gmakstutis/openrag
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Merge remote-tracking branch 'origin/fix/chat_filter_clearing' into all-merges

Browse source
This commit is contained in:
Lucas Oliveira 2025-12-05 17:14:23 -03:00
commit 2c9e8c520e
12 changed files with 1952 additions and 1652 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

8
frontend/app/api/queries/useGetConversationsQuery.ts
View file

@ -4,6 +4,7 @@ import {
useQueryClient,
} from "@tanstack/react-query";
import type { EndpointType } from "@/contexts/chat-context";
import { useChat } from "@/contexts/chat-context";
export interface RawConversation {
response_id: string;
@ -50,6 +51,7 @@ export const useGetConversationsQuery = (
options?: Omit<UseQueryOptions, "queryKey" | "queryFn">,
) => {
const queryClient = useQueryClient();
const { isOnboardingComplete } = useChat();
async function getConversations(context: { signal?: AbortSignal }): Promise<ChatConversation[]> {
try {
@ -95,6 +97,11 @@ export const useGetConversationsQuery = (
}
}
// Extract enabled from options and combine with onboarding completion check
// Query is only enabled if onboarding is complete AND the caller's enabled condition is met
const callerEnabled = options?.enabled ?? true;
const enabled = isOnboardingComplete && callerEnabled;
const queryResult = useQuery(
{
queryKey: ["conversations", endpoint, refreshTrigger],
@ -106,6 +113,7 @@ export const useGetConversationsQuery = (
refetchOnMount: false, // Don't refetch on every mount
refetchOnWindowFocus: false, // Don't refetch when window regains focus
...options,
enabled, // Override enabled after spreading options to ensure onboarding check is applied
},
queryClient,
);

15
frontend/app/api/queries/useGetNudgesQuery.ts
View file

@ -3,6 +3,8 @@ import {
useQuery,
useQueryClient,
} from "@tanstack/react-query";
import { useChat } from "@/contexts/chat-context";
import { useProviderHealthQuery } from "./useProviderHealthQuery";
type Nudge = string;
@ -27,6 +29,13 @@ export const useGetNudgesQuery = (
) => {
const { chatId, filters, limit, scoreThreshold } = params ?? {};
const queryClient = useQueryClient();
const { isOnboardingComplete } = useChat();
// Check if LLM provider is healthy
// If health data is not available yet, assume healthy (optimistic)
// Only disable if health data exists and shows LLM error
const { data: health } = useProviderHealthQuery();
const isLLMHealthy = health === undefined || (health?.status === "healthy" && !health?.llm_error);
function cancel() {
queryClient.removeQueries({
@ -77,6 +86,11 @@ export const useGetNudgesQuery = (
}
}
// Extract enabled from options and combine with onboarding completion and LLM health checks
// Query is only enabled if onboarding is complete AND LLM provider is healthy AND the caller's enabled condition is met
const callerEnabled = options?.enabled ?? true;
const enabled = isOnboardingComplete && isLLMHealthy && callerEnabled;
const queryResult = useQuery(
{
queryKey: ["nudges", chatId, filters, limit, scoreThreshold],
@ -91,6 +105,7 @@ export const useGetNudgesQuery = (
return Array.isArray(data) && data.length === 0 ? 5000 : false;
},
...options,
enabled, // Override enabled after spreading options to ensure onboarding check is applied
},
queryClient,
);

