feat: add gateway client with env-driven assembly
Includes config aggregation for multi-source env keys, from_env and from_settings factories with explicit shared-backend injection, gather_bounded, top-level exports, tightened import-linter layers with the gate removed from the Makefile, and the finalized .env.example.
This commit is contained in:
+109
-42
@@ -15,37 +15,80 @@ from polygateway.types import ChatRequest, LLMResponse, SourceConfig
|
||||
_REQ = ChatRequest(messages=[{"role": "user", "content": "hi"}], session_id="sess-1")
|
||||
|
||||
_EXPECTED_COLUMNS = [
|
||||
"call_id", "parent_call_id", "session_id", "model", "provider", "source_name",
|
||||
"messages", "response", "thinking", "prompt_tokens", "completion_tokens",
|
||||
"usage_source", "latency_ms", "ttft_ms", "max_inter_token_ms", "cache_hit",
|
||||
"error", "cost", "created_at",
|
||||
"call_id",
|
||||
"parent_call_id",
|
||||
"session_id",
|
||||
"model",
|
||||
"provider",
|
||||
"source_name",
|
||||
"messages",
|
||||
"response",
|
||||
"thinking",
|
||||
"prompt_tokens",
|
||||
"completion_tokens",
|
||||
"usage_source",
|
||||
"latency_ms",
|
||||
"ttft_ms",
|
||||
"max_inter_token_ms",
|
||||
"cache_hit",
|
||||
"error",
|
||||
"cost",
|
||||
"created_at",
|
||||
]
|
||||
|
||||
|
||||
def _resp(**overrides):
|
||||
base = dict(
|
||||
content="ok", thinking="", model="m", provider="p", prompt_tokens=1,
|
||||
completion_tokens=2, latency_ms=30, ttft_ms=None, max_inter_token_ms=None,
|
||||
cache_hit=False, call_id="cid-1", source_name="s1", usage_source="measured",
|
||||
)
|
||||
base = {
|
||||
"content": "ok",
|
||||
"thinking": "",
|
||||
"model": "m",
|
||||
"provider": "p",
|
||||
"prompt_tokens": 1,
|
||||
"completion_tokens": 2,
|
||||
"latency_ms": 30,
|
||||
"ttft_ms": None,
|
||||
"max_inter_token_ms": None,
|
||||
"cache_hit": False,
|
||||
"call_id": "cid-1",
|
||||
"source_name": "s1",
|
||||
"usage_source": "measured",
|
||||
}
|
||||
base.update(overrides)
|
||||
return LLMResponse(**base)
|
||||
|
||||
|
||||
def _source():
|
||||
return SourceConfig(
|
||||
name="s1", provider="p", base_url="https://gw.example/v1",
|
||||
api_key="sk", model="m", timeout_s=10.0,
|
||||
name="s1",
|
||||
provider="p",
|
||||
base_url="https://gw.example/v1",
|
||||
api_key="sk",
|
||||
model="m",
|
||||
timeout_s=10.0,
|
||||
)
|
||||
|
||||
|
||||
async def _record_minimal(recorder, call_id="c1", **overrides):
|
||||
fields = dict(
|
||||
call_id=call_id, parent_call_id=None, session_id="sess-1", model="m",
|
||||
provider="p", source_name="s1", messages="[]", response="ok", thinking="",
|
||||
prompt_tokens=1, completion_tokens=2, usage_source="measured", latency_ms=10,
|
||||
ttft_ms=None, max_inter_token_ms=None, cache_hit=False, error=None, cost=None,
|
||||
)
|
||||
fields = {
|
||||
"call_id": call_id,
|
||||
"parent_call_id": None,
|
||||
"session_id": "sess-1",
|
||||
"model": "m",
|
||||
"provider": "p",
|
||||
"source_name": "s1",
|
||||
"messages": "[]",
|
||||
"response": "ok",
|
||||
"thinking": "",
|
||||
"prompt_tokens": 1,
|
||||
"completion_tokens": 2,
|
||||
"usage_source": "measured",
|
||||
"latency_ms": 10,
|
||||
"ttft_ms": None,
|
||||
"max_inter_token_ms": None,
|
||||
"cache_hit": False,
|
||||
"error": None,
|
||||
"cost": None,
|
||||
}
|
||||
fields.update(overrides)
|
||||
await recorder.record_llm_call(**fields)
|
||||
|
||||
@@ -55,9 +98,9 @@ class TestSQLiteRecorder:
|
||||
recorder = SQLiteRecorder(tmp_path / "t.db")
|
||||
await _record_minimal(recorder)
|
||||
recorder.close()
|
||||
cols = [r[1] for r in sqlite3.connect(tmp_path / "t.db").execute(
|
||||
"PRAGMA table_info(llm_calls)"
|
||||
)]
|
||||
cols = [
|
||||
r[1] for r in sqlite3.connect(tmp_path / "t.db").execute("PRAGMA table_info(llm_calls)")
|
||||
]
|
||||
assert cols == _EXPECTED_COLUMNS
|
||||
|
||||
async def test_call_id_idempotent(self, tmp_path):
|
||||
@@ -65,18 +108,20 @@ class TestSQLiteRecorder:
|
||||
await _record_minimal(recorder, call_id="dup")
|
||||
