mirror of
https://github.com/NousResearch/hermes-agent.git
synced 2026-06-27 11:22:03 +00:00
43b8ba4181
After a prolonged outage the in-process network-error ladder escalates to fatal and GatewayRunner._platform_reconnect_watcher rebuilds a fresh adapter that reconnects through the bootstrap path. That path called start_polling(drop_pending_updates=True), discarding every update Telegram queued during the outage — all messages sent while the bot was down were silently lost. The in-process ladder and 409-conflict handler already passed drop_pending_updates=False; only bootstrap did not distinguish a cold first boot from a reconnect. Thread an is_reconnect signal from the watcher through _connect_adapter_with_timeout into adapter.connect(). The base BasePlatformAdapter.connect() gains a keyword-only is_reconnect=False so every adapter inherits a tolerant signature (no per-platform breakage when the runner forwards the kwarg). Telegram translates is_reconnect into drop_pending_updates=not is_reconnect on both the polling and webhook bootstrap calls. Cold boot still drops the stale queue; a watcher reconnect preserves it. Fixes #46621. Co-authored-by: annguyenNous <annguyen@nousresearch.com> Co-authored-by: kyssta-exe <kyssta-exe@users.noreply.github.com> Co-authored-by: Kewe63 <Kewe63@users.noreply.github.com>
99 lines
4 KiB
Python
99 lines
4 KiB
Python
"""Phase 0 regression harness for the relay/connector work.
|
|
|
|
Locks the *behavioral contract* that the future ``RelayAdapter`` must reproduce:
|
|
the gateway's ``stream_consumer`` and ``BasePlatformAdapter`` read per-platform
|
|
capabilities through a small, stable surface. A relay adapter that exposes the
|
|
same surface (``MAX_MESSAGE_LENGTH`` attribute, ``message_len_fn`` property,
|
|
``supports_draft_streaming`` probe, and only the abstract methods) slots into
|
|
the existing consumer with no consumer changes.
|
|
|
|
These are deliberately *behavioral* (construct an adapter, drive the code,
|
|
assert the observable outcome) rather than source-string snapshots, per the
|
|
repo's "don't write change-detector tests" rule. They pass on ``main`` before
|
|
any ``RelayAdapter`` exists — they describe the contract, not the relay.
|
|
"""
|
|
|
|
import inspect
|
|
|
|
from gateway.config import Platform, PlatformConfig
|
|
from gateway.platforms.base import BasePlatformAdapter
|
|
|
|
|
|
class _MinAdapter(BasePlatformAdapter):
|
|
"""Smallest concrete adapter: implements exactly the abstract methods."""
|
|
|
|
async def connect(self, *, is_reconnect: bool = False): # pragma: no cover - not called
|
|
return True
|
|
|
|
async def disconnect(self): # pragma: no cover - not called
|
|
return None
|
|
|
|
async def send(self, *args, **kwargs): # pragma: no cover - not called
|
|
return None
|
|
|
|
async def get_chat_info(self, chat_id): # pragma: no cover - not called
|
|
return {}
|
|
|
|
|
|
def _make() -> BasePlatformAdapter:
|
|
return _MinAdapter(PlatformConfig(), Platform.LOCAL)
|
|
|
|
|
|
def test_abstract_methods_are_the_known_set():
|
|
"""The relay adapter must implement exactly this set of abstract methods.
|
|
|
|
Everything else on BasePlatformAdapter has a default, so ONE generic
|
|
RelayAdapter overriding the right subset is feasible without per-platform
|
|
gateway classes. If a new abstractmethod is added here, the relay design
|
|
(and the cross-repo contract) must be revisited — hence the lock.
|
|
|
|
NOTE: this is four methods, not three — ``get_chat_info`` is abstract too
|
|
(defined far below the connect/disconnect/send cluster in base.py). The
|
|
RelayAdapter must implement it (proxying a chat-info lookup to the
|
|
connector, or returning a descriptor-derived stub).
|
|
"""
|
|
abstract = {
|
|
name
|
|
for name, member in inspect.getmembers(BasePlatformAdapter)
|
|
if getattr(member, "__isabstractmethod__", False)
|
|
}
|
|
assert abstract == {"connect", "disconnect", "send", "get_chat_info"}
|
|
|
|
|
|
def test_message_len_fn_defaults_to_len():
|
|
"""message_len_fn is the per-platform length-unit hook (Telegram overrides
|
|
it for UTF-16). The default is plain ``len``; the relay adapter will
|
|
override it from its negotiated descriptor's ``len_unit``."""
|
|
inst = _make()
|
|
assert inst.message_len_fn("hello") == 5
|
|
|
|
|
|
def test_supports_draft_streaming_defaults_false():
|
|
"""Draft streaming is opt-in per platform; the consumer falls back to the
|
|
edit-based path when False. The relay adapter flips this from its
|
|
descriptor's ``supports_draft_streaming`` flag."""
|
|
inst = _make()
|
|
assert inst.supports_draft_streaming() is False
|
|
|
|
|
|
def test_stream_consumer_reads_max_message_length_by_attribute():
|
|
"""The consumer resolves the per-platform char limit by reading the
|
|
adapter's ``MAX_MESSAGE_LENGTH`` attribute (defaulting to 4096 when
|
|
absent). The relay adapter exposes this as an attribute set from its
|
|
descriptor — so a relay adapter that sets the attribute is chunked
|
|
correctly with no consumer change.
|
|
"""
|
|
from gateway import stream_consumer
|
|
|
|
class _NoLimit:
|
|
pass
|
|
|
|
class _WithLimit:
|
|
MAX_MESSAGE_LENGTH = 1234
|
|
|
|
assert getattr(_NoLimit(), "MAX_MESSAGE_LENGTH", 4096) == 4096
|
|
assert getattr(_WithLimit(), "MAX_MESSAGE_LENGTH", 4096) == 1234
|
|
# The consumer depends on BasePlatformAdapter for the message_len_fn
|
|
# isinstance guard (import-level contract the relay adapter satisfies by
|
|
# subclassing BasePlatformAdapter).
|
|
assert stream_consumer._BasePlatformAdapter is BasePlatformAdapter
|