.dir_name, F.text, free_input(texts)) async def got_dir_name(message: Message, state: FSMContext, member: User) -> None: data = await state.get_data() directory = data.get("commit_path", "") @@ -562,7 +562,7 @@ def build_router( text, kb = await dir_screen(state, member.tg_id, path, 0) await message.answer(text, reply_markup=kb) - @router.message(Flow.message, F.text) + @router.message(Flow.message, F.text, free_input(texts)) async def got_message(message: Message, state: FSMContext, member: User) -> None: try: text = check_message(message.text or "") @@ -579,7 +579,7 @@ def build_router( screen, kb = await draft_screen(state, member.tg_id) await message.answer(screen, reply_markup=kb) - @router.message(Flow.content, F.text | F.document) + @router.message(Flow.content, F.text | F.document, free_input(texts)) async def got_content(message: Message, state: FSMContext, member: User, bot: Bot) -> None: data = await state.get_data() target: str = data["commit_target"] diff --git a/app/bot/routers/common.py b/app/bot/routers/common.py index 439a03f..e5afa5f 100644 --- a/app/bot/routers/common.py +++ b/app/bot/routers/common.py @@ -16,7 +16,6 @@ def build_router(texts: Texts, access: AccessService) -> Router: router = Router(name="common") sections = section_by_button(texts) section_text = { - Section.LOGS: texts.section.logs, Section.ALERTS: texts.section.alerts, } diff --git a/app/bot/routers/logs.py b/app/bot/routers/logs.py new file mode 100644 index 0000000..e6522c7 --- /dev/null +++ b/app/bot/routers/logs.py @@ -0,0 +1,234 @@ +"""«📜 Логи»: pick a container, then its tail or your own grep (SPEC L-1…L-3, L-5, L-6). The agent is #18.""" + +import asyncio +import html +import logging +from datetime import timedelta +from enum import StrEnum + +from aiogram import F, Router +from aiogram.enums import ParseMode +from aiogram.filters.callback_data import CallbackData +from aiogram.fsm.context import FSMContext +from aiogram.fsm.state import State, StatesGroup +from aiogram.types import BufferedInputFile, CallbackQuery, InlineKeyboardMarkup, Message +from aiogram.utils.keyboard import InlineKeyboardBuilder + +from app.bot.menu import Nav, NavAction, free_input +from app.db.models import LogMode, User, utcnow +from app.logs.grep import GrepError, GrepErrorCode, GrepResult, parse_grep, run, to_command +from app.logs.journal import LogJournal +from app.runtime.docker import Container, DockerClient, DockerError +from app.texts import LogsTexts, Texts + +log = logging.getLogger(__name__) + +TAILS = (50, 200) # lines (L-2) +PERIODS = (5, 60, 24 * 60) # minutes (L-2) +GREP_WINDOW = timedelta(hours=24) +MAX_MESSAGE = 3500 # leave room under Telegram's 4096 for the header and HTML markup +MAX_FILE = 5 * 1024 * 1024 # L-5 + + +class LogsAct(StrEnum): + LIST = "l" + OPEN = "o" + TAIL = "t" + SINCE = "s" + GREP = "g" + AGENT = "a" + + +class LogsCb(CallbackData, prefix="lg"): + a: LogsAct + id: str = "" # 12-char container id prefix + n: int = 0 # lines for TAIL, minutes for SINCE + + +class LogsFlow(StatesGroup): + grep = State() + + +def period(t: LogsTexts, minutes: int) -> str: + return t.hours.format(n=minutes // 60) if minutes >= 60 else t.minutes.format(n=minutes) + + +def list_keyboard(texts: Texts, containers: list[Container]) -> InlineKeyboardMarkup: + kb = InlineKeyboardBuilder() + for c in containers: + kb.button(text=f"{c.light.value} {c.name}", callback_data=LogsCb(a=LogsAct.OPEN, id=c.short_id)) + kb.button(text=texts.menu.back, callback_data=Nav(action=NavAction.BACK)) + kb.adjust(1) + return kb.as_markup() + + +def container_keyboard(texts: Texts, short_id: str) -> InlineKeyboardMarkup: + t = texts.logs + kb = InlineKeyboardBuilder() + for n in TAILS: + kb.button(text=t.tail.format(n=n), callback_data=LogsCb(a=LogsAct.TAIL, id=short_id, n=n)) + for minutes in PERIODS: + kb.button(text=period(t, minutes), callback_data=LogsCb(a=LogsAct.SINCE, id=short_id, n=minutes)) + kb.button(text=t.grep_button, callback_data=LogsCb(a=LogsAct.GREP, id=short_id)) + kb.button(text=t.agent_button, callback_data=LogsCb(a=LogsAct.AGENT, id=short_id)) + kb.button(text=t.to_list, callback_data=LogsCb(a=LogsAct.LIST)) + kb.adjust(2, 3, 2, 1) + return kb.as_markup() + + +def notes(t: LogsTexts, result: GrepResult) -> list[str]: + out = [] + if result.truncated: + out.append(t.truncated.format(n=len(result.lines))) + if result.partial: + out.append(t.partial) + if result.skipped: + out.append(t.skipped.format(n=result.skipped)) + return out + + +def grep_error(t: LogsTexts, e: GrepError) -> str: + template = { + GrepErrorCode.EMPTY: t.err_empty, + GrepErrorCode.QUOTES: t.err_quotes, + GrepErrorCode.FLAG: t.err_flag, + GrepErrorCode.CONTEXT: t.err_context, + GrepErrorCode.EXTRA: t.err_extra, + GrepErrorCode.TOO_LONG: t.err_too_long, + GrepErrorCode.REGEX: t.err_regex, + }[e.code] + return template.format(arg=html.escape(e.arg[:100])) + + +async def send_result( + msg: Message, texts: Texts, head: str, body: str, footer: list[str], kb: InlineKeyboardMarkup, filename: str +) -> None: + """Short results inline as
, long ones as a .log document (L-5); the container keyboard rides along."""
