Skip to content

OpenRAG is a comprehensive, single package Retrieval-Augmented Generation platform built on Langflow, Docling, and Opensearch.

Python

refactor: Split settings into package and refactor (#1570)

Edwin Jose4bd10c488deb
diff --git a/pyproject.toml b/pyproject.toml
index e82264c17..44ba46bcb 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -97,8 +97,25 @@ exclude = ["flows/components"]
select = ["E", "F", "I", "B", "UP"]
ignore = ["E501"] # line-length handled by formatter
+# FastAPI's idiomatic dependency-injection pattern uses `Depends(...)` and
+# our `require_permission(...)` as call expressions in default-argument
+# position. B008 flags these as bugs ("function call in argument default")
+# but for FastAPI they are by design — the framework introspects them at
+# request time. Whitelist the call sites instead of sprinkling noqa
+# comments across every handler.
[tool.ruff.lint.flake8-bugbear]
-extend-immutable-calls = ["fastapi.Depends", "fastapi.params.Depends", "Depends"]
+extend-immutable-calls = [
+ "fastapi.Depends",
+ "fastapi.Query",
+ "fastapi.Path",
+ "fastapi.Body",
+ "fastapi.Header",
+ "fastapi.Cookie",
+ "fastapi.File",
+ "fastapi.Form",
+ "fastapi.Security",
+ "dependencies.require_permission",
+]
# `bootstrap` must be the first import in any module that uses it
# (it loads .env and configures structlog before anything else can
@@ -125,4 +142,14 @@ no_strict_optional = true
warn_unused_ignores = true
exclude = ["flows/components"]
+# FastAPI handlers in this module declare a Pydantic response model as
+# their return type but legitimately return `JSONResponse(...)` on error
+# paths. The runtime accepts both, but the annotation can't be widened to
+# `SomeResponse | JSONResponse` because FastAPI then tries to derive an
+# OpenAPI schema from the union and fails. Suppress only the resulting
+# return-value mismatches in this one file.
+[[tool.mypy.overrides]]
+module = "api.settings.endpoints"
+disable_error_code = ["return-value"]
+
diff --git a/src/api/router.py b/src/api/router.py
index 6e219aa4f..3ddd309c1 100644
--- a/src/api/router.py
+++ b/src/api/router.py
@@ -3,18 +3,17 @@
import json
import os
import tempfile
-from typing import List, Optional
from fastapi import Depends, File, Form, UploadFile
from fastapi.responses import JSONResponse
from config.settings import DISABLE_INGEST_WITH_LANGFLOW
from dependencies import (
+ get_current_user,
get_document_service,
get_langflow_file_service,
get_session_manager,
get_task_service,
- get_current_user,
)
from session_manager import User
from utils.logging_config import get_logger
@@ -23,10 +22,10 @@ logger = get_logger(__name__)
async def upload_ingest_router(
- file: List[UploadFile] = File(...),
- session_id: Optional[str] = Form(None),
- settings_json: Optional[str] = Form(None, alias="settings"),
- tweaks_json: Optional[str] = Form(None, alias="tweaks"),
+ file: list[UploadFile] = File(...),
+ session_id: str | None = Form(None),
+ settings_json: str | None = Form(None, alias="settings"),
+ tweaks_json: str | None = Form(None, alias="tweaks"),
replace_duplicates: str = Form("true"),
create_filter: str = Form("false"),
document_service=Depends(get_document_service),
@@ -50,6 +49,7 @@ async def upload_ingest_router(
logger.debug("Routing to traditional OpenRAG upload")
# Route to traditional upload — just take the first file
from api.upload import upload as traditional_upload_fn
+
return await traditional_upload_fn(
file=file[0] if file else None,
document_service=document_service,
@@ -73,7 +73,7 @@ async def upload_ingest_router(
async def _langflow_upload_ingest_task(
- upload_files: List[UploadFile],
+ upload_files: list[UploadFile],
session_id,
settings_json,
tweaks_json,
@@ -124,7 +124,9 @@ async def _langflow_upload_ingest_task(
f.write(content)
temp_file_paths.append(temp_path)
- file_path_to_original_filename = dict(zip(temp_file_paths, original_filenames))
+ file_path_to_original_filename = dict(
+ zip(temp_file_paths, original_filenames, strict=True)
+ )
task_id = await task_service.create_langflow_upload_task(
user_id=user_id,
@@ -154,6 +156,7 @@ async def _langflow_upload_ingest_task(
except Exception:
from utils.file_utils import safe_unlink
+
for temp_path in temp_file_paths:
safe_unlink(temp_path)
raise
@@ -161,6 +164,7 @@ async def _langflow_upload_ingest_task(
except Exception as e:
logger.error("Task-based langflow upload_ingest failed", error=str(e))
import traceback
+
logger.error("Full traceback", traceback=traceback.format_exc())
error_msg = str(e)
if "AuthenticationException" in error_msg or "access denied" in error_msg.lower():
diff --git a/src/api/settings/__init__.py b/src/api/settings/__init__.py
new file mode 100644
index 000000000..8801675b7
--- /dev/null
+++ b/src/api/settings/__init__.py
@@ -0,0 +1,88 @@
+"""Settings/onboarding endpoint package.
+
+The original `src/api/settings.py` (2081 lines) was split into focused
+submodules. This `__init__.py` preserves the public-import contract every
+external caller relies on:
+
+* `src/app/routes/internal.py` registers each endpoint via
+ `from api import settings; app.add_api_route(..., settings.<name>, ...)`.
+* `src/api/v1/settings.py` imports `SettingsUpdateBody` and (lazily)
+ `update_settings`.
+* `src/services/startup_orchestrator.py` lazily imports
+ `reapply_all_settings` for flow-reset re-sync.
+
+Adding a new endpoint to this package: define it in `endpoints.py`, then
+re-export it here.
+"""
+
+from api.settings.endpoints import (
+ get_settings,
+ onboarding,
+ refresh_openrag_docs,
+ rollback_onboarding,
+ update_docling_preset,
+ update_onboarding_state,
+ update_settings,
+)
+from api.settings.langflow_sync import reapply_all_settings
+from api.settings.models import (
+ AgentConfig,
+ AnthropicProviderConfig,
+ AssistantMessage,
+ DoclingPresetBody,
+ DoclingPresetResponse,
+ IngestionDefaultsConfig,
+ KnowledgeConfig,
+ OllamaProviderConfig,
+ OnboardingBody,
+ OnboardingResponse,
+ OnboardingStateBody,
+ OnboardingStateConfig,
+ OnboardingStateResponse,
+ OpenAIProviderConfig,
+ ProvidersConfig,
+ RefreshOpenRAGDocsResponse,
+ RollbackBody,
+ RollbackResponse,
+ SettingsResponse,
+ SettingsUpdateBody,
+ SettingsUpdateResponse,
+ WatsonXProviderConfig,
+)
+
+__all__ = [
+ # Endpoints
+ "get_settings",
+ "update_settings",
+ "onboarding",
+ "update_onboarding_state",
+ "rollback_onboarding",
+ "update_docling_preset",
+ "refresh_openrag_docs",
+ # Internal orchestrator (called from services/startup_orchestrator.py)
+ "reapply_all_settings",
+ # Pydantic models (some are imported externally; re-export all for parity
+ # with the pre-split flat module's surface)
+ "SettingsUpdateBody",
+ "OnboardingBody",
+ "AssistantMessage",
+ "OnboardingStateBody",
+ "DoclingPresetBody",
+ "OnboardingStateConfig",
+ "OpenAIProviderConfig",
+ "AnthropicProviderConfig",
+ "WatsonXProviderConfig",
+ "OllamaProviderConfig",
+ "ProvidersConfig",
+ "KnowledgeConfig",
+ "AgentConfig",
+ "IngestionDefaultsConfig",
+ "SettingsResponse",
+ "OnboardingResponse",
+ "RefreshOpenRAGDocsResponse",
+ "DoclingPresetResponse",
+ "OnboardingStateResponse",
+ "SettingsUpdateResponse",
+ "RollbackResponse",
+ "RollbackBody",
+]
diff --git a/src/api/settings.py b/src/api/settings/endpoints.py
similarity index 61%
rename from src/api/settings.py
rename to src/api/settings/endpoints.py
index 40a532297..24a3f4b93 100644
--- a/src/api/settings.py
+++ b/src/api/settings/endpoints.py
@@ -1,229 +1,93 @@
-from dependencies import get_models_service
+"""FastAPI endpoint handlers for /settings, /onboarding, /onboarding/state,
+/onboarding/rollback, /settings/docling-preset, and /openrag-docs/refresh.
+
+Lifted verbatim from the original `src/api/settings.py`. Models live in
+`api.settings.models`; provider/filter helpers in `api.settings.helpers`;
+Langflow-sync helpers in `api.settings.langflow_sync`. No behavior change.
+"""
+
import asyncio
import json
-import platform
-from fastapi import Depends, Request, HTTPException
+
+from fastapi import Depends, HTTPException, Request
from fastapi.responses import JSONResponse
-from pydantic import BaseModel, Field
-from utils.logging_config import get_logger
-from utils.telemetry import TelemetryClient, Category, MessageId
-from utils.version_utils import OPENRAG_VERSION
+
+from api.provider_validation import validate_provider_setup
+from api.settings.helpers import (
+ _affected_embedding_models,
+ _create_openrag_docs_filter,
+ _embedding_conflict_response,
+ _first_configured_embedding_provider,
+ _first_configured_llm_provider,
+ _get_flows_service,
+)
+from api.settings.langflow_sync import (
+ _background_tasks,
+ _run_async_post_save_langflow_updates,
+ _update_langflow_docling_settings,
+ _update_langflow_global_variables,
+ _update_langflow_model_values,
+ _update_langflow_system_prompt,
+ _update_mcp_server_urls,
+)
+from api.settings.models import (
+ AgentConfig,
+ AnthropicProviderConfig,
+ DoclingPresetBody,
+ DoclingPresetResponse,
+ IngestionDefaultsConfig,
+ KnowledgeConfig,
+ OllamaProviderConfig,
+ OnboardingBody,
+ OnboardingResponse,
+ OnboardingStateBody,
+ OnboardingStateConfig,
+ OnboardingStateResponse,
+ OpenAIProviderConfig,
+ ProvidersConfig,
+ RefreshOpenRAGDocsResponse,
+ RollbackBody,
+ RollbackResponse,
+ SettingsResponse,
+ SettingsUpdateBody,
+ SettingsUpdateResponse,
+ WatsonXProviderConfig,
+)
from config.settings import (
DEFAULT_DOCS_URL,
+ ENVIRONMENT,
INGEST_SAMPLE_DATA,
- LANGFLOW_URL,
LANGFLOW_CHAT_FLOW_ID,
LANGFLOW_INGEST_FLOW_ID,
LANGFLOW_PUBLIC_URL,
+ LANGFLOW_URL,
LOCALHOST_URL,
OPENRAG_INGEST_VIA_CHAT,
SEGMENT_WRITE_KEY,
- ENVIRONMENT,
clients,
- get_openrag_config,
config_manager,
- is_no_auth_mode,
+ get_openrag_config,
)
-from typing import List, Optional, Any, Dict
-
-from api.provider_validation import validate_provider_setup
-from utils.langflow_utils import LangflowNotReadyError, wait_for_langflow
from dependencies import (
- get_session_manager,
- get_flows_service,
- get_task_service,
+ get_chat_service,
get_current_user,
get_document_service,
- get_langflow_file_service,
+ get_flows_service,
get_knowledge_filter_service,
- get_chat_service,
+ get_langflow_file_service,
+ get_models_service,
+ get_session_manager,
+ get_task_service,
require_permission,
)
from services.docling_service import DoclingConfig, get_docling_preset_configs
from session_manager import User
+from utils.langflow_utils import LangflowNotReadyError, wait_for_langflow
+from utils.logging_config import get_logger
+from utils.telemetry import Category, MessageId, TelemetryClient
+from utils.version_utils import OPENRAG_VERSION
logger = get_logger(__name__)
-_background_tasks: set[asyncio.Task] = set()
-
-
-class SettingsUpdateBody(BaseModel):
- llm_model: Optional[str] = Field(None, min_length=1)
- llm_provider: Optional[str] = Field(None, pattern="^(openai|anthropic|watsonx|ollama)$")
- system_prompt: Optional[str] = None
- chunk_size: Optional[int] = Field(None, gt=0)
- chunk_overlap: Optional[int] = Field(None, ge=0)
- table_structure: Optional[bool] = None
- ocr: Optional[bool] = None
- picture_descriptions: Optional[bool] = None
- embedding_model: Optional[str] = Field(None, min_length=1)
- embedding_provider: Optional[str] = Field(None, pattern="^(openai|watsonx|ollama)$")
- index_name: Optional[str] = Field(None, min_length=1)
- openai_api_key: Optional[str] = Field(None, min_length=1)
- anthropic_api_key: Optional[str] = Field(None, min_length=1)
- watsonx_api_key: Optional[str] = Field(None, min_length=1)
- watsonx_endpoint: Optional[str] = Field(None, min_length=1)
- watsonx_project_id: Optional[str] = Field(None, min_length=1)
- ollama_endpoint: Optional[str] = Field(None, min_length=1)
- remove_ollama_config: Optional[bool] = None
- remove_openai_config: Optional[bool] = None
- remove_anthropic_config: Optional[bool] = None
- remove_watsonx_config: Optional[bool] = None
- # Explicit confirmation that the caller accepts removing a provider whose
- # embedding models are still in use by indexed documents. Without this,
- # the backend returns 409 and the frontend prompts the user.
