Files
hermes-agent/tests/plugins/memory/test_supermemory_provider.py
Mahesh Sanikommu 1b76cfff83 fix(supermemory): keep pending turns across session switch until written
A failed flush at on_session_switch() restored the pending buffer and
then cleared it on the next line, so an unavailable service at the switch
boundary still lost every pending turn. Pending turns now carry their own
session_id; _write_turns() batches per session, so a later retry (next
turn, session end, shutdown) writes old-session turns under the old
session's custom_id even after the switch.

Addresses review on #109359.
2026-09-15 11:55:10 +05:30

497 lines
19 KiB
Python

import json
import os
import stat
import threading
import pytest
from plugins.memory.supermemory import (
SupermemoryMemoryProvider,
_capture_custom_id,
_clean_text_for_capture,
_format_connection_summary,
_format_prefetch_context,
_load_supermemory_config,
_probe_supermemory_connection,
_save_supermemory_config,
)
class FakeClient:
def __init__(self, api_key: str, timeout: float, container_tag: str, search_mode: str = "hybrid",
base_url: str = ""):
self.api_key = api_key
self.timeout = timeout
self.container_tag = container_tag
self.search_mode = search_mode
self.base_url = base_url
self.add_calls = []
self.search_results = []
self.profile_response = {"static": [], "dynamic": [], "search_results": []}
self.fail_add = False
self.forgotten_ids = []
self.forget_by_query_response = {"success": True, "message": "Forgot"}
def add_memory(self, content, metadata=None, *, entity_context="",
container_tag=None, custom_id=None):
if self.fail_add:
raise RuntimeError("boom")
self.add_calls.append({
"content": content,
"metadata": metadata,
"entity_context": entity_context,
"container_tag": container_tag,
"custom_id": custom_id,
})
return {"id": "mem_123"}
def search_memories(self, query, *, limit=5, container_tag=None, search_mode=None):
return self.search_results
def get_profile(self, query=None, *, container_tag=None):
return self.profile_response
def forget_memory(self, memory_id, *, container_tag=None):
self.forgotten_ids.append(memory_id)
def forget_by_query(self, query, *, container_tag=None):
return self.forget_by_query_response
@pytest.fixture
def provider(monkeypatch, tmp_path):
monkeypatch.setenv("SUPERMEMORY_API_KEY", "test-key")
monkeypatch.setattr("plugins.memory.supermemory._SupermemoryClient", FakeClient)
p = SupermemoryMemoryProvider()
p.initialize("session-1", hermes_home=str(tmp_path), platform="cli")
return p
def test_is_available_false_without_api_key(monkeypatch):
monkeypatch.delenv("SUPERMEMORY_API_KEY", raising=False)
p = SupermemoryMemoryProvider()
assert p.is_available() is False
def test_load_and_save_config_round_trip(tmp_path):
_save_supermemory_config({"container_tag": "demo-tag", "auto_capture": False}, str(tmp_path))
cfg = _load_supermemory_config(str(tmp_path))
# container_tag is kept raw — sanitization happens in initialize() after template resolution
assert cfg["container_tag"] == "demo-tag"
assert cfg["auto_capture"] is False
assert cfg["auto_recall"] is True
def test_clean_text_for_capture_strips_injected_context():
text = "hello\n<supermemory-context>ignore me</supermemory-context>\nworld"
assert _clean_text_for_capture(text) == "hello\nworld"
def test_clean_text_for_capture_strips_inline_data_uri():
text = "look: data:image/png;base64,iVBORw0KGgoAAAANSUhEUg== ok"
assert _clean_text_for_capture(text) == "look: [image] ok"
def test_format_prefetch_context_deduplicates_overlap():
result = _format_prefetch_context(
static_facts=["Jordan prefers short answers"],
dynamic_facts=["Jordan prefers short answers", "Uses Hermes"],
search_results=[{"memory": "Uses Hermes", "similarity": 0.9}],
max_results=10,
)
assert result.count("Jordan prefers short answers") == 1
assert result.count("Uses Hermes") == 1
assert "<supermemory-context>" in result
def test_prefetch_includes_profile_on_first_turn(provider):
provider._client.profile_response = {
"static": ["Jordan prefers short answers"],
"dynamic": ["Current project is Supermemory provider"],
"search_results": [{"memory": "Working on Hermes memory provider", "similarity": 0.88}],
}
provider.on_turn_start(1, "start")
result = provider.prefetch("what am I working on?")
