CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-01 09:01:57 +00:00
Code Issues Packages Projects Releases Wiki Activity

webui: remove client-side context pre-check and rely on backend for limits (#16506)

Browse Source 
* fix: make SSE client robust to premature [DONE] in agentic proxy chains

* webui: remove client-side context pre-check and rely on backend for limits

Removed the client-side context window pre-check and now simply sends messages
while keeping the dialog imports limited to core components, eliminating the
maximum context alert path

Simplified streaming and non-streaming chat error handling to surface a generic
'No response received from server' error whenever the backend returns no content

Removed the obsolete maxContextError plumbing from the chat store so state
management now focuses on the core message flow without special context-limit cases

* webui: cosmetic rename of error messages

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

Co-authored-by: Aleksander Grygier <aleksander.grygier@gmail.com>

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

Co-authored-by: Aleksander Grygier <aleksander.grygier@gmail.com>

* Update tools/server/webui/src/lib/components/app/chat/ChatScreen/ChatScreen.svelte

Co-authored-by: Aleksander Grygier <aleksander.grygier@gmail.com>

* Update tools/server/webui/src/lib/components/app/chat/ChatScreen/ChatScreen.svelte

Co-authored-by: Aleksander Grygier <aleksander.grygier@gmail.com>

* chore: update webui build output

---------

Co-authored-by: Aleksander Grygier <aleksander.grygier@gmail.com>
This commit is contained in:
Pascal
2025-10-12 18:06:41 +02:00
committed by GitHub
parent c7be9febcb
commit 81d54bbfd5
11 changed files with 141 additions and 366 deletions
Download Patch File Download Diff File Expand all files Collapse all files

