@kreuzberg/liter-llm-node 1.6.4 → 1.7.0

package/index.d.ts

@@ -1,5 +1,5 @@
 // This file is auto-generated by alef — DO NOT EDIT.
-// alef:hash:52effca2513e1c5f8a9a5b90a372d645388de4b92e4f822acc38d97502c322b6
+// alef:hash:bfb3bc291787eb8f5228b353fec6c4ba65c5bcc3438beaff9afb5b0cc9c3aa87
 // To regenerate: alef generate
 // To verify freshness: alef verify --exit-code
 /* eslint-disable */
@@ -154,6 +154,26 @@ export declare function createClient(
  */
 export declare function createClientFromJson(json: string): DefaultClient;
 
+/**
+ * Decode a base64 data URL into [`DecodedDataUrl`].
+ *
+ * Returns `None` for:
+ * - Non-data URLs (strings that do not start with `"data:"`).
+ * - Malformed prefixes (missing `";base64,"` marker).
+ * - Invalid base64 payloads.
+ *
+ * The returned MIME string is extracted verbatim from the URL prefix —
+ * it is not validated or normalised.
+ */
+export declare function decodeDataUrl(url: string): DecodedDataUrl | null;
+
+/**
+ * Encode bytes as a base64 data URL: `data:<mime>;base64,<b64>`.
+ *
+ * `mime` defaults to [`IMAGE_PNG`] when `None`.
+ */
+export declare function encodeDataUrl(bytes: Uint8Array, mime?: string | undefined | null): string;
+
 /**
  * Install the `ring` crypto provider as the rustls process default, idempotently.
  *
@@ -174,10 +194,29 @@ export declare function createClientFromJson(json: string): DefaultClient;
  */
 export declare function ensureCryptoProvider(): void;
 
+/**
+ * Content shape for assistant messages.
+ *
+ * `#[serde(untagged)]` means providers returning a plain scalar string for the
+ * `content` field still deserialise correctly into `AssistantContent::Text(_)`.
+ * Providers returning an array of typed parts (e.g. after an image-generation
+ * or audio-synthesis request) deserialise into `AssistantContent::Parts(_)`.
+ */
+export declare enum AssistantContent {
+  /** Plain text response (the common case for text-only models). */
+  Text = "Text",
+  /** Structured parts — text, refusals, output images, output audio. */
+  Parts = "Parts",
+}
+
 /** Assistant's response to a user message. */
 export interface AssistantMessage {
-  /** The assistant's text response. Absent if tool calls are returned instead. */
-  readonly content?: string;
+  /**
+   * The assistant's response: plain text, structured parts, or absent.
+   *
+   * `None` is valid when the model replies with tool calls only.
+   */
+  readonly content?: AssistantContent;
   /** Optional name for the assistant. */
   readonly name?: string;
   /** Tool calls the model wants to execute, if any. */
@@ -188,6 +227,18 @@ export interface AssistantMessage {
   readonly functionCall?: FunctionCall;
 }
 
+/**
+ * One part of a structured assistant response.
+ *
+ * `#[serde(tag = "type", rename_all = "snake_case")]` matches OpenAI's
+ * parts-spec discriminator (`"type": "text"`, `"type": "output_image"`, …).
+ */
+export type AssistantPart =
+  | { type: "text"; text: string }
+  | { type: "refusal"; refusal: string }
+  | { type: "output_image"; imageUrl: ImageUrl }
+  | { type: "output_audio"; audio: AudioContent };
+
 /** Audio content part for speech-capable models. */
 export interface AudioContent {
   /** Base64-encoded audio data. */
@@ -411,6 +462,13 @@ export interface ChatCompletionRequest {
   readonly seed?: number;
   /** Reasoning effort level (low, medium, high) for extended-thinking models. */
   readonly reasoningEffort?: ReasoningEffort;
+  /**
+   * Output modalities to request from the model.
+   *
+   * For OpenAI audio models, pass `["text", "audio"]`. Vertex AI / Gemini
+   * translates these to `generationConfig.responseModalities` (uppercase).
+   */
+  readonly modalities?: Array<Modality>;
   /**
    * Provider-specific extra parameters merged into the request body.
    * Use for guardrails, safety settings, grounding config, etc.
@@ -591,7 +649,7 @@ export interface CreateTranscriptionRequest {
 export interface CustomProviderConfig {
   /** Unique name for this provider (e.g., "my-provider"). */
   readonly name: string;
-  /** Base URL for the provider's API (e.g., "https://api.my-provider.com/v1"). */
+  /** Base URL for the provider's API (e.g., `<https://api.my-provider.com/v1>`). */
   readonly baseUrl: string;
   /** Authentication header format. */
   readonly authHeader: AuthHeaderFormat;
@@ -599,6 +657,19 @@ export interface CustomProviderConfig {
   readonly modelPrefixes: Array<string>;
 }
 
