pi-openai-toolkit 0.14.4 → 0.14.6

package/src/responses-headers.ts

@@ -1,108 +1,173 @@
-import type { ProviderHeaders } from "@earendil-works/pi-ai";
-import { mergeProviderHeaders } from "./provider-headers";
-
-export type ResponsesHeaderRuntime = {
-	api: string;
-	apiKey: string;
-	headers?: ProviderHeaders;
-	sessionId?: string;
-	currentModel: {
-		headers?: ProviderHeaders;
-	};
-};
-
-function isRecord(value: unknown): value is Record<string, unknown> {
-	return !!value && typeof value === "object" && !Array.isArray(value);
-}
-
-function decodeJwtPayload(token: string): Record<string, unknown> | undefined {
-	const parts = token.split(".");
-	if (parts.length !== 3) return undefined;
-	try {
-		const payload = JSON.parse(Buffer.from(parts[1]!, "base64url").toString("utf8")) as unknown;
-		return isRecord(payload) ? payload : undefined;
-	} catch {
-		return undefined;
-	}
-}
-
-function extractCodexAccountId(token: string): string | undefined {
-	const authClaims = decodeJwtPayload(token)?.["https://api.openai.com/auth"];
-	if (!isRecord(authClaims)) return undefined;
-	const accountId = authClaims.chatgpt_account_id;
-	return typeof accountId === "string" && accountId.trim() ? accountId.trim() : undefined;
-}
-
-function buildCodexUserAgent(): string {
-	const platform = typeof process !== "undefined" ? process.platform : "browser";
-	const arch = typeof process !== "undefined" ? process.arch : "unknown";
-	return `pi (${platform}; ${arch})`;
-}
-
-/**
- * The Codex backend version-gates model availability on the `version` header
- * (the pinned @openai/codex client version). `gpt-6-astra` requires >= 0.153.0;
- * an older or absent pin makes the SKU invisible to discovery and rejects
- * requests that name it. Keep this in sync with the newest gate Oh My Pi pins
- * in @oh-my-pi/pi-wire/codex (`CODEX_CLIENT_VERSION`).
- */
-export const CODEX_CLIENT_VERSION = "0.153.0";
-
-/** Prompt cache keys are clamped by the same host rule (`OPENAI_PROMPT_CACHE_KEY_MAX_LENGTH`). */
-const CODEX_SESSION_ID_MAX_LENGTH = 64;
-
-export function clampCodexSessionId(sessionId: string | undefined): string | undefined {
-	if (!sessionId) return undefined;
-	const chars = Array.from(sessionId);
-	return chars.length <= CODEX_SESSION_ID_MAX_LENGTH ? sessionId : chars.slice(0, CODEX_SESSION_ID_MAX_LENGTH).join("");
-}
-
-export type ResponsesHeaderOptions = {
-	accept: string;
-	contentType?: string;
-	/** Conversation identity for Codex cache-affinity headers; clamped to 64 chars here. */
-	sessionId?: string;
-};
-
-export function buildResponsesRequestHeaders(
-	runtime: ResponsesHeaderRuntime,
-	options: ResponsesHeaderOptions,
-): Headers {
-	const headers = new Headers(
-		mergeProviderHeaders(runtime.currentModel.headers, runtime.headers),
-	);
-	headers.set("accept", options.accept);
-	headers.set("content-type", options.contentType ?? "application/json");
-	if (!headers.has("authorization")) {
-		headers.set("authorization", `Bearer ${runtime.apiKey}`);
-	}
-
-	if (runtime.api === "openai-codex-responses") {
-		const accountId = extractCodexAccountId(runtime.apiKey);
-		if (accountId) headers.set("chatgpt-account-id", accountId);
-		headers.set("originator", "pi");
-		headers.set("user-agent", buildCodexUserAgent());
-		headers.set("openai-beta", "responses=experimental");
-		// Codex backend gating: a request without a new-enough `version` never
-		// reaches model SKUs that arrived behind the gate (gpt-6-astra).
-		headers.set("version", CODEX_CLIENT_VERSION);
-		// Prompt-cache affinity rides on the conversation identity: the backend
-		// keeps a conversation's cache prefix on the node that wrote it, so our
-		// synthetic compaction requests must present the same id the live
-		// transcript requests used, or every compact pays a full cache miss.
