@hebo-ai/gateway 0.9.1 → 0.9.3

package/README.md

@@ -42,7 +42,7 @@ bun install @hebo-ai/gateway
 - Endpoints
   - [/chat/completions](#chatcompletions) | [/embeddings](#embeddings) | [/models](#models) | [/responses](#responses) | [/conversations](#conversations)
 - OpenAI Extensions
-  - [Reasoning](#reasoning) | [Service Tier](#service-tier) | [Prompt Caching](#prompt-caching)
+  - [Reasoning](#reasoning) | [Service Tier](#service-tier) | [Prompt Caching](#prompt-caching) | [Compressed Requests](#compressed-requests)
 - Advanced Usage
   - [Passing Framework State to Hooks](#passing-framework-state-to-hooks) | [Selective Route Mounting](#selective-route-mounting) | [Low-level Schemas & Converters](#low-level-schemas--converters)
 
@@ -792,6 +792,36 @@ Provider behavior:
 - **Google Gemini**: maps `cached_content` to Gemini `cachedContent`.
 - **Amazon Nova (Bedrock)**: maps `cache_control` to Bedrock `cachePoints` and inserts an automatic cache point on a stable prefix when none is provided.
 
+
+### Compressed Requests
+
+The gateway supports gzip and deflate compressed request bodies via the Web Compression Streams API. The `maxBodySize` option controls the maximum *decompressed* body size for these compressed requests, protecting against gzip bombs and oversized payloads.
+
+```ts
+import { gateway } from "@hebo-ai/gateway";
+
+const gw = gateway({
+  // ...
+  // Maximum decompressed body size in bytes (default: 10 MB).
+  // Set to 0 to disable the decompressed size limit.
+  maxBodySize: 10 * 1024 * 1024,
+});
+```
+
+Compressed requests that exceed this limit after decompression receive an HTTP `413 Payload Too Large` response. Unsupported `Content-Encoding` values return HTTP `415 Unsupported Media Type`.
+
+> [!IMPORTANT]
+> **Plain (uncompressed) request body size limits** are *not* enforced by the gateway — they should be configured at the framework or server level. The gateway only enforces `maxBodySize` on decompressed output, since the framework cannot know the decompressed size ahead of time.
+>
+> Framework-level configuration examples:
+>
+> - **Bun** — [`Bun.serve({ maxRequestBodySize: 10_485_760 })`](https://bun.sh/docs/api/http#bun-serve)
+> - **Elysia** — inherits from Bun's `maxRequestBodySize`
+> - **Hono** — [`bodyLimit` middleware](https://hono.dev/docs/middleware/builtin/body-limit): `app.use(bodyLimit({ maxSize: 10 * 1024 * 1024 }))`
+> - **Express** — [`express.json({ limit: '10mb' })`](https://expressjs.com/en/api.html#express.json)
+> - **Fastify** — [`fastify({ bodyLimit: 10485760 })`](https://fastify.dev/docs/latest/Reference/Server/#bodylimit)
+> - **Node.js `http`** — [`server.maxRequestSize`](https://nodejs.org/api/http.html) (v22.6+), or use a reverse proxy like nginx (`client_max_body_size 10m`)
+
 ## 🧪 Advanced Usage
 
 ### Logger Settings
@@ -863,19 +893,37 @@ Attribute names and span & metrics semantics follow OpenTelemetry GenAI semantic
 https://opentelemetry.io/docs/specs/semconv/gen-ai/gen-ai-spans/
 https://opentelemetry.io/docs/specs/semconv/gen-ai/gen-ai-metrics/
 
+For observability integration that is not otel compliant, you can disable built-in telemetry and manually instrument requests during `before` / `after` hooks.
+
+#### Custom Telemetry Attributes
+
+Use `ctx.otel` in any hook to attach attributes to both spans and metrics:
+
+```ts
+hooks: {
+  onRequest: (ctx) => {
+    ctx.otel["app.tenant.id"] = tenantId;
+    ctx.otel["app.user.id"] = userId;
+  },
+}
+```
+
+These attributes appear on the active span and on all metric instruments (request duration, token usage, TPOT, TTFT).
+
 > [!TIP]
 > To populate custom span attributes, the inbound W3C `baggage` header is supported. Keys in the `hebo.` namespace are mapped to span attributes, with the namespace stripped. For example: `baggage: hebo.user_id=u-123` becomes span attribute `user_id=u-123`.  
 > For `/chat/completions` and `/embeddings`, request `metadata` (`Record<string, string>`, key 1-64 chars, value up to 512 chars) is also forwarded to spans as `gen_ai.request.metadata.<key>`.
 
