n9router 0.4.44 → 0.4.46

package/app/open-sse/executors/kiro.js

@@ -1,9 +1,12 @@
 import { BaseExecutor } from "./base.js";
 import { PROVIDERS } from "../config/providers.js";
+import { resolveKiroModel } from "../config/kiroConstants.js";
 import { v4 as uuidv4 } from "uuid";
 import { refreshKiroToken } from "../services/tokenRefresh.js";
 import { proxyAwareFetch } from "../utils/proxyFetch.js";
 import { HTTP_STATUS, DEFAULT_KIRO_RETRY_COUNT, FETCH_CONNECT_TIMEOUT_MS } from "../config/runtimeConfig.js";
+import { SSE_DONE, SSE_HEADERS } from "../utils/sseConstants.js";
+import { getCapabilitiesForModel } from "../providers/capabilities.js";
 
 // SSE keepalive comment. Emitted when raw EventStream bytes arrive but don't yet
 // complete a parseable frame (partial-frame buffering during reasoning prefill).
@@ -38,13 +41,73 @@ export class KiroExecutor extends BaseExecutor {
       "Amz-Sdk-Invocation-Id": uuidv4()
     };
 
-    if (credentials.accessToken) {
+    // API-key auth: the key is stored as accessToken and sent as a bearer token
+    // exactly like an OAuth access token, but with an extra `tokentype: API_KEY`
+    // header so CodeWhisperer treats it as a long-lived API key rather than an
+    // OIDC/social access token. Mirrors the Kiro IDE headless-auth behavior.
+    // Enterprise / Microsoft Entra (external_idp) tokens are OAuth access tokens,
+    // but CodeWhisperer requires TokenType=EXTERNAL_IDP to bind them to profiles.
+    const authMethod = credentials?.providerSpecificData?.authMethod;
+    const isApiKey = authMethod === "api_key";
+    const isExternalIdp = authMethod === "external_idp";
+
+    const apiKey = credentials?.apiKey || (isApiKey ? credentials?.accessToken : null);
+    if (isApiKey && apiKey) {
+      headers["Authorization"] = `Bearer ${apiKey}`;
+      headers["tokentype"] = "API_KEY";
+    } else if (credentials.accessToken) {
       headers["Authorization"] = `Bearer ${credentials.accessToken}`;
+      if (isExternalIdp) {
+        headers["TokenType"] = "EXTERNAL_IDP";
+      }
     }
 
