Files
LLeMbas/tests/test_terminal.py
Homer 3d51ba061e Tests that found things reading did not
The testing pass: 2140 tests to 2283, and four bugs that no amount of
reading had turned up. Three came from driving the JavaScript under a
Node DOM stub, which is the practice CLAUDE.md sets out and this is the
reason it does.

The terminal dropped every keystroke after a reconnect. `onclose` closed
over the module-level socket rather than its own, and close() queues its
event -- so the old socket's close arrived after a new one was assigned
and nulled the live one. Output kept coming, because onmessage is bound
to the object, while every send gates on the variable. It also announced
"Disconnected" about a shell that had just reconnected.

Two scripts were loaded twice on /messages, once by base.html and again
by the page. Each is an IIFE with its own state, so four keyboard
shortcuts toggled their panel twice and therefore did nothing, /help
opened two dialogs, and an @ mention attached its file twice. A sweep
refuses any template re-loading what base.html has.

The microphone had no guard while the permission prompt was up, so each
click opened another stream and only the last was ever stopped. And a
skill shared with you took its name out of your own library: create
checked uniqueness against what is *visible* rather than what is owned,
against a (owner_id, name) constraint, and told you to edit a row you
cannot edit.

--ink-faint failed the contrast minimum in both themes -- 3.85 and 3.19
against 4.5 -- so the smallest text on every screen was the hardest to
read. Measured in a headless browser rather than judged by eye.

And the suite runs on 3.11 and 3.12 now as well as 3.14. It had only ever
run on 3.14 while the image ships 3.12 and the packaging claimed 3.11:
the interpreter most people would run was the one nothing had tested.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-07 14:41:45 +02:00

