The Kotlin Kernel for Jupyter notebooks is a powerful tool that allows you to write and run Kotlin 1.9.23 code within the Jupyter Notebook environment. This Kernel essentially acts as a bridge between Jupyter Notebook and the Kotlin compiler.
The Kotlin Kernel for notebooks supports running code cells to immediately see outputs, basic code completion, error analysis, and other interactive coding features, enhancing the interactive experience provided by the Kotlin REPL.
With the Kotlin Kernel for notebooks, you gain access to a range of features like:
- Accessing APIs within cells and using APIs for handling outputs.
- Retrieving information from previously executed code snippets, allowing quick project exploration.
- Importing various libraries with a single line of code or even integrating new libraries into your project.
You can leverage Kotlin Kernel's benefits in IntelliJ IDEA through the Kotlin Notebook plugin, in your Jupyter Notebook, or in Datalore.
Click here to expand the table of contents.
Start using the Kotlin Kernel for Jupyter notebooks:
- See the introductory notebook guide.
- Check notebook samples.
- Try the sample notebooks online:
- Explore the Kotlin Notebook docs to learn about features, use cases, and tutorials.
Note: The Kotlin Kernel for Jupyter notebooks is in Beta.
The latest version of the Kotlin Kernel for notebooks uses the Kotlin compiler of version 1.9.23.
We tested the Kotlin Kernel for notebooks with the following clients:
| Client | Minimal supported version |
|---|---|
| JupyterLab | 1.2.6 |
| Jupyter Notebook | 6.0.3 |
| Jupyter Console | 6.1.0 |
We tested the Kotlin Kernel for notebooks with all the mentioned clients on the following operating systems:
- Windows
- Ubuntu Linux
- macOS
You can create, open, and work with Kotlin notebooks on various clients:
Our Kotlin Kernel is fully integrated into Kotlin Notebook, which you can use directly within IntelliJ IDEA by installing the Kotlin Notebook plugin.
In Datalore, Kotlin is supported natively, with the Kotlin Kernel already bundled for an out-of-the-box experience.
For other Jupyter clients, you'll need to install the Kotlin Kernel separately using conda, pip, or sources.
If you use IntelliJ IDEA, you need to install the Kotlin Notebook plugin that contains the Kotlin Kernel for Jupyter notebooks.
Install the Kotlin Notebook plugin by downloading its latest version from the JetBrains Marketplace.
Alternatively, access the Kotlin Notebook plugin from Settings | Plugins | Marketplace within IntelliJ IDEA.
Note: For a quick introduction to Kotlin Notebook, see this blog post.
If you use Jupyter Notebook and Conda, run this Conda command to install the stable version package of the Kotlin Kernel for Jupyter notebooks:
conda install -c jetbrains kotlin-jupyter-kernel
Alternatively, run this Conda command to install the package from the developers channel:
conda install -c jetbrains-dev kotlin-jupyter-kernel
To uninstall the Kotlin Kernel for Jupyter notebooks, run this Conda command:
conda remove kotlin-jupyter-kernel
If you use Jupyter Notebook and Pip, run this Pip command to install the stable version package of the Kotlin Kernel for Jupyter notebooks:
pip install kotlin-jupyter-kernel
Alternatively, run this Pip command to install the package from the developers channel:
pip install -i https://test.pypi.org/simple/ kotlin-jupyter-kernel
To uninstall the Kotlin Kernel for Jupyter notebooks, run this Pip command:
pip uninstall kotlin-jupyter-kernel
If you use either IntelliJ IDEA or Jupyter Notebook, you can install the Kotlin Kernel for Jupyter notebooks from sources.
Clone this repository and run the following Gradle command in the root folder:
./gradlew install
The default installation path is ~/.ipython/kernels/kotlin/. You can also install the package in another location using the -PinstallPath= option.
However, Jupyter only looks for the kernel specification files in predefined places. For more details, see Jupyter docs.
To uninstall the Kotlin Kernel for Jupyter notebooks from sources, run this Gradle command:
./gradlew uninstall
When installing the Kotlin Kernel for Jupyter notebooks, issues can occur while detecting the kernel specification file. These issues occur due to different Python environments and installation modes.
If you are using Pip or Conda to install the package, run this post-install fixup script:
python -m kotlin_kernel fix-kernelspec-locationThis script replaces the kernel specification files with the detected user path.
Note: Don't forget to re-run this script when updating the Kotlin Kernel for notebooks.
See how to update the Kotlin Kernel for Jupyter notebooks using the Kotlin Notebook plugin, Conda, Pip, and Datalore.
If you use the Kotlin Notebook plugin, update it to the latest version in Settings | Plugins | Installed within IntelliJ IDEA.
Alternatively, you can download and install the latest plugin version from the JetBrains Marketplace.
If you use Jupyter Notebook and Conda, run this Conda command to update the stable version package:
conda update -c jetbrains kotlin-jupyter-kernel
Alternatively, run this Conda command to update the package from the developers channel:
conda update -c jetbrains-dev kotlin-jupyter-kernel
To change to a specific Kotlin Kernel version, add the ={VERSION} parameter to the kotlin-jupyter-kernel command. In this command,
replace the {VERSION} parameter with the desired PyPi version of the Kotlin Jupyter Kernel (for example, 0.12.0.322).
You can specify the version in both commands the one from the stable package and the one from the developers channel package. For example, for the stable package:
conda install -c jetbrains kotlin-jupyter-kernel={0.12.0.322}
If you use Jupyter Notebook and Pip, run this Pip command to update the stable version package:
pip install kotlin-jupyter-kernel --upgrade
Alternatively, run this Conda command to update the package from the developers channel:
pip install -i https://test.pypi.org/simple/ kotlin-jupyter-kernel --upgrade
To change to a specific Kotlin Kernel version, add the ={VERSION} parameter to the kotlin-jupyter-kernel command. In this command,
replace the {VERSION} parameter with the desired PyPi version of the Kotlin Jupyter Kernel (for example, 0.12.0.322).
You can specify the version in both commands the one from the stable package and the one from the developers channel package. For example, for the stable package:
pip install kotlin-jupyter-kernel=={0.12.0.322} --ignore-installed
To update the Kotlin Kernel for notebooks in Datalore:
-
Add an
environment.ymlfile to the Notebook files containing the following snippet:datalore-env-format-version: "0.2" datalore-package-manager: "pip" datalore-base-env: "default" dependencies: - pip: - kotlin-jupyter-kernel=={VERSION}
-
Replace the
{VERSION}parameter for the latest PyPi version of the Kotlin Jupyter Kernel (for example,0.12.0.322). -
Stop and restart the machine in Datalore.
See how to use the Kotlin Kernel for Jupyter notebooks with the Kotlin Notebook plugin, Jupyter clients, and Datalore.
After installing the Kotlin Notebook plugin in IntelliJ IDEA, create a new notebook by selecting File | New | Kotlin Notebook, or right-click on a folder and select New | Kotlin Notebook.
Now you're good to go!
You can use our Kotlin Kernel through JupyterLab, Jupyter Notebook, and Jupyter Console clients:
- Run one of the following commands in the console:
-
In JupyterLab:
jupyter lab -
In Jupyter Notebook:
jupyter notebook -
In Jupyter Console:
jupyter console --kernel=kotlin
- Create a new notebook and set
kotlinas kernel. This step applies to Jupyter Notebook or JupyterLab, and it's not required for Jupyter Console.
The default kernel uses the JDK that the environment points in the KOTLIN_JUPYTER_JAVA_HOME variable. In case the KOTLIN_JUPYTER_JAVA_HOME variable is not set,
the kernel also uses the JDK from the JAVA_HOME variable.
The kernel uses the arguments that the environment points in the KOTLIN_JUPYTER_JAVA_OPTS variable.
In case the KOTLIN_JUPYTER_JAVA_OPTS variable is not set, the kernel also uses the JVM arguments from the JAVA_OPTS variable.
Additionally, the kernel uses arguments that the environment points in the KOTLIN_JUPYTER_JAVA_OPTS_EXTRA variable.
