fix(compact): route combo compact requests through the failover path (rebase of #3109)

src/adapters/openai-responses.ts

@@ -2286,6 +2286,7 @@ export function createResponsesPassthroughAdapter(provider: OcxProviderConfig):
       let doneText = "";
       let snapshot = "";
       let usage: OcxUsage | undefined;
+      let compactionEncryptedContent: string | undefined;
       for await (const event of decodeServerSentEvents(response.body, { translatorBudget: budget })) {
         let payload: unknown;
         try { payload = JSON.parse(event.data); } catch { continue; }
@@ -2320,6 +2321,17 @@ export function createResponsesPassthroughAdapter(provider: OcxProviderConfig):
             return;
           case "response.completed":
             {
+              const responsePayload = isPlainObject(payload.response) ? payload.response : undefined;
+              const output = Array.isArray(responsePayload?.output) ? responsePayload.output : [];
+              const compaction = output.find(item => isPlainObject(item) && item.type === "compaction");
+              if (isPlainObject(compaction) && typeof compaction.encrypted_content === "string") {
+                const nextEncryptedContent = compaction.encrypted_content;
+                const previousBytes = budgetEncoder.encode(compactionEncryptedContent ?? "").byteLength;
+                const reservation = budget.reserveTransient(budgetEncoder.encode(nextEncryptedContent).byteLength, { kind: "retained_collectors" });
+                compactionEncryptedContent = nextEncryptedContent;
+                reservation.commitRetained();
+                budget.releaseRetained(previousBytes, { kind: "retained_collectors" });
+              }
               const next = responsesPayloadText(payload.response);
               const previousBytes = budgetEncoder.encode(snapshot).byteLength;
               const reservation = budget.reserveTransient(budgetEncoder.encode(next).byteLength, { kind: "retained_collectors" });
@@ -2336,7 +2348,11 @@ export function createResponsesPassthroughAdapter(provider: OcxProviderConfig):
       const text = snapshot || doneText || deltas;
       if (text) yield { type: "text_delta", text };
       budget.releaseRetained(budgetEncoder.encode(deltas).byteLength + budgetEncoder.encode(doneText).byteLength + budgetEncoder.encode(snapshot).byteLength, { kind: "retained_collectors" });
-      yield { type: "done", ...(usage ? { usage } : {}) };
+      yield {
+        type: "done",
+        ...(usage ? { usage } : {}),
+        ...(compactionEncryptedContent ? { compactionEncryptedContent } : {}),
+      };
     },
 
     async parseResponse(response: Response, budget: TranslatorBudget): Promise<AdapterEvent[]> {
@@ -2354,14 +2370,23 @@ export function createResponsesPassthroughAdapter(provider: OcxProviderConfig):
       if (payload.status === "incomplete") {
         return [{ type: "incomplete", reason: responsesErrorMessage(payload) }];
       }
+      const usage = usageFromResponsesPayload(payload);
+      const output = Array.isArray(payload.output) ? payload.output : [];
+      const compaction = output.find(item => isPlainObject(item) && item.type === "compaction");
+      const compactionEncryptedContent = isPlainObject(compaction) && typeof compaction.encrypted_content === "string"
+        ? compaction.encrypted_content
+        : undefined;
       const text = responsesPayloadText(payload);
-      if (!text) {
-        // A completed turn with no usable text cannot become a summary; saying so is
-        // better than installing an empty compaction as replacement history.
+      if (!text && !compactionEncryptedContent) {
+        // A completed turn with neither text nor a native compaction blob cannot become a
+        // replacement-history item. A ciphertext-only native completion is valid, though.
         return [{ type: "error", message: "upstream compaction returned no summary text" }];
       }
-      const usage = usageFromResponsesPayload(payload);
-      return [{ type: "text_delta", text }, { type: "done", ...(usage ? { usage } : {}) }];
+      return [...(text ? [{ type: "text_delta" as const, text }] : []), {
+        type: "done",
+        ...(usage ? { usage } : {}),
+        ...(compactionEncryptedContent ? { compactionEncryptedContent } : {}),
+      }];
     },
   };
 }