assert "User Profile (Persistent)" in result
assert "Recent Context" in result
assert "Relevant Memories" in result
def test_capture_custom_id_buckets_by_four_hours():
from datetime import datetime, timezone
a = _capture_custom_id("session-1", datetime(2026, 9, 12, 3, 59, tzinfo=timezone.utc))
b = _capture_custom_id("session-1", datetime(2026, 9, 12, 4, 0, tzinfo=timezone.utc))
assert a == "session_1_2026-09-12_b0"
assert b == "session_1_2026-09-12_b1"
assert _capture_custom_id("", datetime(2026, 9, 12, 23, 0, tzinfo=timezone.utc)) == "hermes_2026-09-12_b5"
def test_sync_turn_writes_turn_to_session_document(provider):
# Every completed turn is appended to one document per session per 4h window.
provider.sync_turn("hello", "hi there", session_id="session-1")
assert len(provider._client.add_calls) == 1
call = provider._client.add_calls[0]
assert call["custom_id"] == _capture_custom_id("session-1")
assert call["content"] == "[role: user]\nhello\n[user:end]\n[role: assistant]\nhi there\n[assistant:end]"
assert call["metadata"]["type"] == "conversation"
assert call["metadata"]["session_id"] == "session-1"
assert call["entity_context"]
assert provider._pending_turns == []
def test_sync_turn_skips_empty_turn(provider):
provider.sync_turn("", "<supermemory-context>x</supermemory-context>", session_id="session-1")
assert provider._client.add_calls == []
def test_failed_turn_write_is_retried_at_session_end(provider):
provider._client.fail_add = True
provider.sync_turn("hello", "hi there", session_id="session-1")
assert provider._client.add_calls == []
assert provider._pending_turns == [{"user": "hello", "assistant": "hi there", "session_id": "session-1"}]
provider._client.fail_add = False
provider.on_session_end([])
assert len(provider._client.add_calls) == 1
call = provider._client.add_calls[0]
assert call["custom_id"] == _capture_custom_id("session-1")
assert "hello" in call["content"]
assert provider._pending_turns == []
def test_pending_turns_are_batched_with_next_turn(provider):
provider._client.fail_add = True
provider.sync_turn("one", "uno", session_id="session-1")
provider._client.fail_add = False
provider.sync_turn("two", "dos", session_id="session-1")
assert len(provider._client.add_calls) == 1
assert provider._client.add_calls[0]["content"].index("one") < provider._client.add_calls[0]["content"].index("two")
assert provider._pending_turns == []
def test_session_switch_flushes_pending_to_old_session(provider):
provider._client.fail_add = True
provider.sync_turn("hello", "hi", session_id="session-1")
provider._client.fail_add = False
provider.on_session_switch("session-2", reset=True)
assert provider._client.add_calls[0]["custom_id"] == _capture_custom_id("session-1")
assert provider._session_id == "session-2"
assert provider._pending_turns == []
def test_failed_switch_flush_keeps_old_session_turns_for_later_retry(provider):
provider._client.fail_add = True
provider.sync_turn("old turn", "old reply", session_id="session-1")
provider.on_session_switch("session-2", reset=True) # flush fails: service unavailable at the boundary
assert provider._session_id == "session-2"