     return headers;
   }
 
+  /**
+   * Auth-aware endpoint ordering.
+   *
+   * API-key Kiro connections store a raw CodeWhisperer credential (validated
+   * against codewhisperer.us-east-1.amazonaws.com via ListAvailableProfiles).
+   * The Kiro IDE gateway (runtime.*.kiro.dev) expects Kiro OIDC/social tokens
+   * and rejects an `tokentype: API_KEY` token with 401/403 — which
+   * BaseExecutor.execute() returns immediately (only 429 / network errors fall
+   * through to the next host). So for api-key auth we must try the *.amazonaws.com
+   * CodeWhisperer hosts FIRST, mirroring the Kiro-Go reference fork which never
+   * routes api-key traffic through kiro.dev. External IdP enterprise tokens also
+   * use the CodeWhisperer surface, with the `TokenType: EXTERNAL_IDP` header.
+   * Other OAuth methods keep the default order (kiro.dev first) since their
+   * tokens are what that gateway accepts.
+   */
+  getOrderedBaseUrls(credentials) {
+    const baseUrls = this.getBaseUrls();
+    const authMethod = credentials?.providerSpecificData?.authMethod;
+    // IAM Identity Center (idc) tokens are AWS SSO access tokens — the same
+    // family as external_idp/api_key. The kiro.dev gateway rejects them with
+    // 403 "bearer token invalid", so they must hit the CodeWhisperer
+    // *.amazonaws.com surface, and in the region the token was minted in
+    // (the baseUrls are hardcoded us-east-1).
+    const isCodeWhispererSurface =
+      authMethod === "api_key" || authMethod === "external_idp" || authMethod === "idc";
+    if (!isCodeWhispererSurface) return baseUrls;
+
+    const region = (credentials?.providerSpecificData?.region || "us-east-1").trim();
+    const regionalize = (u) =>
+      region && region !== "us-east-1" && u.includes("amazonaws.com")
+        ? u.replace(/([a-z]+)\.[a-z0-9-]+\.amazonaws\.com/, `$1.${region}.amazonaws.com`)
+        : u;
+
+    const amazon = baseUrls.filter((u) => u.includes("amazonaws.com")).map(regionalize);
+    const others = baseUrls.filter((u) => !u.includes("amazonaws.com"));
+    return amazon.length > 0 ? [...amazon, ...others] : baseUrls;
+  }
+
+  buildUrl(model, stream, urlIndex = 0, credentials = null) {
+    const baseUrls = this.getOrderedBaseUrls(credentials);
+    return baseUrls[urlIndex] || baseUrls[0] || this.config.baseUrl;
+  }
+
   transformRequest(model, body, stream, credentials) {
     return body;
   }
@@ -64,9 +127,9 @@ export class KiroExecutor extends BaseExecutor {
    * exponential backoff retry (MITM-style) for transient errors.
    */
   async execute({ model, body, stream, credentials, signal, log, proxyOptions = null, streamWatchdogEnabled = true }) {
-    const url = this.buildUrl(model, stream, 0);
+    const url = this.buildUrl(model, stream, 0, credentials);
     const transformedBody = this.transformRequest(model, body, stream, credentials);
-    
+
     // Resolve max retry count: per-account → global setting → default
     const maxRetries = credentials?.kiroRetryCount ?? DEFAULT_KIRO_RETRY_COUNT;
     let retryAttempts = 0;
@@ -113,7 +176,7 @@ export class KiroExecutor extends BaseExecutor {
         if (response.status === 400) {
           let bodyText = "";
           try { bodyText = await response.clone().text(); } catch { /* ignore */ }
-          
+
           if (!this._isRetryable400(bodyText)) {
             // Genuine validation error — don't retry
             return { response, url, headers, transformedBody };
@@ -172,6 +235,8 @@ export class KiroExecutor extends BaseExecutor {
     let chunkIndex = 0;
     const responseId = `chatcmpl-${Date.now()}`;
     const created = Math.floor(Date.now() / 1000);
+    const capabilityModel = resolveKiroModel(model).upstream;
+    const contextWindow = getCapabilitiesForModel("kiro", capabilityModel).contextWindow || 200000;
     const state = {
       endDetected: false,
       finishEmitted: false,
@@ -179,7 +244,8 @@ export class KiroExecutor extends BaseExecutor {
       hasReasoningContent: false,
       reasoningChunkCount: 0,
       toolCallIndex: 0,
-      seenToolIds: new Map()
+      seenToolIds: new Map(),
+      inThinking: false
     };
 
     const transformStream = new TransformStream({
@@ -218,16 +284,45 @@ export class KiroExecutor extends BaseExecutor {
           if (!event) continue;
 
           const eventType = event.headers[":event-type"] || "";
-          
+
           // Track total content length for token estimation
           if (!state.totalContentLength) state.totalContentLength = 0;
           if (!state.contextUsagePercentage) state.contextUsagePercentage = 0;
 
