webui: preserve partial response on streaming error (ggml-org#23090)

Author: ServeurpersoCom

tools/server/webui/src/lib/constants/agentic.ts

@@ -4,9 +4,6 @@ export const ATTACHMENT_SAVED_REGEX = /\[Attachment saved: ([^\]]+)\]/;
 
 export const NEWLINE_SEPARATOR = '\n';
 
-export const LLM_ERROR_BLOCK_START = '\n\n```\nUpstream LLM error:\n';
-export const LLM_ERROR_BLOCK_END = '\n```\n';
-
 export const DEFAULT_AGENTIC_CONFIG: AgenticConfig = {
 	enabled: true,
 	maxTurns: 100,

tools/server/webui/src/lib/stores/agentic.svelte.ts

@@ -30,12 +30,7 @@ import { ToolSource, ToolPermissionDecision } from '$lib/enums';
 import { SvelteMap } from 'svelte/reactivity';
 import { ToolsService } from '$lib/services/tools.service';
 import { isAbortError } from '$lib/utils';
-import {
-	DEFAULT_AGENTIC_CONFIG,
-	NEWLINE_SEPARATOR,
-	LLM_ERROR_BLOCK_START,
-	LLM_ERROR_BLOCK_END
-} from '$lib/constants';
+import { DEFAULT_AGENTIC_CONFIG, NEWLINE_SEPARATOR } from '$lib/constants';
 import {
 	IMAGE_MIME_TO_EXTENSION,
 	DATA_URI_BASE64_REGEX,
@@ -640,10 +635,9 @@ class AgenticStore {
 					return;
 				}
 				const normalizedError = error instanceof Error ? error : new Error('LLM stream error');
-				// Save error as content in the current turn
-				onChunk?.(`${LLM_ERROR_BLOCK_START}${normalizedError.message}${LLM_ERROR_BLOCK_END}`);
+				// preserve partial output as is, the outer error dialog informs the user separately
 				await onAssistantTurnComplete?.(
-					turnContent + `${LLM_ERROR_BLOCK_START}${normalizedError.message}${LLM_ERROR_BLOCK_END}`,
+					turnContent,
 					turnReasoningContent || undefined,
 					this.buildFinalTimings(capturedTimings, agenticTimings),
 					undefined

tools/server/webui/src/lib/stores/chat.svelte.ts

@@ -814,7 +814,7 @@ class ChatStore {
 					);
 				}
 			},
-			onError: (error: Error) => {
+			onError: async (error: Error) => {
 				this.setStreamingActive(false);
 				if (isAbortError(error)) {
 					cleanupStreamingState();
@@ -826,13 +826,10 @@ class ChatStore {
 					return;
 				}
 				console.error('Streaming error:', error);
+				// keep whatever was streamed so far, the message stays in memory and in DB
+				await this.savePartialResponseIfNeeded(convId);
 				cleanupStreamingState();
 				this.clearPendingMessage(convId);
-				const idx = conversationsStore.findMessageIndex(assistantMessage.id);
-				if (idx !== -1) {
-					const failedMessage = conversationsStore.removeMessageAtIndex(idx);
-					if (failedMessage) DatabaseService.deleteMessage(failedMessage.id).catch(console.error);
-				}
 				const contextInfo = (
 					error as Error & { contextInfo?: { n_prompt_tokens: number; n_ctx: number } }
 				).contextInfo;
@@ -1389,9 +1386,17 @@ class ChatStore {
 						}
 
 						console.error('Continue generation error:', error);
-						conversationsStore.updateMessageAtIndex(idx, { content: originalContent });
-
-						await DatabaseService.updateMessage(msg.id, { content: originalContent });
+						// keep whatever was appended so far, the message stays in memory and in DB
+						await DatabaseService.updateMessage(msg.id, {
+							content: originalContent + appendedContent,
+							reasoningContent: originalReasoning + appendedReasoning || undefined,
+							timestamp: Date.now()
+						});
+						conversationsStore.updateMessageAtIndex(idx, {
+							content: originalContent + appendedContent,
+							reasoningContent: originalReasoning + appendedReasoning || undefined,
+							timestamp: Date.now()
+						});
 
 						this.setChatLoading(msg.convId, false);
 						this.clearChatStreaming(msg.convId);