- force_remove: Optional[bool] = False
-
-
-class OnboardingBody(BaseModel):
- llm_provider: Optional[str] = Field(None, pattern="^(openai|anthropic|watsonx|ollama)$")
- llm_model: Optional[str] = Field(None, min_length=1)
- embedding_provider: Optional[str] = Field(None, pattern="^(openai|watsonx|ollama)$")
- embedding_model: Optional[str] = Field(None, min_length=1)
- openai_api_key: Optional[str] = Field(None, min_length=1)
- anthropic_api_key: Optional[str] = Field(None, min_length=1)
- watsonx_api_key: Optional[str] = Field(None, min_length=1)
- watsonx_endpoint: Optional[str] = Field(None, min_length=1)
- watsonx_project_id: Optional[str] = Field(None, min_length=1)
- ollama_endpoint: Optional[str] = Field(None, min_length=1)
-
-
-class AssistantMessage(BaseModel):
- role: str
- content: str
- timestamp: str
-
-
-class OnboardingStateBody(BaseModel):
- current_step: Optional[int] = None
- assistant_message: Optional[AssistantMessage] = None
- selected_nudge: Optional[str] = None
- card_steps: Optional[Dict[str, Any]] = None
- upload_steps: Optional[Dict[str, Any]] = None
- openrag_docs_filter_id: Optional[str] = None
- user_doc_filter_id: Optional[str] = None
- openrag_docs_ingested_version: Optional[str] = None
- openrag_docs_remote_signature: Optional[str] = None
-
-
-class DoclingPresetBody(BaseModel):
- preset: Optional[str] = None
- table_structure: Optional[bool] = None
- ocr: Optional[bool] = None
- picture_descriptions: Optional[bool] = None
-
-
-class OnboardingStateConfig(BaseModel):
- current_step: Optional[int]
- assistant_message: Optional[AssistantMessage]
- selected_nudge: Optional[str]
- card_steps: Optional[Dict[str, Any]]
- upload_steps: Optional[Dict[str, Any]]
- openrag_docs_filter_id: Optional[str]
- user_doc_filter_id: Optional[str]
- openrag_docs_ingested_version: Optional[str]
- openrag_docs_remote_signature: Optional[str]
-
-class OpenAIProviderConfig(BaseModel):
- has_api_key: bool
- configured: bool
-
-class AnthropicProviderConfig(BaseModel):
- has_api_key: bool
- configured: bool
-
-class WatsonXProviderConfig(BaseModel):
- has_api_key: bool
- endpoint: Optional[str]
- project_id: Optional[str]
- configured: bool
-
-class OllamaProviderConfig(BaseModel):
- endpoint: Optional[str]
- configured: bool
-
-class ProvidersConfig(BaseModel):
- openai: OpenAIProviderConfig
- anthropic: AnthropicProviderConfig
- watsonx: WatsonXProviderConfig
- ollama: OllamaProviderConfig
-
-class KnowledgeConfig(BaseModel):
- embedding_model: Optional[str]
- embedding_provider: Optional[str]
- chunk_size: Optional[int]
- chunk_overlap: Optional[int]
- table_structure: Optional[bool]
- ocr: Optional[bool]
- picture_descriptions: Optional[bool]
- index_name: Optional[str]
-
-class AgentConfig(BaseModel):
- llm_model: Optional[str]
- llm_provider: Optional[str]
- system_prompt: Optional[str]
-
-class IngestionDefaultsConfig(BaseModel):
- chunkSize: Optional[int]
- chunkOverlap: Optional[int]
- separator: Optional[str]
- embeddingModel: Optional[str]
-
-class SettingsResponse(BaseModel):
- langflow_url: str
- flow_id: Optional[str]
- ingest_flow_id: Optional[str]
- langflow_public_url: Optional[str]
- edited: bool
- onboarding: OnboardingStateConfig
- providers: ProvidersConfig
- knowledge: KnowledgeConfig
- agent: AgentConfig
- localhost_url: str
- langflow_edit_url: Optional[str] = None
- langflow_ingest_edit_url: Optional[str] = None
- ingestion_defaults: Optional[IngestionDefaultsConfig] = None
- ingest_via_chat: bool = False
- segment_write_key: Optional[str] = None
- environment: Optional[str] = None
-
-
-class OnboardingResponse(BaseModel):
- message: str
- edited: bool
- sample_data_ingested: bool
- openrag_docs_filter_id: Optional[str] = None
- task_id: Optional[str] = None
-
-
-class RefreshOpenRAGDocsResponse(BaseModel):
- message: str
- refreshed: bool
-
-class DoclingPresetResponse(BaseModel):
- message: str
- settings: dict
- preset_config: DoclingConfig
-
-
-class OnboardingStateResponse(BaseModel):
- message: str
- updated_fields: List[str]
-
-class SettingsUpdateResponse(BaseModel):
- message: str
-
-class RollbackResponse(BaseModel):
- message: str
- cancelled_tasks: int
- deleted_files: int
- reset_flows: int
- deleted_conversations: int
-
-class RollbackBody(BaseModel):
- embedding_only: bool = False
-
-
-# Docling preset configurations
async def get_settings(
@@ -245,7 +109,9 @@ async def get_settings(
langflow_ingest_edit_url = None
if LANGFLOW_PUBLIC_URL and LANGFLOW_INGEST_FLOW_ID:
- langflow_ingest_edit_url = f"{LANGFLOW_PUBLIC_URL.rstrip('/')}/flow/{LANGFLOW_INGEST_FLOW_ID}"
+ langflow_ingest_edit_url = (
+ f"{LANGFLOW_PUBLIC_URL.rstrip('/')}/flow/{LANGFLOW_INGEST_FLOW_ID}"
+ )
ingestion_defaults_obj = None
# Fetch ingestion flow configuration to get actual component defaults
@@ -268,31 +134,29 @@ async def get_settings(
if flow_data.get("data", {}).get("nodes"):
for node in flow_data["data"]["nodes"]:
- node_template = (
- node.get("data", {}).get("node", {}).get("template", {})
- )
+ node_template = node.get("data", {}).get("node", {}).get("template", {})
# Split Text component (SplitText-QIKhg)
if node.get("id") == "SplitText-QIKhg":
if node_template.get("chunk_size", {}).get("value"):
- ingestion_defaults["chunkSize"] = node_template[
- "chunk_size"
- ]["value"]
+ ingestion_defaults["chunkSize"] = node_template["chunk_size"][
+ "value"
+ ]
if node_template.get("chunk_overlap", {}).get("value"):
ingestion_defaults["chunkOverlap"] = node_template[
"chunk_overlap"
]["value"]
if node_template.get("separator", {}).get("value"):
- ingestion_defaults["separator"] = node_template[
- "separator"
- ]["value"]
+ ingestion_defaults["separator"] = node_template["separator"][
+ "value"
+ ]
# OpenAI Embeddings component (OpenAIEmbeddings-joRJ6)
elif node.get("id") == "OpenAIEmbeddings-joRJ6":
if node_template.get("model", {}).get("value"):
- ingestion_defaults["embeddingModel"] = (
- node_template["model"]["value"]
- )
+ ingestion_defaults["embeddingModel"] = node_template["model"][
+ "value"
+ ]
ingestion_defaults_obj = IngestionDefaultsConfig(**ingestion_defaults)
@@ -363,121 +227,7 @@ async def get_settings(
except Exception as e:
logger.error(f"Failed to retrieve settings: {str(e)}")
- return JSONResponse(
- {"error": f"Failed to retrieve settings: {str(e)}"}, status_code=500
- )
-
-
-def _first_configured_llm_provider(config, excluding: str) -> str:
- """Return the first configured LLM provider that isn't `excluding`."""
- for p in ["openai", "anthropic", "watsonx", "ollama"]:
- if p != excluding and getattr(config.providers, p).configured:
- return p
- return "openai"
-
-
-def _first_configured_embedding_provider(config, excluding: str) -> str:
- """Return the first configured embedding provider (openai/watsonx/ollama) that isn't `excluding`."""
- for p in ["openai", "watsonx", "ollama"]:
- if p != excluding and getattr(config.providers, p).configured:
- return p
- return "openai"
-
-
-async def _affected_embedding_models(
- provider: str,
- session_manager,
- user,
- models_service,
-) -> List[Dict[str, Any]]:
- """Find embedding models present in the corpus that belong to ``provider``.
-
- Used to warn users before they remove a provider whose embedding models
- were used to index documents — otherwise semantic search silently breaks
- for those docs. Returns a list of ``{"model": str, "doc_count": int}``.
-
- Conservative on errors (returns empty list) so infra issues don't block
- provider removal.
- """
- from services.models_service import ModelsService
- from config.settings import get_index_name
-
- provider_lower = provider.lower()
- if provider_lower == "anthropic":
- # Anthropic doesn't serve embedding models — nothing to warn about.
- return []
-
- try:
- # Refresh so the registry reflects currently-configured providers
- # before we use it to attribute models.
- await models_service.update_model_registry()
- registry = ModelsService._model_provider_registry
-
- # Use the admin client so DLS does not scope the aggregation to the
- # requesting user's documents. Provider removal is a global operation
- # that affects all tenants, so we must see every document's embedding
- # model regardless of ownership.
- agg_result = await clients.opensearch.search(
- index=get_index_name(),
- body={
- "size": 0,
- "aggs": {
- "embedding_models": {
- "terms": {"field": "embedding_model", "size": 50}
- }
- },
- },
- params={"terminate_after": 0},
- )
- buckets = (
- agg_result.get("aggregations", {})
- .get("embedding_models", {})
- .get("buckets", [])
- )
-
- affected: List[Dict[str, Any]] = []
- for bucket in buckets:
- model = bucket.get("key")
- if not model:
- continue
- mapped = registry.get(model)
- # Narrow fallback: the watsonx registry bootstrap requires the
- # provider still be configured, so models from an about-to-be-
- # removed watsonx can still be attributed via the "ibm/" prefix.
- if mapped is None and provider_lower == "watsonx" and model.startswith("ibm/"):
- mapped = "watsonx"
- if mapped == provider_lower:
- affected.append({"model": model, "doc_count": bucket.get("doc_count", 0)})
- return affected
- except Exception as e:
- logger.warning(
- "Could not determine affected embedding models for provider removal",
- provider=provider,
- error=str(e),
- )
- return []
-
-
-def _embedding_conflict_response(
- provider_label: str, provider_key: str, affected: List[Dict[str, Any]]
-) -> JSONResponse:
- """Shared 409 response when removing a provider whose embedding models are
- still referenced by indexed documents."""
- model_names = [a["model"] for a in affected]
- return JSONResponse(
- {
- "error": (
- f"Removing {provider_label} will disable semantic search on "
- f"documents indexed with: {', '.join(model_names)}. "
- f"Re-ingest affected documents with another embedding model, "
- f"or retry with force_remove=true to proceed anyway."