-For observability integration that is not otel compliant, you can disable built-in telemetry and manually instrument requests during `before` / `after` hooks.
-
 #### Metrics
 
 The Gateway also emits `gen_ai` metrics:
 
 - `gen_ai.server.request.duration` (histogram, seconds)
 - `gen_ai.server.time_per_output_token` (histogram, seconds)
-- `gen_ai.client.token.usage` (histogram, tokens; tagged with `gen_ai.token.type=input|output`)
+- `gen_ai.server.time_to_first_token` (histogram, seconds)
+- `gen_ai.client.token.usage` (histogram, tokens; tagged with `gen_ai.token.type=input|output|cached|reasoning`)
+
+Metric names and attributes follow OpenTelemetry GenAI semantic conventions. Histogram bucket boundaries are tuned for practical dashboards and alerting rather than copied verbatim from upstream recommendations.
 
 To capture them, configure a global `MeterProvider` before creating the gateway:
 
@@ -1073,3 +1121,33 @@ Non-streaming versions are available via `toChatCompletionsResponse`. Equivalent
 
 > [!TIP]
 > Since Zod v4.3 you can generate a JSON Schema from any zod object by calling `z.toJSONSchema(...)`. This is useful for producing OpenAPI documentation from the same source of truth.
+
+
+### Request Body Size
+
+The gateway supports gzip and deflate compressed request bodies via the Web Compression Streams API. The `maxBodySize` option controls the maximum *decompressed* body size for these compressed requests, protecting against gzip bombs and oversized payloads.
+
+```ts
+import { gateway } from "@hebo-ai/gateway";
+
+const gw = gateway({
+  // ...
+  // Maximum decompressed body size in bytes (default: 10 MB).
+  // Set to 0 to disable the decompressed size limit.
+  maxBodySize: 10 * 1024 * 1024,
+});
+```
+
+Compressed requests that exceed this limit after decompression receive an HTTP `413 Payload Too Large` response. Unsupported `Content-Encoding` values return HTTP `415 Unsupported Media Type`.
+
+> [!IMPORTANT]
+> **Plain (uncompressed) request body size limits** are *not* enforced by the gateway — they should be configured at the framework or server level. The gateway only enforces `maxBodySize` on decompressed output, since the framework cannot know the decompressed size ahead of time.
+>
+> Framework-level configuration examples:
+>
+> - **Bun** — [`Bun.serve({ maxRequestBodySize: 10_485_760 })`](https://bun.sh/docs/api/http#bun-serve)
+> - **Elysia** — inherits from Bun's `maxRequestBodySize`
+> - **Hono** — [`bodyLimit` middleware](https://hono.dev/docs/middleware/builtin/body-limit): `app.use(bodyLimit({ maxSize: 10 * 1024 * 1024 }))`
+> - **Express** — [`express.json({ limit: '10mb' })`](https://expressjs.com/en/api.html#express.json)
+> - **Fastify** — [`fastify({ bodyLimit: 10485760 })`](https://fastify.dev/docs/latest/Reference/Server/#bodylimit)
+> - **Node.js `http`** — [`server.maxRequestSize`](https://nodejs.org/api/http.html) (v22.6+), or use a reverse proxy like nginx (`client_max_body_size 10m`)

package/dist/config.js

@@ -3,6 +3,7 @@ import { isLogger, logger, setLoggerInstance } from "./logger";
 import { createDefaultLogger } from "./logger/default";
 import { installAiSdkWarningLogger } from "./telemetry/ai-sdk";
 import { DEFAULT_CHAT_TIMEOUT_MS, kParsed, } from "./types";
+import { DEFAULT_MAX_BODY_SIZE } from "./utils/body";
 export const parseConfig = (config) => {
     // If it has been parsed before, just return.
     if (kParsed in config)
@@ -94,10 +95,23 @@ export const parseConfig = (config) => {
             flex = t.flex;
     }
     const parsedTimeouts = { normal, flex };
+    // Body size limit
+    const rawMax = config.maxBodySize;
+    let maxBodySize;
+    if (typeof rawMax === "number" && Number.isFinite(rawMax) && rawMax >= 0) {
+        maxBodySize = rawMax;
+    }
+    else {
+        maxBodySize = DEFAULT_MAX_BODY_SIZE;
+        if (rawMax !== undefined) {
+            logger.warn(`[config] invalid maxBodySize (${rawMax}), using default ${DEFAULT_MAX_BODY_SIZE}`);
+        }
+    }
     // Return parsed config.
     return {
         ...config,
         timeouts: parsedTimeouts,
+        maxBodySize,
         telemetry: {
             ...config.telemetry,
             enabled: telemetryEnabled,

package/dist/endpoints/chat-completions/converters.d.ts

@@ -1,9 +1,9 @@
 import type { SharedV3ProviderMetadata } from "@ai-sdk/provider";
 import type { GenerateTextResult, StreamTextResult, FinishReason, ToolSet, ModelMessage, UserContent, LanguageModelUsage, TextStreamPart, ReasoningOutput, AssistantModelMessage, ToolModelMessage, UserModelMessage } from "ai";
 import { Output } from "ai";
-import type { ChatCompletionsToolCall, ChatCompletionsTool, ChatCompletionsToolChoice, ChatCompletionsStream, ChatCompletionsContentPart, ChatCompletionsMessage, ChatCompletionsUserMessage, ChatCompletionsAssistantMessage, ChatCompletionsToolMessage, ChatCompletionsFinishReason, ChatCompletionsUsage, ChatCompletionsInputs, ChatCompletions, ChatCompletionsChunk, ChatCompletionsReasoningDetail } from "./schema";
 import type { SseErrorFrame, SseFrame } from "../../utils/stream";
 import { type TextCallOptions, type ToolChoiceOptions } from "../shared/converters";
+import type { ChatCompletionsToolCall, ChatCompletionsTool, ChatCompletionsToolChoice, ChatCompletionsStream, ChatCompletionsContentPart, ChatCompletionsMessage, ChatCompletionsUserMessage, ChatCompletionsAssistantMessage, ChatCompletionsToolMessage, ChatCompletionsFinishReason, ChatCompletionsUsage, ChatCompletionsInputs, ChatCompletions, ChatCompletionsChunk, ChatCompletionsReasoningDetail } from "./schema";
 export declare function convertToTextCallOptions(params: ChatCompletionsInputs): TextCallOptions;
 export declare function convertToModelMessages(messages: ChatCompletionsMessage[]): ModelMessage[];
 export declare function fromChatCompletionsUserMessage(message: ChatCompletionsUserMessage): UserModelMessage;

package/dist/endpoints/chat-completions/converters.js

@@ -1,4 +1,5 @@
 import { Output, jsonSchema, tool } from "ai";
+import { GatewayError } from "../../errors/gateway";
 import { toResponse } from "../../utils/response";
 import { parseJsonOrText, parseReasoningOptions, parsePromptCachingOptions, resolveResponseServiceTier, normalizeToolName, stripEmptyKeys, parseBase64, parseImageInput, extractReasoningMetadata, } from "../shared/converters";
 // --- Request Flow ---
@@ -32,7 +33,7 @@ export function convertToTextCallOptions(params) {
 }
 function convertToOutput(responseFormat) {
     if (!responseFormat || responseFormat.type === "text") {
-        return;
+        return undefined;
     }
     const { name, description, schema } = responseFormat.json_schema;
     return Output.object({
@@ -207,7 +208,7 @@ export function fromChatCompletionsContent(content) {
                 return out;
             }
             default:
-                throw new Error(`Unhandled content part type: ${part.type}`);
+                throw new GatewayError(`Unsupported content part type: ${part.type}`, 400);
         }
     });
 }
@@ -257,7 +258,7 @@ function fromFilePart(base64Data, mediaType, filename, cacheControl) {
 }
 export const convertToToolSet = (tools) => {
     if (!tools) {
-        return;
+        return undefined;
     }
     const toolSet = {};
     for (const t of tools) {