src/bridge.ts

@@ -1223,10 +1223,12 @@ export function bridgeToResponsesSSE(
                 // Exactly one compaction item per turn; codex-rs takes the first and fatals on 0.
                 const item = {
                   type: "compaction", id: `cmp_${uuid()}`,
-                  encrypted_content: encodeCompactionSummary(compactionText),
+                  encrypted_content: event.compactionEncryptedContent ?? encodeCompactionSummary(compactionText),
                 };
                 emit("response.output_item.done", { output_index: outputIndex, item });
-                retainFinishedItem(item as OutputItem, compactionTextBytes);
+                retainFinishedItem(item as OutputItem, event.compactionEncryptedContent
+                  ? bytesOf(event.compactionEncryptedContent)
+                  : compactionTextBytes);
                 outputIndex++;
               }
               // Recognize every adapter's truncation vocabulary, not just the canonical pair.
@@ -1574,6 +1576,7 @@ function buildResponseJSONWithBudget(
   let sawTerminal = false;
   let compactionText = "";
   let compactionTextBytes = 0;
+  let compactionEncryptedContent: string | undefined;
 
   let currentText = "";
   let currentTextBytes = 0;
@@ -1915,6 +1918,7 @@ function buildResponseJSONWithBudget(
         break;
       case "done":
         usage = e.usage;
+        compactionEncryptedContent = e.compactionEncryptedContent;
         sawTerminal = true;
         endTurn = e.endTurn;
         cleanDone = e.stopReason === undefined;
@@ -1967,7 +1971,11 @@ function buildResponseJSONWithBudget(
     && sawTerminal
     && !isTruncatedStopReason(rawStopReason)
   ) {
-    pushOutput({ type: "compaction", id: `cmp_${uuid()}`, encrypted_content: encodeCompactionSummary(compactionText) }, compactionTextBytes);
+   const item = {
+      type: "compaction", id: `cmp_${uuid()}`,
+      encrypted_content: compactionEncryptedContent ?? encodeCompactionSummary(compactionText),
+    };
+    pushOutput(item, compactionEncryptedContent ? bytesOf(compactionEncryptedContent) : compactionTextBytes);
   }
 
   const failure = errorEvent ? adapterFailureFromEvent(errorEvent) : undefined;

src/server/responses/compact.ts

@@ -562,7 +562,10 @@ export async function handleResponsesCompact(
   // Native /responses/compact exists on the canonical ChatGPT backend and on the
   // official OpenAI API. Any other Responses-shaped gateway must take the routed
   // summarizer path below, or compaction fails against an endpoint it never had (#422).
-  if (supportsNativeResponsesCompactEndpoint(route.providerName, route.provider) && !accountGatedCompactWireModel) {
+  // Combo-resolved targets skip native compact so failover can advance through the
+  // combo target list when the picked model returns 429/5xx — the routed path below
+  // dispatches through handleResponses → handleComboResponses with full failover.
+  if (supportsNativeResponsesCompactEndpoint(route.providerName, route.provider) && !accountGatedCompactWireModel && !route.combo) {
     if (req.signal.aborted) {
       return formatErrorResponse(499, "client_cancelled", "Client cancelled compact request");
     }
@@ -1003,8 +1006,10 @@ export async function handleResponsesCompact(
     ...raw,
     // Canonical ChatGPT Responses rejects non-streaming turns. Daybreak cannot use the
     // native compact endpoint either, so run its synthetic compaction as SSE and collapse
-    // the completed event back into the v1 compact JSON contract below.
-    stream: accountGatedCompactWireModel ? true : false,
+    // the completed event back into the v1 compact JSON contract below. Combo-dispatched
+    // turns also go out as SSE: failover can land on a canonical child that rejects a
+    // non-streaming turn, and every combo-capable provider already serves streaming traffic.
+    stream: accountGatedCompactWireModel || route.combo ? true : false,
     input: [...inputItems, { type: "compaction_trigger" }],
   };
   const internalHeaders = new Headers({ "content-type": "application/json" });
@@ -1078,9 +1083,12 @@ export async function handleResponsesCompact(
       `compaction turn produced ${compactionItems.length} compaction items, expected exactly 1`,
     );
   }
-  // The canonical Responses stream returns a real OpenAI-encrypted compaction item. OCX cannot
-  // and should not decrypt it; /responses/compact callers can consume that item directly.
-  if (accountGatedCompactWireModel) {
+  // Native Responses backends return a real opaque OpenAI-encrypted compaction item. OCX cannot
+  // and should not decrypt it; preserve that item for /responses/compact callers. Synthetic
+  // routed summaries are our `ocx1:` envelope and must be decoded into v1 history items.
+  if (typeof compactionItems[0]!.encrypted_content === "string"
+    && compactionItems[0]!.encrypted_content.trim().length > 0
+    && !compactionItems[0]!.encrypted_content.startsWith("ocx1:")) {
     const result = new Response(JSON.stringify({ output: compactionItems }), {
       headers: { "Content-Type": "application/json" },
     });