           // Handle assistantResponseEvent
           if (eventType === "assistantResponseEvent" && event.payload?.content) {
-            const content = event.payload.content;
+            let content = event.payload.content;
+
+            // Kiro Claude models can leak <thinking> blocks into the content stream.
+            // We strip these literal tags to prevent duplication, as the reasoning
+            // is already routed correctly via reasoningContentEvent.
+            if (state.inThinking) {
+              if (content.includes("</thinking>")) {
+                state.inThinking = false;
+                const after = content.split("</thinking>").slice(1).join("</thinking>");
+                content = after.startsWith("\n") ? after.substring(1) : after;
+              } else {
+                content = ""; // Drop entirely while inside thinking block
+              }
+            } else if (content.includes("<thinking>")) {
+              state.inThinking = true;
+              if (content.includes("</thinking>")) {
+                state.inThinking = false;
+                const before = content.split("<thinking>")[0];
+                const after = content.split("</thinking>").slice(1).join("</thinking>");
+                content = before + (after.startsWith("\n") ? after.substring(1) : after);
+              } else {
+                content = content.split("<thinking>")[0];
+              }
+            }
+
+            if (!content && state.hasReasoningContent) {
+              // If we stripped everything, skip emitting an empty content chunk
+              continue;
+            }
+
             state.totalContentLength += content.length;
-            
+
             const chunk = {
               id: responseId,
               object: "chat.completion.chunk",
@@ -414,13 +509,24 @@ export class KiroExecutor extends BaseExecutor {
             if (metrics && typeof metrics === 'object') {
               const inputTokens = metrics.inputTokens || 0;
               const outputTokens = metrics.outputTokens || 0;
-              
+              // ponytail: Amazon Q upstream does not expose cache fields today,
+              // but pick up cache_read_input_tokens / cache_creation_input_tokens
+              // if the event shape grows them so cost tracking stays accurate.
+              const cachedTokens = metrics.cacheReadInputTokens || metrics.cache_read_input_tokens || 0;
+              const cacheCreationInputTokens = metrics.cacheCreationInputTokens || metrics.cache_creation_input_tokens || 0;
+
               if (inputTokens > 0 || outputTokens > 0) {
                 state.usage = {
                   prompt_tokens: inputTokens,
                   completion_tokens: outputTokens,
                   total_tokens: inputTokens + outputTokens
                 };
+                // Kiro is Claude-backed: inputTokens EXCLUDES cache (Claude convention),
+                // not inclusive like OpenAI's cached_tokens. Emit cache_read_input_tokens
+                // (not cached_tokens) so canonicalizeUsage takes the Claude fold path and
+                // correctly adds cache back into prompt_tokens instead of undercharging.
+                if (cachedTokens > 0) state.usage.cache_read_input_tokens = cachedTokens;
+                if (cacheCreationInputTokens > 0) state.usage.cache_creation_input_tokens = cacheCreationInputTokens;
               }
             }
           }
@@ -428,27 +534,26 @@ export class KiroExecutor extends BaseExecutor {
           // Emit final chunk only after receiving BOTH meteringEvent AND contextUsageEvent
           if (state.hasMeteringEvent && state.hasContextUsage && !state.finishEmitted) {
             state.finishEmitted = true;
-            
+
             // Estimate tokens if not available from events
             if (!state.usage) {
               // Estimate output tokens from content length
-              const estimatedOutputTokens = state.totalContentLength > 0 
+              const estimatedOutputTokens = state.totalContentLength > 0
                 ? Math.max(1, Math.floor(state.totalContentLength / 4))
                 : 0;
-              
+
               // Estimate input tokens from contextUsagePercentage
-              // Kiro models typically have 200k context window
               const estimatedInputTokens = state.contextUsagePercentage > 0
-                ? Math.floor(state.contextUsagePercentage * 200000 / 100)
+                ? Math.floor(state.contextUsagePercentage * contextWindow / 100)
                 : 0;
-              
+
               state.usage = {
                 prompt_tokens: estimatedInputTokens,
                 completion_tokens: estimatedOutputTokens,
                 total_tokens: estimatedInputTokens + estimatedOutputTokens
               };
             }
-            
+
             const finishChunk = {
               id: responseId,
               object: "chat.completion.chunk",
@@ -460,12 +565,12 @@ export class KiroExecutor extends BaseExecutor {
                 finish_reason: state.hasToolCalls ? "tool_calls" : "stop"
               }]
             };
-            
+
             // Include usage in final chunk if available
             if (state.usage) {
               finishChunk.usage = state.usage;
             }
-            
+
             controller.enqueue(new TextEncoder().encode(`data: ${JSON.stringify(finishChunk)}\n\n`));
           }
         }
@@ -502,24 +607,20 @@ export class KiroExecutor extends BaseExecutor {
         }
 
         // Send final done message
-        controller.enqueue(new TextEncoder().encode("data: [DONE]\n\n"));
+        controller.enqueue(new TextEncoder().encode(SSE_DONE));
       }
     });
 
