isagellm-core 0.3.0.5__tar.gz → 0.3.0.6__tar.gz

{isagellm_core-0.3.0.5/src/isagellm_core.egg-info → isagellm_core-0.3.0.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: isagellm-core
-Version: 0.3.0.5
+Version: 0.3.0.6
 Summary: sageLLM core runtime with PD separation (MVP)
 Author: IntelliStream Team
 License: Proprietary - IntelliStream
@@ -17,6 +17,8 @@ Requires-Dist: isagellm-protocol<0.4.0,>=0.3.0.2
 Requires-Dist: isagellm-backend<0.4.0,>=0.3.0.5
 Requires-Dist: isagellm-comm<0.4.0,>=0.3.0.1
 Requires-Dist: isagellm-kv-cache<0.4.0,>=0.3.0.1
+Requires-Dist: fastapi>=0.100.0
+Requires-Dist: uvicorn>=0.22.0
 Provides-Extra: dev
 Requires-Dist: pytest>=7.0.0; extra == "dev"
 Requires-Dist: pytest-cov>=4.0.0; extra == "dev"
@@ -79,6 +81,12 @@ sageLLM Core - 引擎协调层与运行时系统
 - ⚙️ 配置系统 - YAML/JSON + Pydantic v2
 - ✅ CPU-First - 无 GPU 测试支持
 
+## MoE（规划中）
+
+- MoE 路由/调度/执行图由 **sagellm-core** 主责
+- 依赖 **sagellm-comm**（all-to-all/拓扑通信）与 **sagellm-backend**（专家算子/内核）
+- 需先在 Protocol 中补充全局字段（如 router 元数据、专家负载指标）
+
 ## 安装
 
 ```bash

{isagellm_core-0.3.0.5 → isagellm_core-0.3.0.6}/README.md

@@ -50,6 +50,12 @@ sageLLM Core - 引擎协调层与运行时系统
 - ⚙️ 配置系统 - YAML/JSON + Pydantic v2
 - ✅ CPU-First - 无 GPU 测试支持
 
+## MoE（规划中）
+
+- MoE 路由/调度/执行图由 **sagellm-core** 主责
+- 依赖 **sagellm-comm**（all-to-all/拓扑通信）与 **sagellm-backend**（专家算子/内核）
+- 需先在 Protocol 中补充全局字段（如 router 元数据、专家负载指标）
+
 ## 安装
 
 ```bash

{isagellm_core-0.3.0.5 → isagellm_core-0.3.0.6}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "isagellm-core"
-version = "0.3.0.5"
+version = "0.3.0.6"
 description = "sageLLM core runtime with PD separation (MVP)"
 readme = "README.md"
 requires-python = "==3.11.*"
@@ -24,6 +24,8 @@ dependencies = [
   "isagellm-backend>=0.3.0.5,<0.4.0",
   "isagellm-comm>=0.3.0.1,<0.4.0",
   "isagellm-kv-cache>=0.3.0.1,<0.4.0",
+  "fastapi>=0.100.0",
+  "uvicorn>=0.22.0",
 ]
 
 [project.optional-dependencies]
@@ -52,6 +54,9 @@ sagellm_core = ["py.typed"]
 [tool.ruff]
 line-length = 100
 
+[project.scripts]
+sage-engine = "sagellm_core.engine_server:main"
+
 [project.entry-points."sagellm.engines"]
 cpu = "sagellm_core.engines.cpu:create_cpu_engine"
 hf-cuda = "sagellm_core.engines.hf_cuda:create_hf_cuda_engine"

{isagellm_core-0.3.0.5 → isagellm_core-0.3.0.6/src/isagellm_core.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: isagellm-core
-Version: 0.3.0.5
+Version: 0.3.0.6
 Summary: sageLLM core runtime with PD separation (MVP)
 Author: IntelliStream Team
 License: Proprietary - IntelliStream
@@ -17,6 +17,8 @@ Requires-Dist: isagellm-protocol<0.4.0,>=0.3.0.2
 Requires-Dist: isagellm-backend<0.4.0,>=0.3.0.5
 Requires-Dist: isagellm-comm<0.4.0,>=0.3.0.1
 Requires-Dist: isagellm-kv-cache<0.4.0,>=0.3.0.1
+Requires-Dist: fastapi>=0.100.0
+Requires-Dist: uvicorn>=0.22.0
 Provides-Extra: dev
 Requires-Dist: pytest>=7.0.0; extra == "dev"
 Requires-Dist: pytest-cov>=4.0.0; extra == "dev"
@@ -79,6 +81,12 @@ sageLLM Core - 引擎协调层与运行时系统
 - ⚙️ 配置系统 - YAML/JSON + Pydantic v2
 - ✅ CPU-First - 无 GPU 测试支持
 