- ),
- "code": "embedding_provider_in_use",
- "affected_provider": provider_key,
- "affected_models": affected,
- },
- status_code=409,
- )
+ return JSONResponse({"error": f"Failed to retrieve settings: {str(e)}"}, status_code=500)
async def update_settings(
@@ -494,9 +244,7 @@ async def update_settings(
# Check if config is marked as edited
if not current_config.edited:
return JSONResponse(
- {
- "error": "Configuration must be marked as edited before updates are allowed"
- },
+ {"error": "Configuration must be marked as edited before updates are allowed"},
status_code=403,
)
@@ -514,7 +262,7 @@ async def update_settings(
"watsonx_project_id",
"ollama_endpoint",
]
-
+
should_validate = any(getattr(body, field) is not None for field in provider_fields)
if should_validate:
@@ -523,8 +271,16 @@ async def update_settings(
# Validate LLM provider if being changed
if body.llm_provider is not None or body.llm_model is not None:
- llm_provider = body.llm_provider if body.llm_provider is not None else current_config.agent.llm_provider
- llm_model = body.llm_model if body.llm_model is not None else current_config.agent.llm_model
+ llm_provider = (
+ body.llm_provider
+ if body.llm_provider is not None
+ else current_config.agent.llm_provider
+ )
+ llm_model = (
+ body.llm_model
+ if body.llm_model is not None
+ else current_config.agent.llm_model
+ )
# Get the provider config (with any updates from the request)
llm_provider_config = current_config.providers.get_provider_config(llm_provider)
@@ -534,7 +290,10 @@ async def update_settings(
endpoint = getattr(llm_provider_config, "endpoint", None)
project_id = getattr(llm_provider_config, "project_id", None)
- if getattr(body, f"{llm_provider}_api_key", None) is not None and getattr(body, f"{llm_provider}_api_key", None).strip():
+ if (
+ getattr(body, f"{llm_provider}_api_key", None) is not None
+ and getattr(body, f"{llm_provider}_api_key", None).strip()
+ ):
api_key = getattr(body, f"{llm_provider}_api_key", None)
if getattr(body, f"{llm_provider}_endpoint", None) is not None:
endpoint = getattr(body, f"{llm_provider}_endpoint", None)
@@ -552,18 +311,31 @@ async def update_settings(
# Validate embedding provider if being changed
if body.embedding_provider is not None or body.embedding_model is not None:
- embedding_provider = body.embedding_provider if body.embedding_provider is not None else current_config.knowledge.embedding_provider
- embedding_model = body.embedding_model if body.embedding_model is not None else current_config.knowledge.embedding_model
+ embedding_provider = (
+ body.embedding_provider
+ if body.embedding_provider is not None
+ else current_config.knowledge.embedding_provider
+ )
+ embedding_model = (
+ body.embedding_model
+ if body.embedding_model is not None
+ else current_config.knowledge.embedding_model
+ )
# Get the provider config (with any updates from the request)
- embedding_provider_config = current_config.providers.get_provider_config(embedding_provider)
+ embedding_provider_config = current_config.providers.get_provider_config(
+ embedding_provider
+ )
# Apply any updates from the request
api_key = getattr(embedding_provider_config, "api_key", None)
endpoint = getattr(embedding_provider_config, "endpoint", None)
project_id = getattr(embedding_provider_config, "project_id", None)
- if getattr(body, f"{embedding_provider}_api_key", None) is not None and getattr(body, f"{embedding_provider}_api_key", None).strip():
+ if (
+ getattr(body, f"{embedding_provider}_api_key", None) is not None
+ and getattr(body, f"{embedding_provider}_api_key", None).strip()
+ ):
api_key = getattr(body, f"{embedding_provider}_api_key", None)
if getattr(body, f"{embedding_provider}_endpoint", None) is not None:
endpoint = getattr(body, f"{embedding_provider}_endpoint", None)
@@ -577,7 +349,9 @@ async def update_settings(
endpoint=endpoint,
project_id=project_id,
)
- logger.info(f"Embedding provider validation successful for {embedding_provider}")
+ logger.info(
+ f"Embedding provider validation successful for {embedding_provider}"
+ )
except Exception as e:
logger.error(f"Provider validation failed: {str(e)}")
@@ -593,8 +367,7 @@ async def update_settings(
current_config.agent.llm_model = body.llm_model
config_updated = True
await TelemetryClient.send_event(
- Category.SETTINGS_OPERATIONS,
- MessageId.ORB_SETTINGS_LLM_MODEL
+ Category.SETTINGS_OPERATIONS, MessageId.ORB_SETTINGS_LLM_MODEL
)
logger.info(f"LLM model changed from {old_model} to {body.llm_model}")
@@ -603,8 +376,7 @@ async def update_settings(
current_config.agent.llm_provider = body.llm_provider
config_updated = True
await TelemetryClient.send_event(
- Category.SETTINGS_OPERATIONS,
- MessageId.ORB_SETTINGS_LLM_PROVIDER
+ Category.SETTINGS_OPERATIONS, MessageId.ORB_SETTINGS_LLM_PROVIDER
)
logger.info(f"LLM provider changed from {old_provider} to {body.llm_provider}")
@@ -612,8 +384,7 @@ async def update_settings(
current_config.agent.system_prompt = body.system_prompt
config_updated = True
await TelemetryClient.send_event(
- Category.SETTINGS_OPERATIONS,
- MessageId.ORB_SETTINGS_SYSTEM_PROMPT
+ Category.SETTINGS_OPERATIONS, MessageId.ORB_SETTINGS_SYSTEM_PROMPT
)
# Also update the chat flow with the new system prompt
@@ -632,8 +403,7 @@ async def update_settings(
current_config.knowledge.embedding_model = new_embedding_model
config_updated = True
await TelemetryClient.send_event(
- Category.SETTINGS_OPERATIONS,
- MessageId.ORB_SETTINGS_EMBED_MODEL
+ Category.SETTINGS_OPERATIONS, MessageId.ORB_SETTINGS_EMBED_MODEL
)
logger.info(f"Embedding model changed from {old_model} to {new_embedding_model}")
@@ -642,17 +412,17 @@ async def update_settings(
current_config.knowledge.embedding_provider = body.embedding_provider
config_updated = True
await TelemetryClient.send_event(
- Category.SETTINGS_OPERATIONS,
- MessageId.ORB_SETTINGS_EMBED_PROVIDER
+ Category.SETTINGS_OPERATIONS, MessageId.ORB_SETTINGS_EMBED_PROVIDER
+ )
+ logger.info(
+ f"Embedding provider changed from {old_provider} to {body.embedding_provider}"
)
- logger.info(f"Embedding provider changed from {old_provider} to {body.embedding_provider}")
if body.table_structure is not None:
current_config.knowledge.table_structure = body.table_structure
config_updated = True
await TelemetryClient.send_event(
- Category.SETTINGS_OPERATIONS,
- MessageId.ORB_SETTINGS_DOCLING_UPDATED
+ Category.SETTINGS_OPERATIONS, MessageId.ORB_SETTINGS_DOCLING_UPDATED
)
# Also update the flow with the new docling settings
@@ -666,8 +436,7 @@ async def update_settings(
current_config.knowledge.ocr = body.ocr
config_updated = True
await TelemetryClient.send_event(
- Category.SETTINGS_OPERATIONS,
- MessageId.ORB_SETTINGS_DOCLING_UPDATED
+ Category.SETTINGS_OPERATIONS, MessageId.ORB_SETTINGS_DOCLING_UPDATED
)
# Also update the flow with the new docling settings
@@ -681,8 +450,7 @@ async def update_settings(
current_config.knowledge.picture_descriptions = body.picture_descriptions
config_updated = True
await TelemetryClient.send_event(
- Category.SETTINGS_OPERATIONS,
- MessageId.ORB_SETTINGS_DOCLING_UPDATED
+ Category.SETTINGS_OPERATIONS, MessageId.ORB_SETTINGS_DOCLING_UPDATED
)
# Also update the flow with the new docling settings
@@ -693,51 +461,51 @@ async def update_settings(
logger.error(f"Failed to update docling settings in flow: {str(e)}")
if body.chunk_size is not None:
- effective_overlap = body.chunk_overlap if body.chunk_overlap is not None else current_config.knowledge.chunk_overlap
+ effective_overlap = (
+ body.chunk_overlap
+ if body.chunk_overlap is not None
+ else current_config.knowledge.chunk_overlap
+ )
if effective_overlap >= body.chunk_size:
raise HTTPException(
- status_code=422,
- detail="chunk_overlap must be less than chunk_size"
+ status_code=422, detail="chunk_overlap must be less than chunk_size"
)
current_config.knowledge.chunk_size = body.chunk_size
config_updated = True
await TelemetryClient.send_event(
- Category.SETTINGS_OPERATIONS,
- MessageId.ORB_SETTINGS_CHUNK_UPDATED
+ Category.SETTINGS_OPERATIONS, MessageId.ORB_SETTINGS_CHUNK_UPDATED
)
# Also update the ingest flow with the new chunk size
try:
flows_service = _get_flows_service()
await flows_service.update_ingest_flow_chunk_size(body.chunk_size)
- logger.info(
- f"Successfully updated ingest flow chunk size to {body.chunk_size}"
- )
+ logger.info(f"Successfully updated ingest flow chunk size to {body.chunk_size}")
except Exception as e:
logger.error(f"Failed to update ingest flow chunk size: {str(e)}")
# Don't fail the entire settings update if flow update fails
# The config will still be saved
if body.chunk_overlap is not None:
- effective_chunk_size = body.chunk_size if body.chunk_size is not None else current_config.knowledge.chunk_size
+ effective_chunk_size = (
+ body.chunk_size
+ if body.chunk_size is not None
+ else current_config.knowledge.chunk_size
+ )
if body.chunk_overlap >= effective_chunk_size:
raise HTTPException(
- status_code=422,
- detail="chunk_overlap must be less than chunk_size"
+ status_code=422, detail="chunk_overlap must be less than chunk_size"
)
current_config.knowledge.chunk_overlap = body.chunk_overlap
config_updated = True
await TelemetryClient.send_event(
- Category.SETTINGS_OPERATIONS,
- MessageId.ORB_SETTINGS_CHUNK_UPDATED
+ Category.SETTINGS_OPERATIONS, MessageId.ORB_SETTINGS_CHUNK_UPDATED
)
# Also update the ingest flow with the new chunk overlap
try:
flows_service = _get_flows_service()
- await flows_service.update_ingest_flow_chunk_overlap(
- body.chunk_overlap
- )
+ await flows_service.update_ingest_flow_chunk_overlap(body.chunk_overlap)
logger.info(
f"Successfully updated ingest flow chunk overlap to {body.chunk_overlap}"
)
@@ -750,14 +518,15 @@ async def update_settings(
current_config.knowledge.index_name = new_index_name
config_updated = True
await TelemetryClient.send_event(
- Category.SETTINGS_OPERATIONS,
- MessageId.ORB_SETTINGS_INDEX_NAME_UPDATED
+ Category.SETTINGS_OPERATIONS, MessageId.ORB_SETTINGS_INDEX_NAME_UPDATED
)
logger.info(f"Index name changed from {old_index_name} to {new_index_name}")
# Also update global variable with new index name
try:
- await clients._create_langflow_global_variable("OPENSEARCH_INDEX_NAME", new_index_name, modify=True)
+ await clients._create_langflow_global_variable(
+ "OPENSEARCH_INDEX_NAME", new_index_name, modify=True
+ )
logger.info(
f"Successfully updated global variable with new index name {new_index_name}"
)
@@ -813,7 +582,9 @@ async def update_settings(
)
if not other_providers_configured:
return JSONResponse(
- {"error": "Cannot remove Ollama configuration: configure another model provider first."},
+ {
+ "error": "Cannot remove Ollama configuration: configure another model provider first."
+ },
status_code=400,
)
if not body.force_remove:
@@ -825,10 +596,14 @@ async def update_settings(
current_config.providers.ollama.endpoint = ""
current_config.providers.ollama.configured = False
if current_config.agent.llm_provider == "ollama":
- current_config.agent.llm_provider = _first_configured_llm_provider(current_config, "ollama")
+ current_config.agent.llm_provider = _first_configured_llm_provider(
+ current_config, "ollama"
+ )
current_config.agent.llm_model = ""
if current_config.knowledge.embedding_provider == "ollama":
- current_config.knowledge.embedding_provider = _first_configured_embedding_provider(current_config, "ollama")
+ current_config.knowledge.embedding_provider = _first_configured_embedding_provider(
+ current_config, "ollama"
+ )
current_config.knowledge.embedding_model = ""
config_updated = True
provider_updated = True
@@ -841,7 +616,9 @@ async def update_settings(
)
if not other_providers_configured:
return JSONResponse(
- {"error": "Cannot remove OpenAI configuration: configure another model provider first."},
+ {
+ "error": "Cannot remove OpenAI configuration: configure another model provider first."