26
frontend/app/api/queries/useProviderHealthQuery.ts
View file

@ -5,6 +5,7 @@ import {
} from "@tanstack/react-query";
import { useChat } from "@/contexts/chat-context";
import { useGetSettingsQuery } from "./useGetSettingsQuery";
import { useGetTasksQuery } from "./useGetTasksQuery";
export interface ProviderHealthDetails {
llm_model: string;
@ -40,11 +41,20 @@ export const useProviderHealthQuery = (
) => {
const queryClient = useQueryClient();
// Get chat error state from context (ChatProvider wraps the entire app in layout.tsx)
const { hasChatError, setChatError } = useChat();
// Get chat error state and onboarding completion from context (ChatProvider wraps the entire app in layout.tsx)
const { hasChatError, setChatError, isOnboardingComplete } = useChat();
const { data: settings = {} } = useGetSettingsQuery();
// Check if there are any active ingestion tasks
const { data: tasks = [] } = useGetTasksQuery();
const hasActiveIngestion = tasks.some(
(task) =>
task.status === "pending" ||
task.status === "running" ||
task.status === "processing",
);
async function checkProviderHealth(): Promise<ProviderHealthResponse> {
try {
const url = new URL("/api/provider/health", window.location.origin);
@ -55,6 +65,7 @@ export const useProviderHealthQuery = (
}
// Add test_completion query param if specified or if chat error exists
// Use the same testCompletion value that's in the queryKey
const testCompletion = params?.test_completion ?? hasChatError;
if (testCompletion) {
url.searchParams.set("test_completion", "true");
@ -101,7 +112,10 @@ export const useProviderHealthQuery = (
}
}
const queryKey = ["provider", "health", params?.test_completion];
// Include hasChatError in queryKey so React Query refetches when it changes
// This ensures the health check runs with test_completion=true when chat errors occur
const testCompletion = params?.test_completion ?? hasChatError;
const queryKey = ["provider", "health", testCompletion, hasChatError];
const failureCountKey = queryKey.join("-");
const queryResult = useQuery(
@ -143,7 +157,11 @@ export const useProviderHealthQuery = (
refetchOnWindowFocus: false, // Disabled to reduce unnecessary calls on tab switches
refetchOnMount: true,
staleTime: 30000, // Consider data stale after 30 seconds
enabled: !!settings?.edited && options?.enabled !== false, // Only run after onboarding is complete
enabled:
!!settings?.edited &&
isOnboardingComplete &&
!hasActiveIngestion && // Disable health checks when ingestion is happening
options?.enabled !== false, // Only run after onboarding is complete
...options,
},
queryClient,

45
frontend/app/chat/page.tsx
View file

@ -10,6 +10,7 @@ import { useTask } from "@/contexts/task-context";
import { useChatStreaming } from "@/hooks/useChatStreaming";
import { FILE_CONFIRMATION, FILES_REGEX } from "@/lib/constants";
import { useLoadingStore } from "@/stores/loadingStore";
import { useGetConversationsQuery } from "../api/queries/useGetConversationsQuery";
import { useGetNudgesQuery } from "../api/queries/useGetNudgesQuery";
import { AssistantMessage } from "./_components/assistant-message";
import { ChatInput, type ChatInputHandle } from "./_components/chat-input";
@ -36,6 +37,7 @@ function ChatPage() {
forkFromResponse,
refreshConversations,
refreshConversationsSilent,
refreshTrigger,
previousResponseIds,
setPreviousResponseIds,
placeholderConversation,
@ -71,6 +73,14 @@ function ChatPage() {
const lastLoadedConversationRef = useRef<string | null>(null);
const { addTask } = useTask();
console.log(endpoint, refreshTrigger);
// Check if chat history is loading
const { isLoading: isConversationsLoading } = useGetConversationsQuery(
endpoint,
refreshTrigger,
);
// Use conversation-specific filter instead of global filter
const selectedFilter = conversationFilter;
@ -116,7 +126,12 @@ function ChatPage() {
if (conversationFilter && typeof window !== "undefined") {
const newKey = `conversation_filter_${responseId}`;
localStorage.setItem(newKey, conversationFilter.id);
console.log("[CHAT] Saved filter association:", newKey, "=", conversationFilter.id);
console.log(
"[CHAT] Saved filter association:",
newKey,
"=",
conversationFilter.id,
);
}
}
},
@ -507,6 +522,9 @@ function ChatPage() {
setTimeout(() => {
chatInputRef.current?.focusInput();
}, 100);
} else if (!conversationData) {
// No conversation selected (new conversation)
lastLoadedConversationRef.current = null;
}
}, [
conversationData,
@ -677,7 +695,7 @@ function ChatPage() {
scoreThreshold: parsedFilterData?.scoreThreshold ?? 0,
},
{
enabled: isOnboardingComplete, // Only fetch nudges after onboarding is complete
enabled: isOnboardingComplete && !isConversationsLoading, // Only fetch nudges after onboarding is complete AND chat history is not loading
},
);
@ -841,7 +859,12 @@ function ChatPage() {
// Store the response ID if present for this endpoint
if (result.response_id) {
console.log("[DEBUG] Received response_id:", result.response_id, "currentConversationId:", currentConversationId);
console.log(
"[DEBUG] Received response_id:",
result.response_id,
"currentConversationId:",
currentConversationId,
);
setPreviousResponseIds((prev) => ({
...prev,
@ -850,11 +873,16 @@ function ChatPage() {
// If this is a new conversation (no currentConversationId), set it now
if (!currentConversationId) {
console.log("[DEBUG] Setting currentConversationId to:", result.response_id);
console.log(
"[DEBUG] Setting currentConversationId to:",
result.response_id,
);
setCurrentConversationId(result.response_id);
refreshConversations(true);
} else {
console.log("[DEBUG] Existing conversation, doing silent refresh");
console.log(
"[DEBUG] Existing conversation, doing silent refresh",
);
// For existing conversations, do a silent refresh to keep backend in sync
refreshConversationsSilent();
}
@ -863,7 +891,12 @@ function ChatPage() {
if (conversationFilter && typeof window !== "undefined") {
const newKey = `conversation_filter_${result.response_id}`;
localStorage.setItem(newKey, conversationFilter.id);
console.log("[DEBUG] Saved filter association:", newKey, "=", conversationFilter.id);
console.log(
"[DEBUG] Saved filter association:",
newKey,
"=",
conversationFilter.id,
);
}
}
} else {

