looker-pdf

This is a tool that can be used to download and combine multiple Looker dashboards into a single PDF.

Installation

We use Conda to manage our Python versions. In this project, we are using Python version 3.7.3. To create the Conda environment, change to the root of this repo and type:

$ conda env create

That will install the Python executable and required Python libraries, and create a new Conda environment named looker-pdf. In order to switch to the new environment:

$ source activate looker-pdf

Using The CLI

Once you've switched to the new environment, you should be able to run the included looker-pdf.py script:

Usage: looker-pdf.py [OPTIONS] CONFIG_FILE OUTPUT_FILE

Options:
  --base-url TEXT       [required]
  --client-id TEXT      [required]
  --client-secret TEXT  [required]
  --debug               Show logging output.
  --help                Show this message and exit.

CONFIG_FILE refers to a JSON file with the arguments to use to generate the dashboard PDFs. OUTPUT_FILE is the name of the output PDF. --base-url, --client-id, and --client-secret are required to communicate with the Looker API. Details for obtaining API credentials can be found here.

Config File Format

The config file should represent an object with two keys: dashboard_args and shared_args. dashboard_args should be a list of objects, each of which has a required dashboard_id number, and optional arguments used to render the dashboard as a PDF:

{
    "dashboard_args": [
        {
            "dashboard_id": 1,
            "dashboard_style": "single_column",
            "dashboard_filters": "foo=bar&baz=qux",
            "width": 2048,
            "height": 1536
        },
        {
            "dashboard_id": 2,
            "width": 2048,
            "height": 1536
        }
    ]
}

shared_args is a single object which can be used to specify options that should be used by all dashboards:

{
    "dashboard_args": [
        {
            "dashboard_id": 1,
            "dashboard_style": "single_column",
            "dashboard_filters": "foo=bar&baz=qux"
        },
        {
            "dashboard_id": 2
        }
    ],
    "shared_args": {
	"width": 2048,
        "height": 1536
    }
}

We've included a sample config file which can be used as a template.

Authenticating with Environment Variables

In order to communicate with the Looker API you must specify a Base URL, a Client ID, and a Client Secret. You can use the --base-url, --client-id, and --client-secret CLI options to do this, but you can also use environment variables LOOKER_BASE_URL, LOOKER_CLIENT_ID, and LOOKER_CLIENT_SECRET. You can set those from the command line, in a shell setup script, or you can use Honcho to read them from a .env file. We've included a sample .env file. If you fill in the authentication details in the sample file and rename it to .env, then you can run the CLI without needing to specify options like this:

honcho run python looker-pdf.py config.json output.pdf

A Note On the Looker API Client SDK

The Looker API Client SDK in Python is included in the looker/ directory of this repo. I followed the directions for building a Looker API Client SDK in Python in the Looker Community Forum. I used the latest release of swagger-codegen (v. 3.0.8). I also ran swagger-codegen-cli.jar with a config.json file which specified the Python module name to use (the module name defaults to swagger_client by default):

{
	"packageName": "looker"
}

So my command for running swagger-codegen-cli.jar looked like this:

$ java -jar ./swagger-codegen/modules/swagger-codegen-cli/target/swagger-codegen-cli.jar generate -i looker_api.json -l python -o python_sdk -c config.json

I moved the looker subdirectory from python_sdk to this repo.

On line 49 of [looker/configuration.py](looker/configuration.py#L49], in the Configuration class, our company base URL was set as the default base URL. I replaced that with None.