feat(video): add support for retrieving video content from OpenAI

Implement new interfaces and methods to handle video content retrieval from OpenAI, including type definitions and API client integration

src/renderer/src/aiCore/index_new.ts

@@ -14,7 +14,13 @@ import { addSpan, endSpan } from '@renderer/services/SpanManagerService'
 import { StartSpanParams } from '@renderer/trace/types/ModelSpanEntity'
 import type { Assistant, GenerateImageParams, Model, Provider } from '@renderer/types'
 import type { AiSdkModel, StreamTextParams } from '@renderer/types/aiCoreTypes'
-import { CreateVideoParams, CreateVideoResult, RetrieveVideoParams, RetrieveVideoResult } from '@renderer/types/video'
+import {
+  CreateVideoParams,
+  CreateVideoResult,
+  RetrieveVideoContentResult,
+  RetrieveVideoParams,
+  RetrieveVideoResult
+} from '@renderer/types/video'
 import { buildClaudeCodeSystemModelMessage } from '@shared/anthropic'
 import { type ImageModel, type LanguageModel, type Provider as AiSdkProvider, wrapLanguageModel } from 'ai'
 
@@ -515,6 +521,13 @@ export default class ModernAiProvider {
     return this.legacyProvider.retrieveVideo(params)
   }
 
+  /**
+   * We manually implement this method before aisdk supports it well
+   */
+  public async retrieveVideoContent(params: RetrieveVideoParams): Promise<RetrieveVideoContentResult> {
+    return this.legacyProvider.retrieveVideoContent(params)
+  }
+
   public getBaseURL(): string {
     return this.legacyProvider.getBaseURL()
   }

src/renderer/src/aiCore/legacy/clients/openai/OpenAIResponseAPIClient.ts

@@ -36,7 +36,7 @@ import {
   OpenAIResponseSdkTool,
   OpenAIResponseSdkToolCall
 } from '@renderer/types/sdk'
-import { CreateVideoParams, RetrieveVideoParams } from '@renderer/types/video'
+import { CreateVideoParams, RetrieveVideoContentParams, RetrieveVideoParams } from '@renderer/types/video'
 import { addImageFileToContents } from '@renderer/utils/formats'
 import {
   isSupportedToolUse,
@@ -163,6 +163,11 @@ export class OpenAIResponseAPIClient extends OpenAIBaseClient<
     return sdk.videos.retrieve(params.videoId, params.options)
   }
 
+  public async retrieveVideoContent(params: RetrieveVideoContentParams): Promise<Response> {
+    const sdk = await this.getSdkInstance()
+    return sdk.videos.downloadContent(params.videoId, params.query, params.options)
+  }
+
   private async handlePdfFile(file: FileMetadata): Promise<OpenAI.Responses.ResponseInputFile | undefined> {
     if (file.size > 32 * MB) return undefined
     try {

src/renderer/src/aiCore/legacy/index.ts

@@ -7,7 +7,13 @@ import { withSpanResult } from '@renderer/services/SpanManagerService'
 import { StartSpanParams } from '@renderer/trace/types/ModelSpanEntity'
 import type { GenerateImageParams, Model, Provider } from '@renderer/types'
 import type { RequestOptions, SdkModel } from '@renderer/types/sdk'
-import { CreateVideoParams, CreateVideoResult, RetrieveVideoParams, RetrieveVideoResult } from '@renderer/types/video'
+import {
+  CreateVideoParams,
+  CreateVideoResult,
+  RetrieveVideoContentResult,
+  RetrieveVideoParams,
+  RetrieveVideoResult
+} from '@renderer/types/video'
 import { isSupportedToolUse } from '@renderer/utils/mcp-tools'
 
 import { AihubmixAPIClient } from './clients/aihubmix/AihubmixAPIClient'
@@ -181,7 +187,7 @@ export default class AiProvider {
   }
 
   public async createVideo(params: CreateVideoParams): Promise<CreateVideoResult> {
-    if (this.apiClient instanceof OpenAIResponseAPIClient) {
+    if (this.apiClient instanceof OpenAIResponseAPIClient && params.type === 'openai') {
       const video = await this.apiClient.createVideo(params)
       return {
         type: 'openai',
@@ -204,6 +210,18 @@ export default class AiProvider {
     }
   }
 
+  public async retrieveVideoContent(params: RetrieveVideoParams): Promise<RetrieveVideoContentResult> {
+    if (this.apiClient instanceof OpenAIResponseAPIClient && params.type === 'openai') {
+      const response = await this.apiClient.retrieveVideoContent(params)
+      return {
+        type: 'openai',
+        response
+      }
+    } else {
+      throw new Error('Video generation is not supported by this provider')
+    }
+  }
+
   public getBaseURL(): string {
     return this.apiClient.getBaseURL()
   }

src/renderer/src/services/ApiService.ts

@@ -16,7 +16,13 @@ import type { StreamTextParams } from '@renderer/types/aiCoreTypes'
 import { type Chunk, ChunkType } from '@renderer/types/chunk'
 import { Message } from '@renderer/types/newMessage'
 import { SdkModel } from '@renderer/types/sdk'
-import { CreateVideoParams, CreateVideoResult, RetrieveVideoParams, RetrieveVideoResult } from '@renderer/types/video'
+import {
+  CreateVideoParams,
+  CreateVideoResult,
+  RetrieveVideoContentResult,
+  RetrieveVideoParams,
+  RetrieveVideoResult
+} from '@renderer/types/video'
 import { removeSpecialCharactersForTopicName, uuid } from '@renderer/utils'
 import { abortCompletion, readyToAbort } from '@renderer/utils/abortController'
 import { isAbortError } from '@renderer/utils/error'
@@ -408,6 +414,11 @@ export async function retrieveVideo(params: RetrieveVideoParams): Promise<Retrie
   return ai.retrieveVideo(params)
 }
 
+export async function retrieveVideoContent(params: RetrieveVideoParams): Promise<RetrieveVideoContentResult> {
+  const ai = new AiProviderNew(params.provider)
+  return ai.retrieveVideoContent(params)
+}
+
 export function hasApiKey(provider: Provider) {
   if (!provider) return false
   if (['ollama', 'lmstudio', 'vertexai', 'cherryai'].includes(provider.id)) return true

src/renderer/src/types/video.ts

@@ -54,3 +54,29 @@ export interface OpenAIRetrieveVideoResult extends RetrieveVideoBaseResult {
 }
 
 export type RetrieveVideoResult = OpenAIRetrieveVideoResult
+
+// Retrieve Video Content
+interface RetrieveVideoContentBaseParams {
+  type: VideoEndpointType
+  provider: Provider
+}
+
+export interface OpenAIRetrieveVideoContentParams extends RetrieveVideoContentBaseParams {
+  type: 'openai'
+  videoId: string
+  query?: OpenAI.Videos.VideoDownloadContentParams
+  options?: OpenAI.RequestOptions
+}
+
+export type RetrieveVideoContentParams = OpenAIRetrieveVideoContentParams
+
+interface RetrieveVideoContentBaseResult {
+  type: VideoEndpointType
+}
+
+export interface OpenAIRetrieveVideoContentResult extends RetrieveVideoContentBaseResult {
+  type: 'openai'
+  response: Response
+}
+
+export type RetrieveVideoContentResult = OpenAIRetrieveVideoContentResult