14
frontend/app/onboarding/_components/ibm-onboarding.tsx
View file

@ -1,8 +1,8 @@
import type { Dispatch, SetStateAction } from "react";
import { useEffect, useState } from "react";
import IBMLogo from "@/components/icons/ibm-logo";
import { LabelInput } from "@/components/label-input";
import { LabelWrapper } from "@/components/label-wrapper";
import IBMLogo from "@/components/icons/ibm-logo";
import { Switch } from "@/components/ui/switch";
import {
Tooltip,
@ -39,14 +39,16 @@ export function IBMOnboarding({
hasEnvApiKey?: boolean;
}) {
const [endpoint, setEndpoint] = useState(
alreadyConfigured ? "" : (existingEndpoint || "https://us-south.ml.cloud.ibm.com"),
alreadyConfigured
? ""
: existingEndpoint || "https://us-south.ml.cloud.ibm.com",
);
const [apiKey, setApiKey] = useState("");
const [getFromEnv, setGetFromEnv] = useState(
hasEnvApiKey && !alreadyConfigured,
);
const [projectId, setProjectId] = useState(
alreadyConfigured ? "" : (existingProjectId || ""),
alreadyConfigured ? "" : existingProjectId || "",
);
const options = [
@ -93,14 +95,12 @@ export function IBMOnboarding({
} = useGetIBMModelsQuery(
{
endpoint: debouncedEndpoint ? debouncedEndpoint : undefined,
apiKey: getFromEnv ? "" : (debouncedApiKey ? debouncedApiKey : undefined),
apiKey: getFromEnv ? "" : debouncedApiKey ? debouncedApiKey : undefined,
projectId: debouncedProjectId ? debouncedProjectId : undefined,
},
{
enabled:
!!debouncedEndpoint ||
!!debouncedApiKey ||
!!debouncedProjectId ||
(!!debouncedEndpoint && !!debouncedApiKey && !!debouncedProjectId) ||
getFromEnv ||
alreadyConfigured,
},

6
frontend/app/onboarding/_components/onboarding-card.tsx
View file

@ -507,7 +507,7 @@ const OnboardingCard = ({
hasEnvApiKey={
currentSettings?.providers?.openai?.has_api_key === true
}
alreadyConfigured={providerAlreadyConfigured}
alreadyConfigured={providerAlreadyConfigured && modelProvider === "openai"}
/>
</TabsContent>
<TabsContent value="watsonx">
@ -517,7 +517,7 @@ const OnboardingCard = ({
setSampleDataset={setSampleDataset}
setIsLoadingModels={setIsLoadingModels}
isEmbedding={isEmbedding}
alreadyConfigured={providerAlreadyConfigured}
alreadyConfigured={providerAlreadyConfigured && modelProvider === "watsonx"}
existingEndpoint={currentSettings?.providers?.watsonx?.endpoint}
existingProjectId={currentSettings?.providers?.watsonx?.project_id}
hasEnvApiKey={currentSettings?.providers?.watsonx?.has_api_key === true}
@ -530,7 +530,7 @@ const OnboardingCard = ({
setSampleDataset={setSampleDataset}
setIsLoadingModels={setIsLoadingModels}
isEmbedding={isEmbedding}
alreadyConfigured={providerAlreadyConfigured}
alreadyConfigured={providerAlreadyConfigured && modelProvider === "ollama"}
existingEndpoint={currentSettings?.providers?.ollama?.endpoint}
/>
</TabsContent>

61
frontend/app/onboarding/_components/onboarding-upload.tsx
View file

@ -21,6 +21,7 @@ const OnboardingUpload = ({ onComplete }: OnboardingUploadProps) => {
const [isUploading, setIsUploading] = useState(false);
const [currentStep, setCurrentStep] = useState<number | null>(null);
const [uploadedFilename, setUploadedFilename] = useState<string | null>(null);
const [uploadedTaskId, setUploadedTaskId] = useState<string | null>(null);
const [shouldCreateFilter, setShouldCreateFilter] = useState(false);
const [isCreatingFilter, setIsCreatingFilter] = useState(false);
@ -43,23 +44,26 @@ const OnboardingUpload = ({ onComplete }: OnboardingUploadProps) => {
// Monitor tasks and call onComplete when file processing is done
useEffect(() => {
if (currentStep === null || !tasks) {
if (currentStep === null || !tasks || !uploadedTaskId) {
return;
}
// Check if there are any active tasks (pending, running, or processing)
const activeTasks = tasks.find(
(task) =>
task.status === "pending" ||
task.status === "running" ||
task.status === "processing",
);
// Find the task by task ID from the upload response
const matchingTask = tasks.find((task) => task.task_id === uploadedTaskId);
// If no active tasks and we have more than 1 task (initial + new upload), complete it
if (
(!activeTasks || (activeTasks.processed_files ?? 0) > 0) &&
tasks.length > 1
) {
// If no matching task found, wait for it to appear
if (!matchingTask) {
return;
}
// Check if the matching task is still active (pending, running, or processing)
const isTaskActive =
matchingTask.status === "pending" ||
matchingTask.status === "running" ||
matchingTask.status === "processing";
// If task is completed or has processed files, complete the onboarding step
if (!isTaskActive || (matchingTask.processed_files ?? 0) > 0) {
// Set to final step to show "Done"
setCurrentStep(STEP_LIST.length);
@ -91,6 +95,7 @@ const OnboardingUpload = ({ onComplete }: OnboardingUploadProps) => {
icon: "file",
});
// Wait for filter creation to complete before proceeding
createFilterMutation
.mutateAsync({
name: displayName,
@ -114,9 +119,16 @@ const OnboardingUpload = ({ onComplete }: OnboardingUploadProps) => {
})
.finally(() => {
setIsCreatingFilter(false);
});
}
// Refetch nudges to get new ones
refetchNudges();
// Wait a bit before completing (after filter is created)
setTimeout(() => {
onComplete();
}, 1000);
});
} else {
// No filter to create, just complete
// Refetch nudges to get new ones
refetchNudges();
@ -125,7 +137,18 @@ const OnboardingUpload = ({ onComplete }: OnboardingUploadProps) => {
onComplete();
}, 1000);
}
}, [tasks, currentStep, onComplete, refetchNudges, shouldCreateFilter, uploadedFilename]);
}
}, [
tasks,
currentStep,
onComplete,
refetchNudges,
shouldCreateFilter,
uploadedFilename,
uploadedTaskId,
createFilterMutation,
isCreatingFilter,
]);
const resetFileInput = () => {
if (fileInputRef.current) {
@ -144,6 +167,11 @@ const OnboardingUpload = ({ onComplete }: OnboardingUploadProps) => {
const result = await uploadFile(file, true, true); // Pass createFilter=true
console.log("Document upload task started successfully");
// Store task ID to track the specific upload task
if (result.taskId) {
setUploadedTaskId(result.taskId);
}
// Store filename and createFilter flag in state to create filter after ingestion succeeds
if (result.createFilter && result.filename) {
setUploadedFilename(result.filename);
@ -176,6 +204,7 @@ const OnboardingUpload = ({ onComplete }: OnboardingUploadProps) => {
// Reset on error
setCurrentStep(null);
setUploadedTaskId(null);
} finally {
setIsUploading(false);
}

16
frontend/components/chat-renderer.tsx
View file

@ -47,8 +47,7 @@ export function ChatRenderer({
refreshConversations,
startNewConversation,
setConversationFilter,
setCurrentConversationId,
setPreviousResponseIds,
setOnboardingComplete,
} = useChat();
// Initialize onboarding state based on local storage and settings
@ -170,12 +169,17 @@ export function ChatRenderer({
localStorage.removeItem(ONBOARDING_UPLOAD_STEPS_KEY);
}
// Clear ALL conversation state so next message starts fresh
await startNewConversation();
// Mark onboarding as complete in context
setOnboardingComplete(true);
// Store the user document filter as default for new conversations and load it
// Store the user document filter as default for new conversations FIRST
// This must happen before startNewConversation() so the filter is available
await storeDefaultFilterForNewConversations(true);
// Clear ALL conversation state so next message starts fresh
// This will pick up the default filter we just set
await startNewConversation();
// Clean up onboarding filter IDs now that we've set the default
if (typeof window !== "undefined") {
localStorage.removeItem(ONBOARDING_OPENRAG_DOCS_FILTER_ID_KEY);
@ -202,6 +206,8 @@ export function ChatRenderer({
localStorage.removeItem(ONBOARDING_CARD_STEPS_KEY);
localStorage.removeItem(ONBOARDING_UPLOAD_STEPS_KEY);
}
// Mark onboarding as complete in context
setOnboardingComplete(true);
// Store the OpenRAG docs filter as default for new conversations
storeDefaultFilterForNewConversations(false);
setShowLayout(true);