assert provider._pending_turns == [{"user": "old turn", "assistant": "old reply", "session_id": "session-1"}]
provider._client.fail_add = False
provider.sync_turn("new turn", "new reply", session_id="session-2")
calls = provider._client.add_calls
assert [c["custom_id"] for c in calls] == [_capture_custom_id("session-1"), _capture_custom_id("session-2")]
assert calls[0]["metadata"]["session_id"] == "session-1" and "old turn" in calls[0]["content"]
assert calls[1]["metadata"]["session_id"] == "session-2" and "new turn" in calls[1]["content"]
assert provider._pending_turns == []
def test_failed_switch_flush_is_retried_at_shutdown(provider):
provider._client.fail_add = True
provider.sync_turn("old turn", "old reply", session_id="session-1")
provider.on_session_switch("session-2", reset=True)
provider._client.fail_add = False
provider.shutdown()
assert provider._client.add_calls[0]["custom_id"] == _capture_custom_id("session-1")
assert provider._pending_turns == []
def test_sync_turn_drops_inline_image_payloads(provider):
blob = "A" * 4096
provider.sync_turn(f"describe this data:image/png;base64,{blob}", "a screenshot", session_id="session-1")
call = provider._client.add_calls[0]
assert "describe this [image]" in call["content"]
assert blob not in json.dumps(call)
def test_merge_metadata_stamps_sm_source():
# sm_source routes Hermes writes into the "Hermes" Space in the Supermemory
# app (functional routing, not telemetry) — must always be present.
from plugins.memory.supermemory import _SupermemoryClient
client = _SupermemoryClient.__new__(_SupermemoryClient)
merged = client._merge_metadata({"type": "explicit_memory"})
assert merged["sm_source"] == "hermes"
assert merged["type"] == "explicit_memory"
# Legacy "source" is migrated into "type" when type is absent.
merged2 = client._merge_metadata({"source": "conversation_turn"})
assert merged2["sm_source"] == "hermes"
assert merged2["type"] == "conversation_turn"
assert "source" not in merged2
def test_shutdown_joins_threads_and_flushes_buffer(provider, monkeypatch):
started = threading.Event()
release = threading.Event()
def slow_add_memory(content, metadata=None, *, entity_context="",
container_tag=None, custom_id=None):
if provider._client.fail_add:
raise RuntimeError("boom")
started.set()
release.wait(timeout=1)
provider._client.add_calls.append({
"content": content,
"metadata": metadata,
"entity_context": entity_context,
"custom_id": custom_id,
})
return {"id": "mem_slow"}
monkeypatch.setattr(provider._client, "add_memory", slow_add_memory)
# A failed turn write stays pending; shutdown retries it.
provider._client.fail_add = True
provider.sync_turn(
"Please remember this request in long-term memory",
"Absolutely, I will keep that in long-term memory.",
session_id="session-1",
)
provider._client.fail_add = False
assert provider._sync_thread is None
assert len(provider._pending_turns) == 1
# on_memory_write still runs on a background thread.
provider.on_memory_write("add", "memory", "Jordan likes concise docs")
assert started.wait(timeout=1)
assert provider._write_thread is not None
release.set()
provider.shutdown()
