2.0 Update

Improvements

• Rewrote module around new statements syntax
• Added a subquery method
• Added __and__ / __or__ support
• Added basic group_by support
• Type hints

Changes

• in_() now takes a single parameter rather than attempting to guess the input

Deprecations

• Removed ftrack_query.entity and other __getattr__ methods, replaced with ftrack_query.attr (entity.parent.task is now attr('parent.task'))
• Removed Comparison.__call__ (because primary keys had to be hardcoded - attr('value')('x') is no longer mapped to attr('value') == 'x')
• Removed ability to join queries with or (too complex and not useful - select('Task').where(name='Task 1') | select('Task').where(name='Task 2') will no longer evaluate to 'select from Task where (name is "Task 1" or name is "Task 2")')
• Removed Session.get() override (the type won't automatically default to TypedContext any more)
• Removed keys() method on an entity type (since it wasn't a clean implementation)
• Removed the custom logger and merged with the default one

Bug fixes and custom exceptions

Improvements

• Raises UnboundSessionError if attempting to execute a query with no attached session.

Fixes

• Fix error when using in_() without any arguments or with an empty generator.

Change component cleanup method to an option

Changes

• The clean_components() method has been moved to .options(remove_components=True)

Fix bug when reusing create statements

Fixes

• Fix bug where reusing create statements would result in duplicate values.

Improved statements

Features

• Add attr function to use with comparisons (attr('x.y') is replacing entity.x.y)
• Add .options() method on queries to set the page size or attach a new session
• Add FTRACK_API_PAGE_SIZE environment variable to set the page size globally, and a page_size parameter to FTrackQuery() to set it for the entire session
• Add session.select() as shorthand for session.execute(select())

Improvements

• .in_() now supports any data types
• Generators are now automatically unpacked when passed to in_()
• Subqueries will not attempt to remap any values (it was confusing behaviour)

New statement syntax

Features

• Added select, update, delete and create statements to work in a similar way to SQLAlchemy.
• Rewrote subquery support - subqueries will be used when the API allows for it, otherwise an error will be raised (the old way would automatically execute the query).

Improvements

• Add a clean_components() method when building a delete statement, which will automatically remove matching components from every location.
• Stop automatically executing the query when checking for truthiness.
• Allow the use of order_by for sorting, and also accept asc/desc as strings.
• Added a contains method, which is the equivelant of .like(f'%{val}%').

Automatically execute query objects in comparisons

Improvements

• Any query objects passed into comparisons will now be executed.
  The following is now valid:
  session.Episode.where(project=query) - Executes query.one().
  session.Episode.where(entity.project == query) - Executes query.one().
  session.Episode.where(entity.project.in_(query)) - Executes query.all().
  session.Episode.where(entity.project.in_(query, query)) - Executes query.one() on each.

Improve populate behaviour

Improvements

• Accept lists in populate function - session.populate(item, (entity.name, entity.project.name)) is now valid

Improvement to primary key query

Fixes

• If the primary key query finds multiple entities with different capitalisation, return the one exactly matching the input

Remove ftrack_api modules as class attributes

Changes

• Remove FTrackQuery.exc (ftrack_api.exception)
• Remove FTrackQuery.symbol (ftrack_api.symbol)
• Remove FTrackQuery.Entity (ftrack_api.entity.base.Entity)