465 lines
16 KiB
Python

"""The shell behind the terminal panel.
Everything here runs against a real SSH server with a real PTY, because the
whole point of this module is what a pseudo-terminal does and a stub would agree
with any design at all.
"""
from __future__ import annotations
import asyncio
import pytest
from lembas.services.agent import terminal as terminal_service
from lembas.services.agent.base import ExecError
from lembas.services.agent.terminal import Session
# Stands up something real -- see the `slow` marker in pyproject.toml.
pytestmark = pytest.mark.slow
asyncssh = pytest.importorskip("asyncssh")
# --- A machine with a shell on it ----------------------------------------------
class _Server(asyncssh.SSHServer):
def begin_auth(self, username: str) -> bool:
return False
@pytest.fixture
async def shell_host():
"""A server whose sessions behave enough like a shell to be tested against.
It announces itself, echoes what is typed at it, and remembers what the PTY
was asked for -- which is the part the client side has to get right.
"""
seen: dict = {}
async def handler(process):
seen["command"] = process.command
seen["term_type"] = process.get_terminal_type()
seen["term_size"] = process.get_terminal_size()
seen["process"] = process
process.stdout.write("READY\n")
while True:
try:
line = await process.stdin.readline()
except asyncssh.TerminalSizeChanged:
# A window change interrupts the read rather than arriving as
# data. A real shell redraws and carries on.
continue
except Exception: # noqa: BLE001 - the session ending is not a failure
break
if not line or line.rstrip("\n") == "exit":
break
process.stdout.write(f"echo:{line.rstrip()}\n")
process.exit(0)
server = await asyncssh.create_server(
_Server,
"127.0.0.1",
0,
server_host_keys=[asyncssh.generate_private_key("ssh-ed25519")],
process_factory=handler,
)
port = next(iter(server.sockets)).getsockname()[1]
from lembas.services.agent import ssh as ssh_service
line, _fingerprint = await ssh_service.capture_host_key("127.0.0.1", port)
try:
yield {"port": port, "host_key": line, "seen": seen}
finally:
# Before the server, always: a client connection still open makes
# `wait_closed` wait for it, and a test that failed mid-way has one.
await terminal_service.shutdown()
server.close()
await server.wait_closed()
def _spec(host) -> dict:
return {
"host": "127.0.0.1",
"port": host["port"],
"username": "tester",
"auth": "password",
"password": "",
"private_key": "",
"key_passphrase": "",
"host_key": host["host_key"],
"connect_timeout": 10,
}
async def _open(host, *, chat_id="chat-1", owner="user-1", project_dir="", **kwargs) -> Session:
return await terminal_service.open_session(
chat_id,
owner_id=owner,
profile_id="profile-1",
label="Test box",
spec=_spec(host),
project_dir=project_dir,
**kwargs,
)
async def _read(viewer, *, timeout: float = 5.0) -> bytes:
return await asyncio.wait_for(viewer.queue.get(), timeout=timeout)
async def _read_until(viewer, needle: bytes, *, timeout: float = 5.0) -> bytes:
"""Frames are whatever the far side wrote, so a line can arrive in pieces.
Starts from the snapshot, because a viewer that attached after the shell
had already said something finds it there rather than on the queue -- which
is the whole design, and would otherwise make this hang.
"""
seen = viewer.snapshot
while needle not in seen:
chunk = await _read(viewer, timeout=timeout)
assert chunk is not None, f"the session closed before {needle!r} arrived"
seen += chunk
return seen
# --- The PTY --------------------------------------------------------------------
async def test_the_pty_is_asked_for_with_a_term_type_and_a_size(shell_host):
session = await _open(shell_host, cols=100, rows=30)
viewer = session.attach(cols=100, rows=30)
await _read_until(viewer, b"READY")
assert shell_host["seen"]["term_type"] == "xterm-256color"
assert shell_host["seen"]["term_size"][:2] == (100, 30)
await session.close()
async def test_a_project_directory_becomes_a_cd_before_the_shell(shell_host):
session = await _open(shell_host, project_dir="/srv/work")
viewer = session.attach()
await _read_until(viewer, b"READY")
command = shell_host["seen"]["command"]
assert command is not None
assert "cd '/srv/work'" in command
assert "exec" in command
await session.close()
async def test_no_project_directory_and_no_integration_is_the_plain_login_shell(shell_host):
"""The original behaviour, kept reachable and kept tested.
Switching integration off has to give back *exactly* what was there before,
byte for byte -- including the None that means "whatever the account logs
in with". A fallback that is nearly the old behaviour is not a fallback.
"""
session = await _open(shell_host, integrate=False)
viewer = session.attach()
await _read_until(viewer, b"READY")
assert shell_host["seen"]["command"] is None
await session.close()
async def test_an_unknown_shell_falls_through_to_the_plain_login_shell(shell_host):