# All tracked threads joined and cleared.
assert provider._sync_thread is None
assert provider._write_thread is None
assert provider._prefetch_thread is None
# Explicit memory write and the retried turn both went through.
assert len(provider._client.add_calls) == 2
flushed = next(c for c in provider._client.add_calls if c.get("custom_id") == _capture_custom_id("session-1"))
assert provider._pending_turns == []
def test_store_tool_returns_saved_payload(provider):
result = json.loads(provider.handle_tool_call("supermemory_store", {"content": "Jordan likes concise docs"}))
assert result["saved"] is True
assert result["id"] == "mem_123"
def test_search_tool_formats_results(provider):
provider._client.search_results = [
{"id": "m1", "memory": "Jordan likes concise docs", "similarity": 0.92}
]
result = json.loads(provider.handle_tool_call("supermemory_search", {"query": "concise docs"}))
assert result["count"] == 1
assert result["results"][0]["similarity"] == 92
def test_forget_tool_by_id(provider):
result = json.loads(provider.handle_tool_call("supermemory_forget", {"id": "m1"}))
assert result == {"forgotten": True, "id": "m1"}
assert provider._client.forgotten_ids == ["m1"]
def test_profile_tool_formats_sections(provider):
provider._client.profile_response = {
"static": ["Jordan prefers concise docs"],
"dynamic": ["Working on Supermemory provider"],
"search_results": [],
}
result = json.loads(provider.handle_tool_call("supermemory_profile", {}))
assert result["static_count"] == 1
assert result["dynamic_count"] == 1
assert "User Profile (Persistent)" in result["profile"]
def test_handle_tool_call_returns_error_when_unconfigured(monkeypatch):
monkeypatch.delenv("SUPERMEMORY_API_KEY", raising=False)
p = SupermemoryMemoryProvider()
result = json.loads(p.handle_tool_call("supermemory_search", {"query": "x"}))
assert "error" in result
# -- Identity template tests --------------------------------------------------
def test_identity_template_resolved_in_container_tag(monkeypatch, tmp_path):
"""container_tag with {identity} resolves to profile-scoped tag."""
monkeypatch.setenv("SUPERMEMORY_API_KEY", "test-key")
monkeypatch.setattr("plugins.memory.supermemory._SupermemoryClient", FakeClient)
_save_supermemory_config({"container_tag": "hermes-{identity}"}, str(tmp_path))
p = SupermemoryMemoryProvider()
p.initialize("s1", hermes_home=str(tmp_path), platform="cli", agent_identity="coder")
assert p._container_tag == "hermes_coder"
def test_container_tag_env_var_override(monkeypatch, tmp_path):
"""SUPERMEMORY_CONTAINER_TAG env var overrides config."""
monkeypatch.setenv("SUPERMEMORY_API_KEY", "test-key")
monkeypatch.setenv("SUPERMEMORY_CONTAINER_TAG", "env-override")
monkeypatch.setattr("plugins.memory.supermemory._SupermemoryClient", FakeClient)
p = SupermemoryMemoryProvider()
p.initialize("s1", hermes_home=str(tmp_path), platform="cli")
assert p._container_tag == "env_override"
# -- Search mode tests --------------------------------------------------------
def test_invalid_search_mode_falls_back_to_default(monkeypatch, tmp_path):
"""Invalid search_mode falls back to 'hybrid'."""
monkeypatch.setenv("SUPERMEMORY_API_KEY", "test-key")
monkeypatch.setattr("plugins.memory.supermemory._SupermemoryClient", FakeClient)
_save_supermemory_config({"search_mode": "invalid_mode"}, str(tmp_path))
p = SupermemoryMemoryProvider()
p.initialize("s1", hermes_home=str(tmp_path), platform="cli")
assert p._search_mode == "hybrid"
# -- Base URL tests -------------------------------------------------------------
def test_base_url_defaults_to_cloud(monkeypatch, tmp_path):
"""Without config or env override, the client targets api.supermemory.ai."""
monkeypatch.setenv("SUPERMEMORY_API_KEY", "test-key")
monkeypatch.delenv("SUPERMEMORY_BASE_URL", raising=False)
monkeypatch.setattr("plugins.memory.supermemory._SupermemoryClient", FakeClient)
p = SupermemoryMemoryProvider()
p.initialize("s1", hermes_home=str(tmp_path), platform="cli")
assert p._base_url == "https://api.supermemory.ai"
assert p._client.base_url == "https://api.supermemory.ai"
def test_client_passes_custom_base_url_to_sdk(monkeypatch):
"""SDK client receives the normalized base URL."""