-		const sessionId = clampCodexSessionId(options.sessionId ?? runtime.sessionId);
-		if (sessionId) {
-			// Header names match the installed host's live-request builder
-			// (`buildSSEHeaders`: `session-id` + `x-client-request-id`) so our
-			// synthetic requests land in the same conversation affinity group.
-			headers.set("session-id", sessionId);
-			headers.set("x-client-request-id", sessionId);
-		}
-	}
-
-	return headers;
-}
-
-export const _responsesHeadersTest = {
-	extractCodexAccountId,
-};
+import type { ProviderHeaders } from "@earendil-works/pi-ai";
+import { mergeProviderHeaders } from "./provider-headers";
+
+export const CODEX_AFFINITY_SCOPE = "codex-session-v1" as const;
+export const CODEX_GATEWAY_ORIGINATOR = "codex_cli_rs" as const;
+
+/** Headers permitted from Pi/provider configuration onto the gateway hop. */
+export const CODEX_GATEWAY_FORWARD_HEADERS = [
+	"accept",
+	"content-type",
+	"session-id",
+	"thread-id",
+	"x-parent-session-id",
+	"x-codex-parent-thread-id",
+	"x-client-request-id",
+	"x-codex-model",
+	"x-codex-affinity-scope",
+	"x-codex-window-id",
+	"x-codex-turn-metadata",
+	"x-codex-beta-features",
+	"x-openai-encrypted-tool-arguments",
+	"x-openai-tool-output-truncation-policy",
+	"x-openai-internal-codex-responses-lite",
+	"version",
+	"originator",
+	"user-agent",
+] as const;
+
+const CODEX_GATEWAY_FORWARD_HEADER_SET = new Set<string>(CODEX_GATEWAY_FORWARD_HEADERS);
+
+export function filterCodexGatewayHeaders(headers: ProviderHeaders | undefined): Record<string, string> {
+	const result: Record<string, string> = {};
+	for (const [name, value] of Object.entries(headers ?? {})) {
+		if (typeof value !== "string" || !CODEX_GATEWAY_FORWARD_HEADER_SET.has(name.toLowerCase())) continue;
+		result[name] = value;
+	}
+	return result;
+}
+
+export type CodexAffinity = {
+	/** Bare model id used by the gateway route selector, never provider/model. */
+	model: string;
+	scope: typeof CODEX_AFFINITY_SCOPE;
+};
+
+export type ResponsesHeaderRuntime = {
+	api: string;
+	apiKey: string;
+	headers?: ProviderHeaders;
+	sessionId?: string;
+	/** Opt-in gateway Codex routing metadata for synthetic requests. */
+	codexAffinity?: CodexAffinity;
+	currentModel: {
+		headers?: ProviderHeaders;
+	};
+};
+
+function isRecord(value: unknown): value is Record<string, unknown> {
+	return !!value && typeof value === "object" && !Array.isArray(value);
+}
+
+function decodeJwtPayload(token: string): Record<string, unknown> | undefined {
+	const parts = token.split(".");
+	if (parts.length !== 3) return undefined;
+	try {
+		const payload = JSON.parse(Buffer.from(parts[1]!, "base64url").toString("utf8")) as unknown;
+		return isRecord(payload) ? payload : undefined;
+	} catch {
+		return undefined;
+	}
+}
+
+function extractCodexAccountId(token: string): string | undefined {
+	const authClaims = decodeJwtPayload(token)?.["https://api.openai.com/auth"];
+	if (!isRecord(authClaims)) return undefined;
+	const accountId = authClaims.chatgpt_account_id;
+	return typeof accountId === "string" && accountId.trim() ? accountId.trim() : undefined;
+}
+
+function buildCodexUserAgent(): string {
+	const platform = typeof process !== "undefined" ? process.platform : "browser";
+	const arch = typeof process !== "undefined" ? process.arch : "unknown";
+	return `pi (${platform}; ${arch})`;
+}
+
+export function buildCodexCliUserAgent(version: string): string {
+	const platform = typeof process !== "undefined" ? process.platform : "browser";
+	const arch = typeof process !== "undefined" ? process.arch : "unknown";
+	return `codex_cli_rs/${version} (${platform}; ${arch})`;
+}
+
+/**
+ * The Codex backend version-gates model availability on the `version` header
+ * (the pinned @openai/codex client version). `gpt-6-astra` requires >= 0.153.0;
+ * an older or absent pin makes the SKU invisible to discovery and rejects
+ * requests that name it. Keep this in sync with the newest gate Oh My Pi pins
+ * in @oh-my-pi/pi-wire/codex (`CODEX_CLIENT_VERSION`).