     // Pipe response body through transform stream
     if (!response.body) {
-      return new Response("data: [DONE]\n\n", { status: response.status, headers: { "Content-Type": "text/event-stream" } });
+      return new Response(SSE_DONE, { status: response.status, headers: { "Content-Type": "text/event-stream" } });
     }
     const transformedStream = response.body.pipeThrough(transformStream);
 
     return new Response(transformedStream, {
       status: response.status,
       statusText: response.statusText,
-      headers: {
-        "Content-Type": "text/event-stream",
-        "Cache-Control": "no-cache",
-        "Connection": "keep-alive"
-      }
+      headers: { ...SSE_HEADERS }
     });
   }
 

package/app/open-sse/executors/mimo-free.js

@@ -0,0 +1,167 @@
+import { BaseExecutor } from "./base.js";
+import { PROVIDERS } from "../config/providers.js";
+import { proxyAwareFetch } from "../utils/proxyFetch.js";
+import { createHash } from "crypto";
+import os from "os";
+
+const BOOTSTRAP_URL = "https://api.xiaomimimo.com/api/free-ai/bootstrap";
+const CHAT_URL = PROVIDERS["mimo-free"].baseUrl;
+const SESSION_AFFINITY_PREFIX = "ses_";
+const SESSION_ID_LENGTH = 24;
+const JWT_FALLBACK_TTL_SEC = 3000;
+const JWT_EXPIRY_BUFFER_MS = 300000;
+const SESSION_CHARS = "abcdefghijklmnopqrstuvwxyz0123456789";
+
+// Anti-abuse gate: upstream rejects requests without a Chrome-like User-Agent with 403 "Illegal access"
+const USER_AGENTS = [
+  "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36",
+  "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36",
+  "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36",
+];
+
+// Anti-abuse gate marker: the free chat endpoint returns 403 "Illegal access"
+// unless a system message contains this exact MiMoCode signature substring.
+export const MIMO_SYSTEM_MARKER =
+  "You are MiMoCode, an interactive CLI tool that helps users with software engineering tasks.";
+
+// In-memory JWT cache (per-process, survives across requests but not restarts)
+let cachedJwt = null;
+let jwtExpiresAt = 0;
+
+// Device fingerprint reused as the bootstrap "client" — stable per machine
+function generateFingerprint() {
+  let username = "unknown-user";
+  try {
+    username = os.userInfo().username;
+  } catch {
+    // ignore
+  }
+  const cpu = (os.cpus()[0]?.model || "unknown-cpu").trim();
+  const seed = `${os.hostname()}|${os.platform()}|${os.arch()}|${cpu}|${username}`;
+  return createHash("sha256").update(seed).digest("hex");
+}
+
+function generateSessionId() {
+  let id = SESSION_AFFINITY_PREFIX;
+  for (let i = 0; i < SESSION_ID_LENGTH; i++) {
+    id += SESSION_CHARS[Math.floor(Math.random() * SESSION_CHARS.length)];
+  }
+  return id;
+}
+
+// Derive expiry from the JWT exp claim; fall back to a fixed TTL when unparseable
+function parseJwtExp(jwt) {
+  try {
+    const payload = JSON.parse(Buffer.from(jwt.split(".")[1], "base64").toString());
+    if (payload.exp) return payload.exp * 1000;
+  } catch {
+    // ignore
+  }
+  return Date.now() + JWT_FALLBACK_TTL_SEC * 1000;
+}
+
+// Ensure the body carries the anti-abuse marker in a system message (idempotent)
+function injectSystemMarker(body) {
+  const messages = body?.messages;
+  if (!Array.isArray(messages)) return body;
+  const hasMarker = messages.some(
+    (m) => m?.role === "system" && typeof m.content === "string" && m.content.includes(MIMO_SYSTEM_MARKER)
+  );
+  if (hasMarker) return body;