+ t = texts.logs
+ tail = "".join(f"\n{html.escape(n)}" for n in footer)
+ if not body:
+ await msg.answer(f"{head}\n{t.nothing}{tail}", parse_mode=ParseMode.HTML, reply_markup=kb)
+ return
+ escaped = html.escape(body)
+ if len(escaped) <= MAX_MESSAGE:
+ await msg.answer(f"{head}\n{escaped}{tail}", parse_mode=ParseMode.HTML, reply_markup=kb)
+ return
+ data = body.encode()[-MAX_FILE:]
+ await msg.answer_document(
+ BufferedInputFile(data, filename),
+ caption=f"{head}\n{t.sent_as_file}{tail}",
+ parse_mode=ParseMode.HTML,
+ reply_markup=kb,
+ )
+
+
+def build_router(texts: Texts, docker: DockerClient | None, journal: LogJournal) -> Router:
+ router = Router(name="logs")
+ t = texts.logs
+
+ def need_docker() -> DockerClient:
+ if docker is None:
+ raise DockerError("docker is not configured (DOCKER_URL)")
+ return docker
+
+ @router.message(F.text == texts.menu.logs)
+ async def open_logs(message: Message, state: FSMContext) -> None:
+ await state.clear()
+ try:
+ containers = await need_docker().containers()
+ except DockerError as e:
+ log.warning("logs: %s", e)
+ await message.answer(t.unavailable)
+ return
+ await message.answer(t.title if containers else t.empty, reply_markup=list_keyboard(texts, containers))
+
+ @router.callback_query(LogsCb.filter())
+ async def on_action(query: CallbackQuery, callback_data: LogsCb, state: FSMContext, member: User) -> None:
+ if not isinstance(query.message, Message):
+ await query.answer()
+ return
+ msg = query.message
+ try:
+ d = need_docker()
+ if callback_data.a is LogsAct.LIST:
+ await state.clear()
+ containers = await d.containers()
+ await msg.edit_text(t.title if containers else t.empty, reply_markup=list_keyboard(texts, containers))
+ await query.answer()
+ return
+ c = await d.get(callback_data.id)
+ kb = container_keyboard(texts, c.short_id)
+ match callback_data.a:
+ case LogsAct.OPEN:
+ await state.clear()
+ await msg.edit_text(t.container.format(name=c.name), reply_markup=kb)
+ case LogsAct.TAIL | LogsAct.SINCE:
+ await query.answer()
+ if callback_data.a is LogsAct.TAIL:
+ n = min(max(callback_data.n, 1), max(TAILS))
+ text = await d.logs(c.id, tail=n)
+ scope, q = t.tail.format(n=n), {"tail": n}
+ else:
+ minutes = min(max(callback_data.n, 1), max(PERIODS))
+ text = await d.logs(c.id, since=utcnow() - timedelta(minutes=minutes))
+ scope, q = t.since.format(period=period(t, minutes)), {"since_minutes": minutes}
+ lines = text.splitlines()
+ await journal.record(member.tg_id, c.name, LogMode.TAIL, q, len(lines))
+ head = t.head_tail.format(name=html.escape(c.name), scope=scope, lines=len(lines))
+ await send_result(msg, texts, head, "\n".join(lines), [], kb, f"{c.name}.log")
+ return
+ case LogsAct.GREP:
+ await state.set_state(LogsFlow.grep)
+ await state.update_data(logs_id=c.short_id)
+ back = InlineKeyboardBuilder()
+ back.button(text=t.to_container, callback_data=LogsCb(a=LogsAct.OPEN, id=c.short_id))
+ await msg.answer(
+ t.ask_grep.format(name=html.escape(c.name)),
+ parse_mode=ParseMode.HTML,
+ reply_markup=back.as_markup(),
+ )
+ case LogsAct.AGENT:
+ await query.answer(t.agent_soon, show_alert=True)
+ return
+ except DockerError as e:
+ log.warning("logs: %s", e)
+ await query.answer(t.unavailable, show_alert=True)
+ return
+ await query.answer()
+
+ @router.message(LogsFlow.grep, F.text, free_input(texts))
+ async def got_grep(message: Message, state: FSMContext, member: User) -> None:
+ try:
+ query = parse_grep(message.text or "")
+ except GrepError as e:
+ await message.answer(grep_error(t, e), parse_mode=ParseMode.HTML)
+ return # stay in the state: fix and resend
+ data = await state.get_data()
+ try:
+ d = need_docker()
+ c = await d.get(data.get("logs_id", ""))
+ text = await d.logs(c.id, since=utcnow() - GREP_WINDOW)
+ except DockerError as e:
+ log.warning("logs: %s", e)
+ await message.answer(t.unavailable)
+ return
+ await state.clear()
+ result = await asyncio.to_thread(run, query, text)
+ await journal.record(member.tg_id, c.name, LogMode.GREP, query.to_dict(), result.matched)
+ head = t.head_grep.format(
+ command=html.escape(to_command(query)), name=html.escape(c.name), matched=result.matched
+ )
+ body = "" if query.count else "\n".join(result.lines)
+ footer = notes(t, result)
+ if query.count: # -c: the number in the header is the answer
+ await message.answer(
+ head + "".join(f"\n{html.escape(n)}" for n in footer),
+ parse_mode=ParseMode.HTML,
+ reply_markup=container_keyboard(texts, c.short_id),
+ )
+ return
+ await send_result(
+ message, texts, head, body, footer, container_keyboard(texts, c.short_id), f"{c.name}-grep.log"
+ )
+
+ return router
diff --git a/app/config.py b/app/config.py
index 35122bd..4dd40ad 100644
--- a/app/config.py
+++ b/app/config.py
@@ -93,7 +93,9 @@ class DockerSettings(BaseSettings):
label: str = "tgit.managed=true"
timeout: float = Field(default=10.0, gt=0)
snapshot_interval: float = Field(default=15.0, gt=0)
+ # One logs request reads at most this much (SPEC L-5: 5 MB); the newest lines win
max_log_bytes: int = Field(default=5 * 1024 * 1024, gt=0)
+ max_log_lines: int = Field(default=50_000, gt=0)
class RuntimeSettings(BaseSettings):
diff --git a/app/logs/__init__.py b/app/logs/__init__.py
new file mode 100644
index 0000000..0642cb3
--- /dev/null
+++ b/app/logs/__init__.py
@@ -0,0 +1 @@
+"""Log search: a grep subset executed in Python, never in a shell (ARCHITECTURE §2.7)."""