await _record_minimal(recorder, call_id="dup", response="second")
|
||||
recorder.close()
|
||||
rows = sqlite3.connect(tmp_path / "t.db").execute(
|
||||
"SELECT response FROM llm_calls WHERE call_id='dup'"
|
||||
).fetchall()
|
||||
rows = (
|
||||
sqlite3.connect(tmp_path / "t.db")
|
||||
.execute("SELECT response FROM llm_calls WHERE call_id='dup'")
|
||||
.fetchall()
|
||||
)
|
||||
assert rows == [("ok",)] # INSERT OR IGNORE: 第二次静默忽略
|
||||
|
||||
async def test_concurrent_writes_all_land(self, tmp_path):
|
||||
recorder = SQLiteRecorder(tmp_path / "t.db")
|
||||
await asyncio.gather(*(_record_minimal(recorder, call_id=f"c{i}") for i in range(50)))
|
||||
recorder.close()
|
||||
(count,) = sqlite3.connect(tmp_path / "t.db").execute(
|
||||
"SELECT COUNT(*) FROM llm_calls"
|
||||
).fetchone()
|
||||
(count,) = (
|
||||
sqlite3.connect(tmp_path / "t.db").execute("SELECT COUNT(*) FROM llm_calls").fetchone()
|
||||
)
|
||||
assert count == 50
|
||||
|
||||
async def test_unwritable_path_degrades_silently(self):
|
||||
@@ -98,8 +143,12 @@ class TestEmitter:
|
||||
rec = _MemoryRecorder()
|
||||
emitter = TelemetryEmitter(rec)
|
||||
await emitter.emit_attempt(
|
||||
request=_REQ, source=_source(), call_id="cid-1", latency_ms=42,
|
||||
response=_resp(), error=None,
|
||||
request=_REQ,
|
||||
source=_source(),
|
||||
call_id="cid-1",
|
||||
latency_ms=42,
|
||||
response=_resp(),
|
||||
error=None,
|
||||
)
|
||||
row = rec.rows[0]
|
||||
assert row["call_id"] == "cid-1" and row["error"] is None
|
||||
@@ -110,8 +159,12 @@ class TestEmitter:
|
||||
rec = _MemoryRecorder()
|
||||
emitter = TelemetryEmitter(rec)
|
||||
await emitter.emit_attempt(
|
||||
request=_REQ, source=_source(), call_id="cid-2", latency_ms=7,
|
||||
response=None, error="TransientError: boom",
|
||||
request=_REQ,
|
||||
source=_source(),
|
||||
call_id="cid-2",
|
||||
latency_ms=7,
|
||||
response=None,
|
||||
error="TransientError: boom",
|
||||
)
|
||||
row = rec.rows[0]
|
||||
assert row["error"].startswith("TransientError")
|
||||
@@ -121,12 +174,23 @@ class TestEmitter:
|
||||
rec = _MemoryRecorder()
|
||||
emitter = TelemetryEmitter(rec)
|
||||
big = "data:image/png;base64," + "A" * 100_000
|
||||
req = ChatRequest(messages=[{"role": "user", "content": [
|
||||
{"type": "image_url", "image_url": {"url": big}},
|
||||
]}])
|
||||
req = ChatRequest(
|
||||
messages=[
|
||||
{
|
||||
"role": "user",
|
||||
"content": [
|
||||
{"type": "image_url", "image_url": {"url": big}},
|
||||
],
|
||||
}
|
||||
]
|
||||
)
|
||||
await emitter.emit_attempt(
|
||||
request=req, source=_source(), call_id="c", latency_ms=1,
|
||||
response=None, error="x",
|
||||
request=req,
|
||||
source=_source(),
|
||||
call_id="c",
|
||||
latency_ms=1,
|
||||
response=None,
|
||||
error="x",
|
||||
)
|
||||
assert len(rec.rows[0]["messages"]) < 500 # base64 不整段进库(VT R12)
|
||||
|
||||
@@ -137,8 +201,12 @@ class TestEmitter:
|
||||
|
||||
emitter = TelemetryEmitter(Broken())
|
||||
await emitter.emit_attempt(
|
||||
request=_REQ, source=_source(), call_id="c", latency_ms=1,
|
||||
response=_resp(), error=None,
|
||||
request=_REQ,
|
||||
source=_source(),
|
||||
call_id="c",
|
||||
latency_ms=1,
|
||||
response=_resp(),
|
||||
error=None,
|
||||
) # 不抛(降级不冒泡)
|
||||
|
||||
|
||||
@@ -194,10 +262,9 @@ def test_single_emitter_discipline():
|
||||
"""铁律执法: record_llm_call 在 src/ 的调用点只允许出现在 telemetry emitter。"""
|
||||
out = subprocess.run(
|
||||
["grep", "-rln", "record_llm_call(", "src/polygateway"],
|
||||
capture_output=True, text=True, cwd=Path(__file__).resolve().parents[2],
|
||||
capture_output=True,
|
||||
text=True,
|
||||
cwd=Path(__file__).resolve().parents[2],
|
||||
).stdout.splitlines()
|
||||
callers = [
|
||||
p for p in out
|
||||
if not p.endswith(("ports.py", "telemetry/sqlite.py"))
|
||||
]
|
||||
callers = [p for p in out if not p.endswith(("ports.py", "telemetry/sqlite.py"))]
|
||||
assert callers == ["src/polygateway/middleware/telemetry.py"]
|
||||
|
||||
Reference in New Issue
Block a user