@vanillaskyai/video 0.6.0 → 0.7.1

package/dist/{reaction-2XCLOJOK.js → reaction-SLEGR3BE.js}

@@ -1,9 +1,9 @@
 import {
   SceneBackground
-} from "./chunk-7BGAU6C3.js";
+} from "./chunk-MXZSDGZQ.js";
 import {
   TemplateText
-} from "./chunk-TCEHK2BW.js";
+} from "./chunk-YJJC4N4D.js";
 import "./chunk-L5BTTCZV.js";
 import "./chunk-FRN6WKHA.js";
 import "./chunk-W3VDIBGR.js";

package/dist/server.d.ts

@@ -1,7 +1,9 @@
-import { n as VideoGenerationContext, o as VideoPlanner, p as VideoRequest, a as VideoCapabilities, j as VideoInput, d as VideoAudio, q as VideoSceneValidator, r as VideoTemplatePacing, s as VideoSnapshotRetention, t as VideoResumeCursor, u as VideoSceneValidationContext, e as VideoScene } from './types-BV9IqExh.js';
-import { a as VideoFinishReason, V as VideoWarning, b as VideoEvent } from './events-BBP30j3c.js';
-export { c as VideoWarningCategory } from './events-BBP30j3c.js';
+import { o as VideoGenerationContext, p as VideoPlanner, q as VideoRequest, a as VideoCapabilities, g as VideoInput, d as VideoAudio, r as VideoSceneValidator, j as VideoTemplatePacing, s as VideoSnapshotRetention, t as VideoResumeCursor, u as VideoSceneValidationContext, e as VideoScene, b as VideoOrientation } from './types-2wHBqtg8.js';
+import { a as VideoFinishReason, V as VideoWarning, b as VideoEvent } from './events-B6qS1Lsb.js';
+export { c as VideoWarningCategory } from './events-B6qS1Lsb.js';
 import { S as SceneTemplateMetadata } from './catalog-types-BIhSpOWK.js';
+import { V as VideoChatWelcomeOptions } from './types-DrABlRa7.js';
+export { a as VideoChatCapabilities, b as VideoChatConversationTurn, c as VideoChatMode, d as VideoChatWelcomePrompt } from './types-DrABlRa7.js';
 
 interface VideoProviderUsage {
     inputTokens?: number;
@@ -133,6 +135,12 @@ interface MediaResolverContext {
     scene: Readonly<VideoScene>;
     templateId: string;
     preferredType: "image" | "video" | "any";
+    /**
+     * The visual language generated media must match, when the application set
+     * one. Append it to a provider prompt; a shot that ignores it is the
+     * mismatch this exists to prevent.
+     */
+    generatedLook?: string;
     signal: AbortSignal;
 }
 type MediaResolver = (query: string, context: MediaResolverContext) => ResolvedMedia | null | Promise<ResolvedMedia | null>;
@@ -150,6 +158,32 @@ interface VideoHandlerOptions extends Omit<VideoStreamHandlerOptions, "generate"
     allowMediaUrl?: Parameters<typeof createTemplateSceneValidator>[0]["allowMediaUrl"];
     /** Resolve bounded semantic media intent through an application-owned provider. */
     resolveMedia?: MediaResolver;
+    /**
+     * How many scenes may resolve media at once. Defaults to one.
+     *
+     * Raise it when resolution is slow enough to be felt - generated video rather
+     * than a stock search. Scenes are still emitted in the order they were
+     * planned; only the waiting overlaps.
+     */
+    mediaConcurrency?: number;
+    /**
+     * How many scenes in one request may resolve media at all.
+     *
+     * Unbounded by default, which is right when media is searched for and wrong
+     * when it is generated: the planner decides the scene count, and every scene
+     * is then a paid clip. Past the ceiling a scene keeps its copy on the brand
+     * gradient, and a `media_budget_reached` warning says it happened.
+     */
+    maxResolvedMedia?: number;
+    /**
+     * Ask the planner to write each scene's spoken line, on the scene itself.
+     *
+     * A narrated video otherwise costs a round trip per scene - handing a model
+     * the scene that was just planned and asking what to say over it - and those
+     * calls have to be chained, because a line is written knowing the ones
+     * before it. The planner already knows the scene and the ones around it.
+     */
+    narrate?: boolean;
 }
 /**
  * Build the secure route for a template kit while leaving provider, model,
@@ -157,4 +191,68 @@ interface VideoHandlerOptions extends Omit<VideoStreamHandlerOptions, "generate"
  */
 declare function createVideoHandler(options: VideoHandlerOptions): VideoStreamHandler;
 