package/dist/endpoints/chat-completions/handler.js

@@ -5,8 +5,9 @@ import { winterCgHandler } from "../../lifecycle";
 import { logger } from "../../logger";
 import { modelMiddlewareMatcher } from "../../middleware/matcher";
 import { resolveProvider } from "../../providers/registry";
-import { getGenAiGeneralAttributes, recordTimePerOutputToken, recordTokenUsage, } from "../../telemetry/gen-ai";
+import { getGenAiGeneralAttributes, recordTimePerOutputToken, recordTimeToFirstToken, recordTokenUsage, } from "../../telemetry/gen-ai";
 import { addSpanEvent, setSpanAttributes } from "../../telemetry/span";
+import { parseRequestBody } from "../../utils/body";
 import { prepareForwardHeaders } from "../../utils/request";
 import { convertToTextCallOptions, toChatCompletions, toChatCompletionsStream } from "./converters";
 import { getChatRequestAttributes, getChatResponseAttributes } from "./otel";
@@ -22,14 +23,8 @@ export const chatCompletions = (config) => {
         if (!ctx.request || ctx.request.method !== "POST") {
             throw new GatewayError("Method Not Allowed", 405);
         }
-        // Parse + validate input.
-        try {
-            // oxlint-disable-next-line no-unsafe-assignment
-            ctx.body = await ctx.request.json();
-        }
-        catch {
-            throw new GatewayError("Invalid JSON", 400);
-        }
+        // Parse + validate input (handles Content-Encoding decompression + body size limits).
+        ctx.body = (await parseRequestBody(ctx.request, cfg.maxBodySize));
         logger.trace({ requestId: ctx.requestId, body: ctx.body }, "[chat] ChatCompletionsBody");
         addSpanEvent("hebo.request.deserialized");
         const parsed = ChatCompletionsBodySchema.safeParse(ctx.body);
@@ -83,6 +78,7 @@ export const chatCompletions = (config) => {
         // Execute request (streaming vs. non-streaming).
         if (stream) {
             addSpanEvent("hebo.ai-sdk.started");
+            let ttft = 0;
             const result = streamText({
                 model: languageModelWithMiddleware,
                 headers: prepareForwardHeaders(ctx.request),
@@ -94,6 +90,12 @@ export const chatCompletions = (config) => {
                     throw new DOMException("The operation was aborted.", "AbortError");
                 },
                 onError: () => { },
+                onChunk: () => {
+                    if (!ttft) {
+                        ttft = performance.now() - start;
+                        recordTimeToFirstToken(ttft, genAiGeneralAttrs, genAiSignalLevel);
+                    }
+                },
                 onFinish: (res) => {
                     addSpanEvent("hebo.ai-sdk.completed");
                     const streamResult = toChatCompletions(res, ctx.resolvedModelId);
@@ -102,7 +104,7 @@ export const chatCompletions = (config) => {
                     const genAiResponseAttrs = getChatResponseAttributes(streamResult, genAiSignalLevel);
                     setSpanAttributes(genAiResponseAttrs);
                     recordTokenUsage(genAiResponseAttrs, genAiGeneralAttrs, genAiSignalLevel);
-                    recordTimePerOutputToken(start, genAiResponseAttrs, genAiGeneralAttrs, genAiSignalLevel);
+                    recordTimePerOutputToken(start, ttft, genAiResponseAttrs, genAiGeneralAttrs, genAiSignalLevel);
                 },
                 experimental_include: {
                     requestBody: false,
@@ -131,6 +133,7 @@ export const chatCompletions = (config) => {
         });
         logger.trace({ requestId: ctx.requestId, result }, "[chat] AI SDK result");
         addSpanEvent("hebo.ai-sdk.completed");
+        recordTimeToFirstToken(performance.now() - start, genAiGeneralAttrs, genAiSignalLevel);
         // Transform result.
         ctx.result = toChatCompletions(result, ctx.resolvedModelId);
         logger.trace({ requestId: ctx.requestId, result: ctx.result }, "[chat] ChatCompletions");
@@ -142,7 +145,7 @@ export const chatCompletions = (config) => {
             ctx.result = (await hooks.after(ctx)) ?? ctx.result;
             addSpanEvent("hebo.hooks.after.completed");
         }
-        recordTimePerOutputToken(start, genAiResponseAttrs, genAiGeneralAttrs, genAiSignalLevel);
+        recordTimePerOutputToken(start, 0, genAiResponseAttrs, genAiGeneralAttrs, genAiSignalLevel);
         return ctx.result;
     };
     return { handler: winterCgHandler(handler, config) };