diff --git a/app/logs/grep.py b/app/logs/grep.py
new file mode 100644
index 0000000..1aa45e4
--- /dev/null
+++ b/app/logs/grep.py
@@ -0,0 +1,201 @@
+"""`grep` for logs: parse a grep-like expression, run it with the `regex` module, show it back as a command.
+
+Nothing here executes anything: `shlex.split` only tokenizes, the pattern goes to `regex` with a per-line timeout
+and an overall time budget, so `(a+)+$` on a long line costs at most the budget, not the bot.
+Supported: PATTERN, -i, -v, -E, -c, -C N (also -C2, -iv, `--`, a leading `grep`). Anything else is an error.
+"""
+
+import shlex
+import time
+from dataclasses import asdict, dataclass, field
+from enum import StrEnum
+from typing import Any
+
+import regex
+
+MAX_PATTERN = 200
+MAX_CONTEXT = 10
+MAX_LINES = 500 # output lines, context and `--` separators included
+LINE_TIMEOUT = 0.05 # seconds per line
+BUDGET = 3.0 # seconds per search; on overrun the result is partial and says so
+
+_BRE_SPECIAL = set("+?|(){}")
+
+
+class GrepErrorCode(StrEnum):
+ EMPTY = "empty"
+ QUOTES = "quotes"
+ FLAG = "flag"
+ CONTEXT = "context"
+ EXTRA = "extra"
+ TOO_LONG = "too_long"
+ REGEX = "regex"
+
+
+class GrepError(ValueError):
+ def __init__(self, code: GrepErrorCode, arg: str = "") -> None:
+ super().__init__(f"{code}: {arg}" if arg else code.value)
+ self.code = code
+ self.arg = arg
+
+
+@dataclass(frozen=True)
+class GrepQuery:
+ pattern: str
+ ignore_case: bool = False
+ invert: bool = False
+ extended: bool = False
+ count: bool = False
+ context: int = 0
+
+ def to_dict(self) -> dict[str, Any]:
+ return asdict(self)
+
+ def compile(self) -> "regex.Pattern[str]":
+ source = self.pattern if self.extended else bre_to_ere(self.pattern)
+ try:
+ return regex.compile(source, regex.IGNORECASE if self.ignore_case else 0)
+ except regex.error as e:
+ raise GrepError(GrepErrorCode.REGEX, str(e)) from e
+
+
+def bre_to_ere(pattern: str) -> str:
+ """Basic grep regex → Python: without -E `+ ? | ( ) { }` are literals and `\\+` etc. are the operators."""
+ out: list[str] = []
+ i, in_class = 0, False
+ while i < len(pattern):
+ ch = pattern[i]
+ if in_class:
+ out.append(ch)
+ if ch == "]" and not (out[-2:-1] == ["["] or out[-3:-1] == ["[", "^"]):
+ in_class = False
+ elif ch == "[":
+ in_class = True
+ out.append(ch)
+ elif ch == "\\" and i + 1 < len(pattern):
+ nxt = pattern[i + 1]
+ out.append(nxt if nxt in _BRE_SPECIAL else ch + nxt)
+ i += 1
+ elif ch in _BRE_SPECIAL:
+ out.append("\\" + ch)
+ else:
+ out.append(ch)
+ i += 1
+ return "".join(out)
+
+
+def _context(value: str) -> int:
+ if not value.isdigit() or int(value) > MAX_CONTEXT:
+ raise GrepError(GrepErrorCode.CONTEXT, value)
+ return int(value)
+
+
+def parse_grep(expr: str) -> GrepQuery:
+ try:
+ tokens = shlex.split(expr)
+ except ValueError as e:
+ raise GrepError(GrepErrorCode.QUOTES) from e
+ if tokens[:1] == ["grep"]:
+ tokens = tokens[1:]
+ opts: dict[str, Any] = {}
+ patterns: list[str] = []
+ i = 0
+ while i < len(tokens):
+ tok = tokens[i]
+ if patterns or tok == "--" or not tok.startswith("-") or tok == "-":
+ if tok == "--" and not patterns:
+ patterns.extend(tokens[i + 1 :])
+ break
+ patterns.append(tok)
+ elif tok.startswith("-C"):
+ if len(tok) > 2:
+ opts["context"] = _context(tok[2:])
+ elif i + 1 < len(tokens):
+ opts["context"] = _context(tokens[i + 1])
+ i += 1
+ else:
+ raise GrepError(GrepErrorCode.CONTEXT, "")
+ else:
+ for j, flag in enumerate(tok[1:], start=1):
+ match flag:
+ case "i":
+ opts["ignore_case"] = True
+ case "v":
+ opts["invert"] = True
+ case "E":
+ opts["extended"] = True
+ case "c":
+ opts["count"] = True
+ case "C" if tok[j + 1 :].isdigit():
+ opts["context"] = _context(tok[j + 1 :])
+ break
+ case _:
+ raise GrepError(GrepErrorCode.FLAG, f"-{flag}")
+ i += 1
+ if not patterns or not patterns[0]:
+ raise GrepError(GrepErrorCode.EMPTY)
+ if len(patterns) > 1:
+ raise GrepError(GrepErrorCode.EXTRA, patterns[1])
+ if len(patterns[0]) > MAX_PATTERN:
+ raise GrepError(GrepErrorCode.TOO_LONG, str(len(patterns[0])))
+ query = GrepQuery(pattern=patterns[0], **opts)
+ query.compile() # syntax errors now, not halfway through the logs
+ return query
+
+
+def to_command(query: GrepQuery) -> str:
+ """The query as a shell-looking line, for the user to see what was searched (SPEC L-15)."""