-export { type MediaResolver, type MediaResolverContext, type ResolvedMedia, SceneTemplateMetadata as ServerTemplateMetadata, type ServerTemplateRegistry, VideoFinishReason, type VideoGenerationSummary, type VideoHandlerOptions, type VideoProviderUsage, VideoWarning, createServerTemplateRegistry, createVideoHandler };
+type VideoChatTextTask = "narration" | "suggestions";
+interface VideoChatTextContext {
+    task: VideoChatTextTask;
+    systemPrompt: string;
+    userPrompt: string;
+    maxOutputTokens: number;
+    signal: AbortSignal;
+}
+type VideoChatTextGenerator = (context: VideoChatTextContext) => string | Promise<string>;
+interface VideoChatSpeechContext {
+    text: string;
+    signal: AbortSignal;
+}
+interface VideoChatSpeechResult {
+    audio: Uint8Array | ArrayBuffer;
+    mediaType?: string;
+}
+type VideoChatSpeechGenerator = (context: VideoChatSpeechContext) => VideoChatSpeechResult | Promise<VideoChatSpeechResult>;
+interface VideoChatTranscriptionContext {
+    audio: Uint8Array;
+    mediaType: string;
+    signal: AbortSignal;
+}
+type VideoChatTranscriber = (context: VideoChatTranscriptionContext) => string | Promise<string>;
+interface VideoChatMediaContext {
+    purpose: "response" | "welcome" | "suggestion";
+    orientation: VideoOrientation;
+    generatedLook?: string;
+    signal: AbortSignal;
+    requestId?: string;
+    scene?: Readonly<VideoScene>;
+    templateId?: string;
+    preferredType?: "image" | "video" | "any";
+}
+type VideoChatMediaResolver = (query: string, context: VideoChatMediaContext) => ResolvedMedia | null | Promise<ResolvedMedia | null>;
+interface VideoChatHandlerOptions extends Omit<VideoHandlerOptions, "allowMediaUrl" | "basePrompt" | "maxResolvedMedia" | "narrate" | "resolveMedia"> {
+    /** Generate the small non-streaming text tasks around the visual plan. */
+    generateText: VideoChatTextGenerator;
+    /** Optional generated speech. Browsers can speak locally when absent. */
+    generateSpeech?: VideoChatSpeechGenerator;
+    /** Optional server transcription used when browser recognition is unavailable. */
+    transcribe?: VideoChatTranscriber;
+    /** Optional stock or application-owned media search. */
+    searchMedia?: VideoChatMediaResolver;
+    /** Optional generated-video provider. Its presence enables paid visual modes. */
+    generateVideo?: VideoChatMediaResolver;
+    /** Trusted application guidance appended to the general-purpose response brief. */
+    instructions?: string;
+    /** Application-owned prompts and visual searches shown before the first turn. */
+    welcome?: VideoChatWelcomeOptions;
+    /** Maximum accepted transcription body. Defaults to 8 MiB. */
+    maxAudioBytes?: number;
+}
+type VideoChatHandler = (request: Request) => Promise<Response>;
+/**
+ * Create the complete provider-neutral server endpoint for a video chat.
+ *
+ * Mount the returned handler once and select operations with the `action`
+ * query parameter. Provider libraries and credentials stay in the application
+ * closures passed here; only bounded prompts, media, and capability booleans
+ * cross the browser boundary.
+ */
+declare function createVideoChatHandler(options: VideoChatHandlerOptions): VideoChatHandler;
+
+export { type MediaResolver, type MediaResolverContext, type ResolvedMedia, SceneTemplateMetadata as ServerTemplateMetadata, type ServerTemplateRegistry, type VideoChatHandlerOptions, VideoChatWelcomeOptions, VideoFinishReason, type VideoGenerationSummary, type VideoHandlerOptions, type VideoProviderUsage, VideoWarning, createServerTemplateRegistry, createVideoChatHandler, createVideoHandler };