"""With integration on the command is no longer None -- but every branch
that does not recognise the shell ends in the same `exec` that was there
before, because a terminal that works without markers is worth more than
markers that break a terminal."""
session = await _open(shell_host)
viewer = session.attach()
await _read_until(viewer, b"READY")
command = shell_host["seen"]["command"]
assert "case ${SHELL##*/} in" in command
assert command.rstrip().endswith("exec ${SHELL:-/bin/sh} -l")
# Every step is silenced, so a full /tmp or a read-only home costs the
# markers and nothing else.
assert "2>/dev/null" in command
await session.close()
async def test_a_single_quote_in_the_directory_cannot_end_the_quoting(shell_host):
session = await _open(shell_host, project_dir="/tmp/it's here; rm -rf /")
viewer = session.attach()
await _read_until(viewer, b"READY")
command = shell_host["seen"]["command"]
assert command.startswith("cd '/tmp/it'\\''s here; rm -rf /'")
await session.close()
async def test_what_is_typed_reaches_the_shell_and_comes_back(shell_host):
session = await _open(shell_host)
viewer = session.attach()
await _read_until(viewer, b"READY")
await session.send(b"hello\n")
assert b"echo:hello" in await _read_until(viewer, b"echo:hello")
await session.close()
async def test_a_resize_reaches_the_far_side(shell_host):
session = await _open(shell_host)
viewer = session.attach(cols=80, rows=24)
await _read_until(viewer, b"READY")
session.resize(viewer, 120, 40)
process = shell_host["seen"]["process"]
for _ in range(50):
if process.get_terminal_size()[:2] == (120, 40):
break
await asyncio.sleep(0.02)
assert process.get_terminal_size()[:2] == (120, 40)
await session.close()
async def test_a_silly_size_is_clamped_rather_than_forwarded(shell_host):
session = await _open(shell_host)
viewer = session.attach(cols=100_000, rows=100_000)
await _read_until(viewer, b"READY")
assert viewer.cols == terminal_service.MAX_COLS
assert viewer.rows == terminal_service.MAX_ROWS
await session.close()
# --- Two tabs, one shell ---------------------------------------------------------
async def test_two_attachments_share_one_shell(shell_host):
session = await _open(shell_host)
first = session.attach()
await _read_until(first, b"READY")
second = session.attach()
await session.send(b"both\n")
assert b"echo:both" in await _read_until(first, b"echo:both")
assert b"echo:both" in await _read_until(second, b"echo:both")
await session.close()
async def test_opening_twice_returns_the_same_session(shell_host):
first = await _open(shell_host)
second = await _open(shell_host)
assert first is second
await first.close()
async def test_the_smaller_viewer_decides_the_size(shell_host):
session = await _open(shell_host)
wide = session.attach(cols=200, rows=60)
await _read_until(wide, b"READY")
session.attach(cols=90, rows=25)
assert (session.cols, session.rows) == (90, 25)
await session.close()
async def test_the_size_lifts_again_when_the_smaller_window_goes(shell_host):
session = await _open(shell_host)
wide = session.attach(cols=200, rows=60)
await _read_until(wide, b"READY")
narrow = session.attach(cols=90, rows=25)
session.detach(narrow)
assert (session.cols, session.rows) == (200, 60)
await session.close()
# --- Lifetime --------------------------------------------------------------------
async def test_detaching_leaves_the_session_running(shell_host):
session = await _open(shell_host)
viewer = session.attach()
await _read_until(viewer, b"READY")
session.detach(viewer)
assert terminal_service.get("chat-1") is session
assert not session.closed
await session.close()
async def test_reattaching_replays_the_scrollback(shell_host):
session = await _open(shell_host)
viewer = session.attach()
await _read_until(viewer, b"READY")
await session.send(b"before\n")
await _read_until(viewer, b"echo:before")
session.detach(viewer)
returning = session.attach()
assert b"READY" in returning.snapshot
assert b"echo:before" in returning.snapshot
await session.close()
async def test_scrollback_is_bounded(shell_host, monkeypatch):
monkeypatch.setattr(terminal_service, "SCROLLBACK_BYTES", 1024)
session = await _open(shell_host)
viewer = session.attach()
await _read_until(viewer, b"READY")
for index in range(200):
session._remember(f"line {index} ".encode() + b"y" * 100)
assert session._scrollback_bytes <= 1024 + 110
assert len(session.attach().snapshot) <= 1024 + 110
await session.close()
async def test_a_slow_reader_is_dropped_rather_than_buffered(monkeypatch):
"""No server needed: fanning out is a plain function over the viewers.
The alternative design -- a pump that waits for a full queue -- would stall
every other viewer behind the one that stopped reading, and buffer without
bound while it did.
"""
monkeypatch.setattr(terminal_service, "VIEWER_QUEUE", 4)
session = Session("chat-1", owner_id="user-1", profile_id="p", label="box")
keeping_up = session.attach()
stalled = session.attach()
for index in range(20):
