Kili

See changelog [here](https://docs.kili-technology.com/changelog/july-6th-2022)

**New**
- We’ve released the Kili command line interface. It features commands to:
- Create a project
- List all the projects in your organization
- Import assets into a project
- Describe a project.
Kili CLI gives you the option to easily manage projects without the need to run complex GraphQL queries or using SDK. For more information, refer to https://python-sdk-docs.kili-technology.com/latest/cli.
- We’ve implemented a new mutation: add_assets_to_review. It enables you to add labeled assets to review.
- The KILI_API_ENDPOINT environment variable is now read when instantiating the Kili client. If it differs from the default one in your installation, you can now use this environment variable instead of passing it to the Kili object.
Changed
- In Python SDK, when you use an invalid API key, you will see an error. This will help you troubleshoot issues.

**Fixed**
- Calling the project_versions function with default parameters used to fail before because a deprecated field was used. We've fixed this issue.

**Upcoming changes**
- From Kili version 2.116, projects of type FRAME will be renamed to VIDEO and projects of type VIDEO will be renamed to VIDEO_LEGACY. They will be deprecated starting Kili version 2.116 and removed in a later release.
- The use of .assets, .labels, .project_users, .issues, and.project_versions without providing a project_id is now deprecated. Providing the project_id will be mandatory from Kili version 2.116.

See changelog [here](https://docs.kili-technology.com/changelog/june-1st-2022)

The field `label_json_response_contains` used in `assets` and `count_assets` queries is now deprecated. It will be removed in version 2.115
Instead, you can use `label_category_search` to filter based on categories in labels. Have a look at the following guide to do so.

"How to filter based on label categories"

The search query is composed of logical expressions following this format:

[job_name].[category_name].count [comparaison_operator] [value]

Where:

- `[job_name]` is the name of the job in the interface
- `[category_name]` is the name of the category in the interface for this job
- `[comparaison_operator]` can be one of: [`==`, `>=`, `<=`, `<`, `>`]
- `[value]` is an integer that represents the count of such objects of the given category in the label

These operations can be separated by OR and AND operators

Examples:

- label_category_search = `JOB_CLASSIF.CATEGORY_A.count > 0`
- label_category_search = `JOB_CLASSIF.CATEGORY_A.count > 0 OR JOB_NER.CATEGORY_B.count > 0`
- label_category_search = `(JOB_CLASSIF.CATEGORY_A.count == 1 OR JOB_NER.CATEGORY_B.count > 0) AND JOB_BBOX.CATEGORY_C.count > 10`

See changelog [here](https://docs.kili-technology.com/changelog/may-18th-2022)

You no longer have to build batch calls when using the `append_many_to_dataset()`, `delete_many_from_dataset()`, `update_properties_in_assets()` and `create_predictions()` mutations with more than 100 assets at once. Instead, you will see an internal paginated loop.