Added

• Added CalendarApi.get_company_calendar() to retrieve fiscal calendars for a company.
• Added support for DerivedSignal objects in ExportApi.signal_query() and ExportApi.batched_signal_query().

Breaking changes

• Renamed the Python package from exabel-data-sdk to exabel.
• Removed legacy setup.py; the project now uses pyproject.toml exclusively.
• Removed deprecated parameters from bulk_import(): resources_batches, threads_for_import_func, threads_for_insert_func, insert_func.
• Removed deprecated parameter threads_for_import from TimeSeriesApi.import_time_series().
• Removed username/password authentication for the Export API (deprecated in 7.0.0).

Changed

• Upgraded protobuf stubs.
• Refactored core data classes (DataSet, Entity, EntityType, Relationship, RelationshipType) to use Python dataclasses.
• Migrated SDK tests from unittest to pytest.

Full Changelog: v7.0.0...v8.0.0

Added

• Added support for authenticating with access tokens. Set environment variable EXABEL_ACCESS_TOKEN or command line argument --access-token.
• Added support for hierarchical models in KPI Service.
• Added support for Custom Calendar Service
• Added support for Python 3.14

Breaking changes

• Renamed exabel_data_sdk.client.api.data_classes.kpi_mapping_group.py to exabel_data_sdk.client.api.data_classes.kpi_mapping_group_reference.
• Removed support for Python 3.9 (EOL)

Deprecated

• Support for username/password authentication when accessing the Export API is now deprecated. This authentication method will be removed in a future release. Use access token instead.

Full Changelog: v6.3.0...v7.0.0

Added

• Add support for KPI API.
• Add support for highlighted signals and derived signals in Data set service.

Changed

• Limit number of concurrent requests to list time series.
• Support signal resource names up to 128 characters.
• Handle missing values when loading time series data.

Full Changelog: v6.2.0...v6.3.0

What's Changed

• Increased maximum --pit-offset to 100 days
• Increased maximum signal identifier to 128 characters

Full Changelog: v6.1.0...v6.2.0

Added

• Added script for deleting time series data points.
• Added arguments --should-optimise / --no-should-optimise for controlling time series storage optimization on time series loading script.

Changed

• Increased max backoff from 1 to 10 minutes for retries when bulk importing resources.

Added

• Add scripts for handling Data Sets (create, read, update, delete and list)
• Add export to CSV file when listing signals
• Add script for loading signals from CSV with columns signal name, display name and description (load_signals_from_csv.py)
• Add security as entity type for checking identifiers in the script check_company_identifiers_in_csv.py

Changed

• Include parsing of resource name in scripts for better handling of resource names.
• Refactor list_time_series.py and list_entities.py scripts.

Fixed

• Improved feedback messages

Removed

• Removed support for Python 3.8 as it no longer receives security support. Please upgrade to Python 3.9 or newer.

Added

• Support Snowflake key pair authentication.

Fixed

• Fix paging iterator bug which could lead to hanging when listing resources.

Full Changelog: v5.2.2...v5.3.0

Fixed

• Make the description attribute for ExportData script optional to maintain compatibility

Changed

• Included RESOURCE_EXHAUSTED as a specific retryable error for gRPC errors
• Improved logging for export

Added

• Support logging of active threads in debug mode
• Support loading ratio and percent as time series units
• Add retry logic for all gRPC API calls using the SDK clients

Changed

• Improve logging during loading of resources (progress tracking and summary of failures)
• When replacing time series, empty time series are also loaded and will replace existing time series

Fixed

• Drop rows for companies that was not possible to map when importing time series
• Drop duplicated rows for time series metadata loading