A reactive notebook for Python — run reproducible experiments, query with SQL, execute as a script, deploy as an app, and version with git. Stored as pure Python. All in a modern, AI-native editor.
Python artificial-intelligencedagdata-sciencedata-visualizationdataflowdeveloper-toolsmachine-learningnotebookspipelinepythonreactivesqlweb-app
/* Copyright 2026 Marimo. All rights reserved. */
import { useState } from "react";
import { Logger } from "@/utils/Logger";
import { useRequestClient } from "../network/requests";
import {
showAddPackageToast,
showPackageRestartToast,
} from "./toast-components";
export function useInstallPackages(): {
handleInstallPackages: (
packages: string[],
onSuccess?: () => void,
) => Promise<void>;
loading: boolean;
} {
const [loading, setLoading] = useState(false);
const { addPackage } = useRequestClient();
const handleInstallPackages = async (
packages: string[],
onSuccess?: () => void,
) => {
setLoading(true);
try {
// Batch all packages into a single install call.
// The worker splits by space and passes the full list to micropip,
// which resolves and downloads in parallel internally.
const response = await addPackage({ package: packages.join(" ") });
// The backend resolves the whole list as a single transaction, so the
// response only carries an aggregate success/error. Report a single
// toast covering all requested packages rather than implying a
// per-package outcome we don't actually have.
if (response.restartRequired) {
showPackageRestartToast();
} else if (response.success) {
showAddPackageToast(packages);
onSuccess?.();
} else {
showAddPackageToast(packages, response.error);
}
} catch (error) {
Logger.error(error);
} finally {
setLoading(false);
}
};
return { loading, handleInstallPackages };
}