mirror of
https://github.com/NousResearch/hermes-agent.git
synced 2026-07-22 16:25:58 +00:00
27b31bb7ff
Follow-up to the truncate_message split-loop floor. Two review points:
- CapabilityDescriptor.from_json trusted the wire max_message_length
verbatim, so a connector advertising 0 ('no limit') — or a buggy one
sending 0/negative — produced a descriptor whose bound flowed straight
into the adapter's MAX_MESSAGE_LENGTH and truncate_message. Normalize
it to the documented 4096 default (mirrors from_platform_entry's
'or 4096' and docs/relay-connector-contract.md), fixing the degenerate
budget at its source rather than only surviving it downstream.
- Document the truncate_message length contract for a budget too small
for one codepoint (max_length=1 with a 2-unit surrogate pair under
utf16_len): the chunk intentionally exceeds max_length by that one
indivisible codepoint, because emitting it whole preserves content
where the alternatives are data loss or an infinite loop.
Tests: from_json normalizes 0 and negative bounds to 4096 and passes a
real positive bound through unchanged; the sub-codepoint budget emits
whole codepoints with no data loss (all emojis preserved) and a chunk
that necessarily exceeds the 1-unit budget.
98 lines
3.6 KiB
Python
98 lines
3.6 KiB
Python
"""Tests for the experimental CapabilityDescriptor (relay Phase 0, Task 0.2)."""
|
|
|
|
from gateway.relay.descriptor import CONTRACT_VERSION, CapabilityDescriptor
|
|
|
|
|
|
def _telegram_descriptor(**overrides) -> CapabilityDescriptor:
|
|
base = dict(
|
|
contract_version=CONTRACT_VERSION,
|
|
platform="telegram",
|
|
label="Telegram",
|
|
max_message_length=4096,
|
|
supports_draft_streaming=False,
|
|
supports_edit=True,
|
|
supports_threads=True,
|
|
markdown_dialect="markdown_v2",
|
|
len_unit="utf16",
|
|
emoji="\u2708\ufe0f",
|
|
platform_hint="You are on Telegram.",
|
|
pii_safe=False,
|
|
)
|
|
base.update(overrides)
|
|
return CapabilityDescriptor(**base)
|
|
|
|
|
|
def test_descriptor_roundtrips_json():
|
|
d = _telegram_descriptor()
|
|
assert CapabilityDescriptor.from_json(d.to_json()) == d
|
|
|
|
|
|
def test_descriptor_is_frozen():
|
|
d = _telegram_descriptor()
|
|
try:
|
|
d.max_message_length = 1 # type: ignore[misc]
|
|
except Exception as exc: # FrozenInstanceError
|
|
assert "cannot assign" in str(exc) or "frozen" in str(exc).lower()
|
|
else: # pragma: no cover
|
|
raise AssertionError("descriptor should be immutable (frozen)")
|
|
|
|
|
|
def test_from_json_ignores_unknown_keys():
|
|
"""A newer connector may send fields this gateway doesn't know — those are
|
|
dropped, not fatal (forward-compat during the experimental phase)."""
|
|
d = _telegram_descriptor()
|
|
raw = d.to_json()[:-1] + ', "future_field": "ignored"}'
|
|
restored = CapabilityDescriptor.from_json(raw)
|
|
assert restored == d
|
|
|
|
|
|
def test_from_json_normalizes_zero_max_message_length_to_default():
|
|
"""A connector may advertise max_message_length 0 ("no limit"). from_json
|
|
must normalize it to the documented 4096 default so the receiver never
|
|
carries a degenerate chunking bound into truncate_message()."""
|
|
raw = (
|
|
'{"contract_version": 1, "platform": "x", "label": "X", '
|
|
'"max_message_length": 0, "supports_draft_streaming": false, '
|
|
'"supports_edit": false, "supports_threads": false, '
|
|
'"markdown_dialect": "plain", "len_unit": "chars"}'
|
|
)
|
|
d = CapabilityDescriptor.from_json(raw)
|
|
assert d.max_message_length == 4096
|
|
|
|
|
|
def test_from_json_normalizes_negative_max_message_length_to_default():
|
|
"""A buggy/hostile connector sending a negative bound is normalized too."""
|
|
raw = (
|
|
'{"contract_version": 1, "platform": "x", "label": "X", '
|
|
'"max_message_length": -5, "supports_draft_streaming": false, '
|
|
'"supports_edit": false, "supports_threads": false, '
|
|
'"markdown_dialect": "plain", "len_unit": "chars"}'
|
|
)
|
|
d = CapabilityDescriptor.from_json(raw)
|
|
assert d.max_message_length == 4096
|
|
|
|
|
|
def test_from_json_keeps_a_real_positive_bound():
|
|
"""A normal positive bound is passed through unchanged."""
|
|
d = CapabilityDescriptor.from_json(_telegram_descriptor(max_message_length=2000).to_json())
|
|
assert d.max_message_length == 2000
|
|
|
|
|
|
def test_from_json_fills_optional_defaults():
|
|
"""Optional fields (emoji/platform_hint/pii_safe) fall back to defaults."""
|
|
minimal = (
|
|
'{"contract_version": 1, "platform": "x", "label": "X", '
|
|
'"max_message_length": 2000, "supports_draft_streaming": false, '
|
|
'"supports_edit": false, "supports_threads": false, '
|
|
'"markdown_dialect": "plain", "len_unit": "chars"}'
|
|
)
|
|
d = CapabilityDescriptor.from_json(minimal)
|
|
assert d.pii_safe is False
|
|
assert d.platform_hint == ""
|
|
assert d.emoji == "\U0001f50c"
|
|
|
|
|
|
def test_module_is_marked_experimental():
|
|
import gateway.relay.descriptor as m
|
|
|
|
assert "EXPERIMENTAL" in (m.__doc__ or "")
|