Skip to main content

Creating a package

This documents explains how to create a simple package that contains an info panel displaying some simple statistics for a text in a selected grid cell.

1. Create a package

Before we start, set up development environment if you haven't done it already.

Now, let's create a package:

grok create TextStats --test

This creates a folder TextStats with the default package structure. For the newly created package, you have to install the dependencies. Run this from the TextStats folder:

npm install

2. Implement the panel function

Add the actual panel's code at TextStats/src/package.ts:

//name: TextStats
//tags: panel, widgets
//input: string str {semType: text}
//output: widget result
export function textStats(str: string) {
// for 'gattaca', produces {"g": 1, "a": 3, "t": 2, "c": 1}
const symbolCounts = Array.from(str).reduce((counts, ch) => {
counts[ch] = (counts[ch] || 0) + 1;
return counts;
}, Object.create(null));
return new DG.Widget(ui.divV([
ui.divText("Counting characters:"),

That's it! What creates a panel is this function plus the annotation contained in the comments preceding it. Datagrok recognizes these comments and makes the function textStats become a panel producing a widget, taking a string as an input.

3. Build

Run webpack from the TextStats folder:

npm run build

Note. The build script in package.json includes commands to process your package, e.g., webpack. You are not supposed to modify this script.

4. Publish

Now, let's publish our package to the dev server:

grok publish dev

The return code should be 0 to indicate a successful deployment.

5. Test

Now go to Datagrok and open any data file with string columns. This could be a demog dataset from our demo datasets. Find any text column, right-click on it, pick Column Properties, then add a tag: quality : text. This tag tells Datagrok that there is a plain text stored inside this column. Navigate to a text cell and find your freshly added panel with text stats on the right side of Datagrok UI. Note, that semantic type text of a column matches {semType: text} in function parameter declaration. Datagrok also can detect semantic types automatically, or with help of Detectors .

We have completed the deployment. Let's navigate to Manage | Packages and find your package in the list. Note it has your name prefixed with a v., which means it's only published for you. This is called a Debug mode. To make it available to the user or a group of interest, you can Share it to the group via right-click menu on the package. Don't forget to publish the package as grok publish public --release: this now makes this package released for these groups of interest.

See also: