@aigne/openai 0.15.4 → 0.16.0-beta

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (7) hide show
  1. package/CHANGELOG.md +17 -0
  2. package/lib/cjs/openai-image-model.d.ts +3 -3
  3. package/lib/cjs/openai-image-model.js +13 -4
  4. package/lib/dts/openai-image-model.d.ts +3 -3
  5. package/lib/esm/openai-image-model.d.ts +3 -3
  6. package/lib/esm/openai-image-model.js +14 -5
  7. package/package.json +4 -4
package/CHANGELOG.md CHANGED
@@ -1,5 +1,22 @@
1
1
  # Changelog
2
2
 
3
+ ## [0.16.0-beta](https://github.com/AIGNE-io/aigne-framework/compare/openai-v0.15.4...openai-v0.16.0-beta) (2025-09-22)
4
+
5
+
6
+ ### Features
7
+
8
+ * improve image model architecture and file handling ([#527](https://github.com/AIGNE-io/aigne-framework/issues/527)) ([4db50aa](https://github.com/AIGNE-io/aigne-framework/commit/4db50aa0387a1a0f045ca11aaa61613e36ca7597))
9
+
10
+
11
+ ### Dependencies
12
+
13
+ * The following workspace dependencies were updated
14
+ * dependencies
15
+ * @aigne/core bumped to 1.61.0-beta
16
+ * @aigne/platform-helpers bumped to 0.6.3-beta
17
+ * devDependencies
18
+ * @aigne/test-utils bumped to 0.5.53-beta
19
+
3
20
  ## [0.15.4](https://github.com/AIGNE-io/aigne-framework/compare/openai-v0.15.3...openai-v0.15.4) (2025-09-18)
4
21
 
5
22
 
package/lib/cjs/openai-image-model.d.ts CHANGED
@@ -1,8 +1,8 @@
1
- import { ImageModel, type ImageModelInput, type ImageModelOptions, type ImageModelOutput } from "@aigne/core";
1
+ import { type AgentInvokeOptions, ImageModel, type ImageModelInput, type ImageModelOptions, type ImageModelOutput } from "@aigne/core";
2
2
  import { type Camelize } from "@aigne/core/utils/camelize.js";
3
3
  import type OpenAI from "openai";
4
4
  import type { ClientOptions } from "openai";
5
- export interface OpenAIImageModelInput extends ImageModelInput, Camelize<Omit<OpenAI.ImageGenerateParams, "prompt" | "model" | "n" | "response_format">> {
5
+ export interface OpenAIImageModelInput extends ImageModelInput, Camelize<Omit<OpenAI.ImageGenerateParams | OpenAI.ImageEditParams, "prompt" | "model" | "n" | "response_format">> {
6
6
  }
7
7
  export interface OpenAIImageModelOutput extends ImageModelOutput {
8
8
  }
@@ -51,5 +51,5 @@ export declare class OpenAIImageModel extends ImageModel<OpenAIImageModelInput,
51
51
  * @param input The input to process
52
52
  * @returns The generated response
53
53
  */
54
- process(input: OpenAIImageModelInput): Promise<OpenAIImageModelOutput>;
54
+ process(input: OpenAIImageModelInput, options: AgentInvokeOptions): Promise<OpenAIImageModelOutput>;
55
55
  }
package/lib/cjs/openai-image-model.js CHANGED
@@ -7,6 +7,7 @@ const type_utils_js_1 = require("@aigne/core/utils/type-utils.js");
7
7
  const zod_1 = require("zod");
8
8
  const openai_js_1 = require("./openai.js");
9
9
  const DEFAULT_MODEL = "dall-e-2";
10
+ const OUTPUT_MIME_TYPE = "image/png";
10
11
  const openAIImageModelInputSchema = core_1.imageModelInputSchema.extend({});
11
12
  const openAIImageModelOptionsSchema = zod_1.z.object({
12
13
  apiKey: zod_1.z.string().optional(),
@@ -57,7 +58,7 @@ class OpenAIImageModel extends core_1.ImageModel {
57
58
  * @param input The input to process
58
59
  * @returns The generated response
59
60
  */
60
- async process(input) {
61
+ async process(input, options) {
61
62
  const model = input.model || this.credential.model;
62
63
  const map = {
63
64
  "dall-e-2": ["prompt", "size", "n"],
@@ -83,13 +84,21 @@ class OpenAIImageModel extends core_1.ImageModel {
83
84
  response_format: responseFormat,
84
85
  model,
85
86
  };
86
- const response = await this.client.images.generate({ ...body, stream: false });
87
+ const response = input.image?.length
88
+ ? await this.client.images.edit({
89
+ ...body,
90
+ image: await Promise.all(input.image.map((image) => this.transformFileOutput(core_1.FileOutputType.file, image, options).then((file) => new File([Buffer.from(file.data, "base64")], file.filename || "image.png", {
91
+ type: file.mimeType,
92
+ })))),
93
+ stream: false,
94
+ })
95
+ : await this.client.images.generate({ ...body, stream: false });
87
96
  return {
88
97
  images: (response.data ?? []).map((image) => {
89
98
  if (image.url)
90
- return { url: image.url };
99
+ return { type: "url", url: image.url, mimeType: OUTPUT_MIME_TYPE };
91
100
  if (image.b64_json)
92
- return { base64: image.b64_json };
101
+ return { type: "file", data: image.b64_json, mimeType: OUTPUT_MIME_TYPE };
93
102
  throw new Error("Image response does not contain a valid URL or base64 data");
94
103
  }),
95
104
  usage: {
package/lib/dts/openai-image-model.d.ts CHANGED
@@ -1,8 +1,8 @@
1
- import { ImageModel, type ImageModelInput, type ImageModelOptions, type ImageModelOutput } from "@aigne/core";
1
+ import { type AgentInvokeOptions, ImageModel, type ImageModelInput, type ImageModelOptions, type ImageModelOutput } from "@aigne/core";
2
2
  import { type Camelize } from "@aigne/core/utils/camelize.js";
3
3
  import type OpenAI from "openai";
4
4
  import type { ClientOptions } from "openai";
5
- export interface OpenAIImageModelInput extends ImageModelInput, Camelize<Omit<OpenAI.ImageGenerateParams, "prompt" | "model" | "n" | "response_format">> {
5
+ export interface OpenAIImageModelInput extends ImageModelInput, Camelize<Omit<OpenAI.ImageGenerateParams | OpenAI.ImageEditParams, "prompt" | "model" | "n" | "response_format">> {
6
6
  }
7
7
  export interface OpenAIImageModelOutput extends ImageModelOutput {
8
8
  }
@@ -51,5 +51,5 @@ export declare class OpenAIImageModel extends ImageModel<OpenAIImageModelInput,
51
51
  * @param input The input to process
52
52
  * @returns The generated response
53
53
  */
54
- process(input: OpenAIImageModelInput): Promise<OpenAIImageModelOutput>;
54
+ process(input: OpenAIImageModelInput, options: AgentInvokeOptions): Promise<OpenAIImageModelOutput>;
55
55
  }
package/lib/esm/openai-image-model.d.ts CHANGED
@@ -1,8 +1,8 @@
1
- import { ImageModel, type ImageModelInput, type ImageModelOptions, type ImageModelOutput } from "@aigne/core";
1
+ import { type AgentInvokeOptions, ImageModel, type ImageModelInput, type ImageModelOptions, type ImageModelOutput } from "@aigne/core";
2
2
  import { type Camelize } from "@aigne/core/utils/camelize.js";
3
3
  import type OpenAI from "openai";
4
4
  import type { ClientOptions } from "openai";
5
- export interface OpenAIImageModelInput extends ImageModelInput, Camelize<Omit<OpenAI.ImageGenerateParams, "prompt" | "model" | "n" | "response_format">> {
5
+ export interface OpenAIImageModelInput extends ImageModelInput, Camelize<Omit<OpenAI.ImageGenerateParams | OpenAI.ImageEditParams, "prompt" | "model" | "n" | "response_format">> {
6
6
  }
7
7
  export interface OpenAIImageModelOutput extends ImageModelOutput {
8
8
  }
@@ -51,5 +51,5 @@ export declare class OpenAIImageModel extends ImageModel<OpenAIImageModelInput,
51
51
  * @param input The input to process
52
52
  * @returns The generated response
53
53
  */
54
- process(input: OpenAIImageModelInput): Promise<OpenAIImageModelOutput>;
54
+ process(input: OpenAIImageModelInput, options: AgentInvokeOptions): Promise<OpenAIImageModelOutput>;
55
55
  }
package/lib/esm/openai-image-model.js CHANGED
@@ -1,9 +1,10 @@
1
- import { ImageModel, imageModelInputSchema, } from "@aigne/core";
1
+ import { FileOutputType, ImageModel, imageModelInputSchema, } from "@aigne/core";
2
2
  import { snakelize } from "@aigne/core/utils/camelize.js";
3
3
  import { checkArguments, pick } from "@aigne/core/utils/type-utils.js";
4
4
  import { z } from "zod";
5
5
  import { CustomOpenAI } from "./openai.js";
6
6
  const DEFAULT_MODEL = "dall-e-2";
7
+ const OUTPUT_MIME_TYPE = "image/png";
7
8
  const openAIImageModelInputSchema = imageModelInputSchema.extend({});
8
9
  const openAIImageModelOptionsSchema = z.object({
9
10
  apiKey: z.string().optional(),
@@ -54,7 +55,7 @@ export class OpenAIImageModel extends ImageModel {
54
55
  * @param input The input to process
55
56
  * @returns The generated response
56
57
  */
57
- async process(input) {
58
+ async process(input, options) {
58
59
  const model = input.model || this.credential.model;
59
60
  const map = {
60
61
  "dall-e-2": ["prompt", "size", "n"],
@@ -80,13 +81,21 @@ export class OpenAIImageModel extends ImageModel {
80
81
  response_format: responseFormat,
81
82
  model,
82
83
  };
83
- const response = await this.client.images.generate({ ...body, stream: false });
84
+ const response = input.image?.length
85
+ ? await this.client.images.edit({
86
+ ...body,
87
+ image: await Promise.all(input.image.map((image) => this.transformFileOutput(FileOutputType.file, image, options).then((file) => new File([Buffer.from(file.data, "base64")], file.filename || "image.png", {
88
+ type: file.mimeType,
89
+ })))),
90
+ stream: false,
91
+ })
92
+ : await this.client.images.generate({ ...body, stream: false });
84
93
  return {
85
94
  images: (response.data ?? []).map((image) => {
86
95
  if (image.url)
87
- return { url: image.url };
96
+ return { type: "url", url: image.url, mimeType: OUTPUT_MIME_TYPE };
88
97
  if (image.b64_json)
89
- return { base64: image.b64_json };
98
+ return { type: "file", data: image.b64_json, mimeType: OUTPUT_MIME_TYPE };
90
99
  throw new Error("Image response does not contain a valid URL or base64 data");
91
100
  }),
92
101
  usage: {
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@aigne/openai",
3
- "version": "0.15.4",
3
+ "version": "0.16.0-beta",
4
4
  "description": "AIGNE OpenAI SDK for integrating with OpenAI's GPT models and API services",
5
5
  "publishConfig": {
6
6
  "access": "public"
@@ -39,8 +39,8 @@
39
39
  "openai": "^5.20.3",
40
40
  "uuid": "^13.0.0",
41
41
  "zod": "^3.25.67",
42
- "@aigne/core": "^1.60.3",
43
- "@aigne/platform-helpers": "^0.6.2"
42
+ "@aigne/platform-helpers": "^0.6.3-beta",
43
+ "@aigne/core": "^1.61.0-beta"
44
44
  },
45
45
  "devDependencies": {
46
46
  "@types/bun": "^1.2.22",
@@ -48,7 +48,7 @@
48
48
  "npm-run-all": "^4.1.5",
49
49
  "rimraf": "^6.0.1",
50
50
  "typescript": "^5.9.2",
51
- "@aigne/test-utils": "^0.5.52"
51
+ "@aigne/test-utils": "^0.5.53-beta"
52
52
  },
53
53
  "scripts": {
54
54
  "lint": "tsc --noEmit",