+  return { ...body, messages: [{ role: "system", content: MIMO_SYSTEM_MARKER }, ...messages] };
+}
+
+function resetJwtCache() {
+  cachedJwt = null;
+  jwtExpiresAt = 0;
+}
+
+async function bootstrapJwt(proxyOptions = null) {
+  if (cachedJwt && Date.now() < jwtExpiresAt - JWT_EXPIRY_BUFFER_MS) {
+    return cachedJwt;
+  }
+
+  const response = await proxyAwareFetch(BOOTSTRAP_URL, {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      "User-Agent": USER_AGENTS[Math.floor(Math.random() * USER_AGENTS.length)],
+    },
+    body: JSON.stringify({ client: generateFingerprint() }),
+  }, proxyOptions);
+
+  if (!response.ok) {
+    throw new Error(`MiMo bootstrap failed: ${response.status}`);
+  }
+
+  const data = await response.json();
+  if (!data.jwt) {
+    throw new Error("MiMo bootstrap returned no JWT");
+  }
+
+  cachedJwt = data.jwt;
+  jwtExpiresAt = parseJwtExp(data.jwt);
+  return cachedJwt;
+}
+
+export class MimoFreeExecutor extends BaseExecutor {
+  constructor() {
+    super("mimo-free", PROVIDERS["mimo-free"]);
+    this.sessionId = generateSessionId();
+  }
+
+  buildUrl() {
+    return CHAT_URL;
+  }
+
+  buildHeaders(credentials, stream = true) {
+    return {
+      "Content-Type": "application/json",
+      "X-Mimo-Source": "mimocode-cli-free",
+      "User-Agent": USER_AGENTS[Math.floor(Math.random() * USER_AGENTS.length)],
+      "x-session-affinity": this.sessionId,
+      "Accept": stream ? "text/event-stream" : "application/json",
+    };
+  }
+
+  transformRequest(model, body) {
+    return injectSystemMarker(body);
+  }
+
+  async execute({ model, body, stream, credentials, signal, log, proxyOptions = null }) {
+    let jwt;
+    try {
+      jwt = await bootstrapJwt(proxyOptions);
+    } catch (error) {
+      log?.error?.("AUTH", `MiMo bootstrap failed: ${error.message}`);
+      throw error;
+    }
+
+    const url = this.buildUrl();
+    const transformedBody = this.transformRequest(model, body);
+    const headers = { ...this.buildHeaders(credentials, stream), "Authorization": `Bearer ${jwt}` };
+    const bodyStr = JSON.stringify(transformedBody);
+    log?.debug?.("FETCH", `MIMO-FREE → ${url} | body=${bodyStr.length}B`);
+
+    const response = await proxyAwareFetch(url, { method: "POST", headers, body: bodyStr, signal }, proxyOptions);
+
+    // On auth failure, invalidate cache and retry once with a fresh JWT
+    if (response.status === 401 || response.status === 403) {
+      log?.debug?.("AUTH", `MiMo auth failed (${response.status}), re-bootstrapping...`);
+      resetJwtCache();
+      jwt = await bootstrapJwt(proxyOptions);
+      headers["Authorization"] = `Bearer ${jwt}`;
+      const retryResponse = await proxyAwareFetch(url, { method: "POST", headers, body: bodyStr, signal }, proxyOptions);
+      return { response: retryResponse, url, headers, transformedBody };
+    }
+
+    return { response, url, headers, transformedBody };
+  }
+}
+
+export const __test__ = {
+  generateFingerprint, generateSessionId, bootstrapJwt, resetJwtCache, parseJwtExp,
+  injectSystemMarker, MIMO_SYSTEM_MARKER, BOOTSTRAP_URL, CHAT_URL, SESSION_AFFINITY_PREFIX,
+};
+
+export default MimoFreeExecutor;

package/app/open-sse/executors/opencode-go.js

@@ -1,9 +1,17 @@
 import { BaseExecutor } from "./base.js";
 import { PROVIDERS } from "../config/providers.js";
 import { injectReasoningContent } from "../utils/reasoningContentInjector.js";
+import { ANTHROPIC_API_VERSION } from "../providers/shared.js";
 