session._fan_out(f"chunk {index}\n".encode())
while not keeping_up.queue.empty():
keeping_up.queue.get_nowait()
assert stalled.dropped
assert stalled.id not in session.viewers
assert keeping_up.id in session.viewers
assert not session.closed
# Woken rather than left waiting: the sentinel is how the socket learns to
# reconnect, and the scrollback is what makes that free.
assert stalled.queue.get_nowait() is None
async def test_the_shell_exiting_closes_the_session(shell_host):
session = await _open(shell_host)
viewer = session.attach()
await _read_until(viewer, b"READY")
await session.send(b"exit\n")
for _ in range(200):
if session.closed:
break
await asyncio.sleep(0.02)
assert session.closed
assert session.closed_reason == terminal_service.CLOSED_EXITED
# The viewer is woken with the sentinel rather than left waiting.
frame = await _read(viewer)
while frame is not None:
frame = await _read(viewer)
async def test_a_closed_session_lingers_with_its_reason(shell_host):
session = await _open(shell_host)
viewer = session.attach()
await _read_until(viewer, b"READY")
await session.close(terminal_service.CLOSED_IDLE)
assert terminal_service.get("chat-1") is None
lingering = terminal_service.peek("chat-1")
assert lingering is not None
assert lingering.closed_reason == terminal_service.CLOSED_IDLE
async def test_an_idle_session_is_reaped(shell_host, monkeypatch):
monkeypatch.setattr(terminal_service, "REAP_INTERVAL", 0.05)
session = await _open(shell_host, idle_timeout=0.1)
viewer = session.attach()
await _read_until(viewer, b"READY")
session.detach(viewer)
for _ in range(200):
if session.closed:
break
await asyncio.sleep(0.02)
assert session.closed
assert session.closed_reason == terminal_service.CLOSED_IDLE
async def test_a_watched_session_is_never_idle(shell_host, monkeypatch):
monkeypatch.setattr(terminal_service, "REAP_INTERVAL", 0.05)
session = await _open(shell_host, idle_timeout=0.1)
viewer = session.attach()
await _read_until(viewer, b"READY")
await asyncio.sleep(0.4)
assert not session.closed
assert session.idle_for == 0.0
await session.close()
async def test_closing_a_profile_ends_its_shells(shell_host):
first = await _open(shell_host, chat_id="chat-1")
second = await _open(shell_host, chat_id="chat-2")
for session in (first, second):
await _read_until(session.attach(), b"READY")
assert await terminal_service.close_for_profile("profile-1") == 2
assert first.closed and second.closed
assert first.closed_reason == terminal_service.CLOSED_REVOKED
async def test_closing_a_chat_ends_only_its_shell(shell_host):
first = await _open(shell_host, chat_id="chat-1")
second = await _open(shell_host, chat_id="chat-2")
assert await terminal_service.close_chat("chat-1")
assert first.closed
assert not second.closed
await second.close()
async def test_shutdown_closes_every_shell(shell_host):
first = await _open(shell_host, chat_id="chat-1")
second = await _open(shell_host, chat_id="chat-2")
await terminal_service.shutdown()
assert first.closed and second.closed
assert first.closed_reason == terminal_service.CLOSED_SHUTDOWN
assert terminal_service.count() == 0
# --- Caps -------------------------------------------------------------------------
async def test_the_instance_cap_refuses_a_new_shell(shell_host):
await _open(shell_host, chat_id="chat-1", max_sessions=1)
with pytest.raises(ExecError, match="as many terminals open as it allows"):
await _open(shell_host, chat_id="chat-2", max_sessions=1)
async def test_the_per_person_cap_counts_only_that_person(shell_host):
await _open(shell_host, chat_id="chat-1", owner="user-1", max_per_user=1)
with pytest.raises(ExecError, match="already have 1 terminal"):
await _open(shell_host, chat_id="chat-2", owner="user-1", max_per_user=1)
# Somebody else is unaffected.
other = await _open(shell_host, chat_id="chat-3", owner="user-2", max_per_user=1)
assert other is terminal_service.get("chat-3")
await terminal_service.shutdown()
async def test_a_closed_session_does_not_count_against_the_cap(shell_host):
session = await _open(shell_host, chat_id="chat-1", max_per_user=1)
await terminal_service.close_chat("chat-1")
assert session.closed
replacement = await _open(shell_host, chat_id="chat-2", max_per_user=1)
assert not replacement.closed
await replacement.close()
# --- Refusals ---------------------------------------------------------------------
async def test_an_unconfirmed_host_key_is_refused_before_anything_is_sent(shell_host):
spec = _spec(shell_host)
spec["host_key"] = ""
session = Session("chat-1", owner_id="user-1", profile_id="p", label="Test box")
with pytest.raises(ExecError, match="host key has not been confirmed"):
await session.start(spec)
async def test_a_different_host_key_is_refused(shell_host):
spec = _spec(shell_host)
spec["host_key"] = f"[127.0.0.1]:{shell_host['port']} ssh-ed25519 {'A' * 68}\n"
session = Session("chat-1", owner_id="user-1", profile_id="p", label="Test box")
with pytest.raises(ExecError):
await session.start(spec)