+ },
status_code=400,
)
if not body.force_remove:
@@ -871,7 +648,9 @@ async def update_settings(
)
if not other_providers_configured:
return JSONResponse(
- {"error": "Cannot remove Anthropic configuration: configure another model provider first."},
+ {
+ "error": "Cannot remove Anthropic configuration: configure another model provider first."
+ },
status_code=400,
)
current_config.providers.anthropic.api_key = ""
@@ -892,7 +671,9 @@ async def update_settings(
)
if not other_providers_configured:
return JSONResponse(
- {"error": "Cannot remove IBM watsonx.ai configuration: configure another model provider first."},
+ {
+ "error": "Cannot remove IBM watsonx.ai configuration: configure another model provider first."
+ },
status_code=400,
)
if not body.force_remove:
@@ -900,9 +681,7 @@ async def update_settings(
"watsonx", session_manager, user, models_service
)
if affected:
- return _embedding_conflict_response(
- "IBM watsonx.ai", "watsonx", affected
- )
+ return _embedding_conflict_response("IBM watsonx.ai", "watsonx", affected)
current_config.providers.watsonx.api_key = ""
current_config.providers.watsonx.endpoint = ""
current_config.providers.watsonx.project_id = ""
@@ -920,20 +699,15 @@ async def update_settings(
if provider_updated:
await TelemetryClient.send_event(
- Category.SETTINGS_OPERATIONS,
- MessageId.ORB_SETTINGS_PROVIDER_CREDS
+ Category.SETTINGS_OPERATIONS, MessageId.ORB_SETTINGS_PROVIDER_CREDS
)
if not config_updated:
- return JSONResponse(
- {"error": "No valid fields provided for update"}, status_code=400
- )
+ return JSONResponse({"error": "No valid fields provided for update"}, status_code=400)
# Save the updated configuration
if not config_manager.save_config_file(current_config):
- return JSONResponse(
- {"error": "Failed to save configuration"}, status_code=500
- )
+ return JSONResponse({"error": "Failed to save configuration"}, status_code=500)
# Refresh patched client immediately so subsequent requests pick up latest config.
await clients.refresh_patched_client()
@@ -962,26 +736,19 @@ async def update_settings(
_background_tasks.add(task)
task.add_done_callback(_background_tasks.discard)
-
set_fields = [k for k, v in body.model_dump().items() if v is not None]
- logger.info(
- "Configuration updated successfully", updated_fields=set_fields
- )
+ logger.info("Configuration updated successfully", updated_fields=set_fields)
await TelemetryClient.send_event(
- Category.SETTINGS_OPERATIONS,
- MessageId.ORB_SETTINGS_UPDATED
+ Category.SETTINGS_OPERATIONS, MessageId.ORB_SETTINGS_UPDATED
)
return SettingsUpdateResponse(message="Configuration updated successfully")
except Exception as e:
logger.error("Failed to update settings", error=str(e))
await TelemetryClient.send_event(
- Category.SETTINGS_OPERATIONS,
- MessageId.ORB_SETTINGS_UPDATE_FAILED
- )
- return JSONResponse(
- {"error": f"Failed to update settings: {str(e)}"}, status_code=500
+ Category.SETTINGS_OPERATIONS, MessageId.ORB_SETTINGS_UPDATE_FAILED
)
+ return JSONResponse({"error": f"Failed to update settings: {str(e)}"}, status_code=500)
async def onboarding(
@@ -1022,7 +789,7 @@ async def onboarding(
await TelemetryClient.send_event(
Category.ONBOARDING,
MessageId.ORB_ONBOARD_LLM_MODEL,
- metadata={"llm_model": llm_model_selected}
+ metadata={"llm_model": llm_model_selected},
)
logger.info(f"LLM model selected during onboarding: {llm_model_selected}")
@@ -1033,7 +800,7 @@ async def onboarding(
await TelemetryClient.send_event(
Category.ONBOARDING,
MessageId.ORB_ONBOARD_LLM_PROVIDER,
- metadata={"llm_provider": llm_provider_selected}
+ metadata={"llm_provider": llm_provider_selected},
)
logger.info(f"LLM provider selected during onboarding: {llm_provider_selected}")
@@ -1048,7 +815,7 @@ async def onboarding(
await TelemetryClient.send_event(
Category.ONBOARDING,
MessageId.ORB_ONBOARD_EMBED_MODEL,
- metadata={"embedding_model": embedding_model_selected}
+ metadata={"embedding_model": embedding_model_selected},
)
logger.info(f"Embedding model selected during onboarding: {embedding_model_selected}")
@@ -1059,9 +826,11 @@ async def onboarding(
await TelemetryClient.send_event(
Category.ONBOARDING,
MessageId.ORB_ONBOARD_EMBED_PROVIDER,
- metadata={"embedding_provider": embedding_provider_selected}
+ metadata={"embedding_provider": embedding_provider_selected},
+ )
+ logger.info(
+ f"Embedding provider selected during onboarding: {embedding_provider_selected}"
)
- logger.info(f"Embedding provider selected during onboarding: {embedding_provider_selected}")
# Update provider-specific credentials
if body.openai_api_key:
@@ -1104,7 +873,12 @@ async def onboarding(
elif llm_provider == "anthropic" and current_config.providers.anthropic.api_key:
current_config.providers.anthropic.configured = True
logger.info("Marked Anthropic as configured (chosen as LLM provider)")
- elif llm_provider == "watsonx" and current_config.providers.watsonx.api_key and current_config.providers.watsonx.endpoint and current_config.providers.watsonx.project_id:
+ elif (
+ llm_provider == "watsonx"
+ and current_config.providers.watsonx.api_key
+ and current_config.providers.watsonx.endpoint
+ and current_config.providers.watsonx.project_id
+ ):
current_config.providers.watsonx.configured = True
logger.info("Marked WatsonX as configured (chosen as LLM provider)")
elif llm_provider == "ollama" and current_config.providers.ollama.endpoint:
@@ -1117,7 +891,12 @@ async def onboarding(
if embedding_provider == "openai" and current_config.providers.openai.api_key:
current_config.providers.openai.configured = True
logger.info("Marked OpenAI as configured (chosen as embedding provider)")
- elif embedding_provider == "watsonx" and current_config.providers.watsonx.api_key and current_config.providers.watsonx.endpoint and current_config.providers.watsonx.project_id:
+ elif (
+ embedding_provider == "watsonx"
+ and current_config.providers.watsonx.api_key
+ and current_config.providers.watsonx.endpoint
+ and current_config.providers.watsonx.project_id
+ ):
current_config.providers.watsonx.configured = True
logger.info("Marked WatsonX as configured (chosen as embedding provider)")
elif embedding_provider == "ollama" and current_config.providers.ollama.endpoint:
@@ -1126,28 +905,23 @@ async def onboarding(
should_ingest_sample_data = INGEST_SAMPLE_DATA
if should_ingest_sample_data:
- await TelemetryClient.send_event(
- Category.ONBOARDING,
- MessageId.ORB_ONBOARD_SAMPLE_DATA
- )
+ await TelemetryClient.send_event(Category.ONBOARDING, MessageId.ORB_ONBOARD_SAMPLE_DATA)
logger.info("Sample data ingestion enabled via environment variable")
if not config_updated:
- return JSONResponse(
- {"error": "No valid fields provided for update"}, status_code=400
- )
+ return JSONResponse({"error": "No valid fields provided for update"}, status_code=400)
# Validate provider setup before initializing OpenSearch index
# Use full validation with completion tests (test_completion=True) to ensure provider health during onboarding
try:
- from api.provider_validation import validate_provider_setup
-
# Validate LLM provider if set
if body.llm_provider or body.llm_model:
llm_provider = current_config.agent.llm_provider.lower()
llm_provider_config = current_config.get_llm_provider_config()
- logger.info(f"Validating LLM provider setup for {llm_provider} (full validation with completion test)")
+ logger.info(
+ f"Validating LLM provider setup for {llm_provider} (full validation with completion test)"
+ )
await validate_provider_setup(
provider=llm_provider,
api_key=getattr(llm_provider_config, "api_key", None),
@@ -1156,14 +930,18 @@ async def onboarding(
project_id=getattr(llm_provider_config, "project_id", None),
test_completion=True, # Full validation with completion test - ensures provider health
)
- logger.info(f"LLM provider setup validation completed successfully for {llm_provider}")
+ logger.info(
+ f"LLM provider setup validation completed successfully for {llm_provider}"
+ )
# Validate embedding provider if set
if body.embedding_provider or body.embedding_model:
embedding_provider = current_config.knowledge.embedding_provider.lower()
embedding_provider_config = current_config.get_embedding_provider_config()
- logger.info(f"Validating embedding provider setup for {embedding_provider} (full validation with completion test)")
+ logger.info(
+ f"Validating embedding provider setup for {embedding_provider} (full validation with completion test)"
+ )
await validate_provider_setup(
provider=embedding_provider,
api_key=getattr(embedding_provider_config, "api_key", None),
@@ -1172,7 +950,9 @@ async def onboarding(
project_id=getattr(embedding_provider_config, "project_id", None),
test_completion=True, # Full validation with completion test - ensures provider health
)
- logger.info(f"Embedding provider setup validation completed successfully for {embedding_provider}")
+ logger.info(
+ f"Embedding provider setup validation completed successfully for {embedding_provider}"
+ )
except Exception as e:
logger.error(f"Provider validation failed: {str(e)}")
return JSONResponse(
@@ -1188,34 +968,55 @@ async def onboarding(
logger.error(message, error=str(e))
return JSONResponse(
- {
- "error": message
- },
+ {"error": message},
status_code=503,
)
# Set Langflow global variables and model values based on provider configuration
try:
# Check if any provider-related fields were provided
- provider_fields_provided = any([
- body.openai_api_key, body.anthropic_api_key,
- body.watsonx_api_key, body.watsonx_endpoint, body.watsonx_project_id,
- body.ollama_endpoint
- ])
+ provider_fields_provided = any(
+ [
+ body.openai_api_key,
+ body.anthropic_api_key,
+ body.watsonx_api_key,
+ body.watsonx_endpoint,
+ body.watsonx_project_id,
+ body.ollama_endpoint,
+ ]
+ )
# Update global variables if any provider fields were provided
# or if existing config has values (for OpenAI/Anthropic that might already be set)
- if (provider_fields_provided or
- current_config.providers.openai.api_key != "" or
- current_config.providers.anthropic.api_key != "" or
- current_config.providers.any_configured()):
+ if (
+ provider_fields_provided
+ or current_config.providers.openai.api_key != ""
+ or current_config.providers.anthropic.api_key != ""
+ or current_config.providers.any_configured()
+ ):
await _update_langflow_global_variables(current_config, flows_service=flows_service)
if body.embedding_provider or body.embedding_model:
- await _update_mcp_server_urls(current_config, session_manager=session_manager, flows_service=flows_service)
+ await _update_mcp_server_urls(
+ current_config,
+ session_manager=session_manager,
+ flows_service=flows_service,
+ )
- if body.llm_provider or body.llm_model or body.embedding_provider or body.embedding_model:
- await _update_langflow_model_values(current_config, flows_service, embedding_model=body.embedding_model, embedding_provider=body.embedding_provider, llm_model=body.llm_model, llm_provider=body.llm_provider)
+ if (
+ body.llm_provider
+ or body.llm_model
+ or body.embedding_provider
+ or body.embedding_model
+ ):
+ await _update_langflow_model_values(
+ current_config,
+ flows_service,
+ embedding_model=body.embedding_model,
+ embedding_provider=body.embedding_provider,
+ llm_model=body.llm_model,
+ llm_provider=body.llm_provider,
+ )
except Exception as e:
logger.error(
@@ -1229,16 +1030,15 @@ async def onboarding(
# Initialize the OpenSearch index if embedding model is configured
if body.embedding_model or body.embedding_provider:
try:
+ from config.settings import IBM_AUTH_ENABLED
+ from config.settings import clients as app_clients
from main import init_index
- from config.settings import IBM_AUTH_ENABLED, clients as app_clients
opensearch_client = None
if IBM_AUTH_ENABLED and user and user.jwt_token:
opensearch_client = app_clients.create_user_opensearch_client(user.jwt_token)
- logger.info(
- "Initializing OpenSearch index after onboarding configuration"
- )
+ logger.info("Initializing OpenSearch index after onboarding configuration")
admin_username = user.user_id if IBM_AUTH_ENABLED and user else None
await init_index(opensearch_client, admin_username=admin_username)
logger.info("OpenSearch index initialization completed successfully")
@@ -1248,9 +1048,7 @@ async def onboarding(
error=str(e),
)
return JSONResponse(
- {
- "error": str(e),
- },
+ {"error": str(e)},
status_code=500,
)
@@ -1262,9 +1060,13 @@ async def onboarding(
if not config_manager.save_config_file(current_config):
logger.error("Failed to save embedding model to config")
- return JSONResponse({"error": "Failed to save configuration"}, status_code=500)
+ return JSONResponse(
+ {"error": "Failed to save configuration"}, status_code=500
+ )
- ingestion_jwt = user.jwt_token if IBM_AUTH_ENABLED and user and user.jwt_token else None
+ ingestion_jwt = (
+ user.jwt_token if IBM_AUTH_ENABLED and user and user.jwt_token else None
+ )
task_id = await ingest_default_documents_when_ready(
document_service,
@@ -1289,12 +1091,10 @@ async def onboarding(
logger.info("Sample data ingestion completed successfully")
except Exception as e:
- logger.error(
- "Failed to complete sample data ingestion", error=str(e)
- )
+ logger.error("Failed to complete sample data ingestion", error=str(e))
return JSONResponse(
{"error": f"Failed to ingest sample documents: {str(e)}"},
- status_code=500
+ status_code=500,
)
if config_manager.save_config_file(current_config):
@@ -1321,23 +1121,18 @@ async def onboarding(
await TelemetryClient.send_event(
Category.ONBOARDING,
MessageId.ORB_ONBOARD_CONFIG_EDITED,
- metadata=onboarding_metadata
+ metadata=onboarding_metadata,
)
await TelemetryClient.send_event(
Category.ONBOARDING,
MessageId.ORB_ONBOARD_COMPLETE,
- metadata=onboarding_metadata
+ metadata=onboarding_metadata,
)
logger.info("Configuration marked as edited after onboarding")
else:
- await TelemetryClient.send_event(
- Category.ONBOARDING,
- MessageId.ORB_ONBOARD_FAILED
- )
- return JSONResponse(
- {"error": "Failed to save configuration"}, status_code=500
- )
+ await TelemetryClient.send_event(Category.ONBOARDING, MessageId.ORB_ONBOARD_FAILED)
+ return JSONResponse({"error": "Failed to save configuration"}, status_code=500)
# Refresh cached patched client so latest credentials take effect immediately
await clients.refresh_patched_client()
@@ -1365,9 +1160,7 @@ async def onboarding(
if not config_manager.save_config_file(current_config):
logger.error("Failed to save openrag_docs_filter_id to config")
except Exception as e:
- logger.error(
- "Failed to create OpenRAG Docs knowledge filter", error=str(e)
- )
+ logger.error("Failed to create OpenRAG Docs knowledge filter", error=str(e))
# Don't fail onboarding if filter creation fails
return OnboardingResponse(
@@ -1380,343 +1173,13 @@ async def onboarding(
except Exception as e:
logger.error("Failed to update onboarding settings", error=str(e))
- await TelemetryClient.send_event(
- Category.ONBOARDING,
- MessageId.ORB_ONBOARD_FAILED
- )
+ await TelemetryClient.send_event(Category.ONBOARDING, MessageId.ORB_ONBOARD_FAILED)
return JSONResponse(
{"error": str(e)},
status_code=500,
)
-async def _create_openrag_docs_filter(
- knowledge_filter_service, session_manager, user
-):
- """Create the OpenRAG Docs knowledge filter for onboarding"""
- import uuid
- import json
- from datetime import datetime
-
- if not knowledge_filter_service:
- logger.error("Knowledge filter service not available")
- return None
-
- # Get JWT token
- jwt_token = user.jwt_token
-
- # Guard against duplicates: check if a filter named "OpenRAG Docs" already exists
- try:
- opensearch_client = session_manager.get_user_opensearch_client(
- user.user_id, jwt_token
- )
- from services.knowledge_filter_service import KNOWLEDGE_FILTERS_INDEX_NAME
- existing = await opensearch_client.search(
- index=KNOWLEDGE_FILTERS_INDEX_NAME,
- body={
- "query": {"match_phrase": {"name": "OpenRAG Docs"}},
- "_source": ["id"],
- "size": 1,
- },
- )
- hits = existing.get("hits", {}).get("hits", [])
- if hits:
- existing_id = hits[0]["_source"].get("id")
- logger.info(
- "OpenRAG Docs filter already exists, skipping creation",
- existing_id=existing_id,
- )
- return existing_id
- except Exception as e:
- logger.warning(
- "Could not check for existing OpenRAG Docs filter, proceeding with creation",
- error=str(e),
- )
-
- # In no-auth mode, set owner to None so filter is visible to all users
- # In auth mode, use the actual user as owner
- if is_no_auth_mode():
- owner_user_id = None
- else:
- owner_user_id = user.user_id
-
- # Create the filter document
- filter_id = str(uuid.uuid4())
- query_data = json.dumps({
- "query": "",
- "filters": {
- # URL-based docs ingestion produces many source URLs.
- # Filter by connector type to target OpenRAG docs only.
- "data_sources": ["*"],
- "document_types": ["*"],
- "owners": ["*"],
- "connector_types": ["openrag_docs"],
- },
- "limit": 10,
- "scoreThreshold": 0,
- "color": "blue",
- "icon": "book",
- })
-
- filter_doc = {
- "id": filter_id,
- "name": "OpenRAG Docs",
- "description": "Filter for OpenRAG documentation",
- "query_data": query_data,
- "owner": owner_user_id,
- "allowed_users": [],
- "allowed_groups": [],
- "created_at": datetime.utcnow().isoformat(),
- "updated_at": datetime.utcnow().isoformat(),
- }
-
- result = await knowledge_filter_service.create_knowledge_filter(
- filter_doc, user_id=user.user_id, jwt_token=jwt_token
- )
-
- if result.get("success"):
- return filter_id
- else:
- logger.error("Failed to create OpenRAG Docs filter", error=result.get("error"))
- return None
-
-
-def _get_flows_service():
- """Helper function to get flows service instance"""
- from services.flows_service import FlowsService
-
- return FlowsService()
-
-
-async def _update_langflow_global_variables(config, flows_service=None):
- """Update Langflow global variables for all configured providers"""
- try:
- # WatsonX global variables
- if config.providers.watsonx.api_key:
- await clients._create_langflow_global_variable(
- "WATSONX_APIKEY", config.providers.watsonx.api_key, modify=True
- )
- logger.info("Set WATSONX_APIKEY global variable in Langflow")
-
- if config.providers.watsonx.project_id:
- await clients._create_langflow_global_variable(
- "WATSONX_PROJECT_ID", config.providers.watsonx.project_id, modify=True
- )
- logger.info("Set WATSONX_PROJECT_ID global variable in Langflow")
-
- if config.providers.watsonx.endpoint:
- await clients._create_langflow_global_variable(
- "WATSONX_URL", config.providers.watsonx.endpoint, modify=True
- )
- logger.info("Set WATSONX_URL global variable in Langflow")
-
- # OpenAI global variables
- if config.providers.openai.api_key:
- await clients._create_langflow_global_variable(
- "OPENAI_API_KEY", config.providers.openai.api_key, modify=True
- )
- logger.info("Set OPENAI_API_KEY global variable in Langflow")
-
- # Anthropic global variables
- if config.providers.anthropic.api_key:
- await clients._create_langflow_global_variable(
- "ANTHROPIC_API_KEY", config.providers.anthropic.api_key, modify=True
- )
- logger.info("Set ANTHROPIC_API_KEY global variable in Langflow")
-
- # Ollama global variables
- if config.providers.ollama.endpoint:
- if not flows_service:
- flows_service = _get_flows_service()
-
- endpoint = await flows_service.resolve_ollama_url(config.providers.ollama.endpoint, force_refresh=True)
- await clients._create_langflow_global_variable(
- "OLLAMA_BASE_URL", endpoint, modify=True
- )
- logger.info("Set OLLAMA_BASE_URL global variable in Langflow")
-
- if config.knowledge.embedding_model:
- await clients._create_langflow_global_variable(
- "SELECTED_EMBEDDING_MODEL", config.knowledge.embedding_model, modify=True
- )
- logger.info(
- f"Set SELECTED_EMBEDDING_MODEL global variable to {config.knowledge.embedding_model}"
- )
-
- except Exception as e:
- logger.error(f"Failed to update Langflow global variables: {str(e)}")
- raise
-
-
-async def _run_async_post_save_langflow_updates(
- session_manager,
- update_mcp_servers: bool,
- update_model_values: bool,
- models_service=None,
-) -> None:
- """Apply post-save Langflow synchronization asynchronously."""
- try:
- current_config = get_openrag_config()
- flows_service = _get_flows_service()
-
- # Refresh model registry so get_litellm_model_name(strict=True) sees the
- # updated provider list — force_remove skips _affected_embedding_models which
- # is the usual registry refresh trigger.
- if models_service is not None:
- await models_service.update_model_registry()
-
- # Update global variables
- await _update_langflow_global_variables(
- current_config, flows_service=flows_service
- )
-
- # Update LLM client credentials when embedding selection changes
- if update_mcp_servers:
- await _update_mcp_server_urls(
- current_config, session_manager, flows_service=flows_service
- )
-
- # Update model values if provider/model changed (including removals/fallbacks)
- if update_model_values:
- await _update_langflow_model_values(
- current_config,
- flows_service,
- llm_model=current_config.agent.llm_model,
- llm_provider=current_config.agent.llm_provider,
- embedding_model=current_config.knowledge.embedding_model,
- embedding_provider=current_config.knowledge.embedding_provider,
- )
-
- logger.info("Completed asynchronous Langflow post-save sync")
- except Exception as e:
- # Do not fail user request if async sync fails; keep parity with existing behavior.
- logger.error(f"Failed to update Langflow settings asynchronously: {str(e)}")
-
-
-async def _update_mcp_server_urls(config, session_manager=None, flows_service=None):
- """Update MCP server URLs (patch localhost and convert to streamable HTTP)."""