12
frontend/components/provider-health-banner.tsx
View file

@ -5,8 +5,8 @@ import { useRouter } from "next/navigation";
import { useProviderHealthQuery } from "@/app/api/queries/useProviderHealthQuery";
import type { ModelProvider } from "@/app/settings/_helpers/model-helpers";
import { Banner, BannerIcon, BannerTitle } from "@/components/ui/banner";
import { cn } from "@/lib/utils";
import { useChat } from "@/contexts/chat-context";
import { cn } from "@/lib/utils";
import { Button } from "./ui/button";
interface ProviderHealthBannerProps {
@ -73,8 +73,14 @@ export function ProviderHealthBanner({ className }: ProviderHealthBannerProps) {
let errorMessage: string;
if (llmError && embeddingError) {
// Both have errors - show combined message
errorMessage = health?.message || "Provider validation failed";
// Both have errors - check if they're the same
if (llmError === embeddingError) {
// Same error for both - show once
errorMessage = llmError;
} else {
// Different errors - show both
errorMessage = `${llmError}; ${embeddingError}`;
}
errorProvider = undefined; // Don't link to a specific provider
} else if (llmError) {
// Only LLM has error

61
frontend/contexts/chat-context.tsx
View file

@ -10,6 +10,7 @@ import {
useRef,
useState,
} from "react";
import { ONBOARDING_STEP_KEY } from "@/lib/constants";
export type EndpointType = "chat" | "langflow";
@ -81,6 +82,8 @@ interface ChatContextType {
setConversationFilter: (filter: KnowledgeFilter | null, responseId?: string | null) => void;
hasChatError: boolean;
setChatError: (hasError: boolean) => void;
isOnboardingComplete: boolean;
setOnboardingComplete: (complete: boolean) => void;
}
const ChatContext = createContext<ChatContextType | undefined>(undefined);
@ -112,6 +115,37 @@ export function ChatProvider({ children }: ChatProviderProps) {
useState<KnowledgeFilter | null>(null);
const [hasChatError, setChatError] = useState(false);
// Check if onboarding is complete (onboarding step key should be null)
const [isOnboardingComplete, setIsOnboardingComplete] = useState(() => {
if (typeof window === "undefined") return false;
return localStorage.getItem(ONBOARDING_STEP_KEY) === null;
});
// Sync onboarding completion state with localStorage
useEffect(() => {
const checkOnboarding = () => {
if (typeof window !== "undefined") {
setIsOnboardingComplete(
localStorage.getItem(ONBOARDING_STEP_KEY) === null,
);
}
};
// Check on mount
checkOnboarding();
// Listen for storage events (for cross-tab sync)
window.addEventListener("storage", checkOnboarding);
return () => {
window.removeEventListener("storage", checkOnboarding);
};
}, []);
const setOnboardingComplete = useCallback((complete: boolean) => {
setIsOnboardingComplete(complete);
}, []);
// Listen for ingestion failures and set chat error flag
useEffect(() => {
const handleIngestionFailed = () => {
@ -228,6 +262,10 @@ export function ChatProvider({ children }: ChatProviderProps) {
const startNewConversation = useCallback(async () => {
console.log("[CONVERSATION] Starting new conversation");
// Check if there's existing conversation data - if so, this is a manual "new conversation" action
// Check state values before clearing them
const hasExistingConversation = conversationData !== null || placeholderConversation !== null;
// Clear current conversation data and reset state
setCurrentConversationId(null);
setPreviousResponseIds({ chat: null, langflow: null });
@ -261,15 +299,22 @@ export function ChatProvider({ children }: ChatProviderProps) {
setConversationFilterState(null);
}
} else {
console.log("[CONVERSATION] No default filter set");
// No default filter in localStorage
if (hasExistingConversation) {
// User is manually starting a new conversation - clear the filter
console.log("[CONVERSATION] Manual new conversation - clearing filter");
setConversationFilterState(null);
}
} else {
setConversationFilterState(null);
// First time after onboarding - preserve existing filter if set
// This prevents clearing the filter when startNewConversation is called multiple times during onboarding
console.log("[CONVERSATION] No default filter set, preserving existing filter if any");
// Don't clear the filter - it may have been set by storeDefaultFilterForNewConversations
}
}
}
// Create a temporary placeholder conversation to show in sidebar
const placeholderConversation: ConversationData = {
const newPlaceholderConversation: ConversationData = {
response_id: "new-conversation-" + Date.now(),
title: "New conversation",
endpoint: endpoint,
@ -284,10 +329,10 @@ export function ChatProvider({ children }: ChatProviderProps) {
last_activity: new Date().toISOString(),
};
setPlaceholderConversation(placeholderConversation);
setPlaceholderConversation(newPlaceholderConversation);
// Force immediate refresh to ensure sidebar shows correct state
refreshConversations(true);
}, [endpoint, refreshConversations]);
}, [endpoint, refreshConversations, conversationData, placeholderConversation]);
const addConversationDoc = useCallback((filename: string) => {
setConversationDocs((prev) => [
@ -375,6 +420,8 @@ export function ChatProvider({ children }: ChatProviderProps) {
setConversationFilter,
hasChatError,
setChatError,
isOnboardingComplete,
setOnboardingComplete,
}),
[
endpoint,
@ -396,6 +443,8 @@ export function ChatProvider({ children }: ChatProviderProps) {
conversationFilter,
setConversationFilter,
hasChatError,
isOnboardingComplete,
setOnboardingComplete,
],
);

3
frontend/lib/upload-utils.ts
View file

@ -12,6 +12,7 @@ export interface UploadFileResult {
raw: unknown;
createFilter?: boolean;
filename?: string;
taskId?: string;
}
export async function duplicateCheck(
@ -158,6 +159,7 @@ export async function uploadFile(
(uploadIngestJson as { upload?: { id?: string } }).upload?.id ||
(uploadIngestJson as { id?: string }).id ||
(uploadIngestJson as { task_id?: string }).task_id;
const taskId = (uploadIngestJson as { task_id?: string }).task_id;
const filePath =
(uploadIngestJson as { upload?: { path?: string } }).upload?.path ||
(uploadIngestJson as { path?: string }).path ||
@ -197,6 +199,7 @@ export async function uploadFile(
raw: uploadIngestJson,
createFilter: shouldCreateFilter,
filename,
taskId,
};
return result;

189
src/api/provider_validation.py
View file

@ -1,5 +1,6 @@
"""Provider validation utilities for testing API keys and models during onboarding."""
import json
import httpx
from utils.container_utils import transform_localhost_url
from utils.logging_config import get_logger
@ -7,6 +8,106 @@ from utils.logging_config import get_logger
logger = get_logger(__name__)
def _parse_json_error_message(error_text: str) -> str:
"""Parse JSON error message and extract just the message field."""
try:
# Try to parse as JSON
error_data = json.loads(error_text)
if isinstance(error_data, dict):
# WatsonX format: {"errors": [{"code": "...", "message": "..."}], ...}
if "errors" in error_data and isinstance(error_data["errors"], list):
errors = error_data["errors"]
if len(errors) > 0 and isinstance(errors[0], dict):
message = errors[0].get("message", "")
if message:
return message
code = errors[0].get("code", "")
if code:
return f"Error: {code}"
# OpenAI format: {"error": {"message": "...", "type": "...", "code": "..."}}
if "error" in error_data:
error_obj = error_data["error"]
if isinstance(error_obj, dict):
message = error_obj.get("message", "")
if message:
return message
# Direct message field
if "message" in error_data:
return error_data["message"]
# Generic format: {"detail": "..."}
if "detail" in error_data:
return error_data["detail"]
except (json.JSONDecodeError, ValueError, TypeError):
pass
# Return original text if not JSON or can't parse
return error_text
def _extract_error_details(response: httpx.Response) -> str:
"""Extract detailed error message from API response."""
try:
# Try to parse JSON error response
error_data = response.json()
# Common error response formats
if isinstance(error_data, dict):
# WatsonX format: {"errors": [{"code": "...", "message": "..."}], ...}
if "errors" in error_data and isinstance(error_data["errors"], list):
errors = error_data["errors"]
if len(errors) > 0 and isinstance(errors[0], dict):
# Extract just the message from the first error
message = errors[0].get("message", "")
if message:
return message
# Fallback to code if no message
code = errors[0].get("code", "")
if code:
return f"Error: {code}"
# OpenAI format: {"error": {"message": "...", "type": "...", "code": "..."}}
if "error" in error_data:
error_obj = error_data["error"]
if isinstance(error_obj, dict):
message = error_obj.get("message", "")
error_type = error_obj.get("type", "")
code = error_obj.get("code", "")
if message:
details = message
if error_type:
details += f" (type: {error_type})"
if code:
details += f" (code: {code})"
return details
# Anthropic format: {"error": {"message": "...", "type": "..."}}
if "message" in error_data:
return error_data["message"]
# Generic format: {"message": "..."}
if "detail" in error_data:
return error_data["detail"]
# If JSON parsing worked but no structured error found, try parsing text
response_text = response.text[:500]
parsed = _parse_json_error_message(response_text)
if parsed != response_text:
return parsed
return response_text
except (json.JSONDecodeError, ValueError):
# If JSON parsing fails, try parsing the text as JSON string
response_text = response.text[:500] if response.text else f"HTTP {response.status_code}"
parsed = _parse_json_error_message(response_text)
if parsed != response_text:
return parsed
return response_text
async def validate_provider_setup(
provider: str,
api_key: str = None,
@ -30,7 +131,7 @@ async def validate_provider_setup(
If False, performs lightweight validation (no credits consumed). Default: False.
Raises:
Exception: If validation fails with message "Setup failed, please try again or select a different provider."