 // Models that use /zen/go/v1/messages (Anthropic/Claude format + x-api-key auth)
-const CLAUDE_FORMAT_MODELS = new Set(["minimax-m2.5", "minimax-m2.7"]);
+const MESSAGES_FORMAT_MODELS = new Set([
+  "minimax-m3",
+  "minimax-m2.7",
+  "minimax-m2.5",
+  "qwen3.7-max",
+  "qwen3.7-plus",
+  "qwen3.6-plus",
+]);
 
 const BASE = "https://opencode.ai/zen/go/v1";
 
@@ -15,7 +23,7 @@ export class OpenCodeGoExecutor extends BaseExecutor {
   // buildUrl runs before buildHeaders in BaseExecutor.execute, cache model here
   buildUrl(model) {
     this._lastModel = model;
-    return CLAUDE_FORMAT_MODELS.has(model)
+    return MESSAGES_FORMAT_MODELS.has(model)
       ? `${BASE}/messages`
       : `${BASE}/chat/completions`;
   }
@@ -24,9 +32,9 @@ export class OpenCodeGoExecutor extends BaseExecutor {
     const key = credentials?.apiKey || credentials?.accessToken;
     const headers = { "Content-Type": "application/json" };
 
-    if (CLAUDE_FORMAT_MODELS.has(this._lastModel)) {
+    if (MESSAGES_FORMAT_MODELS.has(this._lastModel)) {
       headers["x-api-key"] = key;
-      headers["anthropic-version"] = "2023-06-01";
+      headers["anthropic-version"] = ANTHROPIC_API_VERSION;
     } else {
       headers["Authorization"] = `Bearer ${key}`;
     }

package/app/open-sse/executors/opencode.js

@@ -15,7 +15,7 @@ export class OpenCodeExecutor extends BaseExecutor {
   }
 
   buildUrl(model) {
-    const base = "https://opencode.ai";
+    const base = this.config.baseUrl;
     return MESSAGES_MODELS.has(model)
       ? `${base}/zen/v1/messages`
       : `${base}/zen/v1/chat/completions`;

package/app/open-sse/executors/perplexity-web.js

@@ -1,5 +1,7 @@
 import { BaseExecutor } from "./base.js";
 import { PROVIDERS } from "../config/providers.js";
+import { SSE_DONE, SSE_HEADERS_NO_BUFFER } from "../utils/sseConstants.js";
+import { sseChunk } from "../utils/sse.js";
 
 const PPLX_SSE_ENDPOINT = PROVIDERS["perplexity-web"].baseUrl;
 const PPLX_API_VERSION = "2.18";
@@ -289,10 +291,6 @@ async function* extractContent(eventStream, signal) {
   yield { delta: "", answer: fullAnswer, backendUuid: backendUuid ?? undefined, done: true };
 }
 
-function sseChunk(data) {
-  return `data: ${JSON.stringify(data)}\n\n`;
-}
-
 function buildStreamingResponse(eventStream, model, cid, created, history, currentMsg, signal) {
   const encoder = new TextEncoder();
   return new ReadableStream({
@@ -340,7 +338,7 @@ function buildStreamingResponse(eventStream, model, cid, created, history, curre
           id: cid, object: "chat.completion.chunk", created, model, system_fingerprint: null,
           choices: [{ index: 0, delta: {}, finish_reason: "stop", logprobs: null }],
         })));
-        controller.enqueue(encoder.encode("data: [DONE]\n\n"));
+        controller.enqueue(encoder.encode(SSE_DONE));
 
         sessionStore(history, currentMsg, cleanResponse(fullAnswer), respBackendUuid);
       } catch (err) {
@@ -348,7 +346,7 @@ function buildStreamingResponse(eventStream, model, cid, created, history, curre
           id: cid, object: "chat.completion.chunk", created, model, system_fingerprint: null,
           choices: [{ index: 0, delta: { content: `[Stream error: ${err.message || String(err)}]` }, finish_reason: "stop", logprobs: null }],
         })));
-        controller.enqueue(encoder.encode("data: [DONE]\n\n"));
+        controller.enqueue(encoder.encode(SSE_DONE));
       } finally {
         controller.close();
       }
@@ -493,7 +491,7 @@ export class PerplexityWebExecutor extends BaseExecutor {
       const sseStream = buildStreamingResponse(response.body, model, cid, created, parsed.history, parsed.currentMsg, signal);
       finalResponse = new Response(sseStream, {
         status: 200,
-        headers: { "Content-Type": "text/event-stream", "Cache-Control": "no-cache", "X-Accel-Buffering": "no" },
+        headers: { ...SSE_HEADERS_NO_BUFFER },
       });
     } else {
       finalResponse = await buildNonStreamingResponse(response.body, model, cid, created, parsed.history, parsed.currentMsg, signal);

package/app/open-sse/executors/qoder.js

@@ -20,18 +20,20 @@
  *     different model upstream, so a missing entry is a hard error.
  */
 