140
tools/server/webui/src/lib/stores/chat.svelte.ts
View File

@@ -39,7 +39,6 @@ import type { ExportedConversations } from '$lib/types/database';
* - Conversation branching for exploring different response paths
* - Streaming AI responses with real-time content updates
* - File attachment support (images, PDFs, text files, audio)
* - Context window management with error recovery
* - Partial response saving when generation is interrupted
* - Message editing with automatic response regeneration
*/
@@ -48,11 +47,9 @@ class ChatStore {
activeMessages = $state<DatabaseMessage[]>([]);
conversations = $state<DatabaseConversation[]>([]);
currentResponse = $state('');
errorDialogState = $state<{ type: 'timeout' | 'server'; message: string } | null>(null);
isInitialized = $state(false);
isLoading = $state(false);
maxContextError = $state<{ message: string; estimatedTokens: number; maxContext: number } | null>(
null
);
titleUpdateConfirmationCallback?: (currentTitle: string, newTitle: string) => Promise<boolean>;
constructor() {
@@ -69,8 +66,6 @@ class ChatStore {
try {
await this.loadConversations();
this.maxContextError = null;
this.isInitialized = true;
} catch (error) {
console.error('Failed to initialize chat store:', error);
@@ -99,8 +94,6 @@ class ChatStore {
this.activeConversation = conversation;
this.activeMessages = [];
this.maxContextError = null;
await goto(`#/chat/${conversation.id}`);
return conversation.id;
@@ -133,8 +126,6 @@ class ChatStore {
this.activeMessages = await DatabaseStore.getConversationMessages(convId);
}
this.maxContextError = null;
return true;
} catch (error) {
console.error('Failed to load conversation:', error);
@@ -418,56 +409,6 @@ class ChatStore {
return;
}
if (error.name === 'ContextError') {
console.warn('Context error detected:', error.message);
this.isLoading = false;
this.currentResponse = '';
const messageIndex = this.activeMessages.findIndex(
(m: DatabaseMessage) => m.id === assistantMessage.id
);
if (messageIndex !== -1) {
this.activeMessages.splice(messageIndex, 1);
DatabaseStore.deleteMessage(assistantMessage.id).catch(console.error);
}
// Use structured context info from new exceed_context_size_error format if available
const contextInfo = (
error as Error & {
contextInfo?: { promptTokens: number; maxContext: number; estimatedTokens: number };
}
).contextInfo;
let estimatedTokens = 0;
let maxContext = serverStore.serverProps?.default_generation_settings.n_ctx || 8192;
if (contextInfo) {
// Use precise token counts from server response
estimatedTokens = contextInfo.promptTokens;
maxContext = contextInfo.maxContext;
} else {
// Fallback to estimation for older error format
try {
// Rough estimation: ~4 characters per token
const messageContent = JSON.stringify(messages);
estimatedTokens = Math.ceil(messageContent.length / 4);
} catch {
estimatedTokens = 0;
}
}
this.maxContextError = {
message: error.message,
estimatedTokens,
maxContext
};
if (onError) {
onError(error);
}
return;
}
console.error('Streaming error:', error);
this.isLoading = false;
this.currentResponse = '';
@@ -477,9 +418,19 @@ class ChatStore {
);
if (messageIndex !== -1) {
this.activeMessages[messageIndex].content = `Error: ${error.message}`;
const [failedMessage] = this.activeMessages.splice(messageIndex, 1);
if (failedMessage) {
DatabaseStore.deleteMessage(failedMessage.id).catch((cleanupError) => {
console.error('Failed to remove assistant message after error:', cleanupError);
});
}
}
const dialogType = error.name === 'TimeoutError' ? 'timeout' : 'server';
this.showErrorDialog(dialogType, error.message);
if (onError) {
onError(error);
}
@@ -487,6 +438,14 @@ class ChatStore {
});
}
private showErrorDialog(type: 'timeout' | 'server', message: string): void {
this.errorDialogState = { type, message };
}
dismissErrorDialog(): void {
this.errorDialogState = null;
}
/**
* Checks if an error is an abort error (user cancelled operation)
* @param error - The error to check
@@ -574,6 +533,7 @@ class ChatStore {
return;
}
this.errorDialogState = null;
this.isLoading = true;
this.currentResponse = '';
@@ -603,37 +563,23 @@ class ChatStore {
const conversationContext = this.activeMessages.slice(0, -1);
await this.streamChatCompletion(
conversationContext,
assistantMessage,
undefined,
(error: Error) => {
if (error.name === 'ContextError' && userMessage) {
const userMessageIndex = this.findMessageIndex(userMessage.id);
if (userMessageIndex !== -1) {
this.activeMessages.splice(userMessageIndex, 1);
DatabaseStore.deleteMessage(userMessage.id).catch(console.error);
}
}
}
);
await this.streamChatCompletion(conversationContext, assistantMessage);
} catch (error) {
if (this.isAbortError(error)) {
this.isLoading = false;
return;
}
if (error instanceof Error && error.name === 'ContextError' && userMessage) {
const userMessageIndex = this.findMessageIndex(userMessage.id);
if (userMessageIndex !== -1) {
this.activeMessages.splice(userMessageIndex, 1);
DatabaseStore.deleteMessage(userMessage.id).catch(console.error);
}
}
console.error('Failed to send message:', error);
this.isLoading = false;
if (!this.errorDialogState) {
if (error instanceof Error) {
const dialogType = error.name === 'TimeoutError' ? 'timeout' : 'server';
this.showErrorDialog(dialogType, error.message);
} else {
this.showErrorDialog('server', 'Unknown error occurred while sending message');
}
}
}
}
@@ -662,24 +608,6 @@ class ChatStore {
this.currentResponse = '';
}
/**
* Clears the max context error state
* Removes any displayed context limit warnings
*/
clearMaxContextError(): void {
this.maxContextError = null;
}
/**
* Sets the max context error state
* @param error - The context error details or null to clear
*/
setMaxContextError(
error: { message: string; estimatedTokens: number; maxContext: number } | null
): void {
this.maxContextError = error;
}
/**
* Saves partial response if generation was interrupted
* Preserves user's partial content and timing data when generation is stopped early
@@ -1250,7 +1178,6 @@ class ChatStore {
this.activeMessages = [];
this.currentResponse = '';
this.isLoading = false;
this.maxContextError = null;
}
/** Refreshes active messages based on currNode after branch navigation */
@@ -1538,6 +1465,7 @@ class ChatStore {
private async generateResponseForMessage(userMessageId: string): Promise<void> {
if (!this.activeConversation) return;
this.errorDialogState = null;
this.isLoading = true;
this.currentResponse = '';
@@ -1584,7 +1512,7 @@ export const activeMessages = () => chatStore.activeMessages;
export const isLoading = () => chatStore.isLoading;
export const currentResponse = () => chatStore.currentResponse;
export const isInitialized = () => chatStore.isInitialized;
export const maxContextError = () => chatStore.maxContextError;
export const errorDialog = () => chatStore.errorDialogState;
export const createConversation = chatStore.createConversation.bind(chatStore);
export const downloadConversation = chatStore.downloadConversation.bind(chatStore);
@@ -1592,9 +1520,9 @@ export const exportAllConversations = chatStore.exportAllConversations.bind(chat
export const importConversations = chatStore.importConversations.bind(chatStore);
export const deleteConversation = chatStore.deleteConversation.bind(chatStore);
export const sendMessage = chatStore.sendMessage.bind(chatStore);
export const dismissErrorDialog = chatStore.dismissErrorDialog.bind(chatStore);
export const gracefulStop = chatStore.gracefulStop.bind(chatStore);
export const clearMaxContextError = chatStore.clearMaxContextError.bind(chatStore);
export const setMaxContextError = chatStore.setMaxContextError.bind(chatStore);
// Branching operations
export const refreshActiveMessages = chatStore.refreshActiveMessages.bind(chatStore);