Automating Notebook Execution

Practicus AI allows you to execute notebooks in an automated fashion, which can be particularly useful for testing and automated workflows.

Executing Notebooks from Practicus AI

1) Create a Notebook File

Begin by creating a file named sample_notebook.ipynb.

2) Add a Parameters Cell (Optional)

Add a cell with the following code to define parameters that can be updated dynamically later:

# Notebook parameters - can be dynamically updated later
some_param = 1
some_other_param = 2

For Jupyter: 1. Select the newly added cell. 2. In the top-right section of JupyterLab, open the property inspector. 3. Click on Add Tag and enter parameters, then click +. 4. The cell now has the parameters tag.

For VS Code: 1. Select the cell you just created. 2. Click the ... in the cell's upper-right corner. 3. Add a cell tag named parameters. 4. The cell is now tagged.

3) Add Your Code as Usual

Continue writing code in the notebook as you normally would. For example, add the following code to print messages and validate your parameters:

print("Starting to run sample notebook")

if some_param <= 0:
    raise ValueError("some_param must be > 0")
else:
    print("some_param value is acceptable:", some_param)

print("Finished running sample notebook")

4) Execute the Notebook

Once you’ve set up the notebook and its parameters, you can execute it using Practicus AI’s automated run capabilities.

import practicuscore as prt

# This will run just fine, 
# and save the resulting output to sample_notebook_output.ipynb
prt.notebooks.execute_notebook("sample_notebook")

# The below *will FAIL* since some_param cannot be 0
prt.notebooks.execute_notebook(
    "sample_notebook",
    parameters={
        "some_param": 0 
    }
)

Advanced features

# Advanced Notebook automation parameters
default_output_folder="~/tests"  # If none, writes notebook output to same folder as notebook
default_failed_output_folder="~/tests_failed"  # If not none, collects failed notebook results 

# By calling configure you can save notebook results to central location
# Please note that you can do this to a shared/ folder daily where all of our members have access to
prt.notebooks.configure(
    default_output_folder=default_output_folder,
    default_failed_output_folder=default_failed_output_folder,
    add_time_stamp_to_output=True,
)

# This will work
prt.notebooks.execute_notebook("sample_notebook")

# This will fail but does not stop the execution of the notebook
prt.notebooks.execute_notebook(
    "sample_notebook",
    parameters={
        "some_param": 0 
    }
)

# Calling validate_history() will raise an exception IF any of the previous notebooks failed
# This is useful to have a primary "orchestration" notebook that executes other child notebooks,
# And then finally fails itself if there was a mistake. 
# You can then report the result, essentially creating a final report. 
prt.notebooks.validate_history()

# You can view the passed and failed execution results in 
# ~/tests and ~/tests_failed with a time stamp (optional)

Executing notebooks from terminal

You can run the below command to execute a notebook from the terminal or an .sh script.

prtcli execute-notebook -p notebook=my-notebook.ipynb 

---

Previous: Automated Code Quality | Next: Work With Connections