+ */
+export const CODEX_CLIENT_VERSION = "0.153.0";
+
+/** Prompt cache keys are clamped by the same host rule (`OPENAI_PROMPT_CACHE_KEY_MAX_LENGTH`). */
+const CODEX_SESSION_ID_MAX_LENGTH = 64;
+
+export function clampCodexSessionId(sessionId: string | undefined): string | undefined {
+	if (!sessionId) return undefined;
+	const chars = Array.from(sessionId);
+	return chars.length <= CODEX_SESSION_ID_MAX_LENGTH ? sessionId : chars.slice(0, CODEX_SESSION_ID_MAX_LENGTH).join("");
+}
+
+export type ResponsesHeaderOptions = {
+	accept: string;
+	contentType?: string;
+	/** Conversation identity for Codex cache-affinity headers; clamped to 64 chars here. */
+	sessionId?: string;
+};
+
+export function buildResponsesRequestHeaders(
+	runtime: ResponsesHeaderRuntime,
+	options: ResponsesHeaderOptions,
+): Headers {
+	const mergedHeaders = mergeProviderHeaders(runtime.currentModel.headers, runtime.headers);
+	const headers = new Headers(
+		runtime.codexAffinity ? filterCodexGatewayHeaders(mergedHeaders) : mergedHeaders,
+	);
+	headers.set("accept", options.accept);
+	headers.set("content-type", options.contentType ?? "application/json");
+	if (!headers.has("authorization")) {
+		headers.set("authorization", `Bearer ${runtime.apiKey}`);
+	}
+
+	const isNativeCodex = runtime.api === "openai-codex-responses";
+	const gatewayAffinity = runtime.codexAffinity;
+	if (isNativeCodex || gatewayAffinity) {
+		if (gatewayAffinity) {
+			// A gateway request must use the configured data-plane key, never an
+			// inherited OAuth/account header from the model registry.
+			headers.set("authorization", `Bearer ${runtime.apiKey}`);
+			headers.delete("chatgpt-account-id");
+			headers.delete("cookie");
+			headers.delete("x-api-key");
+			headers.set("originator", CODEX_GATEWAY_ORIGINATOR);
+			headers.set("user-agent", buildCodexCliUserAgent(CODEX_CLIENT_VERSION));
+			headers.set("x-codex-affinity-scope", gatewayAffinity.scope);
+			headers.set("x-codex-model", gatewayAffinity.model);
+		} else {
+			const accountId = extractCodexAccountId(runtime.apiKey);
+			if (accountId) headers.set("chatgpt-account-id", accountId);
+			headers.set("originator", "pi");
+			headers.set("user-agent", buildCodexUserAgent());
+			headers.set("openai-beta", "responses=experimental");
+		}
+		// Codex backend gating: a request without a new-enough `version` never
+		// reaches model SKUs that arrived behind the gate (gpt-6-astra).
+		headers.set("version", CODEX_CLIENT_VERSION);
+		// Prompt-cache/session affinity rides on the conversation identity. The
+		// gateway path also uses the bare model hint above so NEWapi can route
+		// context/compact requests without inspecting their body.
+		const sessionId = clampCodexSessionId(options.sessionId ?? runtime.sessionId);
+		if (sessionId) {
+			// Header names match the installed host's live-request builder
+			// (`buildSSEHeaders`: `session-id` + `x-client-request-id`) so our
+			// synthetic requests land in the same conversation affinity group.
+			headers.set("session-id", sessionId);
+			headers.set("x-client-request-id", sessionId);
+		}
+	}
+
+	return headers;
+}
+
+export const _responsesHeadersTest = {
+	extractCodexAccountId,
+};