@ai-sdk/klingai 4.0.0-beta.5 → 4.0.0-beta.52

package/src/klingai-video-model.ts

@@ -1,7 +1,7 @@
 import {
   AISDKError,
-  type Experimental_VideoModelV4,
   NoSuchModelError,
+  type Experimental_VideoModelV4,
   type SharedV4Warning,
 } from '@ai-sdk/provider';
 import {
@@ -9,17 +9,19 @@ import {
   convertUint8ArrayToBase64,
   createJsonResponseHandler,
   delay,
-  type FetchFunction,
   getFromApi,
-  lazySchema,
   parseProviderOptions,
   postJsonToApi,
-  type Resolvable,
   resolve,
-  zodSchema,
+  type FetchFunction,
+  type Resolvable,
 } from '@ai-sdk/provider-utils';
 import { z } from 'zod/v4';
 import { klingaiFailedResponseHandler } from './klingai-error';
+import {
+  klingaiVideoModelOptionsSchema,
+  type KlingAIVideoModelOptions,
+} from './klingai-video-model-options';
 import type { KlingAIVideoModelId } from './klingai-video-settings';
 
 type KlingAIVideoMode = 't2v' | 'i2v' | 'motion-control';
@@ -59,283 +61,6 @@ function getApiModelName(modelId: string, mode: KlingAIVideoMode): string {
   return baseName.replace(/\.0$/, '').replace(/\./g, '-');
 }
 