+ flags = "".join(
+ f for f, on in (("i", query.ignore_case), ("v", query.invert), ("E", query.extended), ("c", query.count)) if on
+ )
+ parts = ["grep"]
+ if flags:
+ parts.append(f"-{flags}")
+ if query.context:
+ parts += ["-C", str(query.context)]
+ if query.pattern.startswith("-"):
+ parts.append("--")
+ parts.append(shlex.quote(query.pattern))
+ return " ".join(parts)
+
+
+@dataclass
+class GrepResult:
+ lines: list[str] = field(default_factory=list)
+ matched: int = 0
+ truncated: bool = False # more output than MAX_LINES
+ partial: bool = False # time budget ran out: not every line was searched
+ skipped: int = 0 # lines where the pattern hit the per-line timeout
+
+
+def run(query: GrepQuery, text: str, max_lines: int = MAX_LINES, budget: float = BUDGET) -> GrepResult:
+ pattern = query.compile()
+ lines = text.splitlines()
+ result = GrepResult()
+ deadline = time.monotonic() + budget
+ last_emitted = -1 # index of the last line already in the output (for context merging)
+ for n, line in enumerate(lines):
+ if time.monotonic() > deadline:
+ result.partial = True
+ break
+ try:
+ hit = pattern.search(line, timeout=LINE_TIMEOUT) is not None
+ except TimeoutError:
+ result.skipped += 1
+ continue
+ if hit == query.invert:
+ continue
+ result.matched += 1
+ if query.count or result.truncated:
+ continue
+ start = max(n - query.context, last_emitted + 1)
+ if query.context and result.lines and start > last_emitted + 1:
+ result.lines.append("--")
+ end = min(n + query.context, len(lines) - 1)
+ for k in range(start, end + 1):
+ if len(result.lines) >= max_lines:
+ result.truncated = True
+ break
+ result.lines.append(lines[k])
+ last_emitted = k
+ return result
diff --git a/app/logs/journal.py b/app/logs/journal.py
new file mode 100644
index 0000000..5060b74
--- /dev/null
+++ b/app/logs/journal.py
@@ -0,0 +1,15 @@
+"""Every log search is recorded: who, which container, what, how many lines came back (SPEC L-6)."""
+
+from typing import Any
+
+from app.db.models import LogMode, LogQuery
+from app.db.session import SessionMaker
+
+
+class LogJournal:
+ def __init__(self, sessions: SessionMaker) -> None:
+ self._sessions = sessions
+
+ async def record(self, user_id: int, container: str, mode: LogMode, query: dict[str, Any], lines: int) -> None:
+ async with self._sessions.begin() as session:
+ session.add(LogQuery(user_id=user_id, container=container, mode=mode, query=query, lines=lines))
diff --git a/app/main.py b/app/main.py
index 352714d..06f2ee0 100644
--- a/app/main.py
+++ b/app/main.py
@@ -11,12 +11,13 @@ from app import __version__
from app.access.service import AccessService
from app.bot.fsm import make_storage
from app.bot.middlewares.access import AccessMiddleware
-from app.bot.routers import commit, common, members, status
+from app.bot.routers import commit, common, logs, members, status
from app.config import Settings
from app.db.maintenance import cleanup_loop
from app.db.migrate import upgrade
from app.db.session import SessionMaker, make_engine, make_sessionmaker
from app.health import heartbeat
+from app.logs.journal import LogJournal
from app.logs_setup import setup_logging
from app.repo.drafts import DraftService
from app.repo.git import UploadsRepo
@@ -48,6 +49,7 @@ def build_dispatcher(
commit.build_router(texts, repo, DraftService(sessions), Publisher(repo, sessions), commit_url_template)
)
dp.include_router(status.build_router(texts, docker, checker))
+ dp.include_router(logs.build_router(texts, docker, LogJournal(sessions)))
dp.include_router(common.build_router(texts, access))
return dp
diff --git a/app/runtime/docker.py b/app/runtime/docker.py
index 3fcaa7b..64ab0f5 100644
--- a/app/runtime/docker.py
+++ b/app/runtime/docker.py
@@ -171,7 +171,7 @@ class DockerClient:
def __init__(self, settings: DockerSettings, transport: httpx.AsyncBaseTransport | None = None) -> None:
self._s = settings
self._http = httpx.AsyncClient(base_url=settings.url, timeout=settings.timeout, transport=transport)
- self._known: set[str] = set() # ids from the last listing: the only ones card/logs/stats may touch
+ self._known: dict[str, Container] = {} # the last listing: the only ids card/logs/stats may touch
async def __aenter__(self) -> Self:
return self
@@ -204,22 +204,22 @@ class DockerClient:
labels=labels,
)
)
- self._known = {c.id for c in result}
+ self._known = {c.id: c for c in result}
return sorted(result, key=lambda c: c.name)
- async def resolve(self, short_id: str) -> str:
- """Full id of a managed container by its 12-char prefix (callback data carries only that)."""