- try:
- from services.langflow_mcp_service import LangflowMCPService
-
- mcp_service = LangflowMCPService()
- result = await mcp_service.update_all_mcp_server_urls()
- logger.info("Updated MCP server URLs after settings change", **result)
-
- except Exception as mcp_error:
- logger.warning(f"Failed to update MCP server URLs after settings change: {str(mcp_error)}")
- # Don't fail the entire settings update if MCP update fails
-
-
-async def _update_langflow_model_values(config, flows_service, llm_model=None, llm_provider=None, embedding_model=None, embedding_provider=None):
- """Update model values across Langflow flows for all configured providers"""
- try:
-
- if llm_model or llm_provider:
- effective_llm_provider = (llm_provider or config.agent.llm_provider).lower()
- if llm_provider and llm_provider.lower() != config.agent.llm_provider.lower():
- effective_llm_model = llm_model # do not fall back; force caller to specify
- else:
- effective_llm_model = llm_model or config.agent.llm_model
- result = await flows_service.change_langflow_model_value(
- effective_llm_provider,
- llm_model=effective_llm_model,
- force_llm_update=True
- )
-
- logger.info(
- f"Successfully updated Langflow flows for LLM provider {effective_llm_provider}",
- result=result
- )
-
- if embedding_model or embedding_provider:
- effective_embedding_provider = (embedding_provider or config.knowledge.embedding_provider).lower()
- if embedding_provider and embedding_provider.lower() != config.knowledge.embedding_provider.lower():
- effective_embedding_model = embedding_model # do not fall back; force caller to specify
- else:
- effective_embedding_model = embedding_model or config.knowledge.embedding_model
- result = await flows_service.change_langflow_model_value(
- effective_embedding_provider,
- embedding_model=effective_embedding_model,
- force_embedding_update=True
- )
-
- logger.info(
- f"Successfully updated Langflow flows for embedding provider {effective_embedding_provider}",
- result=result
- )
-
- if not (embedding_model or embedding_provider or llm_model or llm_provider):
- # 2. Update ALL configured embedding providers
- embedding_providers = []
- if config.providers.openai.configured:
- embedding_providers.append("openai")
- if config.providers.watsonx.configured:
- embedding_providers.append("watsonx")
- if config.providers.ollama.configured:
- embedding_providers.append("ollama")
-
- current_embedding_provider = config.knowledge.embedding_provider.lower()
- for provider in embedding_providers:
- # Use configured model for current provider, or None (first available) for others
- embedding_model = (
- config.knowledge.embedding_model
- if provider == current_embedding_provider
- else None
- )
- await flows_service.change_langflow_model_value(
- provider,
- embedding_model=embedding_model,
- force_embedding_update=True
- )
- logger.info(
- f"Successfully updated Langflow flows for embedding provider {provider}"
- )
- except Exception as e:
- logger.error(f"Failed to update Langflow model values: {str(e)}")
- raise
-
-
-async def _update_langflow_system_prompt(config, flows_service):
- """Update system prompt in chat flow"""
- try:
- llm_provider = config.agent.llm_provider.lower()
- await flows_service.update_chat_flow_system_prompt(
- config.agent.system_prompt, llm_provider
- )
- logger.info("Successfully updated chat flow system prompt")
- except Exception as e:
- logger.error(f"Failed to update chat flow system prompt: {str(e)}")
- raise
-
-
-async def _update_langflow_docling_settings(config, flows_service):
- """Update docling settings in ingest flow"""
- try:
- preset_config = get_docling_preset_configs(
- table_structure=config.knowledge.table_structure,
- ocr=config.knowledge.ocr,
- picture_descriptions=config.knowledge.picture_descriptions,
- )
- await flows_service.update_flow_docling_preset("custom", preset_config)
- logger.info("Successfully updated docling settings in ingest flow")
- except Exception as e:
- logger.error(f"Failed to update docling settings: {str(e)}")
- raise
-
-
-async def _update_langflow_chunk_settings(config, flows_service):
- """Update chunk size and overlap in ingest flow"""
- try:
- await flows_service.update_ingest_flow_chunk_size(config.knowledge.chunk_size)
- logger.info(f"Successfully updated ingest flow chunk size to {config.knowledge.chunk_size}")
-
- await flows_service.update_ingest_flow_chunk_overlap(config.knowledge.chunk_overlap)
- logger.info(f"Successfully updated ingest flow chunk overlap to {config.knowledge.chunk_overlap}")
- except Exception as e:
- logger.error(f"Failed to update chunk settings: {str(e)}")
- raise
-
-
async def update_onboarding_state(
body: OnboardingStateBody,
user: User = Depends(require_permission("config:write")),
@@ -1744,9 +1207,7 @@ async def update_onboarding_state(
)
except json.JSONDecodeError:
- return JSONResponse(
- {"error": "Invalid JSON in request body"}, status_code=400
- )
+ return JSONResponse({"error": "Invalid JSON in request body"}, status_code=400)
except Exception as e:
logger.error(f"Error updating onboarding state: {str(e)}")
return JSONResponse(
@@ -1755,57 +1216,9 @@ async def update_onboarding_state(
)
-async def reapply_all_settings(session_manager = None):
- """
- Reapply all current configuration settings to Langflow flows and global variables.
- This is called when flows are detected to have been reset.
- """
- try:
- config = get_openrag_config()
- flows_service = _get_flows_service()
-
- logger.info("Reapplying all settings to Langflow flows and global variables")
-
- # Update MCP server URLs (patch localhost and convert to streamable HTTP)
- await _update_mcp_server_urls(config, session_manager, flows_service=flows_service)
-
- # Update all Langflow settings using helper functions
- try:
- await _update_langflow_global_variables(config, flows_service=flows_service)
- except Exception as e:
- logger.error(f"Failed to update Langflow global variables: {str(e)}")
- # Continue with other updates even if global variables fail
-
- try:
- await _update_langflow_model_values(config, flows_service)
- except Exception as e:
- logger.error(f"Failed to update Langflow model values: {str(e)}")
-
- try:
- await _update_langflow_system_prompt(config, flows_service)
- except Exception as e:
- logger.error(f"Failed to update Langflow system prompt: {str(e)}")
-
- try:
- await _update_langflow_docling_settings(config, flows_service)
- except Exception as e:
- logger.error(f"Failed to update Langflow docling settings: {str(e)}")
-
- try:
- await _update_langflow_chunk_settings(config, flows_service)
- except Exception as e:
- logger.error(f"Failed to update Langflow chunk settings: {str(e)}")
-
- logger.info("Successfully reapplied all settings to Langflow flows")
-
- except Exception as e:
- logger.error(f"Failed to reapply settings: {str(e)}")
- raise
-
-
async def rollback_onboarding(
request: Request,
- body: Optional[RollbackBody] = None,
+ body: RollbackBody | None = None,
session_manager=Depends(get_session_manager),
task_service=Depends(get_task_service),
knowledge_filter_service=Depends(get_knowledge_filter_service),
@@ -1840,7 +1253,8 @@ async def rollback_onboarding(
# Delete knowledge filters created during onboarding
try:
- async def remove_filter(filter_id: Optional[str]):
+
+ async def remove_filter(filter_id: str | None):
if filter_id and knowledge_filter_service:
try:
result = await knowledge_filter_service.delete_knowledge_filter(
@@ -1850,15 +1264,17 @@ async def rollback_onboarding(
logger.info(f"Deleted knowledge filter {filter_id}")
else:
error_msg = result.get("error") if result else "Unknown error"
- logger.warning(f"Could not delete knowledge filter {filter_id}: {error_msg}")
+ logger.warning(
+ f"Could not delete knowledge filter {filter_id}: {error_msg}"
+ )
except Exception as e:
logger.warning(f"Exception deleting knowledge filter {filter_id}: {str(e)}")
- if getattr(current_config.onboarding, 'openrag_docs_filter_id', None):
+ if getattr(current_config.onboarding, "openrag_docs_filter_id", None):
await remove_filter(current_config.onboarding.openrag_docs_filter_id)
current_config.onboarding.openrag_docs_filter_id = None
-
- if getattr(current_config.onboarding, 'user_doc_filter_id', None):
+
+ if getattr(current_config.onboarding, "user_doc_filter_id", None):
await remove_filter(current_config.onboarding.user_doc_filter_id)
current_config.onboarding.user_doc_filter_id = None
except Exception as e:
@@ -1866,6 +1282,7 @@ async def rollback_onboarding(
# Cancel all active tasks and collect successfully ingested files
from session_manager import AnonymousUser
+
anonymous_user_id = AnonymousUser().user_id
for task_data in all_tasks:
@@ -1882,7 +1299,7 @@ async def rollback_onboarding(
except Exception as e:
logger.error(f"Failed to cancel task {task_id}: {str(e)}")
- # Delete all files associated with any task, regardless of whether
+ # Delete all files associated with any task, regardless of whether
# the task failed or completed, to ensure no partial chunks remain in OpenSearch.
files = task_data.get("files", {})
if isinstance(files, dict):
@@ -1894,28 +1311,37 @@ async def rollback_onboarding(
opensearch_client = session_manager.get_user_opensearch_client(
user.user_id, user.jwt_token
)
- from utils.opensearch_queries import build_filename_delete_body
from config.settings import get_index_name
+ from utils.opensearch_queries import (
+ build_filename_delete_body,
+ )
delete_query = build_filename_delete_body(filename)
result = await opensearch_client.delete_by_query(
index=get_index_name(),
body=delete_query,
- conflicts="proceed"
+ conflicts="proceed",
)
deleted_count = result.get("deleted", 0)
if deleted_count > 0:
deleted_files.append(filename)
- logger.info(f"Deleted {deleted_count} chunks for filename {filename}")
+ logger.info(
+ f"Deleted {deleted_count} chunks for filename {filename}"
+ )
except Exception as e:
logger.error(f"Failed to delete documents for {filename}: {str(e)}")
# Wipe the task completely from memory so the frontend doesn't see it anymore
for check_user_id in [user.user_id, anonymous_user_id]:
- if check_user_id in task_service.task_store and task_id in task_service.task_store[check_user_id]:
+ if (
+ check_user_id in task_service.task_store
+ and task_id in task_service.task_store[check_user_id]
+ ):
task_service._task_locks.pop(task_id, None)
task_service.task_store[check_user_id].pop(task_id, None)
- logger.info(f"Purged task {task_id} completely from task_store for user {check_user_id}")
+ logger.info(
+ f"Purged task {task_id} completely from task_store for user {check_user_id}"
+ )
# 4. Reset Langflow flows to their original state
reset_flows_count = 0
@@ -1926,7 +1352,9 @@ async def rollback_onboarding(
reset_flows_count += 1
logger.info(f"Successfully reset {flow_type} flow during rollback")
else:
- logger.warning(f"Failed to reset {flow_type} flow during rollback: {result.get('error')}")
+ logger.warning(
+ f"Failed to reset {flow_type} flow during rollback: {result.get('error')}"
+ )
except Exception as e:
logger.error(f"Error resetting {flow_type} flow during rollback: {e}")
@@ -1963,6 +1391,7 @@ async def rollback_onboarding(
# Save the rolled back configuration manually
try:
import yaml
+
config_file = config_manager.config_file
# Ensure directory exists
@@ -1975,7 +1404,9 @@ async def rollback_onboarding(
# Update cached config
config_manager._config = current_config
- logger.info(f"Successfully saved rolled back configuration with edited={current_config.edited}")
+ logger.info(
+ f"Successfully saved rolled back configuration with edited={current_config.edited}"
+ )
except Exception as e:
logger.error(f"Failed to save rolled back configuration: {e}")
return JSONResponse(
@@ -1987,24 +1418,19 @@ async def rollback_onboarding(
f"Cancelled {len(cancelled_tasks)} tasks, deleted {len(deleted_files)} files, "
f"reset {reset_flows_count} flows, deleted {deleted_conversations_count} conversations"
)
- await TelemetryClient.send_event(
- Category.ONBOARDING,
- MessageId.ORB_ONBOARD_ROLLBACK
- )
+ await TelemetryClient.send_event(Category.ONBOARDING, MessageId.ORB_ONBOARD_ROLLBACK)
return RollbackResponse(
message="Onboarding configuration rolled back successfully",
cancelled_tasks=len(cancelled_tasks),
deleted_files=len(deleted_files),
reset_flows=reset_flows_count,
- deleted_conversations=deleted_conversations_count
+ deleted_conversations=deleted_conversations_count,
)
except Exception as e:
logger.error("Failed to rollback onboarding configuration", error=str(e))
- return JSONResponse(
- {"error": f"Failed to rollback onboarding: {str(e)}"}, status_code=500
- )
+ return JSONResponse({"error": f"Failed to rollback onboarding: {str(e)}"}, status_code=500)
async def update_docling_preset(
@@ -2044,16 +1470,20 @@ async def update_docling_preset(
if preset not in preset_map:
raise HTTPException(
status_code=400,
- detail=f"Invalid preset '{preset}'. Valid presets: {', '.join(preset_map.keys())}"
+ detail=f"Invalid preset '{preset}'. Valid presets: {', '.join(preset_map.keys())}",
)
settings_toggles = preset_map[preset]
else:
# Support new toggle-based API
settings_toggles = {
- "table_structure": body.table_structure if body.table_structure is not None else False,
+ "table_structure": (
+ body.table_structure if body.table_structure is not None else False
+ ),
"ocr": body.ocr if body.ocr is not None else False,
- "picture_descriptions": body.picture_descriptions if body.picture_descriptions is not None else False,
+ "picture_descriptions": (
+ body.picture_descriptions if body.picture_descriptions is not None else False
+ ),
}
# Get the preset configuration
@@ -2074,7 +1504,9 @@ async def update_docling_preset(
except Exception as e:
logger.error("Failed to update docling settings", error=str(e))
- raise HTTPException(status_code=500, detail=f"Failed to update docling settings: {str(e)}")
+ raise HTTPException(
+ status_code=500, detail=f"Failed to update docling settings: {str(e)}"
+ ) from e
async def refresh_openrag_docs(
@@ -2111,4 +1543,4 @@ async def refresh_openrag_docs(
raise HTTPException(
status_code=500,
detail=f"Failed to refresh OpenRAG docs: {str(e)}",
- )
+ ) from e
diff --git a/src/api/settings/helpers.py b/src/api/settings/helpers.py
new file mode 100644
index 000000000..5464708e4
--- /dev/null
+++ b/src/api/settings/helpers.py
@@ -0,0 +1,192 @@
+"""General helpers for the settings/onboarding endpoints.
