File size: 6,195 Bytes
f56a29b | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 | /**
* Stateless Chat API Endpoint
*
* POST /api/chat - Send message, receive SSE stream
*
* This endpoint:
* 1. Receives full state from client (messages + storeState)
* 2. Runs single-pass generation
* 3. Streams events as SSE (text deltas + tool calls)
*
* Fully stateless: interruption is handled by the client aborting
* the fetch request, which triggers req.signal on the server side.
*/
import { NextRequest } from 'next/server';
import { statelessGenerate } from '@/lib/orchestration/stateless-generate';
import { isProviderKeyRequired } from '@/lib/ai/providers';
import type { StatelessChatRequest, StatelessEvent } from '@/lib/types/chat';
import { apiError } from '@/lib/server/api-response';
import { createLogger } from '@/lib/logger';
import { resolveModel } from '@/lib/server/resolve-model';
import type { ThinkingConfig } from '@/lib/types/provider';
const log = createLogger('Chat API');
// Allow streaming responses up to 60 seconds
export const maxDuration = 60;
/**
* POST /api/chat
* Send a message and receive SSE stream of generation events
*
* Request body: StatelessChatRequest
* {
* messages: UIMessage[],
* storeState: { stage, scenes, currentSceneId, mode },
* config: { agentIds, sessionType? },
* apiKey: string,
* baseUrl?: string,
* model?: string
* }
*
* Response: SSE stream of StatelessEvent
*/
export async function POST(req: NextRequest) {
const encoder = new TextEncoder();
let chatModel: string | undefined;
let chatMessageCount: number | undefined;
try {
const body: StatelessChatRequest = await req.json();
chatModel = body.model;
chatMessageCount = body.messages?.length;
// Validate required fields
if (!body.messages || !Array.isArray(body.messages)) {
return apiError('MISSING_REQUIRED_FIELD', 400, 'Missing required field: messages');
}
if (!body.storeState) {
return apiError('MISSING_REQUIRED_FIELD', 400, 'Missing required field: storeState');
}
if (!body.config || !body.config.agentIds || body.config.agentIds.length === 0) {
return apiError('MISSING_REQUIRED_FIELD', 400, 'Missing required field: config.agentIds');
}
const {
model: languageModel,
apiKey: resolvedApiKey,
providerId,
} = await resolveModel({
modelString: body.model,
apiKey: body.apiKey,
baseUrl: body.baseUrl,
providerType: body.providerType,
});
if (isProviderKeyRequired(providerId) && !resolvedApiKey) {
return apiError('MISSING_API_KEY', 401, 'API Key is required');
}
log.info('Processing request');
log.info(
`Agents: ${body.config.agentIds.join(', ')}, Messages: ${body.messages.length}, Turn: ${body.directorState?.turnCount ?? 0}`,
);
// Use the native request signal for abort propagation
const signal = req.signal;
// Create SSE stream
const { readable, writable } = new TransformStream();
const writer = writable.getWriter();
// Stream generation in background with heartbeat to prevent connection timeout
const HEARTBEAT_INTERVAL_MS = 15_000;
(async () => {
// Heartbeat: periodically send SSE comments to keep the connection alive.
// Proxies / browsers may close idle SSE connections after 30-120s of silence.
let heartbeatTimer: ReturnType<typeof setInterval> | null = null;
const startHeartbeat = () => {
stopHeartbeat();
heartbeatTimer = setInterval(() => {
try {
writer.write(encoder.encode(`:heartbeat\n\n`)).catch(() => stopHeartbeat());
} catch {
stopHeartbeat();
}
}, HEARTBEAT_INTERVAL_MS);
};
const stopHeartbeat = () => {
if (heartbeatTimer) {
clearInterval(heartbeatTimer);
heartbeatTimer = null;
}
};
try {
startHeartbeat();
// Default: thinking disabled for low-latency chat. UI requests send
// `thinkingConfig`; eval harnesses can still opt in via `thinking`.
const thinkingConfig: ThinkingConfig = body.thinkingConfig ??
body.thinking ?? { mode: 'disabled', enabled: false };
const generator = statelessGenerate(
{
...body,
apiKey: resolvedApiKey,
},
signal,
languageModel,
thinkingConfig,
);
for await (const event of generator) {
if (signal.aborted) {
log.info('Request was aborted');
break;
}
const data = `data: ${JSON.stringify(event)}\n\n`;
await writer.write(encoder.encode(data));
}
stopHeartbeat();
await writer.close();
} catch (error) {
stopHeartbeat();
// If aborted, just close the writer silently
if (signal.aborted) {
log.info('Request aborted during streaming');
try {
await writer.close();
} catch {
/* already closed */
}
return;
}
log.error(
`Chat stream error [model=${body.model ?? 'unknown'}, agents=${body.config?.agentIds?.length ?? 0}, messages=${body.messages?.length ?? 0}]:`,
error,
);
// Try to send error event
try {
const errorEvent: StatelessEvent = {
type: 'error',
data: {
message: error instanceof Error ? error.message : String(error),
},
};
await writer.write(encoder.encode(`data: ${JSON.stringify(errorEvent)}\n\n`));
await writer.close();
} catch {
// Writer may already be closed
}
}
})();
return new Response(readable, {
headers: {
'Content-Type': 'text/event-stream',
'Cache-Control': 'no-cache',
Connection: 'keep-alive',
},
});
} catch (error) {
log.error(
`Chat request failed [model=${chatModel ?? 'unknown'}, messages=${chatMessageCount ?? 0}]:`,
error,
);
return apiError(
'INTERNAL_ERROR',
500,
error instanceof Error ? error.message : 'Failed to process request',
);
}
}
|