@samchon/openapi

1flowchart
2  subgraph "OpenAPI Specification"
3    v20("Swagger v2.0") --upgrades--> emended[["OpenAPI v3.1 (emended)"]]
4    v30("OpenAPI v3.0") --upgrades--> emended
5    v31("OpenAPI v3.1") --emends--> emended
6  end
7  subgraph "OpenAPI Generator"
8    emended --normalizes--> migration[["Migration Schema"]]
9    migration --"Artificial Intelligence"--> lfc{{"LLM Function Calling"}}
10    lfc --"OpenAI"--> chatgpt("ChatGPT")
11    lfc --"Anthropic"--> claude("Claude")
12    lfc --"Google"--> gemini("Gemini")
13    lfc --"Meta"--> llama("Llama")
14  end

OpenAPI definitions, converters and LLM function calling application composer.

@samchon/openapi is a collection of OpenAPI types for every versions, and converters for them. In the OpenAPI types, there is an "emended" OpenAPI v3.1 specification, which has removed ambiguous and duplicated expressions for the clarity. Every conversions are based on the emended OpenAPI v3.1 specification.

1. Swagger v2.0
2. OpenAPI v3.0
3. OpenAPI v3.1
4. OpenAPI v3.1 emended

@samchon/openapi also provides LLM (Large Language Model) function calling application composer from the OpenAPI document with many strategies. With the HttpLlm module, you can perform the LLM funtion calling extremely easily just by delivering the OpenAPI (Swagger) document.

• HttpLlm.application()
• IHttpLlmApplication<Model>
• IHttpLlmFunction<Model>
• Supported schemas
  • IChatGptSchema: OpenAI ChatGPT
  • IClaudeSchema: Anthropic Claude
  • IGeminiSchema: Google Gemini
  • ILlamaSchema: Meta Llama
• Midldle layer schemas
  • ILlmSchemaV3: middle layer based on OpenAPI v3.0 specification
  • ILlmSchemaV3_1: middle layer based on OpenAPI v3.1 specification

Setup

1npm install @samchon/openapi

Just install by npm i @samchon/openapi command.

Here is an example code utilizing the @samchon/openapi for LLM function calling purpose.

1import {
2  HttpLlm,
3  IChatGptSchema,
4  IHttpLlmApplication,
5  IHttpLlmFunction,
6  OpenApi,
7  OpenApiV3,
8  OpenApiV3_1,
9  SwaggerV2,
10} from "@samchon/openapi";
11import fs from "fs";
12import typia from "typia";
13
14const main = async (): Promise<void> => {
15  // read swagger document and validate it
16  const swagger:
17    | SwaggerV2.IDocument
18    | OpenApiV3.IDocument
19    | OpenApiV3_1.IDocument = JSON.parse(
20    await fs.promises.readFile("swagger.json", "utf8"),
21  );
22  typia.assert(swagger); // recommended
23
24  // convert to emended OpenAPI document,
25  // and compose LLM function calling application
26  const document: OpenApi.IDocument = OpenApi.convert(swagger);
27  const application: IHttpLlmApplication<"chatgpt"> = HttpLlm.application({
28    model: "chatgpt",
29    document,
30  });
31
32  // Let's imagine that LLM has selected a function to call
33  const func: IHttpLlmFunction<"chatgpt"> | undefined = 
34    application.functions.find(
35      // (f) => f.name === "llm_selected_fuction_name"
36      (f) => f.path === "/bbs/articles" && f.method === "post",
37    );
38  if (func === undefined) throw new Error("No matched function exists.");
39
40  // actual execution is by yourself
41  const article = await HttpLlm.execute({
42    connection: {
43      host: "http://localhost:3000",
44    },
45    application,
46    function: func,
47    arguments: {
48      // arguments composed by LLM 
49      body: {
50        title: "Hello, world!",
51        body: "Let's imagine that this argument is composed by LLM.",
52        thumbnail: null,
53      },
54    },
55  });
56  console.log("article", article);
57};
58main().catch(console.error);

OpenAPI Definitions

1flowchart
2  v20(Swagger v2.0) --upgrades--> emended[["<b><u>OpenAPI v3.1 (emended)</u></b>"]]
3  v30(OpenAPI v3.0) --upgrades--> emended
4  v31(OpenAPI v3.1) --emends--> emended
5  emended --downgrades--> v20d(Swagger v2.0)
6  emended --downgrades--> v30d(Swagger v3.0)

@samchon/openapi support every versions of OpenAPI specifications with detailed TypeScript types.

• Swagger v2.0
• OpenAPI v3.0
• OpenAPI v3.1
• OpenAPI v3.1 emended

Also, @samchon/openapi provides "emended OpenAPI v3.1 definition" which has removed ambiguous and duplicated expressions for clarity. It has emended original OpenAPI v3.1 specification like above. You can compose the "emended OpenAPI v3.1 document" by calling the OpenApi.convert() function.

• Operation
  • Merge OpenApiV3_1.IPathItem.parameters to OpenApi.IOperation.parameters
  • Resolve references of OpenApiV3_1.IOperation members
  • Escape references of OpenApiV3_1.IComponents.examples
• JSON Schema
  • Decompose mixed type: OpenApiV3_1.IJsonSchema.IMixed
  • Resolve nullable property: OpenApiV3_1.IJsonSchema.__ISignificant.nullable
  • Array type utilizes only single OpenAPI.IJsonSchema.IArray.items
  • Tuple type utilizes only OpenApi.IJsonSchema.ITuple.prefixItems
  • Merge OpenApiV3_1.IJsonSchema.IAnyOf to OpenApi.IJsonSchema.IOneOf
  • Merge OpenApiV3_1.IJsonSchema.IRecursiveReference to OpenApi.IJsonSchema.IReference
  • Merge OpenApiV3_1.IJsonSchema.IAllOf to OpenApi.IJsonSchema.IObject

Conversions to another version's OpenAPI document is also based on the "emended OpenAPI v3.1 specification" like above diagram. You can do it through OpenApi.downgrade() function. Therefore, if you want to convert Swagger v2.0 document to OpenAPI v3.0 document, you have to call two functions; OpenApi.convert() and then OpenApi.downgrade().

At last, if you utilize typia library with @samchon/openapi types, you can validate whether your OpenAPI document is following the standard specification or not. Just visit one of below playground links, and paste your OpenAPI document URL address. This validation strategy would be superior than any other OpenAPI validator libraries.

• Playground Links
  • 💻 Type assertion
  • 💻 Detailed validation

1import { OpenApi, OpenApiV3, OpenApiV3_1, SwaggerV2 } from "@samchon/openapi";
2import typia from "typia";
3 
4const main = async (): Promise<void> => {
5  // GET YOUR OPENAPI DOCUMENT
6  const response: Response = await fetch(
7    "https://raw.githubusercontent.com/samchon/openapi/master/examples/v3.0/openai.json"
8  );
9  const document: any = await response.json();
10 
11  // TYPE VALIDATION
12  const result = typia.validate<
13    | OpenApiV3_1.IDocument
14    | OpenApiV3.IDocument
15    | SwaggerV2.IDocument
16  >(document);
17  if (result.success === false) {
18    console.error(result.errors);
19    return;
20  }
21 
22  // CONVERT TO EMENDED
23  const emended: OpenApi.IDocument = OpenApi.convert(document);
24  console.info(emended);
25};
26main().catch(console.error);

LLM Function Calling

Preface

1flowchart TD
2  subgraph "OpenAPI Specification"
3    v20("Swagger v2.0") --upgrades--> emended[["OpenAPI v3.1 (emended)"]]
4    v30("OpenAPI v3.0") --upgrades--> emended
5    v31("OpenAPI v3.1") --emends--> emended
6  end
7  subgraph "OpenAPI Generator"
8    emended --normalizes--> migration[["Migration Schema"]]
9    migration --"Artificial Intelligence"--> lfc{{"<b><u>LLM Function Calling</b></u>"}}
10    lfc --"OpenAI"--> chatgpt("ChatGPT")
11    lfc --"Anthropic"--> claude("Claude")
12    lfc --"Google"--> gemini("Gemini")
13    lfc --"Meta"--> llama("Llama")
14  end

LLM function calling application from OpenAPI document.

@samchon/openapi provides LLM (Large Language Model) funtion calling application from the "emended OpenAPI v3.1 document". Therefore, if you have any HTTP backend server and succeeded to build an OpenAPI document, you can easily make the A.I. chatbot application.

In the A.I. chatbot, LLM will select proper function to remotely call from the conversations with user, and fill arguments of the function automatically. If you actually execute the function call through the HttpLlm.execute() funtion, it is the "LLM function call."

Let's enjoy the fantastic LLM function calling feature very easily with @samchon/openapi.

• Application
  • HttpLlm.application()
  • IHttpLlmApplication
  • IHttpLlmFunction
• Schemas
  • IChatGptSchema: OpenAI ChatGPT
  • IClaudeSchema: Anthropic Claude
  • IGeminiSchema: Google Gemini
  • ILlamaSchema: Meta Llama
  • ILlmSchemaV3: middle layer based on OpenAPI v3.0 specification
  • ILlmSchemaV3_1: middle layer based on OpenAPI v3.1 specification
• Type Checkers
  • ChatGptTypeChecker
  • ClaudeTypeChecker
  • GeminiTypeChecker
  • LlamaTypeChecker
  • LlmTypeCheckerV3
  • LlmTypeCheckerV3_1

[!NOTE]

You also can compose ILlmApplication from a class type with typia.

https://typia.io/docs/llm/application

1import { ILlmApplication } from "@samchon/openapi";
2import typia from "typia";
3
4const app: ILlmApplication<"chatgpt"> =
5  typia.llm.application<YourClassType, "chatgpt">();

[!TIP]

LLM selects proper function and fill arguments.

In nowadays, most LLM (Large Language Model) like OpenAI are supporting "function calling" feature. The "LLM function calling" means that LLM automatically selects a proper function and fills parameter values from conversation with the user (may by chatting text).

https://platform.openai.com/docs/guides/function-calling

Execution

Actual function call execution is by yourself.

LLM (Large Language Model) providers like OpenAI selects a proper function to call from the conversations with users, and fill arguments of it. However, function calling feature supported by LLM providers do not perform the function call execution. The actual execution responsibility is on you.

In @samchon/openapi, you can execute the LLM function calling by HttpLlm.execute() (or HttpLlm.propagate()) function. Here is an example code executing the LLM function calling through the HttpLlm.execute() function. As you can see, to execute the LLM function call, you have to deliver these informations:

• Connection info to the HTTP server
• Application of the LLM fuction calling
• LLM function schema to call
• Arguments for the function call (maybe composed by LLM)

Here is the example code executing the LLM function call with @samchon/openapi.

• Example Code: test/examples/chatgpt-function-call-to-sale-create.ts
• Prompt describing the produc to create: Microsoft Surface Pro 9
• Result of the Function Calling: examples/arguments/chatgpt.microsoft-surface-pro-9.input.json

1import {
2  HttpLlm,
3  IChatGptSchema,
4  IHttpLlmApplication,
5  IHttpLlmFunction,
6  OpenApi,
7  OpenApiV3,
8  OpenApiV3_1,
9  SwaggerV2,
10} from "@samchon/openapi";
11import OpenAI from "openai";
12import typia from "typia";
13
14const main = async (): Promise<void> => {
15  // Read swagger document and validate it
16  const swagger:
17    | SwaggerV2.IDocument
18    | OpenApiV3.IDocument
19    | OpenApiV3_1.IDocument = JSON.parse(
20    await fetch(
21      "https://github.com/samchon/shopping-backend/blob/master/packages/api/swagger.json",
22    ).then((r) => r.json()),
23  );
24  typia.assert(swagger); // recommended
25
26  // convert to emended OpenAPI document,
27  // and compose LLM function calling application
28  const document: OpenApi.IDocument = OpenApi.convert(swagger);
29  const application: IHttpLlmApplication<"chatgpt"> = HttpLlm.application({
30    model: "chatgpt",
31    document,
32  });
33
34  // Let's imagine that LLM has selected a function to call
35  const func: IHttpLlmFunction<"chatgpt"> | undefined =
36    application.functions.find(
37      // (f) => f.name === "llm_selected_fuction_name"
38      (f) => f.path === "/shoppings/sellers/sale" && f.method === "post",
39    );
40  if (func === undefined) throw new Error("No matched function exists.");
41
42  // Get arguments by ChatGPT function calling
43  const client: OpenAI = new OpenAI({
44    apiKey: "<YOUR_OPENAI_API_KEY>",
45  });
46  const completion: OpenAI.ChatCompletion =
47    await client.chat.completions.create({
48      model: "gpt-4o",
49      messages: [
50        {
51          role: "assistant",
52          content:
53            "You are a helpful customer support assistant. Use the supplied tools to assist the user.",
54        },
55        {
56          role: "user",
57          content: "<DESCRIPTION ABOUT THE SALE>",
58          // https://github.com/samchon/openapi/blob/master/examples/function-calling/prompts/microsoft-surface-pro-9.md
59        },
60      ],
61      tools: [
62        {
63          type: "function",
64          function: {
65            name: func.name,
66            description: func.description,
67            parameters: func.parameters as Record<string, any>,
68          },
69        },
70      ],
71    });
72  const toolCall: OpenAI.ChatCompletionMessageToolCall =
73    completion.choices[0].message.tool_calls![0];
74
75  // Actual execution by yourself
76  const article = await HttpLlm.execute({
77    connection: {
78      host: "http://localhost:37001",
79    },
80    application,
81    function: func,
82    input: JSON.parse(toolCall.function.arguments),
83  });
84  console.log("article", article);
85};
86main().catch(console.error);

Separation

Arguments from both Human and LLM sides.

When composing parameter arguments through the LLM (Large Language Model) function calling, there can be a case that some parameters (or nested properties) must be composed not by LLM, but by Human. File uploading feature, or sensitive information like secret key (password) cases are the representative examples.

In that case, you can configure the LLM function calling schemas to exclude such Human side parameters (or nested properties) by IHttpLlmApplication.options.separate property. Instead, you have to merge both Human and LLM composed parameters into one by calling the HttpLlm.mergeParameters() before the LLM function call execution of HttpLlm.execute() function.

Here is the example code separating the file uploading feature from the LLM function calling schema, and combining both Human and LLM composed parameters into one before the LLM function call execution.

• Example Code: test/examples/claude-function-call-separate-to-sale-create.ts
• Prompt describing the produc to create: Microsoft Surpace Pro 9
• Result of the Function Calling: examples/arguments/claude.microsoft-surface-pro-9.input.json

1import Anthropic from "@anthropic-ai/sdk";
2import {
3  ClaudeTypeChecker,
4  HttpLlm,
5  IClaudeSchema,
6  IHttpLlmApplication,
7  IHttpLlmFunction,
8  OpenApi,
9  OpenApiV3,
10  OpenApiV3_1,
11  SwaggerV2,
12} from "@samchon/openapi";
13import typia from "typia";
14
15const main = async (): Promise<void> => {
16  // Read swagger document and validate it
17  const swagger:
18    | SwaggerV2.IDocument
19    | OpenApiV3.IDocument
20    | OpenApiV3_1.IDocument = JSON.parse(
21    await fetch(
22      "https://github.com/samchon/shopping-backend/blob/master/packages/api/swagger.json",
23    ).then((r) => r.json()),
24  );
25  typia.assert(swagger); // recommended
26
27  // convert to emended OpenAPI document,
28  // and compose LLM function calling application
29  const document: OpenApi.IDocument = OpenApi.convert(swagger);
30  const application: IHttpLlmApplication<"claude"> = HttpLlm.application({
31    model: "claude",
32    document,
33    options: {
34      reference: true,
35      separate: (schema) =>
36        ClaudeTypeChecker.isString(schema) &&
37        !!schema.contentMediaType?.startsWith("image"),
38    },
39  });
40
41  // Let's imagine that LLM has selected a function to call
42  const func: IHttpLlmFunction<"claude"> | undefined =
43    application.functions.find(
44      // (f) => f.name === "llm_selected_fuction_name"
45      (f) => f.path === "/shoppings/sellers/sale" && f.method === "post",
46    );
47  if (func === undefined) throw new Error("No matched function exists.");
48
49  // Get arguments by ChatGPT function calling
50  const client: Anthropic = new Anthropic({
51    apiKey: "<YOUR_ANTHROPIC_API_KEY>",
52  });
53  const completion: Anthropic.Message = await client.messages.create({
54    model: "claude-3-5-sonnet-latest",
55    max_tokens: 8_192,
56    messages: [
57      {
58        role: "assistant",
59        content:
60          "You are a helpful customer support assistant. Use the supplied tools to assist the user.",
61      },
62      {
63        role: "user",
64        content: "<DESCRIPTION ABOUT THE SALE>",
65        // https://github.com/samchon/openapi/blob/master/examples/function-calling/prompts/microsoft-surface-pro-9.md
66      },
67    ],
68    tools: [
69      {
70        name: func.name,
71        description: func.description,
72        input_schema: func.separated!.llm as any,
73      },
74    ],
75  });
76  const toolCall: Anthropic.ToolUseBlock = completion.content.filter(
77    (c) => c.type === "tool_use",
78  )[0]!;
79
80  // Actual execution by yourself
81  const article = await HttpLlm.execute({
82    connection: {
83      host: "http://localhost:37001",
84    },
85    application,
86    function: func,
87    input: HttpLlm.mergeParameters({
88      function: func,
89      llm: toolCall.input as any,
90      human: {
91        // Human composed parameter values
92        content: {
93          files: [],
94          thumbnails: [
95            {
96              name: "thumbnail",
97              extension: "jpeg",
98              url: "https://serpapi.com/searches/673d3a37e45f3316ecd8ab3e/images/1be25e6e2b1fb7509f1af89c326cb41749301b94375eb5680b9bddcdf88fabcb.jpeg",
99            },
100            // ...
101          ],
102        },
103      },
104    }),
105  });
106  console.log("article", article);
107};
108main().catch(console.error);