nezumi/9router
1
0
Fork 0
mirror of https://github.com/Nezumi-2711/9router.git synced 2026-09-22 20:00:47 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(codex): harden streaming timeouts + Responses terminal events

Browse Source 
Raise stall/connect timeouts to 60s (configurable per-provider), accept
codex response.done, and always emit a terminal response.failed + [DONE]
for Responses passthrough when a stream closes, stalls, or aborts before
a terminal event — preventing codex clients from hanging.

Co-authored-by: jonathanli12 <jonathanli12@users.noreply.github.com>
Co-authored-by: rifuki <rifuki@users.noreply.github.com>
Co-authored-by: nguyenha935 <nguyenha935@users.noreply.github.com>
Co-authored-by: trananhtung <trananhtung@users.noreply.github.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
decolua
2026-06-06 16:07:43 +07:00
co-authored by jonathanli12 rifuki nguyenha935 trananhtung Cursor
parent f161b295a5
commit 9caea88528
15 changed files with 311 additions and 30 deletions
Download Patch File Download Diff File Expand all files Collapse all files
open-sse/utils/stream.js
+61 -3
View File
@@ -3,6 +3,7 @@ import { FORMATS } from "../translator/formats.js";
import { trackPendingRequest, appendRequestLog } from "@/lib/usageDb.js";
import { extractUsage, hasValidUsage, estimateUsage, logUsage, addBufferToUsage, filterUsageForFormat, COLORS } from "./usageTracking.js";
import { parseSSELine, hasValuableContent, fixInvalidId, formatSSE } from "./streamHelpers.js";
import { getOpenAIResponsesEventName, isOpenAIResponsesTerminalEvent, formatIncompleteOpenAIResponsesStreamFailure } from "./responsesStreamHelpers.js";
import { dbg, isDebugEnabled } from "./debugLog.js";
export { COLORS, formatSSE };
@@ -63,6 +64,11 @@ export function createSSEStream(options = {}) {
let sseEmittedCount = 0;
const eventTypeCounts = {};
// Track Responses API event framing for same-format passthrough (codex)
let currentOpenAIResponsesEvent = null;
let openAIResponsesTerminalSeen = false;
let openAIResponsesDoneSent = false;
return new TransformStream({
transform(chunk, controller) {
if (!ttftAt) ttftAt = Date.now();
@@ -83,6 +89,11 @@ export function createSSEStream(options = {}) {
}
}
// Capture Responses API event name to preserve framing in same-format passthrough
if (mode === STREAM_MODE.TRANSLATE && targetFormat === FORMATS.OPENAI_RESPONSES && trimmed.startsWith("event:")) {
currentOpenAIResponsesEvent = trimmed.slice(6).trim();
}
// Passthrough mode: normalize and forward
if (mode === STREAM_MODE.PASSTHROUGH) {
let output;
@@ -174,12 +185,33 @@ export function createSSEStream(options = {}) {
const parsed = parseSSELine(trimmed, targetFormat);
if (!parsed) continue;
// Responses API same-format passthrough: preserve event framing + track terminal state
const isOpenAIResponsesStream = targetFormat === FORMATS.OPENAI_RESPONSES;
const keepsOpenAIResponsesFormat = isOpenAIResponsesStream && sourceFormat === FORMATS.OPENAI_RESPONSES;
const openAIResponsesEventName = isOpenAIResponsesStream
? getOpenAIResponsesEventName(currentOpenAIResponsesEvent, parsed)
: null;
if (isOpenAIResponsesStream && isOpenAIResponsesTerminalEvent(openAIResponsesEventName, parsed)) {
openAIResponsesTerminalSeen = true;
}
// For Ollama: done=true is the final chunk with finish_reason/usage, must translate
// For other formats: done=true is the [DONE] sentinel, skip
if (parsed && parsed.done && targetFormat !== FORMATS.OLLAMA) {
// Synthesize response.failed if the Responses stream never sent a terminal event
if (keepsOpenAIResponsesFormat && !openAIResponsesTerminalSeen) {
const failedOutput = formatIncompleteOpenAIResponsesStreamFailure();
reqLogger?.appendConvertedChunk?.(failedOutput);
controller.enqueue(sharedEncoder.encode(failedOutput));
openAIResponsesTerminalSeen = true;
sseEmittedCount++;
}
const output = "data: [DONE]\n\n";
reqLogger?.appendConvertedChunk?.(output);
controller.enqueue(sharedEncoder.encode(output));
if (keepsOpenAIResponsesFormat) openAIResponsesDoneSent = true;
continue;
}
@@ -224,6 +256,18 @@ export function createSSEStream(options = {}) {
const extracted = extractUsage(parsed);
if (extracted) state.usage = extracted; // Keep original usage for logging
// Responses same-format passthrough: re-emit with original event framing
if (keepsOpenAIResponsesFormat && openAIResponsesEventName) {
const output = formatSSE({ event: openAIResponsesEventName, data: parsed }, sourceFormat);
reqLogger?.appendConvertedChunk?.(output);
controller.enqueue(sharedEncoder.encode(output));
currentOpenAIResponsesEvent = null;
sseEmittedCount++;
continue;
}
currentOpenAIResponsesEvent = null;
// Translate: targetFormat -> openai -> sourceFormat
const translated = translateResponse(targetFormat, sourceFormat, parsed, state);
@@ -237,6 +281,7 @@ export function createSSEStream(options = {}) {
if (translated?.length > 0) {
for (const item of translated) {
if (item === null || item === undefined) continue;
// Filter empty chunks
if (!hasValuableContent(item, sourceFormat)) {
continue; // Skip this empty chunk
@@ -322,6 +367,7 @@ export function createSSEStream(options = {}) {
if (translated?.length > 0) {
for (const item of translated) {
if (item === null || item === undefined) continue;
const output = formatSSE(item, sourceFormat);
reqLogger?.appendConvertedChunk?.(output);
controller.enqueue(sharedEncoder.encode(output));
@@ -341,15 +387,27 @@ export function createSSEStream(options = {}) {
if (flushed?.length > 0) {
for (const item of flushed) {
if (item === null || item === undefined) continue;
const output = formatSSE(item, sourceFormat);
reqLogger?.appendConvertedChunk?.(output);
controller.enqueue(sharedEncoder.encode(output));
}
}
const doneOutput = "data: [DONE]\n\n";
reqLogger?.appendConvertedChunk?.(doneOutput);
controller.enqueue(sharedEncoder.encode(doneOutput));
// Synthesize response.failed if a Responses passthrough stream never reached a terminal event
const keepsOpenAIResponsesFormat = targetFormat === FORMATS.OPENAI_RESPONSES && sourceFormat === FORMATS.OPENAI_RESPONSES;
if (keepsOpenAIResponsesFormat && !openAIResponsesTerminalSeen) {
const failedOutput = formatIncompleteOpenAIResponsesStreamFailure();
reqLogger?.appendConvertedChunk?.(failedOutput);
controller.enqueue(sharedEncoder.encode(failedOutput));
openAIResponsesTerminalSeen = true;
}
if (!keepsOpenAIResponsesFormat || !openAIResponsesDoneSent) {
const doneOutput = "data: [DONE]\n\n";
reqLogger?.appendConvertedChunk?.(doneOutput);
controller.enqueue(sharedEncoder.encode(doneOutput));
}
if (!hasValidUsage(state?.usage) && totalContentLength > 0) {
state.usage = estimateUsage(body, totalContentLength, sourceFormat);