package/dist/endpoints/chat-completions/otel.d.ts

@@ -1,5 +1,5 @@
 import type { Attributes } from "@opentelemetry/api";
-import type { ChatCompletions, ChatCompletionsBody } from "./schema";
 import { type TelemetrySignalLevel } from "../../types";
+import type { ChatCompletions, ChatCompletionsBody } from "./schema";
 export declare const getChatRequestAttributes: (body: ChatCompletionsBody, signalLevel?: TelemetrySignalLevel) => Attributes;
 export declare const getChatResponseAttributes: (completions: ChatCompletions, signalLevel?: TelemetrySignalLevel) => Attributes;

package/dist/endpoints/chat-completions/otel.js

@@ -79,6 +79,9 @@ const toUserParts = (content) => {
                 parts.push(filePart);
                 break;
             }
+            default:
+                parts.push({ type: part.type, content: "[UNHANDLED_CONTENT_PART]" });
+                break;
         }
     }
     return parts;
@@ -96,7 +99,7 @@ const toMessageParts = (message) => {
         case "system":
             return toTextParts(message.content);
         default:
-            throw new Error(`Unhandled content part type: ${message.role}`);
+            return [{ type: message.role, content: "[UNHANDLED_ROLE]" }];
     }
 };
 export const getChatRequestAttributes = (body, signalLevel) => {

package/dist/endpoints/conversations/handler.js

@@ -4,8 +4,9 @@ import { GatewayError } from "../../errors/gateway";
 import { winterCgHandler } from "../../lifecycle";
 import { logger } from "../../logger";
 import { addSpanEvent } from "../../telemetry/span";
-import { ConversationCreateParamsSchema, ConversationItemsAddBodySchema, ConversationUpdateBodySchema, ConversationItemListParamsSchema, ConversationListParamsSchema, } from "./schema";
+import { parseRequestBody } from "../../utils/body";
 import { toConversation, toConversationItem, toConversationDeleted } from "./converters";
+import { ConversationCreateParamsSchema, ConversationItemsAddBodySchema, ConversationUpdateBodySchema, ConversationItemListParamsSchema, ConversationListParamsSchema, } from "./schema";
 export const conversations = (config) => {
     const parsedConfig = parseConfig(config);
     const storage = parsedConfig.storage;
@@ -42,13 +43,7 @@ export const conversations = (config) => {
         };
     }
     async function create(ctx) {
-        let body;
-        try {
-            body = await ctx.request.json();
-        }
-        catch {
-            throw new GatewayError("Invalid JSON", 400);
-        }
+        const body = await parseRequestBody(ctx.request, parsedConfig.maxBodySize);
         addSpanEvent("hebo.request.deserialized");
         const parsed = ConversationCreateParamsSchema.safeParse(body);
         if (!parsed.success) {
@@ -72,13 +67,7 @@ export const conversations = (config) => {
         return toConversation(entity);
     }
     async function update(ctx, conversationId) {
-        let body;
-        try {
-            body = await ctx.request.json();
-        }
-        catch {
-            throw new GatewayError("Invalid JSON", 400);
-        }
+        const body = await parseRequestBody(ctx.request, parsedConfig.maxBodySize);
         addSpanEvent("hebo.request.deserialized");
         const parsed = ConversationUpdateBodySchema.safeParse(body);
         if (!parsed.success) {
@@ -150,13 +139,7 @@ export const conversations = (config) => {
         };
     }
     async function addItems(ctx, conversationId) {
-        let body;
-        try {
-            body = await ctx.request.json();
-        }
-        catch {
-            throw new GatewayError("Invalid JSON", 400);
-        }
+        const body = await parseRequestBody(ctx.request, parsedConfig.maxBodySize);
         addSpanEvent("hebo.request.deserialized");
         const parsed = ConversationItemsAddBodySchema.safeParse(body);
         if (!parsed.success) {