+## MoE（规划中）
+
+- MoE 路由/调度/执行图由 **sagellm-core** 主责
+- 依赖 **sagellm-comm**（all-to-all/拓扑通信）与 **sagellm-backend**（专家算子/内核）
+- 需先在 Protocol 中补充全局字段（如 router 元数据、专家负载指标）
+
 ## 安装
 
 ```bash

{isagellm_core-0.3.0.5 → isagellm_core-0.3.0.6}/src/isagellm_core.egg-info/SOURCES.txt

@@ -15,6 +15,7 @@ src/sagellm_core/config.pyc
 src/sagellm_core/demo.pyc
 src/sagellm_core/engine.pyc
 src/sagellm_core/engine_factory.pyc
+src/sagellm_core/engine_server.pyc
 src/sagellm_core/factory.pyc
 src/sagellm_core/health.pyc
 src/sagellm_core/pd_executor.pyc
@@ -24,14 +25,19 @@ src/sagellm_core/runner.pyc
 src/sagellm_core/runtime.pyc
 src/sagellm_core/workload.pyc
 src/sagellm_core/__pycache__/__init__.cpython-311.pyc
+src/sagellm_core/__pycache__/base_engine.cpython-311.pyc
 src/sagellm_core/__pycache__/config.cpython-311.pyc
 src/sagellm_core/__pycache__/demo.cpython-311.pyc
 src/sagellm_core/__pycache__/engine.cpython-311.pyc
 src/sagellm_core/__pycache__/engine_factory.cpython-311.pyc
+src/sagellm_core/__pycache__/engine_server.cpython-311.pyc
 src/sagellm_core/__pycache__/factory.cpython-311.pyc
 src/sagellm_core/__pycache__/health.cpython-311.pyc
+src/sagellm_core/__pycache__/mock_engine.cpython-311.pyc
+src/sagellm_core/__pycache__/pd_executor.cpython-311.pyc
 src/sagellm_core/__pycache__/plugins.cpython-311.pyc
 src/sagellm_core/__pycache__/runner.cpython-311.pyc
+src/sagellm_core/__pycache__/runtime.cpython-311.pyc
 src/sagellm_core/__pycache__/workload.cpython-311.pyc
 src/sagellm_core/engines/__init__.py
 src/sagellm_core/engines/__init__.pyc
@@ -42,9 +48,11 @@ src/sagellm_core/engines/hf_cuda.pyc
 src/sagellm_core/engines/pytorch.pyc
 src/sagellm_core/engines/pytorch_engine.pyc
 src/sagellm_core/engines/__pycache__/__init__.cpython-311.pyc
+src/sagellm_core/engines/__pycache__/ascend.cpython-311.pyc
 src/sagellm_core/engines/__pycache__/cpu.cpython-311.pyc
 src/sagellm_core/engines/__pycache__/embedding.cpython-311.pyc
 src/sagellm_core/engines/__pycache__/hf_cuda.cpython-311.pyc
+src/sagellm_core/engines/__pycache__/mock.cpython-311.pyc
 src/sagellm_core/engines/__pycache__/pytorch.cpython-311.pyc
 src/sagellm_core/engines/__pycache__/pytorch_engine.cpython-311.pyc
 tests/test_ci_smoke.py
@@ -56,6 +64,7 @@ tests/test_engine.py
 tests/test_engine_behavior_parity.py
 tests/test_engine_contract.py
 tests/test_engine_contract_simplified.py
+tests/test_engine_server.py
 tests/test_pd_separation.py
 tests/test_pytorch_engine.py
 tests/test_streaming_pd.py

{isagellm_core-0.3.0.5 → isagellm_core-0.3.0.6}/src/isagellm_core.egg-info/entry_points.txt

@@ -1,3 +1,6 @@
+[console_scripts]
+sage-engine = sagellm_core.engine_server:main
+
 [sagellm.engines]
 ascend = sagellm_core.engines.ascend:create_ascend_engine
 cpu = sagellm_core.engines.cpu:create_cpu_engine

{isagellm_core-0.3.0.5 → isagellm_core-0.3.0.6}/src/isagellm_core.egg-info/requires.txt

@@ -4,6 +4,8 @@ isagellm-protocol<0.4.0,>=0.3.0.2
 isagellm-backend<0.4.0,>=0.3.0.5
 isagellm-comm<0.4.0,>=0.3.0.1
 isagellm-kv-cache<0.4.0,>=0.3.0.1
+fastapi>=0.100.0
+uvicorn>=0.22.0
 
 [dev]
 pytest>=7.0.0

{isagellm_core-0.3.0.5 → isagellm_core-0.3.0.6}/src/sagellm_core/__init__.py

@@ -12,7 +12,7 @@
 
 from __future__ import annotations
 
-__version__ = "0.3.0.5"
+__version__ = "0.3.0.6"
 
 from sagellm_core.engine import BaseEngine, EngineInstanceConfig
 from sagellm_core.config import (
@@ -45,6 +45,10 @@ from sagellm_core.runner import DemoRunner, RunnerContext
 from sagellm_core.runtime import DistributedConfig, DistributedRuntime, RuntimeState
 from sagellm_core.pd_executor import PDExecutionContext, PDSeparatedExecutor
 
+# Engine HTTP Server
+from sagellm_core.engine_server import app as engine_server_app
+from sagellm_core.engine_server import main as serve_engine
+
 # PyTorch engine (optional, loaded lazily)
 PyTorchEngine = None
 PyTorchEngineConfig = None
@@ -88,7 +92,7 @@ try:
 except Exception:
     pass  # torch or transformers not available
 
-__version__ = "0.3.0.5"
+__version__ = "0.3.0.6"
 
 __all__ = [
     # Version
@@ -136,4 +140,7 @@ __all__ = [
     "RuntimeState",
     "PDExecutionContext",
     "PDSeparatedExecutor",
+    # Engine HTTP Server
+    "engine_server_app",
+    "serve_engine",
 ]