- if not self._known:
- await self.containers()
- for full in self._known:
- if full.startswith(short_id):
- return full
- await self.containers() # maybe recreated since the last listing
- for full in self._known:
- if full.startswith(short_id):
- return full
+ async def get(self, short_id: str) -> Container:
+ """A managed container by its 12-char id prefix (callback data carries only that)."""
+ for attempt in range(2):
+ if attempt or not self._known:
+ await self.containers() # first use, or recreated since the last listing
+ for full, container in self._known.items():
+ if short_id and full.startswith(short_id):
+ return container
raise DockerError(f"container {short_id} is not managed by tgit")
+ async def resolve(self, short_id: str) -> str:
+ return (await self.get(short_id)).id
+
async def inspect(self, container_id: str) -> Details:
self._guard(container_id)
d = await self._get_json(f"/containers/{container_id}/json")
@@ -246,15 +246,18 @@ class DockerClient:
return parse_stats(await self._get_json(f"/containers/{container_id}/stats", params={"stream": "false"}))
async def logs(self, container_id: str, since: datetime | None = None, tail: int | None = None) -> str:
+ """stdout+stderr in order; at most max_log_lines lines and max_log_bytes — the newest ones are kept."""
self._guard(container_id)
- params = {"stdout": "1", "stderr": "1"}
+ lines = min(tail, self._s.max_log_lines) if tail is not None else self._s.max_log_lines
+ params = {"stdout": "1", "stderr": "1", "tail": str(lines)}
if since is not None:
params["since"] = str(int(since.timestamp()))
- if tail is not None:
- params["tail"] = str(tail)
response = await self._request(f"/containers/{container_id}/logs", params)
- raw = response.content[: self._s.max_log_bytes]
- return demux_logs(raw)
+ text = demux_logs(response.content)
+ if len(text) > self._s.max_log_bytes:
+ cut = text[-self._s.max_log_bytes :]
+ text = cut[cut.find("\n") + 1 :] # drop the torn first line
+ return text
async def ping(self) -> bool:
try:
diff --git a/app/texts.py b/app/texts.py
index fd30dbc..99cde01 100644
--- a/app/texts.py
+++ b/app/texts.py
@@ -33,7 +33,6 @@ class HelpTexts(_Strict):
class SectionTexts(_Strict):
commit: str
- logs: str
alerts: str
members: str
cancelled: str
@@ -152,6 +151,37 @@ class StatusTexts(_Strict):
seconds: str
+class LogsTexts(_Strict):
+ title: str
+ empty: str
+ unavailable: str
+ container: str
+ tail: str
+ since: str
+ minutes: str
+ hours: str
+ grep_button: str
+ agent_button: str
+ agent_soon: str
+ to_list: str
+ to_container: str
+ ask_grep: str
+ head_tail: str
+ head_grep: str
+ nothing: str
+ truncated: str
+ partial: str
+ skipped: str
+ sent_as_file: str
+ err_empty: str
+ err_quotes: str
+ err_flag: str
+ err_context: str
+ err_extra: str
+ err_too_long: str
+ err_regex: str
+
+
class Texts(_Strict):
menu: MenuTexts
start: StartTexts
@@ -160,6 +190,7 @@ class Texts(_Strict):
members: MembersTexts
commit: CommitTexts
status: StatusTexts
+ logs: LogsTexts
def load_texts(config_dir: Path) -> Texts:
diff --git a/config/texts.yaml b/config/texts.yaml
index f8e30fa..a255c3d 100644
--- a/config/texts.yaml
+++ b/config/texts.yaml
@@ -30,7 +30,6 @@ help:
section:
commit: "📝 Коммит\n\nРаздел в разработке."
- logs: "📜 Логи\n\nРаздел в разработке."
alerts: "🚨 Алерты\n\nРаздел в разработке."
members: "👥 Участники\n\nРаздел в разработке."
cancelled: "Отменено. Главное меню — кнопками внизу."
@@ -167,3 +166,40 @@ status:
hours: "ч"
minutes: "мин"
seconds: "с"
+
+logs:
+ title: "📜 Логи · выберите контейнер"
+ empty: "Контейнеров tgit не найдено."
+ unavailable: "Docker недоступен, попробуйте через минуту."
+ container: |-
+ 📜 Логи · {name}
+
+ Хвост, свой grep по последним 24 ч или вопрос агенту.
+ tail: "{n} строк"
+ since: "за {period}"
+ minutes: "{n} мин"
+ hours: "{n} ч"
+ grep_button: "🔎 Свой grep"
+ agent_button: "🤖 Спросить агента"
+ agent_soon: "Агент — в следующей версии."
+ to_list: "⬅️ К контейнерам"
+ to_container: "⬅️ К контейнеру"
+ ask_grep: |-
+ 🔎 Выражение grep — ищу по логу {name} за 24 ч.
+
+ Например: -i -C 2 error
+ Флаги: -i без регистра, -v кроме, -E расширенный regex, -c только число, -C N контекст.
+ head_tail: "📜 {name} · {scope} · строк: {lines}"
+ head_grep: "🔎 {command}\n{name} · за 24 ч · совпадений: {matched}"
+ nothing: "Пусто."
+ truncated: "Показаны первые {n} строк."
+ partial: "Поиск остановлен по времени — просмотрена часть лога."
+ skipped: "Пропущено строк из-за тяжёлого шаблона: {n}."
+ sent_as_file: "Результат длинный — отправил файлом."
+ err_empty: "Нужен шаблон, например: -i error"
+ err_quotes: "Не закрыта кавычка."
+ err_flag: "Флаг {arg} не поддерживается. Можно: -i -v -E -c -C N"
+ err_context: "После -C нужно число от 0 до 10."