+
+Provider-fallback selection, embedding-model conflict detection (used to
+warn before removing a provider whose embeddings are still indexed), the
+OpenRAG-docs knowledge-filter creator, and a flows-service factory.
+
+Lifted verbatim from the original `src/api/settings.py` (lines 371–480 and
+1388–1455). No behavior change.
+"""
+
+from typing import Any
+
+from fastapi.responses import JSONResponse
+
+from config.settings import clients, is_no_auth_mode
+from utils.logging_config import get_logger
+
+logger = get_logger(__name__)
+
+
+def _first_configured_llm_provider(config, excluding: str) -> str:
+ """Return the first configured LLM provider that isn't `excluding`."""
+ for p in ["openai", "anthropic", "watsonx", "ollama"]:
+ if p != excluding and getattr(config.providers, p).configured:
+ return p
+ return "openai"
+
+
+def _first_configured_embedding_provider(config, excluding: str) -> str:
+ """Return the first configured embedding provider (openai/watsonx/ollama) that isn't `excluding`."""
+ for p in ["openai", "watsonx", "ollama"]:
+ if p != excluding and getattr(config.providers, p).configured:
+ return p
+ return "openai"
+
+
+async def _affected_embedding_models(
+ provider: str,
+ session_manager,
+ user,
+ models_service,
+) -> list[dict[str, Any]]:
+ """Find embedding models present in the corpus that belong to ``provider``.
+
+ Used to warn users before they remove a provider whose embedding models
+ were used to index documents — otherwise semantic search silently breaks
+ for those docs. Returns a list of ``{"model": str, "doc_count": int}``.
+
+ Conservative on errors (returns empty list) so infra issues don't block
+ provider removal.
+ """
+ from config.settings import get_index_name
+ from services.models_service import ModelsService
+
+ provider_lower = provider.lower()
+ if provider_lower == "anthropic":
+ # Anthropic doesn't serve embedding models — nothing to warn about.
+ return []
+
+ try:
+ # Refresh so the registry reflects currently-configured providers
+ # before we use it to attribute models.
+ await models_service.update_model_registry()
+ registry = ModelsService._model_provider_registry
+
+ # Use the admin client so DLS does not scope the aggregation to the
+ # requesting user's documents. Provider removal is a global operation
+ # that affects all tenants, so we must see every document's embedding
+ # model regardless of ownership.
+ agg_result = await clients.opensearch.search(
+ index=get_index_name(),
+ body={
+ "size": 0,
+ "aggs": {"embedding_models": {"terms": {"field": "embedding_model", "size": 50}}},
+ },
+ params={"terminate_after": 0},
+ )
+ buckets = agg_result.get("aggregations", {}).get("embedding_models", {}).get("buckets", [])
+
+ affected: list[dict[str, Any]] = []
+ for bucket in buckets:
+ model = bucket.get("key")
+ if not model:
+ continue
+ mapped = registry.get(model)
+ # Narrow fallback: the watsonx registry bootstrap requires the
+ # provider still be configured, so models from an about-to-be-
+ # removed watsonx can still be attributed via the "ibm/" prefix.
+ if mapped is None and provider_lower == "watsonx" and model.startswith("ibm/"):
+ mapped = "watsonx"
+ if mapped == provider_lower:
+ affected.append({"model": model, "doc_count": bucket.get("doc_count", 0)})
+ return affected
+ except Exception as e:
+ logger.warning(
+ "Could not determine affected embedding models for provider removal",
+ provider=provider,
+ error=str(e),
+ )
+ return []
+
+
+def _embedding_conflict_response(
+ provider_label: str, provider_key: str, affected: list[dict[str, Any]]
+) -> JSONResponse:
+ """Shared 409 response when removing a provider whose embedding models are
+ still referenced by indexed documents."""
+ model_names = [a["model"] for a in affected]
+ return JSONResponse(
+ {
+ "error": (
+ f"Removing {provider_label} will disable semantic search on "
+ f"documents indexed with: {', '.join(model_names)}. "
+ f"Re-ingest affected documents with another embedding model, "
+ f"or retry with force_remove=true to proceed anyway."
+ ),
+ "code": "embedding_provider_in_use",
+ "affected_provider": provider_key,
+ "affected_models": affected,
+ },
+ status_code=409,
+ )
+
+
+async def _create_openrag_docs_filter(knowledge_filter_service, session_manager, user):
+ """Create the OpenRAG Docs knowledge filter for onboarding"""
+ import json
+ import uuid
+ from datetime import datetime
+
+ if not knowledge_filter_service:
+ logger.error("Knowledge filter service not available")
+ return None
+
+ # Get JWT token
+ jwt_token = user.jwt_token
+
+ # In no-auth mode, set owner to None so filter is visible to all users
+ # In auth mode, use the actual user as owner
+ if is_no_auth_mode():
+ owner_user_id = None
+ else:
+ owner_user_id = user.user_id
+
+ # Create the filter document
+ filter_id = str(uuid.uuid4())
+ query_data = json.dumps(
+ {
+ "query": "",
+ "filters": {
+ # URL-based docs ingestion produces many source URLs.
+ # Filter by connector type to target OpenRAG docs only.
+ "data_sources": ["*"],
+ "document_types": ["*"],
+ "owners": ["*"],
+ "connector_types": ["openrag_docs"],
+ },
+ "limit": 10,
+ "scoreThreshold": 0,
+ "color": "blue",
+ "icon": "book",
+ }
+ )
+
+ filter_doc = {
+ "id": filter_id,
+ "name": "OpenRAG Docs",
+ "description": "Filter for OpenRAG documentation",
+ "query_data": query_data,
+ "owner": owner_user_id,
+ "allowed_users": [],
+ "allowed_groups": [],
+ "created_at": datetime.utcnow().isoformat(),
+ "updated_at": datetime.utcnow().isoformat(),
+ }
+
+ result = await knowledge_filter_service.create_knowledge_filter(
+ filter_doc, user_id=user.user_id, jwt_token=jwt_token
+ )
+
+ if result.get("success"):
+ return filter_id
+ else:
+ logger.error("Failed to create OpenRAG Docs filter", error=result.get("error"))
+ return None
+
+
+def _get_flows_service():
+ """Helper function to get flows service instance"""
+ from services.flows_service import FlowsService
+
+ return FlowsService()
diff --git a/src/api/settings/langflow_sync.py b/src/api/settings/langflow_sync.py
new file mode 100644
index 000000000..0c24f12a1
--- /dev/null
+++ b/src/api/settings/langflow_sync.py
@@ -0,0 +1,308 @@
+"""Langflow synchronization helpers for settings/onboarding flows.
+
+Pushes the latest user config (provider creds, model selection, system
+prompt, docling toggles, chunk sizes, MCP server URLs) into Langflow
+flows and global variables. Also exposes `reapply_all_settings`, called
+from `services/startup_orchestrator.py` when flow-reset detection
+triggers a full re-sync.
+
+Lifted verbatim from the original `src/api/settings.py` (lines 46,
+1458–1684, 1725–1770). No behavior change.
+"""
+
+import asyncio
+
+from api.settings.helpers import _get_flows_service
+from config.settings import clients, get_openrag_config
+from services.docling_service import get_docling_preset_configs
+from utils.logging_config import get_logger
+
+logger = get_logger(__name__)
+
+# Strong refs to in-flight async post-save sync tasks so they aren't
+# garbage-collected mid-flight when the originating request returns.
+_background_tasks: set[asyncio.Task] = set()
+
+
+async def _update_langflow_global_variables(config, flows_service=None):
+ """Update Langflow global variables for all configured providers"""
+ try:
+ # WatsonX global variables
+ if config.providers.watsonx.api_key:
+ await clients._create_langflow_global_variable(
+ "WATSONX_APIKEY", config.providers.watsonx.api_key, modify=True
+ )
+ logger.info("Set WATSONX_APIKEY global variable in Langflow")
+
+ if config.providers.watsonx.project_id:
+ await clients._create_langflow_global_variable(
+ "WATSONX_PROJECT_ID", config.providers.watsonx.project_id, modify=True
+ )
+ logger.info("Set WATSONX_PROJECT_ID global variable in Langflow")
+
+ if config.providers.watsonx.endpoint:
+ await clients._create_langflow_global_variable(
+ "WATSONX_URL", config.providers.watsonx.endpoint, modify=True
+ )
+ logger.info("Set WATSONX_URL global variable in Langflow")
+
+ # OpenAI global variables
+ if config.providers.openai.api_key:
+ await clients._create_langflow_global_variable(
+ "OPENAI_API_KEY", config.providers.openai.api_key, modify=True
+ )
+ logger.info("Set OPENAI_API_KEY global variable in Langflow")
+
+ # Anthropic global variables
+ if config.providers.anthropic.api_key:
+ await clients._create_langflow_global_variable(
+ "ANTHROPIC_API_KEY", config.providers.anthropic.api_key, modify=True
+ )
+ logger.info("Set ANTHROPIC_API_KEY global variable in Langflow")
+
+ # Ollama global variables
+ if config.providers.ollama.endpoint:
+ if not flows_service:
+ flows_service = _get_flows_service()
+
+ endpoint = await flows_service.resolve_ollama_url(
+ config.providers.ollama.endpoint, force_refresh=True
+ )
+ await clients._create_langflow_global_variable("OLLAMA_BASE_URL", endpoint, modify=True)
+ logger.info("Set OLLAMA_BASE_URL global variable in Langflow")
+
+ if config.knowledge.embedding_model:
+ await clients._create_langflow_global_variable(
+ "SELECTED_EMBEDDING_MODEL", config.knowledge.embedding_model, modify=True
+ )
+ logger.info(
+ f"Set SELECTED_EMBEDDING_MODEL global variable to {config.knowledge.embedding_model}"
+ )
+
+ except Exception as e:
+ logger.error(f"Failed to update Langflow global variables: {str(e)}")
+ raise
+
+
+async def _run_async_post_save_langflow_updates(
+ session_manager,
+ update_mcp_servers: bool,
+ update_model_values: bool,
+ models_service=None,
+) -> None:
+ """Apply post-save Langflow synchronization asynchronously."""
+ try:
+ current_config = get_openrag_config()
+ flows_service = _get_flows_service()
+
+ # Refresh model registry so get_litellm_model_name(strict=True) sees the
+ # updated provider list — force_remove skips _affected_embedding_models which
+ # is the usual registry refresh trigger.
+ if models_service is not None:
+ await models_service.update_model_registry()
+
+ # Update global variables
+ await _update_langflow_global_variables(current_config, flows_service=flows_service)
+
+ # Update LLM client credentials when embedding selection changes
+ if update_mcp_servers:
+ await _update_mcp_server_urls(
+ current_config, session_manager, flows_service=flows_service
+ )
+
+ # Update model values if provider/model changed (including removals/fallbacks)
+ if update_model_values:
+ await _update_langflow_model_values(
+ current_config,
+ flows_service,
+ llm_model=current_config.agent.llm_model,
+ llm_provider=current_config.agent.llm_provider,
+ embedding_model=current_config.knowledge.embedding_model,
+ embedding_provider=current_config.knowledge.embedding_provider,
+ )
+
+ logger.info("Completed asynchronous Langflow post-save sync")
+ except Exception as e:
+ # Do not fail user request if async sync fails; keep parity with existing behavior.
+ logger.error(f"Failed to update Langflow settings asynchronously: {str(e)}")
+
+
+async def _update_mcp_server_urls(config, session_manager=None, flows_service=None):
+ """Update MCP server URLs (patch localhost and convert to streamable HTTP)."""