src/types/request.ts

@@ -333,6 +333,8 @@ export type AdapterEvent =
   | {
       type: "done";
       usage?: OcxUsage;
+      /** Native opaque compaction ciphertext returned by a Responses backend. */
+      compactionEncryptedContent?: string;
       stopReason?: string;
       endTurn?: boolean;
       providerState?: OcxProviderContinuationState;

tests/responses-compaction.test.ts

@@ -1,6 +1,7 @@
 import { describe, expect, test } from "bun:test";
 import { bridgeToResponsesSSE, buildResponseJSON } from "../src/bridge";
 import { createResponsesPassthroughAdapter as createResponsesPassthroughAdapterProduction } from "../src/adapters/openai-responses";
+import { createTranslatorBudget } from "../src/lib/translator-budget";
 import { CODEX_FORWARD_BASE_URL } from "../src/providers/openai-tiers";
 import { parseRequest } from "../src/responses/parser";
 import {
@@ -165,6 +166,70 @@ describe("buildResponseJSON compaction mode", () => {
   });
 });
 
+describe("native Responses compaction passthrough", () => {
+  const provider = {
+    adapter: "openai-responses",
+    baseUrl: "https://responses.example/v1",
+    authMode: "key" as const,
+    apiKey: "test-key",
+  };
+
+  test("buffered ciphertext-only completion yields done without a text delta", async () => {
+    const adapter = createResponsesPassthroughAdapterProduction(provider);
+    const encryptedContent = "gAAAAABm-native-buffered-ciphertext";
+    const budget = createTranslatorBudget();
+    try {
+      const events = await adapter.parseResponse!(Response.json({
+        status: "completed",
+        output: [{ type: "compaction", encrypted_content: encryptedContent }],
+      }), budget);
+
+      expect(events).toEqual([{ type: "done", compactionEncryptedContent: encryptedContent }]);
+    } finally {
+      budget.dispose();
+    }
+  });
+
+  test("streaming ciphertext is charged before the compaction item takes ownership", async () => {
+    const adapter = createResponsesPassthroughAdapterProduction(provider);
+    const encryptedContent = "gAAAAABm-native-streaming-ciphertext";
+    const budget = createTranslatorBudget();
+    try {
+      const events: AdapterEvent[] = [];
+      const stream = [
+        "event: response.completed",
+        `data: ${JSON.stringify({
+          type: "response.completed",
+          response: {
+            status: "completed",
+            output: [{ type: "compaction", encrypted_content: encryptedContent }],
+          },
+        })}`,
+        "",
+        "",
+      ].join("\n");
+      for await (const event of adapter.parseStream(new Response(stream, {
+        headers: { "content-type": "text/event-stream" },
+      }), budget)) events.push(event);
+
+      expect(events).toEqual([{ type: "done", compactionEncryptedContent: encryptedContent }]);
+      expect(budget.snapshot().currentBytes).toBe(Buffer.byteLength(encryptedContent));
+
+      const json = buildResponseJSON(events, "test/model", {
+        compaction: true,
+        translatorBudget: budget,
+      }) as { output: Array<{ type: string; encrypted_content?: string }> };
+      expect(json.output).toEqual([expect.objectContaining({
+        type: "compaction",
+        encrypted_content: encryptedContent,
+      })]);
+      expect(budget.snapshot().currentBytes).toBe(Buffer.byteLength(JSON.stringify(json.output[0])));
+    } finally {
+      budget.dispose();
+    }
+  });
+});
+
 describe("COMPACT_PROMPT", () => {
   test("mirrors the codex-rs checkpoint instruction", () => {
     expect(COMPACT_PROMPT).toContain("CONTEXT CHECKPOINT COMPACTION");