orchestrator-core 4.6.0rc2__tar.gz → 4.6.0rc4__tar.gz

{orchestrator_core-4.6.0rc2 → orchestrator_core-4.6.0rc4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: orchestrator-core
-Version: 4.6.0rc2
+Version: 4.6.0rc4
 Summary: This is the orchestrator workflow engine.
 Author-email: SURF <automation-beheer@surf.nl>
 Requires-Python: >=3.11,<3.14
@@ -119,21 +119,26 @@ Configure the database URI in your local environment:
 export DATABASE_URI=postgresql://nwa:nwa@localhost:5432/orchestrator-core
 ```
 
-### Step 3 - Create main.py
+### Step 3 - Create main.py and wsgi.py
 
-Create a `main.py` file.
+Create a `main.py` file for running the CLI.
 
 ```python
-from orchestrator import OrchestratorCore
 from orchestrator.cli.main import app as core_cli
-from orchestrator.settings import AppSettings
-
-app = OrchestratorCore(base_settings=AppSettings())
 
 if __name__ == "__main__":
     core_cli()
 ```
 
+Create a `wsgi.py` file for running the web server.
+
+```python
+from orchestrator import OrchestratorCore
+from orchestrator.settings import AppSettings
+
+app = OrchestratorCore(base_settings=AppSettings())
+```
+
 ### Step 4 - Run the database migrations
 
 Initialize the migration environment and database tables.
@@ -147,7 +152,7 @@ python main.py db upgrade heads
 
 ```shell
 export OAUTH2_ACTIVE=False
-uvicorn --reload --host 127.0.0.1 --port 8080 main:app
+uvicorn --reload --host 127.0.0.1 --port 8080 wsgi:app
 ```
 
 Visit the [ReDoc](http://127.0.0.1:8080/api/redoc) or [OpenAPI](http://127.0.0.1:8080/api/docs) page to view and interact with the API.

{orchestrator_core-4.6.0rc2 → orchestrator_core-4.6.0rc4}/README.md

@@ -42,21 +42,26 @@ Configure the database URI in your local environment:
 export DATABASE_URI=postgresql://nwa:nwa@localhost:5432/orchestrator-core
 ```
 
-### Step 3 - Create main.py
+### Step 3 - Create main.py and wsgi.py
 
-Create a `main.py` file.
+Create a `main.py` file for running the CLI.
 
 ```python
-from orchestrator import OrchestratorCore
 from orchestrator.cli.main import app as core_cli
-from orchestrator.settings import AppSettings
-
-app = OrchestratorCore(base_settings=AppSettings())
 
 if __name__ == "__main__":
     core_cli()
 ```
 
+Create a `wsgi.py` file for running the web server.
+
+```python
+from orchestrator import OrchestratorCore
+from orchestrator.settings import AppSettings
+
+app = OrchestratorCore(base_settings=AppSettings())
+```
+
 ### Step 4 - Run the database migrations
 
 Initialize the migration environment and database tables.
@@ -70,7 +75,7 @@ python main.py db upgrade heads
 
 ```shell
 export OAUTH2_ACTIVE=False
-uvicorn --reload --host 127.0.0.1 --port 8080 main:app
+uvicorn --reload --host 127.0.0.1 --port 8080 wsgi:app
 ```
 
 Visit the [ReDoc](http://127.0.0.1:8080/api/redoc) or [OpenAPI](http://127.0.0.1:8080/api/docs) page to view and interact with the API.

{orchestrator_core-4.6.0rc2 → orchestrator_core-4.6.0rc4}/orchestrator/__init__.py

@@ -1,4 +1,4 @@
-# Copyright 2019-2025 SURF, GÉANT.
+# Copyright 2019-2025 SURF, GÉANT, ESnet.
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
 # You may obtain a copy of the License at
@@ -13,7 +13,7 @@
 
 """This is the orchestrator workflow engine."""
 
-__version__ = "4.6.0rc2"
+__version__ = "4.6.0rc4"
 
 
 from structlog import get_logger

{orchestrator_core-4.6.0rc2 → orchestrator_core-4.6.0rc4}/orchestrator/agentic_app.py

@@ -26,7 +26,7 @@ from orchestrator.cli.main import app as cli_app
 from orchestrator.llm_settings import LLMSettings, llm_settings
 
 if TYPE_CHECKING:
-    from pydantic_ai.models.openai import OpenAIModel
+    from pydantic_ai.models.openai import OpenAIChatModel
 
 logger = get_logger(__name__)
 
@@ -36,7 +36,7 @@ class LLMOrchestratorCore(OrchestratorCore):
         self,
         *args: Any,
         llm_settings: LLMSettings = llm_settings,
-        agent_model: "OpenAIModel | str | None" = None,
+        agent_model: "OpenAIChatModel | str | None" = None,
         **kwargs: Any,
     ) -> None:
         """Initialize the `LLMOrchestratorCore` class.

{orchestrator_core-4.6.0rc2 → orchestrator_core-4.6.0rc4}/orchestrator/api/api_v1/endpoints/agent.py

@@ -20,7 +20,6 @@ from pydantic_ai.agent import Agent
 from starlette.responses import Response
 from structlog import get_logger
 
-from orchestrator.llm_settings import llm_settings
 from orchestrator.search.agent import build_agent_instance
 from orchestrator.search.agent.state import SearchState
 
@@ -29,12 +28,12 @@ logger = get_logger(__name__)
 
 
 @cache
-def get_agent() -> Agent[StateDeps[SearchState], str]:
+def get_agent(request: Request) -> Agent[StateDeps[SearchState], str]:
     """Dependency to provide the agent instance.
 