-import { qoderEncodeBody } from "@/lib/qoder/encoding.js";
-import { buildCosyHeaders } from "@/lib/qoder/cosy.js";
+import { qoderEncodeBody } from "../shared/qoder/encoding.js";
+import { buildCosyHeaders } from "../shared/qoder/cosy.js";
 import { v4 as uuidv4 } from "uuid";
 import { createHash } from "crypto";
 
 import { BaseExecutor } from "./base.js";
 import { PROVIDERS } from "../config/providers.js";
 import { proxyAwareFetch } from "../utils/proxyFetch.js";
+import { SSE_DONE } from "../utils/sseConstants.js";
+import { FETCH_CONNECT_TIMEOUT_MS } from "../config/runtimeConfig.js";
 import {
   QODER_CHAT_URL_ENCODED,
   QODER_MODEL_MAP,
-} from "@/lib/qoder/constants.js";
+} from "../shared/qoder/constants.js";
 import { getQoderModelConfig, resolveQoderModels } from "../services/qoderModels.js";
 
 /**
@@ -240,7 +242,7 @@ function wrapQoderSSE(response, model) {
 
     const data = trimmed.slice(5).trimStart();
     if (data === "[DONE]") {
-      controller.enqueue(encoder.encode("data: [DONE]\n\n"));
+      controller.enqueue(encoder.encode(SSE_DONE));
       doneEmitted = true;
       return;
     }
@@ -259,13 +261,13 @@ function wrapQoderSSE(response, model) {
         choices: [{ index: 0, delta: { content: `\n[qoder error ${statusVal}: ${truncate(msg, 200)}]` }, finish_reason: "stop" }],
       });
       controller.enqueue(encoder.encode(`data: ${errChunk}\n\n`));
-      controller.enqueue(encoder.encode("data: [DONE]\n\n"));
+      controller.enqueue(encoder.encode(SSE_DONE));
       doneEmitted = true;
       return;
     }
     if (!inner) return;
     if (inner === "[DONE]") {
-      controller.enqueue(encoder.encode("data: [DONE]\n\n"));
+      controller.enqueue(encoder.encode(SSE_DONE));
       doneEmitted = true;
       return;
     }
@@ -300,7 +302,7 @@ function wrapQoderSSE(response, model) {
         buffer = "";
       }
       if (!doneEmitted) {
-        controller.enqueue(encoder.encode("data: [DONE]\n\n"));
+        controller.enqueue(encoder.encode(SSE_DONE));
         doneEmitted = true;
       }
     },
@@ -407,15 +409,21 @@ export class QoderExecutor extends BaseExecutor {
       ...cosyHeaders,
     };
 
+    // Abort if upstream doesn't return response headers within connect timeout.
+    const timeoutMs = this.config?.timeoutMs || FETCH_CONNECT_TIMEOUT_MS;
+    const connectCtrl = new AbortController();
+    const connectTimer = setTimeout(() => connectCtrl.abort(new Error("fetch connect timeout")), timeoutMs);
+    const mergedSignal = signal ? AbortSignal.any([signal, connectCtrl.signal]) : connectCtrl.signal;
+
     let response;
     try {
       response = await proxyAwareFetch(
         url,
-        { method: "POST", headers, body: encodedBodyBuf, signal },
+        { method: "POST", headers, body: encodedBodyBuf, signal: mergedSignal },
         proxyOptions,
       );
-    } catch (err) {
-      throw err;
+    } finally {
+      clearTimeout(connectTimer);
     }
 
     if (!response.ok) {