package/dist/endpoints/conversations/storage/dialects/mysql.js

@@ -36,7 +36,7 @@ function createMysql2Executor(pool) {
         async run(sql, params) {
             const [res] = await pool.execute(sql, mapParams(params));
             const header = res;
-            return { changes: Number(header.affectedRows ?? 0) };
+            return { changes: header.affectedRows ?? 0 };
         },
         async transaction(fn) {
             const conn = await pool.getConnection();
@@ -53,7 +53,7 @@ function createMysql2Executor(pool) {
                 async run(sql, params) {
                     const [res] = await conn.execute(sql, mapParams(params));
                     const header = res;
-                    return { changes: Number(header.affectedRows ?? 0) };
+                    return { changes: header.affectedRows ?? 0 };
                 },
                 transaction(txCallback) {
                     return txCallback(txExecutor);
@@ -87,7 +87,7 @@ function createBunMysqlExecutor(sql) {
         async run(query, params) {
             const res = (await sql.unsafe(query, mapParams(params)));
             const result = res;
-            return { changes: Number(result.affectedRows ?? result.count ?? 0) };
+            return { changes: result.affectedRows ?? result.count ?? 0 };
         },
         transaction(fn) {
             return sql.transaction((tx) => {

package/dist/endpoints/conversations/storage/dialects/postgres.js

@@ -59,7 +59,7 @@ function createPgExecutor(pool, mapParams) {
         async run(sql, params) {
             const p = mapParams(params);
             const res = await pool.query(getQuery(sql, p?.length > 0 ? p : undefined));
-            return { changes: Number(res.rowCount ?? 0) };
+            return { changes: res.rowCount ?? 0 };
         },
         async transaction(fn) {
             const client = await pool.connect();
@@ -78,7 +78,7 @@ function createPgExecutor(pool, mapParams) {
                 async run(sql, params) {
                     const p = mapParams(params);
                     const res = await client.query(getQuery(sql, p?.length > 0 ? p : undefined));
-                    return { changes: Number(res.rowCount ?? 0) };
+                    return { changes: res.rowCount ?? 0 };
                 },
                 transaction(txCallback) {
                     return txCallback(txExecutor);
@@ -115,7 +115,7 @@ function createPostgresJsExecutor(sql, mapParams) {
             const p = mapParams(params);
             const res = await sql.unsafe(query, (p?.length > 0 ? p : undefined), { prepare: true });
             const result = res;
-            return { changes: Number(result.count ?? 0) };
+            return { changes: result.count ?? 0 };
         },
         async transaction(fn) {
             return (await sql.begin((tx) => {
@@ -151,7 +151,7 @@ function createBunPostgresExecutor(sql, mapParams) {
                 if (!isNaN(parsed))
                     changes = parsed;
             }
-            return { changes: Number(changes) };
+            return { changes };
         },
         transaction(fn) {
             return sql.transaction((tx) => {

package/dist/endpoints/conversations/storage/dialects/sqlite.js

@@ -97,7 +97,7 @@ function createLibsqlExecutor(client) {
         },
         async run(sql, params) {
             const rs = await client.execute({ sql, args: mapParams(params) ?? [] });
-            return { changes: Number(rs.rowsAffected) };
+            return { changes: rs.rowsAffected };
         },
         async transaction(fn) {
             const tx = await client.transaction("deferred");
@@ -112,7 +112,7 @@ function createLibsqlExecutor(client) {
                 },
                 async run(sql, params) {
                     const rs = await tx.execute({ sql, args: mapParams(params) ?? [] });
-                    return { changes: Number(rs.rowsAffected) };
+                    return { changes: rs.rowsAffected };
                 },
                 transaction(txCallback) {
                     return txCallback(txExecutor);
@@ -143,7 +143,7 @@ function createBunSqliteExecutor(sql) {
         async run(query, params) {
             const res = (await sql.unsafe(query, mapParams(params)));
             const result = res;
-            return { changes: Number(result.affectedRows ?? result.count ?? 0) };
+            return { changes: result.affectedRows ?? result.count ?? 0 };
         },
         transaction(fn) {
             return sql.transaction((tx) => {

