-
Notifications
You must be signed in to change notification settings - Fork 8
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge pull request #285 from ToposInstitute/jupyter-refactor
Reactive helpers for Jupyter kernels
- Loading branch information
Showing
2 changed files
with
139 additions
and
95 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,110 @@ | ||
| import type { ServerConnection } from "@jupyterlab/services"; | ||
| import type { IKernelConnection, IKernelOptions } from "@jupyterlab/services/lib/kernel/kernel"; | ||
| import { | ||
| type Accessor, | ||
| type Resource, | ||
| type ResourceReturn, | ||
| createResource, | ||
| onCleanup, | ||
| } from "solid-js"; | ||
|
|
||
| type ResourceRefetch<T> = ResourceReturn<T>[1]["refetch"]; | ||
|
|
||
| type ServerSettings = Parameters<typeof ServerConnection.makeSettings>[0]; | ||
|
|
||
| /** Create a Jupyter kernel in a reactive context. | ||
| Returns a kernel as a Solid.js resource and a callback to restart the kernel. | ||
| The kernel is automatically shut down when the component is unmounted. | ||
| */ | ||
| export function createKernel( | ||
| serverOptions: ServerSettings, | ||
| kernelOptions: IKernelOptions, | ||
| ): [Resource<IKernelConnection>, ResourceRefetch<IKernelConnection>] { | ||
| const [kernel, { refetch: restartKernel }] = createResource(async () => { | ||
| const jupyter = await import("@jupyterlab/services"); | ||
|
|
||
| const serverSettings = jupyter.ServerConnection.makeSettings(serverOptions); | ||
|
|
||
| const kernelManager = new jupyter.KernelManager({ serverSettings }); | ||
| const kernel = await kernelManager.startNew(kernelOptions); | ||
|
|
||
| return kernel; | ||
| }); | ||
|
|
||
| onCleanup(() => kernel()?.shutdown()); | ||
|
|
||
| return [kernel, restartKernel]; | ||
| } | ||
|
|
||
| /** Create a Julia kernel in a reactive context. */ | ||
| export function createJuliaKernel(serverOptions: ServerSettings) { | ||
| return createKernel(serverOptions, { | ||
| // XXX: Do I have to specify the Julia version? | ||
| name: "julia-1.11", | ||
| }); | ||
| } | ||
|
|
||
| /** Execute code in a Jupyter kernel and retrieve JSON data reactively. | ||
| Assumes that the computation will return JSON data using the "application/json" | ||
| MIME type in Jupyter. Returns the post-processed data as a Solid.js resource and | ||
| a callback to rerun the computation. | ||
| The resource depends reactively on the kernel: if the kernel changes, the code | ||
| will be automatically re-executed. It does *not* depend reactively on the code. | ||
| If the code changes, it must be rerun manually. | ||
| */ | ||
| export function executeAndRetrieve<S, T>( | ||
| kernel: Accessor<IKernelConnection | undefined>, | ||
| executeCode: Accessor<string | undefined>, | ||
| postprocess: (data: S) => T, | ||
| ): [Resource<T | undefined>, ResourceRefetch<T>] { | ||
| const [data, { refetch: reexecute }] = createResource(kernel, async (kernel) => { | ||
| // Request that kernel execute code, if defined. | ||
| const code = executeCode(); | ||
| if (code === undefined) { | ||
| return undefined; | ||
| } | ||
| const future = kernel.requestExecute({ code }); | ||
|
|
||
| // Set up handler for result from kernel. | ||
| let result: { data: S } | undefined; | ||
| future.onIOPub = (msg) => { | ||
| if ( | ||
| msg.header.msg_type === "execute_result" && | ||
| msg.parent_header.msg_id === future.msg.header.msg_id | ||
| ) { | ||
| const content = msg.content as JsonDataContent<S>; | ||
| const data = content["data"]?.["application/json"]; | ||
| if (data !== undefined) { | ||
| result = { data }; | ||
| } | ||
| } | ||
| }; | ||
|
|
||
| // Wait for execution to finish and process result. | ||
| const reply = await future.done; | ||
| if (reply.content.status === "abort") { | ||
| throw new Error("Execution was aborted"); | ||
| } | ||
| if (reply.content.status === "error") { | ||
| // Trackback list already includes `reply.content.evalue`. | ||
| const msg = reply.content.traceback.join("\n"); | ||
| throw new Error(msg); | ||
| } | ||
| if (result === undefined) { | ||
| throw new Error("Data was not received from the kernel"); | ||
| } | ||
| return postprocess(result.data); | ||
| }); | ||
|
|
||
| return [data, reexecute]; | ||
| } | ||
|
|
||
| /** JSON data returned from a Jupyter kernel. */ | ||
| type JsonDataContent<T> = { | ||
| data?: { | ||
| "application/json"?: T; | ||
| }; | ||
| }; |