+/**
+ * Result of decoding a `data:` URL — MIME type and the decoded byte payload.
+ *
+ * Named struct (rather than a tuple) so polyglot bindings can extract
+ * `decode_data_url` with a typed return rather than a sanitized scalar.
+ */
+export interface DecodedDataUrl {
+  /** MIME type extracted from the URL prefix (verbatim, not normalised). */
+  readonly mime?: string;
+  /** Decoded base64 payload. */
+  readonly data?: Uint8Array;
+}
+
 /**
  * Default client implementation backed by `reqwest`.
  *
@@ -944,6 +1015,21 @@ export type Message =
   | { role: "developer"; 0: DeveloperMessage }
   | { role: "function"; 0: FunctionMessage };
 
+/**
+ * Output modality requested from the model.
+ *
+ * Passed as `modalities: ["text", "audio"]` (OpenAI) or translated to
+ * `generationConfig.responseModalities` (Gemini / Vertex AI).
+ */
+export declare enum Modality {
+  /** Text output (the default for all providers). */
+  Text = "text",
+  /** Audio / speech output. */
+  Audio = "audio",
+  /** Image output (Gemini Imagen, gpt-image-1). */
+  Image = "image",
+}
+
 /** A model available from the API. */
 export interface ModelObject {
   /** Model ID (e.g., `"gpt-4o"`, `"claude-3-5-sonnet"`). */
@@ -1194,7 +1280,7 @@ export interface ProviderConfig {
    *
    * Each entry maps an OpenAI-spec field name (e.g. `"max_completion_tokens"`)
    * to the name this provider expects (e.g. `"max_tokens"`).  Applied
-   * automatically by [`ConfigDrivenProvider::transform_request`].
+   * automatically by `ConfigDrivenProvider::transform_request`.
    */
   readonly paramMappings?: Record<string, string>;
 }
@@ -1264,7 +1350,24 @@ export interface RerankResultDocument {
   readonly text: string;
 }
 
-/** Response format constraint. */
+/**
+ * Wire format for the chat completions `response_format` field.
+ *
+ * # Provider mapping
+ *
+ * - **OpenAI** (and OpenAI-compatible providers): emitted verbatim as
+ *   `{"type": "json_schema", "json_schema": {...}}` per the
+ *   chat-completions spec.
+ * - **Gemini / Vertex AI**: translated to
+ *   `generationConfig.responseMimeType = "application/json"` and
+ *   `generationConfig.responseSchema = <schema>`. The `name`,
+ *   `description`, and `strict` fields are dropped — Gemini's
+ *   structured-output API does not consume them.
+ * - **Anthropic**: no native JSON mode. A system instruction is
+ *   prepended asking the model to respond with valid JSON.
+ *   `strict` is advisory only; callers should still validate the
+ *   returned JSON if the schema is load-bearing.
+ */
 export type ResponseFormat =
   | { type: "text" }
   | { type: "json_object" }
@@ -1447,8 +1550,13 @@ export interface StreamToolCall {
 
 /** System message guiding model behavior for the entire conversation. */
 export interface SystemMessage {
-  /** Instructions or context that apply throughout the conversation. */
-  readonly content?: string;
+  /**
+   * Instructions or context that apply throughout the conversation.
+   *
+   * Accepts either a plain text string or an array of content parts,
+   * mirroring [`UserContent`] so that `Message::system_with_parts` works.
+   */
+  readonly content?: UserContent;
   /** Optional name for the system message source. */
   readonly name?: string;
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@kreuzberg/liter-llm-node",
-  "version": "1.6.4",
+  "version": "1.7.0",
   "description": "Universal LLM API client with Rust-powered polyglot bindings.",
   "license": "MIT",
   "repository": {
@@ -33,12 +33,12 @@
     "@napi-rs/cli": "^3.6.2"
   },
   "optionalDependencies": {
-    "@kreuzberg/liter-llm-node-darwin-arm64": "1.6.4",
-    "@kreuzberg/liter-llm-node-darwin-x64": "1.6.4",
-    "@kreuzberg/liter-llm-node-linux-arm64-gnu": "1.6.4",
-    "@kreuzberg/liter-llm-node-linux-x64-gnu": "1.6.4",
-    "@kreuzberg/liter-llm-node-win32-arm64-msvc": "1.6.4",
-    "@kreuzberg/liter-llm-node-win32-x64-msvc": "1.6.4"
+    "@kreuzberg/liter-llm-node-darwin-arm64": "1.7.0",
+    "@kreuzberg/liter-llm-node-darwin-x64": "1.7.0",
+    "@kreuzberg/liter-llm-node-linux-arm64-gnu": "1.7.0",
+    "@kreuzberg/liter-llm-node-linux-x64-gnu": "1.7.0",
+    "@kreuzberg/liter-llm-node-win32-arm64-msvc": "1.7.0",
+    "@kreuzberg/liter-llm-node-win32-x64-msvc": "1.7.0"
   },
   "napi": {
     "binaryName": "liter-llm-node",