The arguments are parsed using the Python shlex.split() function.
To create a Kotlin notebook in Datalore, click on New notebook and select Kotlin as kernel.
You can create a custom Kotlin Kernel for Jupyter Notebook. This allows you to tailor the kernel's environment to your specific requirements, such as using a particular JDK, setting JVM arguments, or defining environment variables.
To create a custom Kotlin Kernel for Jupyter Notebook, use the add-kernel command from the installed kotlin_kernel python package:
python -m kotlin_kernel add-kernel [--name name] [--jdk jdk_home_dir] [--set-jvm-args] [--jvm-arg arg]* [--env KEY VALUE]* [--force]In the add-kernel script, the name argument is required if the jdk argument is not specified. Alternatively, if the jdk argument is specified,
but the name argument is not, then the name is taken from the JDK $vendor $version argument, which is detected from the JDK.
Regardless of how the name is determined, the format of the kernel name is Kotlin ($name),
and the format of the directory name is kotlin_$name. The directory name includes spaces in name, replaced by underscores. Ensure
this format is compatible with your file system.
JVM arguments are joined with a space (' '), supporting multiple arguments within the same entry.
The new arguments are added to existing ones, unless the --set-jvm-args flag is used. In this case, JVM
arguments are set to the KOTLIN_JUPYTER_JAVA_OPTS variable. Both adding and setting arguments work alongside the KOTLIN_JUPYTER_JAVA_OPTS_EXTRA variable.
While Jupyter Kernel environment variable substitutions are supported in the env argument, no replacement occurs
if the used environment variable doesn't exist.
The add-kernel script utilizes the argparse Python library, supporting the --help, @argfile (you don't need the @ symbol in PowerShell), and --opt=value arguments.
The --jvm-arg=arg argument is required when passing JVM arguments that start with the - symbol.
Here's an example of an add-kernel script to create a custom Kotlin Kernel for Jupyter Notebook:
python -m kotlin_kernel add-kernel --name "JDK 15 Big 2 GPU" --jdk ~/.jdks/openjdk-15.0.2 --jvm-arg=-Xmx8G --env CUDA_VISIBLE_DEVICES 0,1Explore the sections below to learn about the features of the Kotlin Kernel for Jupyter notebooks. You can leverage these features using Kotlin Notebook in IntelliJ IDEA, Datalore, or other Jupyter Notebook clients.
The features of the Kotlin Kernel for Jupyter notebooks include:
- REPL commands
- Dependencies resolving
- Default repositories
- Line magics
- Supported libraries
- Rich output
- Rendering
- Autocompletion
- Error analysis
Click here to expand the features.
Our Kotlin Kernel for notebooks comes with a set of REPL commands that let you explore your notebook environment. The following REPL commands are supported:
| Command | Description |
|---|---|
:help |
Displays help information with details of the notebook version, line magics, and supported libraries. |
:classpath |
Displays the current classpath of your notebook environment, showing a list of locations where the notebook searches for libraries and resources. |
:vars |
Displays information about the declared variables and their values. |
You can easily add dynamic dependencies to your notebook from a remote Maven repository or local ones (local JARs). You can add dependencies through annotations or Gradle-like syntax.
You can add dynamic dependencies to the notebook using the following annotations:
-
@file:DependsOn(<coordinates>): In this annotation, you need to specify the coordinates of the dependency. This annotation adds artifacts (like JAR files) to the notebook's classpath. It supports absolute and relative paths to class directories or JARs, as well as Ivy and Maven artifacts:@file:DependsOn(“io.ktor:ktor-client-core-jvm:$ktorVersion“)
-
@file:Repository(<absolute-path>): In this annotation, you need to specify the absolute path of the dependency. This annotation adds a directory or an Ivy or Maven repository to the notebook environment. To specify a Maven local repository, use:@file:Repository("*mavenLocal")
You can load any library from the Maven repository using Gradle-like syntax in any cell, specifying repositories, locations, and so on:
USE {
repositories {
maven {
url = "https://my.secret.repo/maven/"
credentials {
username = USER
password = TOKEN
}
}
}
dependencies {
val ktorVersion = "2.0.3"
implementation("my.secret:artifact:1.0-beta")
implementation("io.ktor:ktor-client-core:$ktorVersion")
implementation("io.ktor:ktor-client-apache:$ktorVersion")
}
}Note: You can use the same Gradle-like syntax to integrate new libraries.
