# Functions on objects

You can write functions that interact with the Ontology using the Python Ontology SDK.

## Generate a Python Ontology SDK

To generate a Python Ontology SDK client, navigate to the [**Resource imports** sidebar](../resource-imports-sidebar/index.md) and select
**Add > Ontology**. From there, select your desired Ontology and import any objects and links you would like to interact with in
your functions. After saving to confirm your selections, a banner will appear to indicate that a corresponding OSDK has not yet been created.

Navigate to the **SDK Generation** tab to generate and install the OSDK.

:::callout{theme="warning"}
If SDK generation returns an import error such as `cannot import name 'Equipment2' from 'ontology_sdk.ontology.objects'`, confirm that you selected an Ontology in the blue bar at the bottom of the interface. Importing object types does not select the Ontology required for SDK generation.
:::

![Create a new SDK.](../../../assets/773f5b30053492085122b22d11ad2ddeb66b17a3ff62206083fa5b3c35c86dd1.webp)

If no OSDK has been generated, you will be prompted to enter a name for the generated package.
The package name cannot be changed after the first version has been generated.

After selecting **Create new version**, you can monitor the generation progress from this view.

![The SDK package being generated.](../../../assets/567454a6cb89a4e79303d93dad2e4478d101580068436d24a6f45fc3aede549d.webp)

Once generation is complete, you will need to install the newly generated version with the <img src="../../../assets/f0924038f4bdc9fe48468a13515772faf57c6155bd970f00ff171e528dfd2c0b.webp" alt="Install"> button.

![The generated SDK package is ready to install.](../../../assets/d30b8ccb8529bd4b82a809260bce3b99b945bb8e33d408f6c88265395317fd5e.webp)

This will trigger an interactive install in the task runner panel.
Once that task completes successfully (the Task Runner will display `BUILD SUCCESSFUL`), code completion for the OSDK will be available in your code assist session.

The `meta.yml` file will also be updated to include a reference to the generated package.
You can manually update `meta.yml` instead of using the installation helper, but if you manually update `meta.yml`, you will need to rebuild your code assist session to pick up the changes.

![The meta.yml includes the installed SDK package.](../../../assets/ed1f2db7930bec54d02cfa6256d1fcefdc9ea181c6bbc3b7829f46d4d64e83fd.webp)

Any time you import additional resources in the sidebar you will be prompted to generate and install a new version of the OSDK that includes these resources.
Additionally, if you modify imported resources (for instance, adding a new property to an already imported object type), you will need to generate a new OSDK version to pick up these changes.

The generated OSDK also provides classes for struct properties. Import each struct property class from its object-specific module:

```python
from ontology_sdk.ontology.objects import MyObjectType
from ontology_sdk.ontology.objects.my_object_type import MyStructPropertyClass
```

For an example of using a generated struct property class, refer to [Edit struct properties](../python-ontology-edits/index.md#edit-struct-properties).

## Examples

For an example object type named `Aircraft` with properties `brand` and `capacity`, you could write a
function that accepts an `Aircraft` object and summarizes it like so:

```python
from functions.api import function
from ontology_sdk.ontology.objects import Aircraft

@function
def aircraft_input_example(aircraft: Aircraft) -> str:
    return f"{aircraft.brand} aircraft, holds {aircraft.capacity} passengers"
```

Furthermore, if you wanted to search for `Aircraft` objects satisfying a certain capacity threshold, you could write the
following:

```python
from functions.api import function
from ontology_sdk import FoundryClient
from ontology_sdk.ontology.objects import Aircraft
from ontology_sdk.ontology.object_sets import AircraftObjectSet

@function
def aircraft_search_example() -> AircraftObjectSet:
    client = FoundryClient()
    return client.ontology.objects.Aircraft.where(Aircraft.object_type.capacity > 100)
```

The Python OSDK also offers beta features such as interoperability with pandas DataFrames:

```python
from functions.api import function
from ontology_sdk.ontology.object_sets import AircraftObjectSet

@function
def aircraft_dataframe_example(aircrafts: AircraftObjectSet) -> int:
    df = aircrafts.to_dataframe()
    return df['capacity'].sum()
```

Review the [Python Ontology SDK documentation](../../ontology-sdk/python-osdk/index.md) for more information.