     The agent is built once and cached for the lifetime of the application.
     """
-    return build_agent_instance(llm_settings.AGENT_MODEL)
+    return build_agent_instance(request.app.agent_model)
 
 
 @router.post("/")

{orchestrator_core-4.6.0rc2 → orchestrator_core-4.6.0rc4}/orchestrator/app.py

@@ -5,7 +5,7 @@ This module contains the main `OrchestratorCore` class for the `FastAPI` backend
 provides the ability to run the CLI.
 """
 
-# Copyright 2019-2020 SURF, ESnet, GÉANT.
+# Copyright 2019-2025 SURF, ESnet, GÉANT.
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
 # You may obtain a copy of the License at

{orchestrator_core-4.6.0rc2 → orchestrator_core-4.6.0rc4}/orchestrator/cli/main.py

@@ -1,4 +1,4 @@
-# Copyright 2019-2020 SURF.
+# Copyright 2019-2025 SURF, ESnet, GÉANT.
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
 # You may obtain a copy of the License at

{orchestrator_core-4.6.0rc2 → orchestrator_core-4.6.0rc4}/orchestrator/db/models.py

@@ -711,7 +711,7 @@ class SearchQueryTable(BaseModel):
     parameters = mapped_column(pg.JSONB, nullable=False)
 
     # Query embedding for semantic search (pgvector)
-    query_embedding = mapped_column(Vector(1536), nullable=True)
+    query_embedding = mapped_column(Vector(llm_settings.EMBEDDING_DIMENSION), nullable=True)
 
     executed_at = mapped_column(UtcTimestamp, server_default=text("current_timestamp()"), nullable=False)
 

{orchestrator_core-4.6.0rc2 → orchestrator_core-4.6.0rc4}/orchestrator/llm_settings.py

@@ -10,14 +10,31 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+from typing import Annotated
+
 from pydantic import Field, field_validator
 from pydantic_settings import BaseSettings
 from structlog import get_logger
 
 logger = get_logger(__name__)
 
+EMBEDDING_DIMENSION_MIN = 100
+EMBEDDING_DIMENSION_MAX = 2000
+EMBEDDING_DIMENSION_DEFAULT = 1536
+
+EMBEDDING_DIMENSION_FIELD = Annotated[
+    int,
+    Field(
+        ge=EMBEDDING_DIMENSION_MIN,
+        le=EMBEDDING_DIMENSION_MAX,
+        default=EMBEDDING_DIMENSION_DEFAULT,
+        description="Embedding dimension: when embeddings are generated at a higher resolution than this setting, the least significant numbers will be truncated",
+    ),
+]
+
 
 class LLMSettings(BaseSettings):
+
     # Feature flags for LLM functionality
     SEARCH_ENABLED: bool = False  # Enable search/indexing with embeddings
     AGENT_ENABLED: bool = False  # Enable agentic functionality
@@ -27,7 +44,7 @@ class LLMSettings(BaseSettings):
     AGENT_MODEL_VERSION: str = "2025-01-01-preview"
     OPENAI_API_KEY: str = ""  # Change per provider (Azure, etc).
     # Embedding settings
-    EMBEDDING_DIMENSION: int = 1536
+    EMBEDDING_DIMENSION: EMBEDDING_DIMENSION_FIELD = 1536
     EMBEDDING_MODEL: str = "openai/text-embedding-3-small"  # See litellm docs for supported models.
     EMBEDDING_SAFE_MARGIN_PERCENT: float = Field(
         0.1, description="Safety margin as a percentage (e.g., 0.1 for 10%) for token budgeting.", ge=0, le=1

{orchestrator_core-4.6.0rc2 → orchestrator_core-4.6.0rc4}/orchestrator/search/agent/agent.py

@@ -16,7 +16,7 @@ from typing import Any
 import structlog
 from pydantic_ai.ag_ui import StateDeps
 from pydantic_ai.agent import Agent
-from pydantic_ai.models.openai import OpenAIModel
+from pydantic_ai.models.openai import OpenAIChatModel
 from pydantic_ai.settings import ModelSettings
 from pydantic_ai.toolsets import FunctionToolset
 
@@ -28,12 +28,12 @@ logger = structlog.get_logger(__name__)
 
 
 def build_agent_instance(
-    model: str | OpenAIModel, agent_tools: list[FunctionToolset[Any]] | None = None
+    model: str | OpenAIChatModel, agent_tools: list[FunctionToolset[Any]] | None = None
 ) -> Agent[StateDeps[SearchState], str]:
     """Build and configure the search agent instance.
 
     Args:
-        model: The LLM model to use (string or OpenAIModel instance)
+        model: The LLM model to use (string or OpenAIChatModel instance)
         agent_tools: Optional list of additional toolsets to include
 
     Returns:

{orchestrator_core-4.6.0rc2 → orchestrator_core-4.6.0rc4}/orchestrator/search/core/embedding.py

@@ -42,7 +42,7 @@ class EmbeddingIndexer:
                 max_retries=llm_settings.LLM_MAX_RETRIES,
             )
             data = sorted(resp.data, key=lambda e: e["index"])
-            return [row["embedding"] for row in data]
+            return [row["embedding"][: llm_settings.EMBEDDING_DIMENSION] for row in data]
         except (llm_exc.APIError, llm_exc.APIConnectionError, llm_exc.RateLimitError, llm_exc.Timeout) as e:
             logger.error("Embedding request failed", error=str(e))
             return [[] for _ in texts]
@@ -67,7 +67,7 @@ class QueryEmbedder:
                 timeout=5.0,
                 max_retries=0,  # No retries, prioritize speed.
             )
-            return resp.data[0]["embedding"]
+            return resp.data[0]["embedding"][: llm_settings.EMBEDDING_DIMENSION]
         except Exception as e:
             logger.error("Async embedding generation failed", error=str(e))
             return []