+ err_extra: "Нужен один шаблон, а «{arg}» лишнее. Шаблон с пробелами — в кавычках."
+ err_too_long: "Шаблон длиннее 200 символов."
+ err_regex: "Шаблон не разбирается: {arg}"
diff --git a/tests/bot/test_commit_edit.py b/tests/bot/test_commit_edit.py
index 47601f1..593256c 100644
--- a/tests/bot/test_commit_edit.py
+++ b/tests/bot/test_commit_edit.py
@@ -160,3 +160,15 @@ async def test_protected_file_cannot_be_replaced(tg: TgClient, texts: Texts, rep
answer = tg.session.sent(AnswerCallbackQuery)[-1]
assert answer.text == texts.commit.protected and answer.show_alert
+
+
+@pytest.mark.parametrize("escape", ["menu_status", "/menu", "/help"])
+async def test_menu_and_commands_escape_input_states(tg: TgClient, texts: Texts, escape: str) -> None:
+ text = texts.menu.status if escape == "menu_status" else escape
+ await to_hello(tg, texts)
+ await tg.press(ADMIN, cb(Act.NEW_FILE))
+ await tg.send(ADMIN, text)
+
+ reply = last_text(tg)
+ assert reply.text != texts.commit.bad_name
+ assert not reply.text.startswith(texts.commit.ask_content.split("{")[0])
diff --git a/tests/bot/test_common.py b/tests/bot/test_common.py
index cf28242..08f8705 100644
--- a/tests/bot/test_common.py
+++ b/tests/bot/test_common.py
@@ -46,7 +46,7 @@ async def test_outsider_stale_button_is_swallowed(tg: TgClient) -> None:
assert len(tg.session.sent(AnswerCallbackQuery)) == 1
-@pytest.mark.parametrize("section", ["logs", "alerts"])
+@pytest.mark.parametrize("section", ["alerts"])
async def test_menu_button_opens_section_with_nav(tg: TgClient, texts: Texts, section: str) -> None:
await tg.send(ADMIN, getattr(texts.menu, section))
diff --git a/tests/bot/test_logs.py b/tests/bot/test_logs.py
new file mode 100644
index 0000000..bdd78af
--- /dev/null
+++ b/tests/bot/test_logs.py
@@ -0,0 +1,159 @@
+from collections.abc import AsyncIterator
+
+import pytest
+from aiogram.methods import AnswerCallbackQuery, EditMessageText, SendDocument, SendMessage
+from sqlalchemy import select
+
+from app.access.service import AccessService
+from app.bot.routers.logs import LogsAct, LogsCb
+from app.db.models import LogMode, LogQuery
+from app.db.session import SessionMaker
+from app.main import build_dispatcher
+from app.repo.git import UploadsRepo
+from app.runtime.docker import DockerClient
+from app.texts import Texts
+from tests.conftest import ADMIN
+from tests.fakes import TgClient
+from tests.runtime.fake_docker import FakeDocker, now_minus
+
+
+@pytest.fixture
+def fake() -> FakeDocker:
+ fake = FakeDocker()
+ fake.add(
+ "crash",
+ logs=[
+ (now_minus(120), "boot"),
+ (now_minus(30), "connect db"),
+ (now_minus(29), "ERROR connection refused"),
+ (now_minus(28), "retry"),
+ (now_minus(2), "html-ish & stuff"),
+ ],
+ )
+ fake.add("vobot-api-1", managed=False, logs=[(now_minus(1), "SECRET")])
+ return fake
+
+
+@pytest.fixture
+async def docker(fake: FakeDocker) -> AsyncIterator[DockerClient]:
+ async with fake.client() as client:
+ yield client
+
+
+@pytest.fixture
+def tg(
+ texts: Texts, sessions: SessionMaker, access: AccessService, repo: UploadsRepo, docker: DockerClient
+) -> TgClient:
+ return TgClient(build_dispatcher(texts, sessions, access, repo, docker=docker))
+
+
+def cid(fake: FakeDocker, name: str) -> str:
+ return next(c.id[:12] for c in fake.containers.values() if c.name == name)
+
+
+async def journal(sessions: SessionMaker) -> list[LogQuery]:
+ async with sessions() as s:
+ return list(await s.scalars(select(LogQuery).order_by(LogQuery.id)))
+
+
+async def test_open_lists_only_managed(tg: TgClient, texts: Texts) -> None:
+ await tg.send(ADMIN, texts.menu.logs)
+
+ [reply] = tg.session.sent(SendMessage)
+ assert reply.text == texts.logs.title
+ labels = [b.text for row in reply.reply_markup.inline_keyboard for b in row]
+ assert "🟢 crash" in labels and not any("vobot" in label for label in labels)
+
+
+async def test_tail_is_escaped_and_journaled(tg: TgClient, fake: FakeDocker, sessions: SessionMaker) -> None:
+ await tg.press(ADMIN, LogsCb(a=LogsAct.TAIL, id=cid(fake, "crash"), n=50).pack())
+
+ [reply] = tg.session.sent(SendMessage)
+ assert "строк: 5" in reply.text
+ assert "<b>html-ish</b> & stuff" in reply.text
+ [row] = await journal(sessions)
+ assert (row.mode, row.container, row.query, row.lines) == (LogMode.TAIL, "crash", {"tail": 50}, 5)
+
+
+async def test_since_period(tg: TgClient, fake: FakeDocker) -> None:
+ await tg.press(ADMIN, LogsCb(a=LogsAct.SINCE, id=cid(fake, "crash"), n=5).pack())