When adding dependencies, consider the following:
- Dependencies in remote repositories are resolved via Maven resolver.
- Caches are stored in the
~/.m2/repositoryfolder by default. However, due to network issues or several artifact resolutions running in parallel, caches may become corrupted. - If you have issues with artifacts resolution, remove caches, restart kernel, and try again.
- While you can utilize Gradle-like syntax, Gradle is not running under the hood, and Gradle metadata is not resolved.
Therefore, advanced dependency configurations are not available. For example, top-level Multiplatform dependencies are not supported.
In these cases, you need to use the
-jvmvariant manually.
The following Maven repositories are included by default:
You can directly use libraries and dependencies from these repositories within your Kotlin code running in the Jupyter Notebook environment.
Line magics are special commands, starting with the % character, that interact with the notebook on a per-line basis. Line magics allow you to import libraries, configure output settings, and perform more operations.
You can use the following line magics in your notebooks using the Kotlin Kernel:
| Magic | Description | Usage example |
|---|---|---|
%use |
Imports supported libraries and injects code from these libraries(artifact resolution, default imports, initialization code, and type renderers). | %use klaxon(5.5), lets-plot |
%trackClasspath |
Logs any changes of the current classpath. This command is useful for debugging artifact resolution failures. | %trackClasspath [on/off] |
%trackExecution |
Logs pieces of code to be executed. This command is useful for debugging libraries support. | %trackExecution [all/generated/off] |
%useLatestDescriptors |
Sets the latest versions of available library descriptors instead of bundled descriptors (used by default). Note that bundled descriptors are preferred because the current kernel version might not support the latest descriptors. For better notebook stability, use bundled descriptors. | %useLatestDescriptors [on/off] |
%output |
Configures the output capturing settings. | %output --max-cell-size=1000 --no-stdout --max-time=100 --max-buffer=400 |
%logLevel |
Sets logging level. | %logLevel [off/error/warn/info/debug] |
Note: For more information, see Line magics.
Kotlin Kernel for Jupyter notebooks comes with a set of integrated libraries, which you can import into your notebook by running the %use line magic before the library's name within a cell.
When you import a library using the %use line magic, the following functionality is added to the notebook:
- Repositories to search for library artifacts
- Artifact dependencies
- Default imports
- Library initialization code
- Renderers for special types. For example, charts and data frames
This behavior is defined by the JSON library descriptor, which provides a set of properties with default values that can be overridden when the library is imported.
To check the descriptors of all supported libraries, see the libraries repository.
The major use case for library properties is to specify a particular library version. If the descriptor has only one property, the library version can be defined without naming:
%use dataframe(0.10)
If the descriptor has more than one property, you need to use the property name:
%use spark(scala=2.11.10, spark=2.4.2)
You can include several libraries in a single %use statement, separated by commas (,):
%use kandy, dataframe, mysql(8.0.15)
You can also specify the source of the library descriptor. By default, it's taken from the libraries repository. To try a descriptor from another revision, use the following syntax:
// Uses a specific version from the default repository
%use [email protected]
// Uses a specific commit of the library
%use lets-plot@ref[24a040fe22335648885b106e2f4ddd63b4d49469]
// Uses a specific version of the library from a custom repository along with library arguments
%use dataframe@dev(0.10)
Note: Using a fixed version of a library is preferred over using the
%useLatestDescriptorsline magic.