Exception: If validation fails, raises the original exception with the actual error message.
"""
provider_lower = provider.lower()
@ -70,7 +171,8 @@ async def validate_provider_setup(
except Exception as e:
logger.error(f"Validation failed for provider {provider_lower}: {str(e)}")
raise Exception("Setup failed, please try again or select a different provider.")
# Preserve the original error message instead of replacing it with a generic one
raise
async def test_lightweight_health(
@ -155,8 +257,9 @@ async def _test_openai_lightweight_health(api_key: str) -> None:
)
if response.status_code != 200:
logger.error(f"OpenAI lightweight health check failed: {response.status_code}")
raise Exception(f"OpenAI API key validation failed: {response.status_code}")
error_details = _extract_error_details(response)
logger.error(f"OpenAI lightweight health check failed: {response.status_code} - {error_details}")
raise Exception(f"OpenAI API key validation failed: {error_details}")
logger.info("OpenAI lightweight health check passed")
@ -225,8 +328,9 @@ async def _test_openai_completion_with_tools(api_key: str, llm_model: str) -> No
)
if response.status_code != 200:
logger.error(f"OpenAI completion test failed: {response.status_code} - {response.text}")
raise Exception(f"OpenAI API error: {response.status_code}")
error_details = _extract_error_details(response)
logger.error(f"OpenAI completion test failed: {response.status_code} - {error_details}")
raise Exception(f"OpenAI API error: {error_details}")
logger.info("OpenAI completion with tool calling test passed")
@ -260,8 +364,9 @@ async def _test_openai_embedding(api_key: str, embedding_model: str) -> None:
)
if response.status_code != 200:
logger.error(f"OpenAI embedding test failed: {response.status_code} - {response.text}")
raise Exception(f"OpenAI API error: {response.status_code}")
error_details = _extract_error_details(response)
logger.error(f"OpenAI embedding test failed: {response.status_code} - {error_details}")
raise Exception(f"OpenAI API error: {error_details}")
data = response.json()
if not data.get("data") or len(data["data"]) == 0:
@ -300,8 +405,9 @@ async def _test_watsonx_lightweight_health(
)
if token_response.status_code != 200:
logger.error(f"IBM IAM token request failed: {token_response.status_code}")
raise Exception("Failed to authenticate with IBM Watson - invalid API key")
error_details = _extract_error_details(token_response)
logger.error(f"IBM IAM token request failed: {token_response.status_code} - {error_details}")
raise Exception(f"Failed to authenticate with IBM Watson: {error_details}")
bearer_token = token_response.json().get("access_token")
if not bearer_token:
@ -335,8 +441,9 @@ async def _test_watsonx_completion_with_tools(
)
if token_response.status_code != 200:
logger.error(f"IBM IAM token request failed: {token_response.status_code}")
raise Exception("Failed to authenticate with IBM Watson")
error_details = _extract_error_details(token_response)
logger.error(f"IBM IAM token request failed: {token_response.status_code} - {error_details}")
raise Exception(f"Failed to authenticate with IBM Watson: {error_details}")
bearer_token = token_response.json().get("access_token")
if not bearer_token:
@ -388,8 +495,11 @@ async def _test_watsonx_completion_with_tools(
)
if response.status_code != 200:
logger.error(f"IBM Watson completion test failed: {response.status_code} - {response.text}")
raise Exception(f"IBM Watson API error: {response.status_code}")
error_details = _extract_error_details(response)
logger.error(f"IBM Watson completion test failed: {response.status_code} - {error_details}")
# If error_details is still JSON, parse it to extract just the message
parsed_details = _parse_json_error_message(error_details)
raise Exception(f"IBM Watson API error: {parsed_details}")
logger.info("IBM Watson completion with tool calling test passed")
@ -398,6 +508,13 @@ async def _test_watsonx_completion_with_tools(
raise Exception("Request timed out")
except Exception as e:
logger.error(f"IBM Watson completion test failed: {str(e)}")
# If the error message contains JSON, parse it to extract just the message
error_str = str(e)
if "IBM Watson API error: " in error_str:
json_part = error_str.split("IBM Watson API error: ", 1)[1]
parsed_message = _parse_json_error_message(json_part)
if parsed_message != json_part:
raise Exception(f"IBM Watson API error: {parsed_message}")
raise