+
+ [reply] = tg.session.sent(SendMessage)
+ assert "за 5 мин · строк: 1" in reply.text
+
+
+async def test_grep_flow(tg: TgClient, texts: Texts, fake: FakeDocker, sessions: SessionMaker) -> None:
+ await tg.press(ADMIN, LogsCb(a=LogsAct.GREP, id=cid(fake, "crash")).pack())
+ await tg.send(ADMIN, "-i -C 1 error")
+
+ ask, result = tg.session.sent(SendMessage)
+ assert ask.text.startswith("🔎 Выражение grep")
+ assert "grep -i -C 1 error" in result.text
+ assert "совпадений: 1" in result.text
+ assert "connect db\nERROR connection refused\nretry" in result.text
+ [row] = await journal(sessions)
+ assert row.mode is LogMode.GREP and row.query["pattern"] == "error" and row.lines == 1
+
+ tg.session.requests.clear()
+ await tg.send(ADMIN, "error") # state is cleared after one search
+ assert not any("совпадений" in (m.text or "") for m in tg.session.sent(SendMessage))
+
+
+async def test_grep_error_keeps_waiting(tg: TgClient, fake: FakeDocker) -> None:
+ await tg.press(ADMIN, LogsCb(a=LogsAct.GREP, id=cid(fake, "crash")).pack())
+ await tg.send(ADMIN, "-r error")
+ await tg.send(ADMIN, "-c retry")
+
+ _, err, result = tg.session.sent(SendMessage)
+ assert "Флаг -r не поддерживается" in err.text
+ assert "совпадений: 1" in result.text and "" not in result.text
+
+
+async def test_grep_nothing_found(tg: TgClient, texts: Texts, fake: FakeDocker) -> None:
+ await tg.press(ADMIN, LogsCb(a=LogsAct.GREP, id=cid(fake, "crash")).pack())
+ await tg.send(ADMIN, "nope")
+
+ assert tg.session.sent(SendMessage)[-1].text.endswith(texts.logs.nothing)
+
+
+async def test_menu_button_escapes_grep_state(tg: TgClient, texts: Texts, fake: FakeDocker) -> None:
+ await tg.press(ADMIN, LogsCb(a=LogsAct.GREP, id=cid(fake, "crash")).pack())
+ await tg.send(ADMIN, texts.menu.logs)
+
+ assert tg.session.sent(SendMessage)[-1].text == texts.logs.title
+
+
+async def test_long_result_goes_as_file(tg: TgClient, fake: FakeDocker) -> None:
+ crash = next(c for c in fake.containers.values() if c.name == "crash")
+ crash.logs = [(now_minus(1), f"line {i} " + "x" * 50) for i in range(200)]
+
+ await tg.press(ADMIN, LogsCb(a=LogsAct.TAIL, id=cid(fake, "crash"), n=200).pack())
+
+ [doc] = tg.session.sent(SendDocument)
+ assert doc.document.filename == "crash.log"
+ assert doc.document.data.decode().count("\n") == 199
+
+
+async def test_foreign_container_refused(tg: TgClient, texts: Texts, fake: FakeDocker) -> None:
+ await tg.press(ADMIN, LogsCb(a=LogsAct.TAIL, id=cid(fake, "vobot-api-1"), n=50).pack())
+
+ assert not tg.session.sent(SendMessage)
+ assert tg.session.sent(AnswerCallbackQuery)[0].text == texts.logs.unavailable
+
+
+async def test_open_container_and_back_to_list(tg: TgClient, texts: Texts, fake: FakeDocker) -> None:
+ await tg.press(ADMIN, LogsCb(a=LogsAct.OPEN, id=cid(fake, "crash")).pack())
+ await tg.press(ADMIN, LogsCb(a=LogsAct.LIST).pack())
+ await tg.press(ADMIN, LogsCb(a=LogsAct.AGENT, id=cid(fake, "crash")).pack())
+
+ card, listing = tg.session.sent(EditMessageText)
+ assert card.text.startswith("📜 Логи · crash")
+ assert listing.text == texts.logs.title
+ assert tg.session.sent(AnswerCallbackQuery)[-1].text == texts.logs.agent_soon
+
+
+async def test_without_docker(texts: Texts, sessions: SessionMaker, access: AccessService, repo: UploadsRepo) -> None:
+ tg = TgClient(build_dispatcher(texts, sessions, access, repo))
+ await tg.send(ADMIN, texts.menu.logs)
+
+ assert [m.text for m in tg.session.sent(SendMessage)] == [texts.logs.unavailable]
diff --git a/tests/logs/__init__.py b/tests/logs/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/tests/logs/test_grep.py b/tests/logs/test_grep.py
new file mode 100644
index 0000000..33fecce
--- /dev/null
+++ b/tests/logs/test_grep.py
@@ -0,0 +1,151 @@
+import time
+
+import pytest
+
+from app.logs.grep import (
+ MAX_PATTERN,
+ GrepError,
+ GrepErrorCode,
+ GrepQuery,
+ bre_to_ere,
+ parse_grep,
+ run,
+ to_command,
+)
+
+LOG = "\n".join(
+ [
+ "08:00 start",
+ "08:01 connect db",
+ "08:02 ERROR connection refused",
+ "08:03 retry",
+ "08:04 ok",
+ "08:05 idle",
+ "08:06 idle",
+ "08:07 error: timeout",
+ "08:08 done",
+ ]
+)
+
+
+@pytest.mark.parametrize(
+ ("expr", "expected"),
+ [
+ ("error", GrepQuery("error")),
+ ("grep -i error", GrepQuery("error", ignore_case=True)),
+ ("-iv 'two words'", GrepQuery("two words", ignore_case=True, invert=True)),
+ ("-E -c 'a|b'", GrepQuery("a|b", extended=True, count=True)),