package/dist/endpoints/conversations/storage/sql.d.ts

@@ -1,5 +1,5 @@
-import type { ConversationStorage, ConversationEntity, ConversationItemEntity, ConversationMetadata, ConversationItemInput, ConversationQueryOptions } from "./types";
 import type { SqlDialect } from "./dialects/types";
+import type { ConversationStorage, ConversationEntity, ConversationItemEntity, ConversationMetadata, ConversationItemInput, ConversationQueryOptions } from "./types";
 export declare class SqlStorage implements ConversationStorage {
     readonly dialect: SqlDialect;
     constructor(options: SqlDialect | {

package/dist/endpoints/conversations/storage/sql.js

@@ -138,14 +138,13 @@ export class SqlStorage {
             args.push(after);
         }
         sqlParts.push(`ORDER BY c.${q("created_at")} ${dir}, c.${q("id")} ${dir}`);
-        const limitVal = Number(limit);
-        if (!isNaN(limitVal)) {
+        if (!Number.isNaN(limit)) {
             if (limitAsLiteral) {
-                sqlParts.push(`LIMIT ${limitVal}`);
+                sqlParts.push(`LIMIT ${limit}`);
             }
             else {
                 sqlParts.push(`LIMIT ${p(nextIdx++)}`);
-                args.push(limitVal);
+                args.push(limit);
             }
         }
         const query = sqlParts.join(" ");
@@ -168,7 +167,7 @@ export class SqlStorage {
             //    to deduplicate the row.
             const conversation = await this.getConversationInternal(id, tx);
             if (!conversation)
-                return;
+                return conversation;
             const createdAt = conversation.created_at;
             const pk = ["id"];
             const updateCols = ["metadata"];
@@ -195,7 +194,7 @@ export class SqlStorage {
             if (!skipCheck) {
                 const conversation = await this.getConversationInternal(conversationId, tx);
                 if (!conversation)
-                    return;
+                    return conversation;
             }
             const { placeholder: p, quote: q } = this.config;
             const columns = ["id", "conversation_id", "type", "data", "created_at"];
@@ -253,14 +252,13 @@ export class SqlStorage {
             args.push(after, conversationId);
         }
         sqlParts.push(`ORDER BY c.${q("created_at")} ${dir}, c.${q("id")} ${dir}`);
-        const limitVal = Number(limit);
-        if (!isNaN(limitVal)) {
+        if (!Number.isNaN(limit)) {
             if (limitAsLiteral) {
-                sqlParts.push(`LIMIT ${limitVal}`);
+                sqlParts.push(`LIMIT ${limit}`);
             }
             else {
                 sqlParts.push(`LIMIT ${p(nextIdx++)}`);
-                args.push(limitVal);
+                args.push(limit);
             }
         }
         const query = sqlParts.join(" ");

