Learn how to access locations using the API and write your own location plugins.
If you are just looking to get started with ftrack and to track your files, consider using the storage scenario instead. With a storage scenario a location plugin will be automatically configured for both clients python and python (legacy).
Collaborating across locations
Successful collaboration is at the heart of great work and we aim to make it easier than ever to share data across teams and timezones.
As part of this, allow us to introduce our ‘locations’ feature as a way to easily track and manage data (files, image sequences etc.) using ftrack across different locations.
With locations it is possible to see where published data is in the world and also to transfer data automatically between different locations, even different storage mechanisms, by defining a few simple Python plugins. By keeping track of the size of the data it also helps manage storage capacity better. In addition, the intrinsic links to production information makes assigning work to others and transferring only the relevant data much simpler as well as greatly reducing the burden on those responsible for archiving finished work.
Before continuing, it is important to emphasise the current scope of the locations feature.
At present, it only deals with data (files on disk) and is not an all purpose solution to distributed working, but rather a part of it. Eventually, the locations concept will extend to be relevant in all parts of the system, but for now it is limited to managing components.
As such, most interaction and configuration with the feature is through the API.
The system is implemented in layers using a few key concepts in order to provide a balance between out of the box functionality and custom configuration.
Data locations can be varied in scope and meaning - a facility, a laptop, a specific drive. As such, rather than place a hard limit on what can be considered a location, ftrack simply requires that a location be identifiable by a string and that string be unique to that location.
A global company with facilities in many different parts of the world might follow a location naming convention similar to the following:
Whereas, for a looser setup, the following might suit better:
When tracking data across several locations it is important to be able to quickly find out where data is available and where it is not. As such, ftrack provides simple mechanisms for retrieving information on the availability of a component in each location through the API.
For a single file, the availability with be either 0% or 100%. For containers, such as file sequences, each file is tracked separately and the availability of the container calculated as an overall percentage (e.g. 47%).
This information allows for better decisions to be made on when work can start. For example, a compositor working on a new sequence may only need some of the frames available in order to start work.
Due to the flexibility of what can be considered a location, the system must be able to cope with locations that represent different ways of storing data. For example, data might be stored on a local hard drive, a cloud service or even in a database.
In addition, the method of accessing that storage can change depending on perspective - local filesystem, FTP, S3 API etc.
To handle this, ftrack introduces the idea of an accessor that provides access to the data in a standard way. An accessor is implemented as a Python plugin following a set interface and can be loaded at runtime to configure relevant access to a location.
With an accessor configured for a location, it becomes possible to not only track data, but also manage it through ftrack by using the accessor to add and remove data from the location.
At present, ftrack includes a
disk accessor for local filesystem access and an
S3 accessor for data stored on Amazon S3. More will be added over time and developers are encouraged to contribute their own.
[Link to code example]
Another important consideration for locations is how data should be structured in the location (folder structure and naming conventions). For example, different facilities may want to use different folder structures, or different storage mechanisms may use different paths for the data.
For this, ftrack supports the use of a Python structure plugin. This plugin is called when adding a component to a location in order to determine the correct structure to use.
[Link to code example]
A structure plugin accepts an ftrack entity as its input and so can be reused for generating general structures as well. For example, an action callback could be implemented to create the base folder structure for some selected shots by reusing a structure plugin.
When a component can be linked to multiple locations it becomes necessary to store information about the relationship on the link rather than directly on the component itself. The most important information is the path to the data in that location.
However, as seen above, not all locations may be filesystem based or accessed using standard filesystem protocols. For this reason, and to help avoid confusion, this path is referred to as a resource identifier and no limitations are placed on the format. Keep in mind though that accessors use this information (retrieved from the database) in order to work out how to access the data, so the format used must be compatible with all the accessors used for any one location. For this reason, most resource identifiers should ideally look like relative filesystem paths.
To further support custom formats for resource identifiers, it is also possible to configure a resource identifier transformer plugin which will convert the identifiers before they are stored centrally and after they are retrieved.
A possible use case of this might be to store JSON encoded metadata about a path in the database and convert this to an actual filesystem path on retrieval.
[Link to code example]
You can read more about the implementation of location plugins here: