Construct Your Personal ChatGPT Clone with React and the OpenAI API — SitePoint

On this tutorial, we’ll stroll by easy methods to construct a customized Chatbot software that may permit us to ask questions and obtain high-quality solutions. The bot will bear in mind earlier prompts, simulating context-aware dialog.

Chatbots have grow to be indispensable instruments for companies and builders looking for to enhance buyer interactions and streamline person experiences in at this time’s quickly evolving digital panorama.

OpenAI’s ChatGPT has reworked from a cutting-edge experiment right into a powerhouse in chatbot improvement. Its meteoric rise to success is nothing in need of outstanding, fascinating customers worldwide.

The demo code of this mission is accessible on CodeSandbox. You’ll have to supply your individual OpenAI API key within the .env file to check it reside. To get one, create an account on the OpenAI, log in, navigate to the API keys and generate a brand new API key.

Planning Options and UI

Our software shall be primarily based on React, and we’ll use OpenAI API to entry the information and use CSS modules for styling.

Using React will permit us to create a dynamic and responsive person interface, enhancing the general person expertise.

The OpenAI API will allow us to acquire entry to superior language processing capabilities, offering information for creating insightful interactions.

Moreover, CSS modules will permit us to keep up a modular design, facilitating environment friendly improvement and customization of the app.

The options we’ll be implementing embody:

• A chosen enter space the place customers will be capable to craft prompts, inviting contextually related inquiries.
• A Submit button that may permit customers to submit their prompts to the API, initiating the dialog course of.
• Message objects that shall be showcased as chat-style messages inside the dialog window, enhancing the interactive chat expertise.
• Message objects to show ChatGPT replies that may present a conversational move.
• A Historical past function that may checklist all the person’s latest prompts. This can even permit customers to revisit earlier conversations.
• A Clear button that may permit the elimination of generated content material, providing a clear slate for brand spanking new conversations.

The picture under reveals our component-based wireframe.

The entire software shall be wrapped in the primary container, which can maintain all the parts collectively. It will likely be additional divided right into a two-column structure.

The primary column will embody all the messages from the person and ChatGPT. On the backside of the column, there shall be an enter space and a button for submitting the immediate.

The second column will maintain the historical past of all the latest prompts. On the backside of the column, there shall be a Clear button that may permit the person to wipe the generated content material.

Choosing a Coloration Scheme

The appliance design will prioritize the convenience of content material notion. This can permit us to supply a few vital advantages:

• Customers will be capable to rapidly comprehend the introduced info, resulting in a extra intuitive and user-friendly expertise.
• It’ll additionally improve accessibility, making certain that people of various backgrounds and skills will be capable to simply navigate and interact with the content material.

The picture under reveals our colour scheme.

The background of the appliance shall be black, whereas the messages, historical past objects, and enter kind shall be darkish grey.

The textual content on the messages and enter backgrounds shall be white, offering a pleasant distinction and make textual content simple to learn.

To offer the app some highlights, the column titles, Submit button, and response message avatars will use a brilliant, lime-green tone.

To accent the Clear button, a light crimson tone shall be used. This can even assist customers keep away from clicking the button by accident.

Setting Up the React App

We’ll use create-react-app to create our software. Run npx create-react-app react-chatgpt to create a brand new React mission.

Look forward to a minute for the setup to finish, after which change the working listing to the newly created folder by cd react-chatgpt and run npm begin to start out the developer server.

This could open up our mission in our default browser. If not, navigate to http://localhost:3000 to open it manually. We needs to be introduced with the React welcome display, as pictured under.

Including International Types

We’ll add world styling to determine a constant and unified visible look throughout all elements of the appliance.

Open index.css and embody the next styling guidelines:

@import url("https://fonts.googleapis.com/css2?household=Varela+Spherical&show=swap");

* {
  margin: 0;
  padding: 0;
  box-sizing: border-box;
  font-family: "Varela Spherical", sans-serif;
}

physique {
  background-color: #121212;
}

First, we import the Varela Spherical font and set the entire app to make use of it.

We additionally take away any pre-defined margins and paddings, in addition to set box-sizing to border-box so the app appears the identical on completely different browsers.

Lastly, we set the background of the physique to a darkish tone, which permits us to spotlight the content material of the appliance.

We’ll want a few avatars to signify the authors of the messages from the person and OpenAI API. This fashion, they’ll be simpler to tell apart.

Create a brand new icons folder contained in the src listing and embody the bot.png and person.png icons.

You’ll be able to obtain samples from icons listing right here, or you should utilize customized ones from websites like FlatIcon or Icons8, so long as you retain the above file names.

Constructing the Elements

First, we want a well-organized file construction that matches the wireframe design.

We’ll use the terminal to create the required folder and element information. Every element may have its personal JavaScript file for performance and CSS file for styling.

Change the working listing within the src folder by working cd src after which run the next command:

mkdir elements && cd elements && contact Message.js Message.module.css Enter.js Enter.module.css Historical past.js Historical past.module.css Clear.js Clear.module.css

The command above will first create a /elements/ folder, then change the working listing to it, and create all the required information inside it.

The Message element

The Message element will show person prompts and API responses inside the dialog, facilitating the real-time change of knowledge between the person and the chatbot.

Open the Message.js file and embody the next code:

import bot from "../icons/bot.png";
import person from "../icons/person.png";

import types from "./Message.module.css";

export default operate Message({ position, content material }) {
  return (
    <div className={types.wrapper}>
      <div>
        <img
          src={position === "assistant" ? bot : person}
          className={types.avatar}
          alt="profile avatar"
        />
      </div>
      <div>
        <p>{content material}</p>
      </div>
    </div>
  );
}

First, we import the downloaded icons for avatars after which import the exterior CSS guidelines for styling.

After that, we create the wrapper for the Message element, which can comprise each icons and textual content content material.

We use the position prop within the conditional to show the suitable avatar because the picture src.

We additionally use the content material prop, which shall be handed in because the textual content response from the OpenAI API and person enter immediate.

Now let’s type the element so it appears like a chat message! Open the Message.module.css file and embody the next guidelines:

.wrapper {
  show: grid;
  grid-template-columns: 60px auto;
  min-height: 60px;
  padding: 20px;
  margin-bottom: 20px;
  border-radius: 10px;
  background-color: #1b1b1d;
}

.avatar {
  width: 40px;
  peak: 40px;
}

We divide the structure into two columns, with the avatars proven within the fixed-width container on the appropriate and the textual content on the left.

Subsequent, we add some padding and margin to the underside of the message. We additionally type the message to have spherical borders and set the background to darkish grey.

Lastly, we set the avatar icon to a set width and peak.

The Enter element

The Enter element shall be an interface component designed to seize person queries, serving because the means by which customers work together and interact with the chatbot.

Open the Enter.js file and embody the next code:

import types from "./Enter.module.css";

export default operate Enter({ worth, onChange, onClick }) {
  return (
    <div className={types.wrapper}>
      <enter
        className={types.textual content}
        placeholder="Your immediate right here..."
        worth={worth}
        onChange={onChange}
      />
      <button className={types.btn} onClick={onClick}>
        Go
      </button>
    </div>
  );
}

We first import the exterior stylesheet to type the element.

We return the element wrapper that features the enter area for the person prompts and the button to submit it to the API.

We set the placeholder worth to be displayed when the enter kind is empty, and create the worth prop to carry the entered immediate, in addition to the onChange prop that shall be known as as soon as the enter worth modifications.

For the button, the onClick prop shall be known as as soon as the person clicks on the button.

Now let’s type the element in order that the enter space appears lovely and the person is inspired to supply prompts! Open the Enter.module.css file and embody the next guidelines:

.wrapper {
  show: grid;
  grid-template-columns: auto 100px;
  peak: 60px;
  border-radius: 10px;
  background-color: #323236;
}

.textual content {
  border: none;
  define: none;
  background: none;
  padding: 20px;
  colour: white;
  font-size: 16px;
}

.btn {
  border: none;
  border-radius: 0 10px 10px 0;
  font-size: 16px;
  font-weight: daring;
  background-color: rgb(218, 255, 170);
}

.btn:hover {
  cursor: pointer;
  background-color: rgb(200, 253, 130);
}

We set the wrapper to be divided into two columns, with a set width for the button and the remainder of the out there width devoted to the enter space.

We additionally outline the precise peak of the element, set the rounded borders for it, and set the background to darkish grey.

For the enter space, we take away the default border, define, background and add some padding. We set the textual content colour to white and set a particular font measurement.

The Historical past element

The Historical past element will show the sequence of previous person and chatbot interactions, offering customers with a contextual reference of their dialog.

Open the Historical past.js file and embody the next code:

import types from "./Historical past.module.css";

export default operate Historical past({ query, onClick }) {
  return (
    <div className={types.wrapper} onClick={onClick}>
      <p>{query.substring(0, 15)}...</p>
    </div>
  );
}

We first import the exterior type guidelines for the element. Then we return the wrapper that may embody the textual content.

The textual content worth shall be handed in as a query prop from the person immediate, and solely the primary 15 characters of the textual content string shall be displayed.

Customers shall be allowed to click on on the historical past objects, and we’ll cross the onClick prop to manage the clicking habits.

Now let’s type the element to make sure it’s visually interesting and matches nicely within the sidebar! Open the Historical past.module.css file and embody the next guidelines:

.wrapper {
  padding: 20px;
  margin-bottom: 20px;
  border-radius: 10px;
  background-color: #1b1b1d;
}

.wrapper:hover {
  cursor: pointer;
  background-color: #323236;
}

We set some padding, add the margin to the underside, and set the rounded corners for the historical past objects. We additionally set the background colour to darkish grey.

As soon as the person hovers over the merchandise, the cursor will change to a pointer and the background colour will change to a lighter shade of grey.

The Clear element

The Clear element shall be a UI component designed to reset or clear the continuing dialog, offering customers with a fast option to begin a brand new interplay with out navigating away from the present interface.

Open the Clear.js file and embody the next code:

import types from "./Clear.module.css";

export default operate Clear({ onClick }) {
  return (
    <button className={types.wrapper} onClick={onClick}>
      Clear
    </button>
  );
}

We first import the exterior stylesheet to type the element.

We return the button that may permit customers to clear the content material of the appliance. We’ll cross the onClick prop to attain the specified habits.

Now let’s type the element to make it stand out and scale back the possibilities of customers urgent it by accident! Open the Clear.module.css file and embody the next guidelines:

.wrapper {
  width: 100%;
  peak: 60px;
  background-color: #ff9d84;
  border: none;
  border-radius: 10px;
  font-size: 16px;
  font-weight: daring;
}

.wrapper:hover {
  cursor: pointer;
  background-color: #ff886b;
}

We set the button to fill the out there width of the column, set the precise peak, and set the background colour to delicate crimson.

We additionally take away the default border, set the rounded corners, set a particular font measurement, and make it daring.

On hover, the cursor will change to a pointer and the background colour will change to a darker shade of crimson.

Constructing the Consumer Interface

Within the earlier part, we constructed all the crucial elements. Now let’s put them collectively and construct the person interface for the appliance.

We’ll configure their performance to create a practical and interactive chatbot interface with organized and reusable code.

Open the App.js file and embody the next code:

import { useState } from "react";

import Message from "./elements/Message";
import Enter from "./elements/Enter";
import Historical past from "./elements/Historical past";
import Clear from "./elements/Clear";

import "./types.css";

export default operate App() {
  const [input, setInput] = useState("");
  const [messages, setMessages] = useState([]);
  const [history, setHistory] = useState([]);

  return (
    <div className="App">
      <div className="Column">
        <h3 className="Title">Chat Messages</h3>
        <div className="Content material">
          {messages.map((el, i) => {
            return <Message key={i} position={el.position} content material={el.content material} />;
          })}
        </div>
        <Enter
          worth={enter}
          onChange={(e) => setInput(e.goal.worth)}
          onClick={enter ? handleSubmit : undefined}
        />
      </div>
      <div className="Column">
        <h3 className="Title">Historical past</h3>
        <div className="Content material">
          {historical past.map((el, i) => {
            return (
              <Historical past
                key={i}
                query={el.query}
                onClick={() =>
                  setMessages([
                    { role: "user", content: history[i].query },
                    { position: "assistant", content material: historical past[i].reply },
                  ])
                }
              />
            );
          })}
        </div>
        <Clear onClick={clear} />
      </div>
    </div>
  );
}

First, we import the useState hook that we’ll use to trace the information state for the appliance. Then we import all of the elements we constructed and the exterior stylesheet for styling.

Then we create the enter state variable to retailer the person immediate enter, messages to retailer the dialog between the person and ChatGPT, and historical past to retailer the historical past of person prompts.

We additionally create the primary wrapper for the entire app that may maintain two columns.

Every column may have a title and content material wrapper that may embody the dialog messages, enter space, and Submit button for the primary column and historical past objects and the Clear button for the second column.

The dialog messages shall be generated by mapping by the messages state variable and the historical past objects — by mapping by the historical past state variable.

We set the enter onChange prop to replace the enter state variable every time person enters any worth within the enter kind.

As soon as the person clicks the Ship button, the person immediate shall be despatched to the OpenAI API to course of and obtain the reply.

For the historical past objects, we set the onClick prop in order that the messages state variable will get up to date to the precise immediate and reply.

Lastly, for the Clear button, we cross the onClick prop a operate that may clear each the message and historical past values, clearing the appliance information.

Creating the App Format

On this part, we’ll organize the person interface elements to create an intuitive construction for efficient person interplay.

Open App.css and embody the next styling guidelines:

.App {
  show: grid;
  grid-template-columns: auto 200px;
  hole: 20px;
  max-width: 1000px;
  margin: 0 auto;
  min-height: 100vh;
  padding: 20px;
}

.Column {
  colour: white;
}

.Title {
  padding: 20px;
  margin-bottom: 20px;
  border-radius: 10px;
  colour: black;
  background-color: rgb(218, 255, 170);
}

.Content material {
  peak: calc(100vh - 200px);
  overflow-y: scroll;
  margin-bottom: 20px;
}

::-webkit-scrollbar {
  show: none;
}

We break up the primary app wrapper into two columns, separated by a spot by utilizing CSS grid structure, and we set the left column for historical past objects to a set width.

Subsequent, we set the wrapper to by no means exceed a sure width, middle it on the display, make it use all the display viewport peak, and add some padding inside it.

For every column’s contents, we set the textual content colour to white.

For the column titles, we set some padding, add the underside margin, and set the rounded corners. We additionally set the title component background colour to lime-green and set the textual content colour to black.

We additionally type the columns themselves by setting the rule that the content material shouldn’t exceed a sure peak and set the content material to be scrollable if it reaches outdoors the peak. We additionally add a margin to the underside.

We additionally cover the scrollbars, in order that we don’t should type them to override the default values for every browser. This rule is optionally available and we may depart it out.

Getting the API Key from OpenAI

If you happen to haven’t already arrange your individual API key for the Sandbox within the introduction of this tutorial, be sure that to create an account on the OpenAI web site.

Subsequent, log in and navigate to the API keys and generate a brand new API key.

Copy the important thing to the clipboard and open your mission.

Create a brand new .env file in your mission root and paste the worth for the next key like so:

REACT_APP_OPENAI_API_KEY=paste-your-code-here

Making ready the Request Name to OpenAI API

By the OpenAI API, our chatbot will be capable to ship textual prompts to the OpenAI server, which can then course of the enter and generate human-like responses.

That is achieved by leveraging a robust language mannequin that’s been skilled on numerous textual content sources. By offering the mannequin with a dialog historical past and the present person immediate, our chatbot will obtain context-aware responses from the API.

On this part, we’ll put together the request and implement the decision to the API to obtain the response and set the information to the state variable we outlined earlier.

Open the App.js once more and add the next code:

export default operate App() {
  

  const handleSubmit = async () => {
    const immediate = {
      position: "person",
      content material: enter,
    };

    setMessages([...messages, prompt]);

    await fetch("https://api.openai.com/v1/chat/completions", {
      methodology: "POST",
      headers: {
        Authorization: `Bearer ${course of.env.REACT_APP_OPENAI_API_KEY}`,
        "Content material-Kind": "software/json",
      },
      physique: JSON.stringify({
        mannequin: "gpt-3.5-turbo",
        messages: [...messages, prompt],
      }),
    })
      .then((information) => information.json())
      .then((information) => {
        const res = information.selections[0].message.content material;
        setMessages((messages) => [
          ...messages,
          {
            role: "assistant",
            content: res,
          },
        ]);
        setHistory((historical past) => [...history, { question: input, answer: res }]);
        setInput("");
      });
  };

  const clear = () => {
    setMessages([]);
    setHistory([]);
  };

  return <div className="App">
}

First, we create a separate handleSubmit operate, which shall be executed as soon as the person has entered the immediate within the enter kind and clicks the Submit button.

Inside handleSubmit, we first create the immediate variable that may maintain the position person and the immediate itself as an object. The position is vital as a result of, when storing our messages, we’ll must know which of them are person messages.

Then we replace the messages state variable with the person immediate.

Subsequent, we make an precise fetch name to the api.openai.com/v1/chat/completions endpoint to entry the information from the OpenAI API.

We specify that it’s a POST request, and set the headers with the authorization token and the content material kind. For the physique parameters, we specify which API mannequin to make use of, and we cross the messages variable because the content material from the person.

As soon as the response is acquired, we retailer it within the res variable. We add the article consisting of the position assistant and the response itself to the message state variable.

We additionally replace the historical past state variable with the article, with the query and corresponding reply because the keys.

After the response is acquired and state variables are up to date, we clear the enter state variable to organize the enter kind for the subsequent person immediate.

Lastly, we create a easy clear operate to clear the messages and historical past state variables, permitting the person to clear the information of the appliance.

Testing the Utility

At this level, we must always have created a totally practical chat software! The very last thing left to do is to check it.

First, let’s attempt to ask ChatGPT a single query.

The animation above reveals a query being submitted and a solution being acquired.

Now let’s attempt to create a dialog.

As proven within the animation above, the chatbot remembers the context from the earlier messages, so we are able to communicate with it whereas being absolutely context-aware.

Now let’s see what occurs as soon as we click on on the Historical past button.

Discover how the chat switches to the respective person immediate and reply. This may very well be helpful if we need to resume the dialog from a particular level.

Lastly, let’s click on on the Clear button.

As anticipated, the contents of the app are cleared. This can be a helpful choice when there’s plenty of content material and the person needs to start out contemporary.

Conclusion

On this tutorial, we’ve discovered easy methods to create an easy-to-use person interface, easy methods to construction our code through elements, easy methods to work with states, easy methods to make API calls, and easy methods to course of the acquired information.

With the mix of superior pure language processing capabilities of the OpenIAI API and the pliability of React, you’ll now be capable to create subtle chatbot functions which you could customise additional to your liking.

Discover that this tutorial shops the API key on the frontend, which could not be safe for manufacturing. If you wish to deploy the mission, it will be advisable to create an Categorical server and use the API key there.

Additionally, if you would like the historical past prompts to be out there after the subsequent preliminary launch, you could possibly retailer after which learn them from native storage, and even join a database to your app and retailer and skim information from there.