package/dist/endpoints/embeddings/handler.js

@@ -7,6 +7,7 @@ import { modelMiddlewareMatcher } from "../../middleware/matcher";
 import { resolveProvider } from "../../providers/registry";
 import { getGenAiGeneralAttributes, recordTimePerOutputToken, recordTokenUsage, } from "../../telemetry/gen-ai";
 import { addSpanEvent, setSpanAttributes } from "../../telemetry/span";
+import { parseRequestBody } from "../../utils/body";
 import { prepareForwardHeaders } from "../../utils/request";
 import { convertToEmbedCallOptions, toEmbeddings } from "./converters";
 import { getEmbeddingsRequestAttributes, getEmbeddingsResponseAttributes } from "./otel";
@@ -22,14 +23,8 @@ export const embeddings = (config) => {
         if (!ctx.request || ctx.request.method !== "POST") {
             throw new GatewayError("Method Not Allowed", 405);
         }
-        // Parse + validate input.
-        try {
-            // oxlint-disable-next-line no-unsafe-assignment
-            ctx.body = await ctx.request.json();
-        }
-        catch {
-            throw new GatewayError("Invalid JSON", 400);
-        }
+        // Parse + validate input (handles Content-Encoding decompression + body size limits).
+        ctx.body = (await parseRequestBody(ctx.request, cfg.maxBodySize));
         logger.trace({ requestId: ctx.requestId, result: ctx.body }, "[chat] EmbeddingsBody");
         addSpanEvent("hebo.request.deserialized");
         const parsed = EmbeddingsBodySchema.safeParse(ctx.body);
@@ -97,7 +92,7 @@ export const embeddings = (config) => {
             ctx.result = (await hooks.after(ctx)) ?? ctx.result;
             addSpanEvent("hebo.hooks.after.completed");
         }
-        recordTimePerOutputToken(start, genAiResponseAttrs, genAiGeneralAttrs, genAiSignalLevel);
+        recordTimePerOutputToken(start, 0, genAiResponseAttrs, genAiGeneralAttrs, genAiSignalLevel);
         return ctx.result;
     };
     return { handler: winterCgHandler(handler, config) };

package/dist/endpoints/embeddings/otel.d.ts

@@ -1,5 +1,5 @@
 import type { Attributes } from "@opentelemetry/api";
-import type { Embeddings, EmbeddingsBody } from "./schema";
 import { type TelemetrySignalLevel } from "../../types";
+import type { Embeddings, EmbeddingsBody } from "./schema";
 export declare const getEmbeddingsRequestAttributes: (body: EmbeddingsBody, signalLevel?: TelemetrySignalLevel) => Attributes;
 export declare const getEmbeddingsResponseAttributes: (embeddings: Embeddings, signalLevel?: TelemetrySignalLevel) => Attributes;

package/dist/endpoints/responses/converters.d.ts

@@ -1,8 +1,8 @@
 import type { GenerateTextResult, StreamTextResult, ToolSet, ModelMessage, LanguageModelUsage, TextStreamPart } from "ai";
 import { Output } from "ai";
-import type { ResponsesInputItem, ResponsesInputs, Responses, ResponsesUsage, ResponsesStream, ResponsesStreamEvent, ResponsesToolChoice, ResponsesTool } from "./schema";
 import type { SseErrorFrame } from "../../utils/stream";
 import { type TextCallOptions, type ToolChoiceOptions } from "../shared/converters";
+import type { ResponsesInputItem, ResponsesInputs, Responses, ResponsesUsage, ResponsesStream, ResponsesStreamEvent, ResponsesToolChoice, ResponsesTool } from "./schema";
 export declare function convertToTextCallOptions(params: ResponsesInputs): TextCallOptions;
 export declare function convertToModelMessages(input: string | ResponsesInputItem[], instructions?: string): ModelMessage[];
 export declare const convertToToolSet: (tools: ResponsesTool[] | undefined) => ToolSet | undefined;

package/dist/endpoints/responses/converters.js

@@ -43,7 +43,7 @@ export function convertToTextCallOptions(params) {
 function convertToOutput(text) {
     if (!text?.format || text.format.type === "text") {
         // FUTURE: Support text.verbosity when AI SDK adds top-level support
-        return;
+        return undefined;
     }
     const { name, description, schema } = text.format;
     return Output.object({
@@ -147,6 +147,7 @@ function fromMessageItem(item) {
         case "assistant":
             return fromAssistantMessageItem(item);
     }
+    throw new GatewayError("Unsupported message role", 400);
 }
 function fromUserMessageItem(item) {
     const out = {
@@ -236,6 +237,8 @@ function fromInputContent(content) {
                 result.push(out);
                 break;
             }
+            default:
+                throw new GatewayError(`Unsupported content part type: ${part.type}`, 400);
         }
     }
     return result;
@@ -359,7 +362,7 @@ function fromFunctionCallOutputItem(item, toolOutputByCallId) {
 }
 export const convertToToolSet = (tools) => {
     if (!tools) {
-        return;
+        return undefined;
     }
     const toolSet = {};
     for (const t of tools) {