isagellm_core-0.3.0.6/tests/test_engine_server.py

@@ -0,0 +1,154 @@
+"""Engine Server 单元测试.
+
+测试 Engine HTTP Server 的端点和功能。
+
+注意：需要真实引擎环境的测试已标记为 @pytest.mark.integration，
+默认跳过，可通过 pytest -m integration 运行。
+"""
+
+from __future__ import annotations
+
+import pytest
+
+from sagellm_protocol.types import Request
+
+
+# ─────────────────────────────────────────────────────────────────
+# CompletionRequest Model Tests (不需要引擎)
+# ─────────────────────────────────────────────────────────────────
+
+
+def test_completion_request_to_protocol():
+    """测试 CompletionRequest 转换为 Protocol Request."""
+    from sagellm_core.engine_server import CompletionRequest
+
+    req = CompletionRequest(
+        request_id="req-1",
+        trace_id="trace-1",
+        model="test-model",
+        prompt="Hello",
+        max_tokens=100,
+        stream=False,
+        temperature=0.7,
+        top_p=0.9,
+    )
+
+    protocol_req = req.to_protocol_request()
+
+    assert protocol_req.request_id == "req-1"
+    assert protocol_req.trace_id == "trace-1"
+    assert protocol_req.model == "test-model"
+    assert protocol_req.prompt == "Hello"
+    assert protocol_req.max_tokens == 100
+    assert protocol_req.stream is False
+    assert protocol_req.temperature == 0.7
+    assert protocol_req.top_p == 0.9
+
+
+def test_completion_request_minimal():
+    """测试 CompletionRequest 最小必需字段."""
+    from sagellm_core.engine_server import CompletionRequest
+
+    req = CompletionRequest(
+        request_id="req-1",
+        trace_id="trace-1",
+        model="test-model",
+        prompt="Hello",
+        max_tokens=50,
+    )
+
+    assert req.stream is False  # 默认值
+    assert req.temperature is None
+    assert req.top_p is None
+
+
+def test_completion_request_with_metadata():
+    """测试 CompletionRequest 带 metadata."""
+    from sagellm_core.engine_server import CompletionRequest
+
+    req = CompletionRequest(
+        request_id="req-1",
+        trace_id="trace-1",
+        model="test-model",
+        prompt="Hello",
+        max_tokens=50,
+        metadata={"user_id": "user-123", "session_id": "sess-456"},
+    )
+
+    protocol_req = req.to_protocol_request()
+    assert protocol_req.metadata == {"user_id": "user-123", "session_id": "sess-456"}
+
+
+def test_completion_request_validation():
+    """测试 CompletionRequest 字段验证."""
+    from sagellm_core.engine_server import CompletionRequest
+    from pydantic import ValidationError
+
+    # max_tokens 必须 > 0
+    with pytest.raises(ValidationError):
+        CompletionRequest(
+            request_id="req-1",
+            trace_id="trace-1",
+            model="test-model",
+            prompt="Hello",
+            max_tokens=0,  # Invalid
+        )
+
+    # temperature 必须在 (0, 2] 范围内
+    with pytest.raises(ValidationError):
+        CompletionRequest(
+            request_id="req-1",
+            trace_id="trace-1",
+            model="test-model",
+            prompt="Hello",
+            max_tokens=50,
+            temperature=3.0,  # Invalid, > 2
+        )
+
+
+# ─────────────────────────────────────────────────────────────────
+# Module Import Tests
+# ─────────────────────────────────────────────────────────────────
+
+
+def test_engine_server_imports():
+    """测试 engine_server 模块可以正常导入."""
+    from sagellm_core.engine_server import (
+        app,
+        main,
+        get_engine,
+        set_engine,
+        CompletionRequest,
+    )
+
+    assert app is not None
+    assert callable(main)
+    assert callable(get_engine)
+    assert callable(set_engine)
+
+
+def test_engine_server_app_routes():
+    """测试 FastAPI app 包含所需路由."""
+    from sagellm_core.engine_server import app
+
+    routes = [route.path for route in app.routes]
+
+    assert "/" in routes
+    assert "/health" in routes
+    assert "/info" in routes
+    assert "/v1/completions" in routes
+    assert "/v1/completions/stream" in routes
+
+
+# ─────────────────────────────────────────────────────────────────
+# CLI Tests
+# ─────────────────────────────────────────────────────────────────
+
+
+def test_cli_argparse():
+    """测试 CLI 参数解析."""
+    import argparse
+    from sagellm_core.engine_server import main
+
+    # 验证 main 是可调用的
+    assert callable(main)