+ ("-C 2 x", GrepQuery("x", context=2)),
+ ("-C2 x", GrepQuery("x", context=2)),
+ ("-iC3 x", GrepQuery("x", ignore_case=True, context=3)),
+ ("-- -v", GrepQuery("-v")),
+ ('"привет мир"', GrepQuery("привет мир")),
+ ],
+)
+def test_parse(expr: str, expected: GrepQuery) -> None:
+ assert parse_grep(expr) == expected
+
+
+@pytest.mark.parametrize(
+ ("expr", "code"),
+ [
+ ("", GrepErrorCode.EMPTY),
+ ("-i", GrepErrorCode.EMPTY),
+ ("'unclosed", GrepErrorCode.QUOTES),
+ ("-r error", GrepErrorCode.FLAG),
+ ("--include=*.py x", GrepErrorCode.FLAG),
+ ("-C x y", GrepErrorCode.CONTEXT),
+ ("-C 99 y", GrepErrorCode.CONTEXT),
+ ("-C", GrepErrorCode.CONTEXT),
+ ("error /etc/passwd", GrepErrorCode.EXTRA),
+ ("x" * (MAX_PATTERN + 1), GrepErrorCode.TOO_LONG),
+ ("-E '(unclosed'", GrepErrorCode.REGEX),
+ ],
+)
+def test_parse_errors(expr: str, code: GrepErrorCode) -> None:
+ with pytest.raises(GrepError) as e:
+ parse_grep(expr)
+ assert e.value.code is code
+
+
+def test_shell_is_never_involved() -> None:
+ # metacharacters are just pattern text
+ q = parse_grep("'$(rm -rf /); `id`'")
+ assert q.pattern == "$(rm -rf /); `id`"
+ assert to_command(q) == "grep '$(rm -rf /); `id`'"
+
+
+def test_basic_vs_extended_regex() -> None:
+ assert bre_to_ere("a+b") == r"a\+b"
+ assert bre_to_ere(r"a\+b") == "a+b"
+ assert bre_to_ere(r"\(x\|y\)") == "(x|y)"
+ assert bre_to_ere("[+?]x") == "[+?]x"
+ assert bre_to_ere("[]a]+") == r"[]a]\+"
+ assert run(parse_grep("1+1"), "1+1\n11").lines == ["1+1"]
+ assert run(parse_grep("-E 1+1"), "1+1\n11").lines == ["11"]
+
+
+def test_ignore_case_and_invert() -> None:
+ assert run(parse_grep("-i error"), LOG).lines == ["08:02 ERROR connection refused", "08:07 error: timeout"]
+ assert run(parse_grep("-v idle"), LOG).matched == 7
+
+
+def test_count() -> None:
+ result = run(parse_grep("-ci error"), LOG)
+ assert result.matched == 2
+ assert result.lines == []
+
+
+def test_context_merges_and_separates_groups() -> None:
+ result = run(parse_grep("-i -C 1 error"), LOG)
+ assert result.lines == [
+ "08:01 connect db",
+ "08:02 ERROR connection refused",
+ "08:03 retry",
+ "--",
+ "08:06 idle",
+ "08:07 error: timeout",
+ "08:08 done",
+ ]
+ overlapping = run(parse_grep("-C 2 idle"), LOG)
+ assert overlapping.lines == LOG.splitlines()[3:] # 08:03…08:08, no duplicates, no separator
+
+
+def test_output_is_capped() -> None:
+ text = "\n".join(f"line {i}" for i in range(2000))
+ result = run(parse_grep("line"), text, max_lines=500)
+ assert len(result.lines) == 500
+ assert result.truncated
+ assert result.matched == 2000 # counting goes on
+
+
+@pytest.mark.parametrize("pattern", ["(a+)+$", "(a|aa)+$"]) # regex optimizes the first; the second really blows up
+def test_redos_does_not_hang(pattern: str) -> None:
+ evil = parse_grep(f"-E '{pattern}'")
+ text = "\n".join(["a" * 40 + "!"] * 200 + ["fine"])
+ started = time.monotonic()
+
+ result = run(evil, text, budget=1.0)
+
+ assert time.monotonic() - started < 2.0
+ assert result.matched == 0
+
+
+def test_big_input() -> None:
+ text = "\n".join(f"{i} request ok" if i % 1000 else f"{i} ERROR" for i in range(1, 100_001))
+ result = run(parse_grep("ERROR"), text)
+ assert result.matched == 100
+ assert not result.partial
+
+
+def test_to_command() -> None:
+ assert to_command(parse_grep("-i -C 2 -E 'conn(ect)?'")) == "grep -iE -C 2 'conn(ect)?'"
+ assert to_command(GrepQuery("-x")) == "grep -- -x"
+ assert parse_grep(to_command(parse_grep("-vc 'a b'"))) == parse_grep("-vc 'a b'")
+
+
+def test_catastrophic_lines_are_skipped_then_budget_stops() -> None:
+ text = "\n".join(["a" * 40 + "!"] * 100)
+ result = run(parse_grep("-E '(a|aa)+$'"), text, budget=0.3)
+ assert result.skipped >= 1
+ assert result.partial
diff --git a/tests/runtime/fake_docker.py b/tests/runtime/fake_docker.py
index bd5f9ed..5112a8f 100644
--- a/tests/runtime/fake_docker.py
+++ b/tests/runtime/fake_docker.py
@@ -97,7 +97,8 @@ class FakeDocker:
return httpx.Response(200, json=self._stats(c))
case "logs":
since = datetime.fromtimestamp(int(q.get("since", "0")), UTC)
- body = b"".join(frame(line + "\n", 2 if "ERR" in line else 1) for at, line in c.logs if at >= since)
+ picked = [line for at, line in c.logs if at >= since][-int(q.get("tail", "1000000")) :]
+ body = b"".join(frame(line + "\n", 2 if "ERR" in line else 1) for line in picked)
return httpx.Response(200, content=body)
return httpx.Response(404, json={"message": "No such container"})