Additionally, you can try resolving the library descriptor from a local file or a remote URL:
// Loads the library from a file
%use mylib@file[/home/user/lib.json]
// Loads the library from a file, and the Kernel detects it's a file
%use @/home/user/libs/lib.json
// Specifies a directory and a file name without extension (the extension file should be JSON)
%use lib@/home/user/libs
// Loads the library descriptor from a remote URL
%use herlib@url[https://site.com/lib.json]
// Loads the library descriptor from a remote URL. The `url[]` part can be omitted if the URL responds with 200(OK)
%use @https://site.com/lib.json
// Loads the library dependencies from a specified JSON file. The library name and URL resolution can be omitted
%use @file[lib.json]
Here you can find all the supported libraries you can use in you Kotlin notebooks through the %use line magic.
Click to see the list of supported libraries.
- 2p-kt - Kotlin Multi-Platform ecosystem for symbolic AI
- biokotlin - BioKotlin aims to be a high-performance bioinformatics library that brings the power and speed of compiled programming languages to scripting and big data environments.
- combinatoricskt - A combinatorics library for Kotlin
- coroutines - Asynchronous programming and reactive streams support
- dataframe - Kotlin framework for structured data processing
- datetime - Kotlin date/time library
- deeplearning4j - Deep learning library for the JVM
- deeplearning4j-cuda - Deep learning library for the JVM (CUDA support)
- default - Default imports: dataframe and Kandy libraries
- develocity-api-kotlin - A library to use the Develocity API in Kotlin scripts or projects
- exposed - Kotlin SQL framework
- fuel - HTTP networking library
- gradle-enterprise-api-kotlin - [Deprecated: use develocity-api-kotlin] A library to use the Gradle Enterprise API in Kotlin scripts or projects
- gral - Java library for displaying plots
- intellij-sdk - IntelliJ SDK that could be used in embedded mode of Kotlin Notebook
- jdsp - Java library for signal processing
- jupyter-js - Experimental
%javascript/%typescript/%jsxline magic support - kalasim - Discrete event simulator
- kaliningraph - Graph library with a DSL for constructing graphs and visualizing the behavior of graph algorithms
- kandy - Kotlin plotting DSL for Lets-Plot
- kandy-echarts - Kotlin plotting DSL for Apache ECharts
- khttp - HTTP networking library
- klaxon - JSON parser for Kotlin
- kmath - Experimental Kotlin algebra-based mathematical library
- kotlin-dl - KotlinDL library which provides Keras-like API for deep learning
- kraphviz - Graphviz wrapper for JVM
- kravis - Kotlin grammar for data visualization
- ksl - KSL - Kotlin Simulation Library for Monte Carlo and Discrete-Event Simulation
- kt-math - Kotlin multi-platform port of java.math.*
- ktor-client - Asynchronous HTTP client
- lets-plot - Kotlin API for Lets-Plot: multiplatform plotting library based on Grammar of Graphics
- lets-plot-gt - Lets-Plot visualisation for GeoTools toolkit
- lib-ext - Extended functionality for Jupyter kernel
- londogard-nlp-toolkit - A Natural Language Processing (NLP) toolkit for Kotlin on the JVM
- multik - Multidimensional array library for Kotlin
- mysql - MySql JDBC Connector
- openai - OpenAI API for Jupyter Notebooks
- plotly - [beta] Plotly.kt jupyter integration for static plots.
- plotly-server - [beta] Plotly.kt jupyter integration for dynamic plots.
- rdkit - Open-Source Cheminformatics Software
- reflection - Imports for Kotlin Reflection
- roboquant - Algorithmic trading platform written in Kotlin
- serialization - Deserialize JSON content using kotlinx.serialization and automatically generate classes for it
- smile - Statistical Machine Intelligence and Learning Engine
- spark - Kotlin API for Apache Spark: unified analytics engine for large-scale data processing
- spark-streaming - Kotlin API for Apache Spark Streaming: scalable, high-throughput, fault-tolerant stream processing of live data streams
- webtau - WebTau end-to-end testing across layers
By default, our Kotlin Kernel for Jupyter notebooks displays return values in text form. However, you can enrich the output by rendering graphics, HTML, or other MIME-encoded data format.