-/**
- * Provider-specific options for KlingAI video generation.
- *
- * Not all options are supported by every model version and video mode (T2V, I2V,
- * motion control). See the KlingAI capability map for detailed compatibility:
- * https://app.klingai.com/global/dev/document-api/apiReference/model/skillsMap
- */
-export type KlingAIVideoModelOptions = {
-  /**
-   * Video generation mode.
-   *
-   * - `'std'`: Standard mode — cost-effective.
-   * - `'pro'`: Professional mode — higher quality but longer generation time.
-   */
-  mode?: 'std' | 'pro' | null;
-
-  /**
-   * Polling interval in milliseconds for checking task status.
-   * Default: 5000 (5 seconds).
-   */
-  pollIntervalMs?: number | null;
-
-  /**
-   * Maximum time in milliseconds to wait for video generation.
-   * Default: 600000 (10 minutes).
-   */
-  pollTimeoutMs?: number | null;
-
-  // --- T2V and I2V options ---
-
-  /**
-   * Negative text prompt to specify what to avoid.
-   * Cannot exceed 2500 characters.
-   */
-  negativePrompt?: string | null;
-
-  /**
-   * Whether to generate sound simultaneously when generating videos.
-   * Only V2.6 and subsequent versions support this parameter,
-   * and requires `mode: 'pro'`.
-   */
-  sound?: 'on' | 'off' | null;
-
-  /**
-   * Flexibility in video generation. The higher the value, the lower the
-   * model's flexibility, and the stronger the relevance to the user's prompt.
-   * Value range: [0, 1]. Kling-v2.x models do not support this parameter.
-   */
-  cfgScale?: number | null;
-
-  /**
-   * Camera movement control. If not specified, the model will intelligently
-   * match based on the input text/images.
-   */
-  cameraControl?: {
-    type:
-      | 'simple'
-      | 'down_back'
-      | 'forward_up'
-      | 'right_turn_forward'
-      | 'left_turn_forward';
-    config?: {
-      horizontal?: number | null;
-      vertical?: number | null;
-      pan?: number | null;
-      tilt?: number | null;
-      roll?: number | null;
-      zoom?: number | null;
-    } | null;
-  } | null;
-
-  // --- I2V-specific options ---
-
-  /**
-   * End frame image for I2V start+end frame control.
-   * Supports image URL or raw base64-encoded image data.
-   * Requires `mode: 'pro'` for most models.
-   */
-  imageTail?: string | null;
-
-  /**
-   * Static brush mask image for I2V motion brush.
-   * Supports image URL or raw base64-encoded image data.
-   */
-  staticMask?: string | null;
-
-  /**
-   * Dynamic brush configurations for I2V motion brush.
-   * Up to 6 groups, each with a mask and motion trajectories.
-   */
-  dynamicMasks?: Array<{
-    mask: string;
-    trajectories: Array<{ x: number; y: number }>;
-  }> | null;
-
-  // --- v3.0 multi-shot options (T2V and I2V) ---
-
-  /**
-   * Enable multi-shot video generation (Kling v3.0+).
-   * When true, the video is split into up to 6 storyboard shots
-   * with individual prompts and durations.
-   *
-   * When multiShot is true with shotType 'customize', multiPrompt is required.
-   * When multiShot is true, the main prompt parameter is ignored by the API.
-   */
-  multiShot?: boolean | null;
-
-  /**
-   * Storyboard method for multi-shot video generation (Kling v3.0+).
-   * Required when multiShot is true.
-   *
-   * - `'customize'`: User-defined shots via multiPrompt.
-   * - `'intelligence'`: Model auto-segments based on the main prompt.
-   */
-  shotType?: 'customize' | 'intelligence' | null;
-
-  /**
-   * Per-shot details for multi-shot video generation (Kling v3.0+).
-   * Required when multiShot is true and shotType is 'customize'.
-   *
-   * Up to 6 shots. Each shot has an index, prompt (max 512 chars),
-   * and duration in seconds. Shot durations must sum to the total duration.
-   */
-  multiPrompt?: Array<{
-    index: number;
-    prompt: string;
-    duration: string;
-  }> | null;
-
-  // --- v3.0 element control (I2V and Motion Control) ---
-
-  /**
-   * Reference elements for element control (Kling v3.0+).
-   * Supports video character elements and multi-image elements.
-   *
-   * - I2V: Up to 3 reference elements. Cannot coexist with voiceList.
-   * - Motion Control: Currently only 1 element supported.
-   *   When referencing an element, the generated video can only
-   *   refer to the orientation of the person in the video.
-   */
-  elementList?: Array<{
-    element_id: number;
-  }> | null;
-
-  // --- v3.0 voice control (T2V and I2V) ---
-
-  /**
-   * Voice references for voice control (Kling v3.0+).
-   * Up to 2 voice references. Referenced via `<<<voice_1>>>` template
-   * syntax in the prompt.
-   *
-   * When voiceList is used and the prompt references voice IDs,
-   * sound must be set to 'on'.
-   * Cannot coexist with elementList on the I2V endpoint.
-   */
-  voiceList?: Array<{
-    voice_id: string;
-  }> | null;
-
-  // --- Shared options ---
-
-  /**
-   * Whether to generate watermarked results simultaneously.
-   */
-  watermarkEnabled?: boolean | null;
-
-  // --- Motion-control-specific options ---
-
-  /**
-   * URL of the reference video. The character actions in the generated video
-   * are consistent with the reference video.
-   *
-   * Supports .mp4/.mov, max 100MB, side lengths 340px–3850px,
-   * duration 3–30 seconds (depends on `characterOrientation`).
-   */
-  videoUrl?: string | null;
-
-  /**
-   * Orientation of the characters in the generated video.
-   *
-   * - `'image'`: Same orientation as the person in the image.
-   *   Reference video duration max 10 seconds.
-   * - `'video'`: Same orientation as the person in the video.
-   *   Reference video duration max 30 seconds.
-   */
-  characterOrientation?: 'image' | 'video' | null;
-
-  /**
-   * Whether to keep the original sound of the reference video.
-   * Default: `'yes'`.
-   */
-  keepOriginalSound?: 'yes' | 'no' | null;
-
-  [key: string]: unknown; // For passthrough
-};
-
-const klingaiVideoModelOptionsSchema = lazySchema(() =>
-  zodSchema(
-    z
-      .object({
-        mode: z.enum(['std', 'pro']).nullish(),
-        pollIntervalMs: z.number().positive().nullish(),
-        pollTimeoutMs: z.number().positive().nullish(),
-        // T2V and I2V
-        negativePrompt: z.string().nullish(),
-        sound: z.enum(['on', 'off']).nullish(),
-        cfgScale: z.number().nullish(),
-        cameraControl: z
-          .object({
-            type: z.enum([
-              'simple',
-              'down_back',
-              'forward_up',
-              'right_turn_forward',
-              'left_turn_forward',
-            ]),
-            config: z
-              .object({
-                horizontal: z.number().nullish(),
-                vertical: z.number().nullish(),
-                pan: z.number().nullish(),
-                tilt: z.number().nullish(),
-                roll: z.number().nullish(),
-                zoom: z.number().nullish(),
-              })
-              .nullish(),
-          })
-          .nullish(),
-        // v3.0 multi-shot
-        multiShot: z.boolean().nullish(),
-        shotType: z.enum(['customize', 'intelligence']).nullish(),
-        multiPrompt: z
-          .array(
-            z.object({
-              index: z.number(),
-              prompt: z.string(),
-              duration: z.string(),
-            }),
-          )
-          .nullish(),
-        // v3.0 element control (I2V)
-        elementList: z
-          .array(
-            z.object({
-              element_id: z.number(),
-            }),
-          )
-          .nullish(),
-        // v3.0 voice control
-        voiceList: z
-          .array(
-            z.object({
-              voice_id: z.string(),
-            }),
-          )
-          .nullish(),
-        // I2V-specific
-        imageTail: z.string().nullish(),
-        staticMask: z.string().nullish(),
-        dynamicMasks: z
-          .array(
-            z.object({
-              mask: z.string(),
-              trajectories: z.array(z.object({ x: z.number(), y: z.number() })),
-            }),
-          )
-          .nullish(),
-        // Motion-control-specific
-        videoUrl: z.string().nullish(),
-        characterOrientation: z.enum(['image', 'video']).nullish(),
-        keepOriginalSound: z.enum(['yes', 'no']).nullish(),
-        watermarkEnabled: z.boolean().nullish(),
-      })
-      .passthrough(),
-  ),
-);
-
 /**
  * Known provider option keys that are handled explicitly and should not be
  * passed through to the API body.
@@ -594,8 +319,14 @@ export class KlingAIVideoModel implements Experimental_VideoModelV4 {
       body.negative_prompt = klingaiOptions.negativePrompt;
     }
 
-    if (klingaiOptions?.sound != null) {
-      body.sound = klingaiOptions.sound;
+    const sound =
+      options.generateAudio != null
+        ? options.generateAudio
+          ? 'on'
+          : 'off'
+        : klingaiOptions?.sound;
+    if (sound != null) {
+      body.sound = sound;
     }
 
     if (klingaiOptions?.cfgScale != null) {
@@ -692,8 +423,14 @@ export class KlingAIVideoModel implements Experimental_VideoModelV4 {
       body.negative_prompt = klingaiOptions.negativePrompt;
     }
 
-    if (klingaiOptions?.sound != null) {
-      body.sound = klingaiOptions.sound;
+    const sound =
+      options.generateAudio != null
+        ? options.generateAudio
+          ? 'on'
+          : 'off'
+        : klingaiOptions?.sound;
+    if (sound != null) {
+      body.sound = sound;
     }
 
     if (klingaiOptions?.cfgScale != null) {

package/src/version.ts

@@ -1,3 +1,6 @@
-declare const __PACKAGE_VERSION__: string;
-
-export const VERSION = __PACKAGE_VERSION__;
+// Version string of this package injected at build time.
+declare const __PACKAGE_VERSION__: string | undefined;
+export const VERSION: string =
+  typeof __PACKAGE_VERSION__ !== 'undefined'
+    ? __PACKAGE_VERSION__
+    : '0.0.0-test';

package/dist/index.d.mts

@@ -1,211 +0,0 @@
-import { ProviderV4, Experimental_VideoModelV4 } from '@ai-sdk/provider';
-import { FetchFunction } from '@ai-sdk/provider-utils';
-
-type KlingAIVideoModelId = 'kling-v1-t2v' | 'kling-v1.6-t2v' | 'kling-v2-master-t2v' | 'kling-v2.1-master-t2v' | 'kling-v2.5-turbo-t2v' | 'kling-v2.6-t2v' | 'kling-v3.0-t2v' | 'kling-v1-i2v' | 'kling-v1.5-i2v' | 'kling-v1.6-i2v' | 'kling-v2-master-i2v' | 'kling-v2.1-i2v' | 'kling-v2.1-master-i2v' | 'kling-v2.5-turbo-i2v' | 'kling-v2.6-i2v' | 'kling-v3.0-i2v' | 'kling-v2.6-motion-control' | 'kling-v3.0-motion-control' | (string & {});
-
-interface KlingAIProviderSettings {
-    /**
-     * KlingAI Access key. Default value is taken from the `KLINGAI_ACCESS_KEY`
-     * environment variable.
-     */
-    accessKey?: string;
-    /**
-     * KlingAI Secret key. Default value is taken from the `KLINGAI_SECRET_KEY`
-     * environment variable.
-     */
-    secretKey?: string;
-    /**
-     * Base URL for the API calls.
-     */
-    baseURL?: string;
-    /**
-     * Custom headers to include in the requests.
-     */
-    headers?: Record<string, string>;
-    /**
-     * Custom fetch implementation. You can use it as a middleware to intercept
-     * requests, or to provide a custom fetch implementation for e.g. testing.
-     */
-    fetch?: FetchFunction;
-}
-interface KlingAIProvider extends ProviderV4 {
-    /**
-     * Creates a model for video generation.
-     */
-    video(modelId: KlingAIVideoModelId): Experimental_VideoModelV4;
-    /**
-     * Creates a model for video generation.
-     */
-    videoModel(modelId: KlingAIVideoModelId): Experimental_VideoModelV4;
-}
-/**
- * Create a KlingAI provider instance.
- */
-declare function createKlingAI(options?: KlingAIProviderSettings): KlingAIProvider;
-/**
- * Default KlingAI provider instance.
- */
-declare const klingai: KlingAIProvider;
-
-/**
- * Provider-specific options for KlingAI video generation.
- *
- * Not all options are supported by every model version and video mode (T2V, I2V,
- * motion control). See the KlingAI capability map for detailed compatibility:
- * https://app.klingai.com/global/dev/document-api/apiReference/model/skillsMap
- */
-type KlingAIVideoModelOptions = {
-    /**
-     * Video generation mode.
-     *
-     * - `'std'`: Standard mode — cost-effective.
-     * - `'pro'`: Professional mode — higher quality but longer generation time.
-     */
-    mode?: 'std' | 'pro' | null;
-    /**
-     * Polling interval in milliseconds for checking task status.
-     * Default: 5000 (5 seconds).
-     */
-    pollIntervalMs?: number | null;
-    /**
-     * Maximum time in milliseconds to wait for video generation.
-     * Default: 600000 (10 minutes).
-     */
-    pollTimeoutMs?: number | null;
-    /**
-     * Negative text prompt to specify what to avoid.
-     * Cannot exceed 2500 characters.
-     */
-    negativePrompt?: string | null;
-    /**
-     * Whether to generate sound simultaneously when generating videos.
-     * Only V2.6 and subsequent versions support this parameter,
-     * and requires `mode: 'pro'`.
-     */
-    sound?: 'on' | 'off' | null;
-    /**
-     * Flexibility in video generation. The higher the value, the lower the
-     * model's flexibility, and the stronger the relevance to the user's prompt.
-     * Value range: [0, 1]. Kling-v2.x models do not support this parameter.
-     */
-    cfgScale?: number | null;
-    /**
-     * Camera movement control. If not specified, the model will intelligently
-     * match based on the input text/images.
-     */
-    cameraControl?: {
-        type: 'simple' | 'down_back' | 'forward_up' | 'right_turn_forward' | 'left_turn_forward';
-        config?: {
-            horizontal?: number | null;
-            vertical?: number | null;
-            pan?: number | null;
-            tilt?: number | null;
-            roll?: number | null;
-            zoom?: number | null;
-        } | null;
-    } | null;
-    /**
-     * End frame image for I2V start+end frame control.
-     * Supports image URL or raw base64-encoded image data.
-     * Requires `mode: 'pro'` for most models.
-     */
-    imageTail?: string | null;
-    /**
-     * Static brush mask image for I2V motion brush.
-     * Supports image URL or raw base64-encoded image data.
-     */
-    staticMask?: string | null;
-    /**
-     * Dynamic brush configurations for I2V motion brush.
-     * Up to 6 groups, each with a mask and motion trajectories.
-     */
-    dynamicMasks?: Array<{
-        mask: string;
-        trajectories: Array<{
-            x: number;
-            y: number;
-        }>;
-    }> | null;
-    /**
-     * Enable multi-shot video generation (Kling v3.0+).
-     * When true, the video is split into up to 6 storyboard shots
-     * with individual prompts and durations.
-     *
-     * When multiShot is true with shotType 'customize', multiPrompt is required.
-     * When multiShot is true, the main prompt parameter is ignored by the API.
-     */
-    multiShot?: boolean | null;
-    /**
-     * Storyboard method for multi-shot video generation (Kling v3.0+).
-     * Required when multiShot is true.
-     *
-     * - `'customize'`: User-defined shots via multiPrompt.
-     * - `'intelligence'`: Model auto-segments based on the main prompt.
-     */
-    shotType?: 'customize' | 'intelligence' | null;
-    /**
-     * Per-shot details for multi-shot video generation (Kling v3.0+).
-     * Required when multiShot is true and shotType is 'customize'.
-     *
-     * Up to 6 shots. Each shot has an index, prompt (max 512 chars),
-     * and duration in seconds. Shot durations must sum to the total duration.
-     */
-    multiPrompt?: Array<{
-        index: number;
-        prompt: string;
-        duration: string;
-    }> | null;
-    /**
-     * Reference elements for element control (Kling v3.0+).
-     * Supports video character elements and multi-image elements.
-     *
-     * - I2V: Up to 3 reference elements. Cannot coexist with voiceList.
-     * - Motion Control: Currently only 1 element supported.
-     *   When referencing an element, the generated video can only
-     *   refer to the orientation of the person in the video.
-     */
-    elementList?: Array<{
-        element_id: number;
-    }> | null;
-    /**
-     * Voice references for voice control (Kling v3.0+).
-     * Up to 2 voice references. Referenced via `<<<voice_1>>>` template
-     * syntax in the prompt.
-     *
-     * When voiceList is used and the prompt references voice IDs,
-     * sound must be set to 'on'.
-     * Cannot coexist with elementList on the I2V endpoint.
-     */
-    voiceList?: Array<{
-        voice_id: string;
-    }> | null;
-    /**
-     * Whether to generate watermarked results simultaneously.
-     */
-    watermarkEnabled?: boolean | null;
-    /**
-     * URL of the reference video. The character actions in the generated video
-     * are consistent with the reference video.
-     *
-     * Supports .mp4/.mov, max 100MB, side lengths 340px–3850px,
-     * duration 3–30 seconds (depends on `characterOrientation`).
-     */
-    videoUrl?: string | null;
-    /**
-     * Orientation of the characters in the generated video.
-     *
-     * - `'image'`: Same orientation as the person in the image.
-     *   Reference video duration max 10 seconds.
-     * - `'video'`: Same orientation as the person in the video.
-     *   Reference video duration max 30 seconds.
-     */
-    characterOrientation?: 'image' | 'video' | null;
-    /**
-     * Whether to keep the original sound of the reference video.
-     * Default: `'yes'`.
-     */
-    keepOriginalSound?: 'yes' | 'no' | null;
-    [key: string]: unknown;
-};
-
-export { type KlingAIProvider, type KlingAIProviderSettings, type KlingAIVideoModelId, type KlingAIVideoModelOptions, type KlingAIVideoModelOptions as KlingAIVideoProviderOptions, createKlingAI, klingai };