cleaner implementation for image edits route

Author: narengogi

src/handlers/responseHandlers.ts

@@ -13,6 +13,7 @@ import {
   handleOctetStreamResponse,
   handleStreamingMode,
   handleTextResponse,
+  handleTextStreamResponse,
 } from './streamHandler';
 import { HookSpan } from '../middlewares/hooks';
 import { env } from 'hono/adapter';
@@ -152,6 +153,10 @@ export async function responseHandler(
     return { response: textResponse, responseJson: null };
   }
 
+  if (responseContentType?.startsWith(CONTENT_TYPES.EVENT_STREAM)) {
+    return { response: handleTextStreamResponse(response), responseJson: null };
+  }
+
   if (!responseContentType && response.status === 204) {
     return {
       response: new Response(response.body, response),

src/handlers/streamHandler.ts

@@ -282,6 +282,10 @@ export function handleOctetStreamResponse(response: Response) {
   return new Response(response.body, response);
 }
 
+export function handleTextStreamResponse(response: Response) {
+  return new Response(response.body, response);
+}
+
 export function handleImageResponse(response: Response) {
   return new Response(response.body, response);
 }

src/providers/azure-ai-inference/api.ts

@@ -50,9 +50,12 @@ const AzureAIInferenceAPI: ProviderAPIConfig = {
       ...(azureDeploymentName && {
         'azureml-model-deployment': azureDeploymentName,
       }),
-      ...(['createTranscription', 'createTranslation', 'uploadFile'].includes(
-        fn
-      )
+      ...([
+        'createTranscription',
+        'createTranslation',
+        'uploadFile',
+        'imageEdit',
+      ].includes(fn)
         ? {
             'Content-Type': 'multipart/form-data',
           }
@@ -119,6 +122,7 @@ const AzureAIInferenceAPI: ProviderAPIConfig = {
       embed: '/embeddings',
       realtime: '/realtime',
       imageGenerate: '/images/generations',
+      imageEdit: '/images/edits',
       createSpeech: '/audio/speech',
       createTranscription: '/audio/transcriptions',
       createTranslation: '/audio/translations',
@@ -165,6 +169,7 @@ const AzureAIInferenceAPI: ProviderAPIConfig = {
       }
       case 'realtime':
       case 'imageGenerate':
+      case 'imageEdit':
       case 'createSpeech':
       case 'createTranscription':
       case 'createTranslation':

src/providers/azure-ai-inference/index.ts

@@ -20,6 +20,7 @@ import { AzureOpenAICreateBatchConfig } from '../azure-openai/createBatch';
 import { AzureAIInferenceGetBatchOutputRequestHandler } from './getBatchOutput';
 import { OpenAIFileUploadRequestTransform } from '../openai/uploadFile';
 import {
+  AzureAIInferenceCreateImageEditResponseTransform,
   AzureAIInferenceCreateSpeechResponseTransform,
   AzureAIInferenceCreateTranscriptionResponseTransform,
   AzureAIInferenceCreateTranslationResponseTransform,
@@ -32,6 +33,7 @@ const AzureAIInferenceAPIConfig: ProviderConfigs = {
   api: AzureAIInferenceAPI,
   chatComplete: AzureAIInferenceChatCompleteConfig,
   imageGenerate: AzureOpenAIImageGenerateConfig,
+  imageEdit: {},
   createSpeech: AzureOpenAICreateSpeechConfig,
   createFinetune: OpenAICreateFinetuneConfig,
   createTranscription: {},
@@ -52,6 +54,7 @@ const AzureAIInferenceAPIConfig: ProviderConfigs = {
       AzureAIInferenceChatCompleteResponseTransform(AZURE_AI_INFERENCE),
     embed: AzureAIInferenceEmbedResponseTransform(AZURE_AI_INFERENCE),
     imageGenerate: AzureAIInferenceResponseTransform,
+    imageEdit: AzureAIInferenceCreateImageEditResponseTransform,
     createSpeech: AzureAIInferenceCreateSpeechResponseTransform,
     createTranscription: AzureAIInferenceCreateTranscriptionResponseTransform,
     createTranslation: AzureAIInferenceCreateTranslationResponseTransform,

src/providers/azure-ai-inference/utils.ts

@@ -1,6 +1,5 @@
 import { AZURE_AI_INFERENCE } from '../../globals';
 import { OpenAIErrorResponseTransform } from '../openai/utils';
-import { ErrorResponse } from '../types';
 
 export const AzureAIInferenceResponseTransform = (
   response: any,
@@ -35,6 +34,17 @@ export const AzureAIInferenceCreateTranscriptionResponseTransform = (
   return { ...response, provider: AZURE_AI_INFERENCE };
 };
 
+export const AzureAIInferenceCreateImageEditResponseTransform = (
+  response: any,
+  responseStatus: number
+) => {
+  if (responseStatus !== 200 && 'error' in response) {
+    return OpenAIErrorResponseTransform(response, AZURE_AI_INFERENCE);
+  }
+
+  return { ...response, provider: AZURE_AI_INFERENCE };
+};
+
 export const AzureAIInferenceCreateTranslationResponseTransform = (
   response: any,
   responseStatus: number

src/providers/azure-openai/api.ts

@@ -45,7 +45,8 @@ const AzureOpenAIAPIConfig: ProviderAPIConfig = {
     if (
       fn === 'createTranscription' ||
       fn === 'createTranslation' ||
-      fn === 'uploadFile'
+      fn === 'uploadFile' ||
+      fn === 'imageEdit'
     ) {
       headersObj['Content-Type'] = 'multipart/form-data';
     }

src/providers/azure-openai/imageEdits.ts

@@ -1,63 +1,6 @@
 import { AZURE_OPEN_AI } from '../../globals';
 import { OpenAIErrorResponseTransform } from '../openai/utils';
-import { ErrorResponse, ImageGenerateResponse, ProviderConfig } from '../types';
-
-export const AzureOpenAIImageEditConfig: ProviderConfig = {
-  image: {
-    param: 'image',
-    required: true,
-  },
-  prompt: {
-    param: 'prompt',
-    required: true,
-  },
-  background: {
-    param: 'background',
-  },
-  input_fidelity: {
-    param: 'input_fidelity',
-  },
-  mask: {
-    param: 'mask',
-  },
-  model: {
-    param: 'model',
-    default: 'dall-e-2',
-  },
-  n: {
-    param: 'n',
-    min: 1,
-    max: 10,
-  },
-  output_compression: {
-    param: 'output_compression',
-    min: 0,
-    max: 100,
-  },
-  output_format: {
-    param: 'output_format',
-  },
-  partial_images: {
-    param: 'partial_images',
-    min: 0,
-    max: 3,
-  },
-  quality: {
-    param: 'quality',
-  },
-  response_format: {
-    param: 'response_format',
-  },
-  size: {
-    param: 'size',
-  },
-  stream: {
-    param: 'stream',
-  },
-  user: {
-    param: 'user',
-  },
-};
+import { ErrorResponse, ImageGenerateResponse } from '../types';
 
 interface AzureOpenAIImageObject {
   b64_json?: string; // The base64-encoded JSON of the generated image, if response_format is b64_json.

src/providers/azure-openai/index.ts

@@ -36,17 +36,14 @@ import {
   OpenAIListInputItemsResponseTransformer,
 } from '../open-ai-base';
 import { AZURE_OPEN_AI } from '../../globals';
-import {
-  AzureOpenAIImageEditConfig,
-  AzureOpenAIImageEditResponseTransform,
-} from './imageEdits';
+import { AzureOpenAIImageEditResponseTransform } from './imageEdits';
 
 const AzureOpenAIConfig: ProviderConfigs = {
   complete: AzureOpenAICompleteConfig,
   embed: AzureOpenAIEmbedConfig,
   api: AzureOpenAIAPIConfig,
   imageGenerate: AzureOpenAIImageGenerateConfig,
-  imageEdit: AzureOpenAIImageEditConfig,
+  imageEdit: {},
   chatComplete: AzureOpenAIChatCompleteConfig,
   createSpeech: AzureOpenAICreateSpeechConfig,
   createFinetune: OpenAICreateFinetuneConfig,

src/providers/openai/api.ts

@@ -17,7 +17,8 @@ const OpenAIAPIConfig: ProviderAPIConfig = {
     if (
       fn === 'createTranscription' ||
       fn === 'createTranslation' ||
-      fn === 'uploadFile'
+      fn === 'uploadFile' ||
+      fn === 'imageEdit'
     )
       headersObj['Content-Type'] = 'multipart/form-data';
 

src/providers/openai/imageEdits.ts

@@ -1,64 +1,7 @@
 import { OPEN_AI } from '../../globals';
-import { ErrorResponse, ImageGenerateResponse, ProviderConfig } from '../types';
+import { ErrorResponse, ImageGenerateResponse } from '../types';
 import { OpenAIErrorResponseTransform } from './utils';
 
-export const OpenAIImageEditConfig: ProviderConfig = {
-  image: {
-    param: 'image',
-    required: true,
-  },
-  prompt: {
-    param: 'prompt',
-    required: true,
-  },
-  background: {
-    param: 'background',
-  },
-  input_fidelity: {
-    param: 'input_fidelity',
-  },
-  mask: {
-    param: 'mask',
-  },
-  model: {
-    param: 'model',
-    default: 'dall-e-2',
-  },
-  n: {
-    param: 'n',
-    min: 1,
-    max: 10,
-  },
-  output_compression: {
-    param: 'output_compression',
-    min: 0,
-    max: 100,
-  },
-  output_format: {
-    param: 'output_format',
-  },
-  partial_images: {
-    param: 'partial_images',
-    min: 0,
-    max: 3,
-  },
-  quality: {
-    param: 'quality',
-  },
-  response_format: {
-    param: 'response_format',
-  },
-  size: {
-    param: 'size',
-  },
-  stream: {
-    param: 'stream',
-  },
-  user: {
-    param: 'user',
-  },
-};
-
 interface OpenAIImageObject {
   b64_json?: string; // The base64-encoded JSON of the generated image, if response_format is b64_json.
   url?: string; // The URL of the generated image, if response_format is url (default).