+ try:
+ from services.langflow_mcp_service import LangflowMCPService
+
+ mcp_service = LangflowMCPService()
+ result = await mcp_service.update_all_mcp_server_urls()
+ logger.info("Updated MCP server URLs after settings change", **result)
+
+ except Exception as mcp_error:
+ logger.warning(f"Failed to update MCP server URLs after settings change: {str(mcp_error)}")
+ # Don't fail the entire settings update if MCP update fails
+
+
+async def _update_langflow_model_values(
+ config,
+ flows_service,
+ llm_model=None,
+ llm_provider=None,
+ embedding_model=None,
+ embedding_provider=None,
+):
+ """Update model values across Langflow flows for all configured providers"""
+ try:
+ if llm_model or llm_provider:
+ effective_llm_provider = (llm_provider or config.agent.llm_provider).lower()
+ if llm_provider and llm_provider.lower() != config.agent.llm_provider.lower():
+ effective_llm_model = llm_model # do not fall back; force caller to specify
+ else:
+ effective_llm_model = llm_model or config.agent.llm_model
+ result = await flows_service.change_langflow_model_value(
+ effective_llm_provider, llm_model=effective_llm_model, force_llm_update=True
+ )
+
+ logger.info(
+ f"Successfully updated Langflow flows for LLM provider {effective_llm_provider}",
+ result=result,
+ )
+
+ if embedding_model or embedding_provider:
+ effective_embedding_provider = (
+ embedding_provider or config.knowledge.embedding_provider
+ ).lower()
+ if (
+ embedding_provider
+ and embedding_provider.lower() != config.knowledge.embedding_provider.lower()
+ ):
+ effective_embedding_model = (
+ embedding_model # do not fall back; force caller to specify
+ )
+ else:
+ effective_embedding_model = embedding_model or config.knowledge.embedding_model
+ result = await flows_service.change_langflow_model_value(
+ effective_embedding_provider,
+ embedding_model=effective_embedding_model,
+ force_embedding_update=True,
+ )
+
+ logger.info(
+ f"Successfully updated Langflow flows for embedding provider {effective_embedding_provider}",
+ result=result,
+ )
+
+ if not (embedding_model or embedding_provider or llm_model or llm_provider):
+ # 2. Update ALL configured embedding providers
+ embedding_providers = []
+ if config.providers.openai.configured:
+ embedding_providers.append("openai")
+ if config.providers.watsonx.configured:
+ embedding_providers.append("watsonx")
+ if config.providers.ollama.configured:
+ embedding_providers.append("ollama")
+
+ current_embedding_provider = config.knowledge.embedding_provider.lower()
+ for provider in embedding_providers:
+ # Use configured model for current provider, or None (first available) for others
+ embedding_model = (
+ config.knowledge.embedding_model
+ if provider == current_embedding_provider
+ else None
+ )
+ await flows_service.change_langflow_model_value(
+ provider, embedding_model=embedding_model, force_embedding_update=True
+ )
+ logger.info(
+ f"Successfully updated Langflow flows for embedding provider {provider}"
+ )
+ except Exception as e:
+ logger.error(f"Failed to update Langflow model values: {str(e)}")
+ raise
+
+
+async def _update_langflow_system_prompt(config, flows_service):
+ """Update system prompt in chat flow"""
+ try:
+ llm_provider = config.agent.llm_provider.lower()
+ await flows_service.update_chat_flow_system_prompt(config.agent.system_prompt, llm_provider)
+ logger.info("Successfully updated chat flow system prompt")
+ except Exception as e:
+ logger.error(f"Failed to update chat flow system prompt: {str(e)}")
+ raise
+
+
+async def _update_langflow_docling_settings(config, flows_service):
+ """Update docling settings in ingest flow"""
+ try:
+ preset_config = get_docling_preset_configs(
+ table_structure=config.knowledge.table_structure,
+ ocr=config.knowledge.ocr,
+ picture_descriptions=config.knowledge.picture_descriptions,
+ )
+ await flows_service.update_flow_docling_preset("custom", preset_config)
+ logger.info("Successfully updated docling settings in ingest flow")
+ except Exception as e:
+ logger.error(f"Failed to update docling settings: {str(e)}")
+ raise
+
+
+async def _update_langflow_chunk_settings(config, flows_service):
+ """Update chunk size and overlap in ingest flow"""
+ try:
+ await flows_service.update_ingest_flow_chunk_size(config.knowledge.chunk_size)
+ logger.info(f"Successfully updated ingest flow chunk size to {config.knowledge.chunk_size}")
+
+ await flows_service.update_ingest_flow_chunk_overlap(config.knowledge.chunk_overlap)
+ logger.info(
+ f"Successfully updated ingest flow chunk overlap to {config.knowledge.chunk_overlap}"
+ )
+ except Exception as e:
+ logger.error(f"Failed to update chunk settings: {str(e)}")
+ raise
+
+
+async def reapply_all_settings(session_manager=None):
+ """
+ Reapply all current configuration settings to Langflow flows and global variables.
+ This is called when flows are detected to have been reset.
+ """
+ try:
+ config = get_openrag_config()
+ flows_service = _get_flows_service()
+
+ logger.info("Reapplying all settings to Langflow flows and global variables")
+
+ # Update MCP server URLs (patch localhost and convert to streamable HTTP)
+ await _update_mcp_server_urls(config, session_manager, flows_service=flows_service)
+
+ # Update all Langflow settings using helper functions
+ try:
+ await _update_langflow_global_variables(config, flows_service=flows_service)
+ except Exception as e:
+ logger.error(f"Failed to update Langflow global variables: {str(e)}")
+ # Continue with other updates even if global variables fail
+
+ try:
+ await _update_langflow_model_values(config, flows_service)
+ except Exception as e:
+ logger.error(f"Failed to update Langflow model values: {str(e)}")
+
+ try:
+ await _update_langflow_system_prompt(config, flows_service)
+ except Exception as e:
+ logger.error(f"Failed to update Langflow system prompt: {str(e)}")
+
+ try:
+ await _update_langflow_docling_settings(config, flows_service)
+ except Exception as e:
+ logger.error(f"Failed to update Langflow docling settings: {str(e)}")
+
+ try:
+ await _update_langflow_chunk_settings(config, flows_service)
+ except Exception as e:
+ logger.error(f"Failed to update Langflow chunk settings: {str(e)}")
+
+ logger.info("Successfully reapplied all settings to Langflow flows")
+
+ except Exception as e:
+ logger.error(f"Failed to reapply settings: {str(e)}")
+ raise
diff --git a/src/api/settings/models.py b/src/api/settings/models.py
new file mode 100644
index 000000000..b19d82c72
--- /dev/null
+++ b/src/api/settings/models.py
@@ -0,0 +1,202 @@
+"""Pydantic request and response models for the settings/onboarding endpoints.
+
+Lifted verbatim from the original `src/api/settings.py` (lines 49–223). No
+shape changes — every external caller sees the same fields and validators
+they did before. See `src/api/settings/__init__.py` for re-exports.
+"""
+
+from typing import Any
+
+from pydantic import BaseModel, Field
+
+from services.docling_service import DoclingConfig
+
+
+class SettingsUpdateBody(BaseModel):
+ llm_model: str | None = Field(None, min_length=1)
+ llm_provider: str | None = Field(None, pattern="^(openai|anthropic|watsonx|ollama)$")
+ system_prompt: str | None = None
+ chunk_size: int | None = Field(None, gt=0)
+ chunk_overlap: int | None = Field(None, ge=0)
+ table_structure: bool | None = None
+ ocr: bool | None = None
+ picture_descriptions: bool | None = None
+ embedding_model: str | None = Field(None, min_length=1)
+ embedding_provider: str | None = Field(None, pattern="^(openai|watsonx|ollama)$")
+ index_name: str | None = Field(None, min_length=1)
+ openai_api_key: str | None = Field(None, min_length=1)
+ anthropic_api_key: str | None = Field(None, min_length=1)
+ watsonx_api_key: str | None = Field(None, min_length=1)
+ watsonx_endpoint: str | None = Field(None, min_length=1)
+ watsonx_project_id: str | None = Field(None, min_length=1)
+ ollama_endpoint: str | None = Field(None, min_length=1)
+ remove_ollama_config: bool | None = None
+ remove_openai_config: bool | None = None
+ remove_anthropic_config: bool | None = None
+ remove_watsonx_config: bool | None = None
+ # Explicit confirmation that the caller accepts removing a provider whose
+ # embedding models are still in use by indexed documents. Without this,
+ # the backend returns 409 and the frontend prompts the user.
+ force_remove: bool | None = False
+
+
+class OnboardingBody(BaseModel):
+ llm_provider: str | None = Field(None, pattern="^(openai|anthropic|watsonx|ollama)$")
+ llm_model: str | None = Field(None, min_length=1)
+ embedding_provider: str | None = Field(None, pattern="^(openai|watsonx|ollama)$")
+ embedding_model: str | None = Field(None, min_length=1)
+ openai_api_key: str | None = Field(None, min_length=1)
+ anthropic_api_key: str | None = Field(None, min_length=1)
+ watsonx_api_key: str | None = Field(None, min_length=1)
+ watsonx_endpoint: str | None = Field(None, min_length=1)
+ watsonx_project_id: str | None = Field(None, min_length=1)
+ ollama_endpoint: str | None = Field(None, min_length=1)
+
+
+class AssistantMessage(BaseModel):
+ role: str
+ content: str
+ timestamp: str
+
+
+class OnboardingStateBody(BaseModel):
+ current_step: int | None = None
+ assistant_message: AssistantMessage | None = None
+ selected_nudge: str | None = None
+ card_steps: dict[str, Any] | None = None
+ upload_steps: dict[str, Any] | None = None
+ openrag_docs_filter_id: str | None = None
+ user_doc_filter_id: str | None = None
+ openrag_docs_ingested_version: str | None = None
+ openrag_docs_remote_signature: str | None = None
+
+
+class DoclingPresetBody(BaseModel):
+ preset: str | None = None
+ table_structure: bool | None = None
+ ocr: bool | None = None
+ picture_descriptions: bool | None = None
+
+
+class OnboardingStateConfig(BaseModel):
+ current_step: int | None
+ assistant_message: AssistantMessage | None
+ selected_nudge: str | None
+ card_steps: dict[str, Any] | None
+ upload_steps: dict[str, Any] | None
+ openrag_docs_filter_id: str | None
+ user_doc_filter_id: str | None
+ openrag_docs_ingested_version: str | None
+ openrag_docs_remote_signature: str | None
+
+
+class OpenAIProviderConfig(BaseModel):
+ has_api_key: bool
+ configured: bool
+
+
+class AnthropicProviderConfig(BaseModel):
+ has_api_key: bool
+ configured: bool
+
+
+class WatsonXProviderConfig(BaseModel):
+ has_api_key: bool
+ endpoint: str | None
+ project_id: str | None
+ configured: bool
+
+
+class OllamaProviderConfig(BaseModel):
+ endpoint: str | None
+ configured: bool
+
+
+class ProvidersConfig(BaseModel):
+ openai: OpenAIProviderConfig
+ anthropic: AnthropicProviderConfig
+ watsonx: WatsonXProviderConfig
+ ollama: OllamaProviderConfig
+
+
+class KnowledgeConfig(BaseModel):
+ embedding_model: str | None
+ embedding_provider: str | None
+ chunk_size: int | None
+ chunk_overlap: int | None
+ table_structure: bool | None
+ ocr: bool | None
+ picture_descriptions: bool | None
+ index_name: str | None
+
+
+class AgentConfig(BaseModel):
+ llm_model: str | None
+ llm_provider: str | None
+ system_prompt: str | None
+
+
+class IngestionDefaultsConfig(BaseModel):
+ chunkSize: int | None
+ chunkOverlap: int | None
+ separator: str | None
+ embeddingModel: str | None
+
+
+class SettingsResponse(BaseModel):
+ langflow_url: str
+ flow_id: str | None
+ ingest_flow_id: str | None
+ langflow_public_url: str | None
+ edited: bool
+ onboarding: OnboardingStateConfig
+ providers: ProvidersConfig
+ knowledge: KnowledgeConfig
+ agent: AgentConfig
+ localhost_url: str
+ langflow_edit_url: str | None = None
+ langflow_ingest_edit_url: str | None = None
+ ingestion_defaults: IngestionDefaultsConfig | None = None
+ ingest_via_chat: bool = False
+ segment_write_key: str | None = None
+ environment: str | None = None
+
+
+class OnboardingResponse(BaseModel):
+ message: str
+ edited: bool
+ sample_data_ingested: bool
+ openrag_docs_filter_id: str | None = None
+ task_id: str | None = None
+
+
+class RefreshOpenRAGDocsResponse(BaseModel):
+ message: str
+ refreshed: bool
+
+
+class DoclingPresetResponse(BaseModel):
+ message: str
+ settings: dict
+ preset_config: DoclingConfig
+
+
+class OnboardingStateResponse(BaseModel):
+ message: str
+ updated_fields: list[str]
+
+
+class SettingsUpdateResponse(BaseModel):
+ message: str
+
+
+class RollbackResponse(BaseModel):
+ message: str
+ cancelled_tasks: int
+ deleted_files: int
+ reset_flows: int
+ deleted_conversations: int
+
+
+class RollbackBody(BaseModel):
+ embedding_only: bool = False