@ -419,8 +536,9 @@ async def _test_watsonx_embedding(
)
if token_response.status_code != 200:
logger.error(f"IBM IAM token request failed: {token_response.status_code}")
raise Exception("Failed to authenticate with IBM Watson")
error_details = _extract_error_details(token_response)
logger.error(f"IBM IAM token request failed: {token_response.status_code} - {error_details}")
raise Exception(f"Failed to authenticate with IBM Watson: {error_details}")
bearer_token = token_response.json().get("access_token")
if not bearer_token:
@ -450,8 +568,11 @@ async def _test_watsonx_embedding(
)
if response.status_code != 200:
logger.error(f"IBM Watson embedding test failed: {response.status_code} - {response.text}")
raise Exception(f"IBM Watson API error: {response.status_code}")
error_details = _extract_error_details(response)
logger.error(f"IBM Watson embedding test failed: {response.status_code} - {error_details}")
# If error_details is still JSON, parse it to extract just the message
parsed_details = _parse_json_error_message(error_details)
raise Exception(f"IBM Watson API error: {parsed_details}")
data = response.json()
if not data.get("results") or len(data["results"]) == 0:
@ -464,6 +585,13 @@ async def _test_watsonx_embedding(
raise Exception("Request timed out")
except Exception as e:
logger.error(f"IBM Watson embedding test failed: {str(e)}")
# If the error message contains JSON, parse it to extract just the message
error_str = str(e)
if "IBM Watson API error: " in error_str:
json_part = error_str.split("IBM Watson API error: ", 1)[1]
parsed_message = _parse_json_error_message(json_part)
if parsed_message != json_part:
raise Exception(f"IBM Watson API error: {parsed_message}")
raise
@ -483,8 +611,9 @@ async def _test_ollama_lightweight_health(endpoint: str) -> None:
)
if response.status_code != 200:
logger.error(f"Ollama lightweight health check failed: {response.status_code}")
raise Exception(f"Ollama endpoint not responding: {response.status_code}")
error_details = _extract_error_details(response)
logger.error(f"Ollama lightweight health check failed: {response.status_code} - {error_details}")
raise Exception(f"Ollama endpoint not responding: {error_details}")
logger.info("Ollama lightweight health check passed")
@ -537,8 +666,9 @@ async def _test_ollama_completion_with_tools(llm_model: str, endpoint: str) -> N
)
if response.status_code != 200:
logger.error(f"Ollama completion test failed: {response.status_code} - {response.text}")
raise Exception(f"Ollama API error: {response.status_code}")
error_details = _extract_error_details(response)
logger.error(f"Ollama completion test failed: {response.status_code} - {error_details}")
raise Exception(f"Ollama API error: {error_details}")
logger.info("Ollama completion with tool calling test passed")
@ -569,8 +699,9 @@ async def _test_ollama_embedding(embedding_model: str, endpoint: str) -> None:
)
if response.status_code != 200:
logger.error(f"Ollama embedding test failed: {response.status_code} - {response.text}")
raise Exception(f"Ollama API error: {response.status_code}")
error_details = _extract_error_details(response)
logger.error(f"Ollama embedding test failed: {response.status_code} - {error_details}")
raise Exception(f"Ollama API error: {error_details}")
data = response.json()
if not data.get("embedding"):
@ -616,8 +747,9 @@ async def _test_anthropic_lightweight_health(api_key: str) -> None:
)
if response.status_code != 200:
logger.error(f"Anthropic lightweight health check failed: {response.status_code}")
raise Exception(f"Anthropic API key validation failed: {response.status_code}")
error_details = _extract_error_details(response)
logger.error(f"Anthropic lightweight health check failed: {response.status_code} - {error_details}")
raise Exception(f"Anthropic API key validation failed: {error_details}")
logger.info("Anthropic lightweight health check passed")
@ -672,8 +804,9 @@ async def _test_anthropic_completion_with_tools(api_key: str, llm_model: str) ->
)
if response.status_code != 200:
logger.error(f"Anthropic completion test failed: {response.status_code} - {response.text}")
raise Exception(f"Anthropic API error: {response.status_code}")
error_details = _extract_error_details(response)
logger.error(f"Anthropic completion test failed: {response.status_code} - {error_details}")
raise Exception(f"Anthropic API error: {error_details}")
logger.info("Anthropic completion with tool calling test passed")