xarray Objects for VS Code

Inspect xarray objects directly from Python notebooks in VS Code. The extension finds DataArray, Dataset, and DataTree variables in Jupyter and marimo notebooks, adds quick actions to notebook hovers, and provides an xarray Objects panel plus a rendered xarray Detail view.

Notebook Workflow

Inspect xarray variables

Work in a notebook with ordinary xarray variables. The extension recognizes DataArray, Dataset, and DataTree objects from the active kernel and keeps the Objects panel in sync with the active notebook.

Use hover actions

Hover over an xarray variable name in a Python cell to see its object type, dimensions, and quick actions.

Use Details to open the rendered xarray HTML representation, or Pin to keep an important variable visible in the xarray Objects panel. When erlab is available, DataArray hovers can also show additional quick actions for connecting to erlab tools.

Browse the xarray Objects panel

The xarray Objects panel lists notebook variables by xarray type, including generic DataArray, Dataset, and DataTree entries.

From the panel you can:

• Select an object to open it in the Detail view.
• Filter the list by xarray type.
• Pin important variables so they stay at the top.
• Jump back to the notebook definition for a variable.
• Refresh the object list after notebook state changes.

Open the xarray Detail view

The xarray Detail view renders the object's HTML representation inside VS Code, so you can inspect dimensions, coordinates, data variables, and attributes without leaving the notebook.

You can open Details from a hover, from the xarray Objects panel, or from the Jupyter Variables view.

Optional ERLab Integration

The core xarray workflow works with any Python kernel that has xarray installed. If the kernel also has erlab, additional DataArray tools appear where they apply:

• Watch and Unwatch connect notebook DataArrays to the ImageTool Manager.
• ImageTool opens a DataArray in erlab's ImageTool.
• More tools provides access to ktool, dtool, restool, meshtool, ftool, and goldtool.
• Reconnect Watches reconnects saved ImageTool Manager rows to matching notebook variables.

Usage

1. Open a Jupyter (.ipynb) notebook or a marimo notebook with a Python kernel.
2. Create or load xarray DataArray, Dataset, or DataTree variables.
3. Hover over variable names in Python cells for quick actions.
4. Open the erlab panel and browse xarray Objects.
5. Select an object, open Details, pin variables, filter by type, or jump to definitions.

Commands

• erlab.openDetail - Open the xarray Detail view from the command palette or the Jupyter Variables view.
• erlab.showOutput - Show the extension output channel.

ERLab Commands

These commands appear when the active notebook kernel has the erlab package installed:

• erlab.watch - Watch a DataArray, or show it if it is already watched.
• erlab.unwatch - Stop watching a DataArray.
• erlab.restoreWatches - Reconnect saved ImageTool Manager rows whose variables exist in the active notebook.
• erlab.itool - Open the DataArray in ImageTool.
• erlab.ktool - Open the DataArray in ktool.
• erlab.dtool - Open the DataArray in dtool.
• erlab.restool - Open the DataArray in restool.
• erlab.meshtool - Open the DataArray in meshtool.
• erlab.ftool - Open the DataArray in ftool.
• erlab.goldtool - Open the DataArray in goldtool.
• erlab.xarray.otherTools - Show a picker for additional ERLab tools.

Settings

• erlab.xarray.displayExpandAttrs (default: true) - Expand the attributes section in xarray HTML representations.
• erlab.xarray.displayExpandCoords (default: true) - Expand the coordinates section in xarray HTML representations.
• erlab.xarray.displayExpandData (default: false) - Expand the data section in xarray HTML representations.
• erlab.itool.useManager (default: true) - Open in the ImageTool Manager when it is already running. Otherwise, open in a new ImageTool window bound to the current kernel. If set to false, always open in the current kernel.
• erlab.kernel.commandTimeoutMs (default: 10000) - Timeout in milliseconds for erlab command executions. Set to 0 to disable the timeout.

Requirements

• VS Code or a compatible editor with notebook support.
• The Jupyter extension (ms-toolsai.jupyter) for Jupyter notebooks, or the marimo extension (marimo-team.vscode-marimo) for marimo notebooks.
• A Python kernel with xarray >=2024.10.
• Optional: erlab for ImageTool and related interactive tools. A Qt backend, PyQt6 or PySide6, is required for erlab GUI tools. Watch and unwatch require erlab>=3.20.0; reconnecting saved watched variables requires erlab>=3.22.0.

Notes

• Hover, panel, and context actions query the active notebook kernel.
• You may be prompted to allow notebook code execution the first time the extension inspects kernel variables.
• ERLab GUI tool commands can block if the kernel has not enabled Qt event-loop integration. In IPython/Jupyter kernels, run %gui qt before opening GUI tools.