import sys
import types
from plugins.memory.supermemory import _SupermemoryClient
captured = {}
class StubSupermemory:
def __init__(self, **kwargs):
captured.update(kwargs)
module = types.ModuleType("supermemory")
module.Supermemory = StubSupermemory
monkeypatch.setitem(sys.modules, "supermemory", module)
monkeypatch.setattr("tools.lazy_deps.ensure", lambda *args, **kwargs: None)
client = _SupermemoryClient(
api_key="test-key",
timeout=1.0,
container_tag="hermes",
base_url="http://localhost:6767/",
)
assert client._base_url == "http://localhost:6767"
assert captured["base_url"] == "http://localhost:6767"
# -- Multi-container tests ----------------------------------------------------
def test_multi_container_disabled_by_default(provider):
"""Multi-container is off by default; schemas have no container_tag param."""
assert provider._enable_custom_containers is False
schemas = provider.get_tool_schemas()
for s in schemas:
assert "container_tag" not in s["parameters"]["properties"]
def test_get_config_schema_minimal():
"""get_config_schema only returns the API key field."""
p = SupermemoryMemoryProvider()
schema = p.get_config_schema()
assert len(schema) == 1
assert schema[0]["key"] == "api_key"
assert schema[0]["secret"] is True
def test_probe_supermemory_connection_missing_key(tmp_path):
status = _probe_supermemory_connection("", str(tmp_path))
assert status["ok"] is False
assert status["error"] == "SUPERMEMORY_API_KEY not set"
assert status["container_tag"] == "hermes"
def _stub_supermemory_importable(monkeypatch):
"""Make ``__import__("supermemory")`` succeed without the real package.
``_probe_supermemory_connection`` guards on ``__import__("supermemory")``
before using the (mocked) client, so tests that mock ``_SupermemoryClient``
must also satisfy that import guard — otherwise they only pass in an
environment where the optional ``supermemory`` package happens to be
installed (and fail on a clean checkout / CI). Mirrors the inverse stub in
``test_is_available_false_when_import_missing``.
"""
import builtins
import types
real_import = builtins.__import__
def fake_import(name, *args, **kwargs):
if name == "supermemory":
return types.ModuleType("supermemory")
return real_import(name, *args, **kwargs)
monkeypatch.setattr(builtins, "__import__", fake_import)
def test_post_setup_writes_config_and_prints_summary(monkeypatch, tmp_path, capsys):
config: dict = {"memory": {}}
monkeypatch.setenv("SUPERMEMORY_API_KEY", "")
monkeypatch.setattr(
"hermes_cli.memory_setup._prompt",
lambda label, secret=True, default=None: "new-api-key",
)
monkeypatch.setattr(
"plugins.memory.supermemory._probe_supermemory_connection",
lambda api_key, hermes_home, **kwargs: {
"ok": True,
"container_tag": "hermes",
"profile_facts": 3,
"auto_recall": True,
"auto_capture": True,
},
)
saved: dict = {}
def fake_save_config(cfg):
saved.update(cfg)
monkeypatch.setattr("hermes_cli.config.save_config", fake_save_config)
SupermemoryMemoryProvider().post_setup(str(tmp_path), config)
assert config["memory"]["provider"] == "supermemory"
assert saved["memory"]["provider"] == "supermemory"
env_text = (tmp_path / ".env").read_text(encoding="utf-8")
assert "SUPERMEMORY_API_KEY=new-api-key" in env_text
out = capsys.readouterr().out
assert "✓ Connected" in out
assert "3 profile facts" in out
assert "Memory provider: supermemory" in out
@pytest.mark.skipif(os.name == "nt", reason="POSIX mode bits not enforced on Windows")
def test_save_config_sets_owner_only_permissions(tmp_path):
"""supermemory.json must be written with 0o600 so API key is not world-readable."""
_save_supermemory_config({"api_key": "sm-test-key"}, str(tmp_path))
config_file = tmp_path / "supermemory.json"
assert config_file.exists()
mode = stat.S_IMODE(config_file.stat().st_mode)
assert mode == 0o600, f"Expected 0o600 (owner-only), got {oct(mode)}"