One approach is to send MIME-encoded results to the client using the MIME helper function:
fun MIME(vararg mimeToData: Pair<String, String>): MimeTypedResult For example:
MIME("text/html" to "<p>Some <em>HTML</em></p>", "text/plain" to "No HTML for text clients")Another approach is to use the HTML helper function, which provides a simpler way to display HTML content directly:
fun HTML(text: String): MimeTypedResultFor example:
HTML("<p>This is an example of <strong>HTML</strong> content rendered using the HTML helper function.</p>")Rendering is the procedure of transforming a value to a form that is appropriate for displaying it in the Jupyter client. The Kotlin Kernel for Jupyter notebook supports various features and mechanisms for rendering values:
-
Renderers: Transform values into other representations. Renderers are controlled via the
RenderersProcessormethod, and you can access it with the notebook API entry point. The Kotlin kernel iterates through a list of available renderers, trying to find one that can handle the given data. A library can define one or more renderers. -
DisplayResultandRenderable: Objects implementingDisplayResultandRenderableinterfaces are rendered to output JSON. -
Text rendering: Render objects to strings using text renderers. Text renderers are controlled via the
TextRenderersProcessormethod, and you can access the method with the notebook API entry point. A library can define one or more renderers. The Kotlin Kernel iterates until at least one renderer returns a non-null string for a passed argument. This kind of renderer can be easily composed with each other. For example, a text renderer for iterables can render its elements with a text renderer processor recursively. -
Throwables rendering: Throwable renderers behave as regular renderers but handle exceptions and errors generated during cell execution.
Successful value evaluation triggers a rendering process. Initially, the RenderersProcessor attempts to convert the value
into a Renderable or DisplayResult object. If successful, the result is transformed into JSON output using the toJson() method.
For Unit values, no output is generated.
If the value cannot be rendered as Renderable or DisplayResult, the TextRenderersProcessor takes over.
It iterates to render the value to a string using the defined text renderers, seeking a non-null string representation. If no suitable renderer is found,
the value is transformed into a string using the toString() method. The resulting string is wrapped in a text/plain MIME JSON.
Upon execution failure, an exception is generated. The first applicable throwable renderer
is chosen for this exception, and the exception is passed to this renderer's render() method so the returned value
is displayed. If no applicable throwable renderer is found, the exception's message and stack trace are printed to standard error.
When working with the Kotlin notebooks, press TAB to get the list of suggested items for completion.
In Jupyter Notebook, you don't need to press TAB. Completion is requested automatically.
Completion works for both globally defined symbols and local symbols, which were loaded into the notebook during cell evaluation.
If you use Jupyter Notebook, you'll notice that compilation errors and warnings are underlined in red and yellow, correspondingly. If you hover the cursor over underlined text, you'll get an error message that can help you to fix the error.
This error analysis is achieved by the kernel-level extension of Jupyter Notebook. The extension sends error-analysis requests to the kernel and renders their results.
- Run the
./gradlew installDebugGradle command. The debugger port is selected automatically. The default port is 1044, and if it's unavailable, the consequent ports are used. If you want an exact port, specify the-PdebugPort=<port>Gradle option. - Run the corresponding command to open the desired notebook client.
- Attach a remote debugger to the JVM with the corresponding port (the debug port number is printed in the terminal when the kernel starts).
Read this article if you want to integrate new JVM libraries in the Kotlin Kernel for Jupyter notebooks.
To learn more, explore the available documentation:
- Docs site with rendered KDoc comments from the codebase.
- Docs about integrating new libraries. If you are a library author, you may be interested in the
apimodule in our project. There is also alibmodule that contains entities available from the Notebook cells, and ashared-compilermodule for Jupyter REPL integration into a standalone application or IDEA plugin. - Explore the Kotlin Notebook docs to learn about features, use cases, and tutorials.
We welcome contributions to further enhance our project! If you come across any issues or have feature requests, please don't hesitate to file an issue.
For issues specifically related to the Kotlin Notebook plugin, utilize this tracker.
Pull requests are highly appreciated! When submitting a pull request, ensure it corresponds to an existing issue.
If you are planning a substantial change, we recommend discussing it with a project maintainer. You can reach out to me through email, Kotlin Slack, or Telegram.
We look forward to your contributions!