SQLAlchemy 1.2 Documentation
A custom list that manages index/position information for contained elements.
orderinglist is a helper for mutable ordered relationships. It will
intercept list operations performed on a
automatically synchronize changes in list position onto a target scalar
slide table, where each row refers to zero or more entries
in a related
bullet table. The bullets within a slide are
displayed in order based on the value of the
position column in the
bullet table. As entries are reordered in memory, the value of the
position attribute should be updated to reflect the new sort order:
Base = declarative_base() class Slide(Base): __tablename__ = 'slide' id = Column(Integer, primary_key=True) name = Column(String) bullets = relationship("Bullet", order_by="Bullet.position") class Bullet(Base): __tablename__ = 'bullet' id = Column(Integer, primary_key=True) slide_id = Column(Integer, ForeignKey('slide.id')) position = Column(Integer) text = Column(String)
The standard relationship mapping will produce a list-like attribute on each
Slide containing all related
but coping with changes in ordering is not handled automatically.
When appending a
attribute will remain unset until manually assigned. When the
is inserted into the middle of the list, the following
will also need to be renumbered.
from sqlalchemy.ext.orderinglist import ordering_list Base = declarative_base() class Slide(Base): __tablename__ = 'slide' id = Column(Integer, primary_key=True) name = Column(String) bullets = relationship("Bullet", order_by="Bullet.position", collection_class=ordering_list('position')) class Bullet(Base): __tablename__ = 'bullet' id = Column(Integer, primary_key=True) slide_id = Column(Integer, ForeignKey('slide.id')) position = Column(Integer) text = Column(String)
With the above mapping the
Bullet.position attribute is managed:
s = Slide() s.bullets.append(Bullet()) s.bullets.append(Bullet()) s.bullets.position >>> 1 s.bullets.insert(1, Bullet()) s.bullets.position >>> 2
OrderingList construct only works with changes to a
collection, and not the initial load from the database, and requires that the
list be sorted when loaded. Therefore, be sure to specify
order_by on the
relationship() against the target ordering attribute, so that the
ordering is correct when first loaded.
OrderingList only provides limited functionality when a primary
key column or unique column is the target of the sort. Operations
that are unsupported or are problematic include:
- two entries must trade values. This is not supported directly in the case of a primary key or unique constraint because it means at least one row would need to be temporarily removed first, or changed to a third, neutral value while the switch occurs.
- an entry must be deleted in order to make room for a new entry. SQLAlchemy’s unit of work performs all INSERTs before DELETEs within a single flush. In the case of a primary key, it will trade an INSERT/DELETE of the same primary key for an UPDATE statement in order to lessen the impact of this limitation, however this does not take place for a UNIQUE column. A future feature will allow the “DELETE before INSERT” behavior to be possible, allevating this limitation, though this feature will require explicit configuration at the mapper level for sets of columns that are to be handled in this way.
ordering_list() takes the name of the related object’s ordering
attribute as an argument. By default, the zero-based integer index of the
object’s position in the
ordering_list() is synchronized with the
ordering attribute: index 0 will get position 0, index 1 position 1, etc. To
start numbering at 1 or some other integer, provide
ordering_list(attr, count_from=None, **kw)¶
OrderingListfactory for use in mapper definitions.
Returns an object suitable for use as an argument to a Mapper relationship’s
from sqlalchemy.ext.orderinglist import ordering_list class Slide(Base): __tablename__ = 'slide' id = Column(Integer, primary_key=True) name = Column(String) bullets = relationship("Bullet", order_by="Bullet.position", collection_class=ordering_list('position'))
- attr¶ – Name of the mapped attribute to use for storage and retrieval of ordering information
- count_from¶ – Set up an integer-based ordering, starting at
count_from. For example,
ordering_list('pos', count_from=1)would create a 1-based list in SQL, storing the value in the ‘pos’ column. Ignored if
Additional arguments are passed to the
Numbering function: consecutive integers starting at 0.
Numbering function: consecutive integers starting at 1.
Numbering function: consecutive integers starting at arbitrary start.
OrderingList(ordering_attr=None, ordering_func=None, reorder_on_append=False)¶
A custom list that manages position information for its children.
__init__(ordering_attr=None, ordering_func=None, reorder_on_append=False)¶
A custom list that manages position information for its children.
collection_classlist implementation that syncs position in a Python list with a position attribute on the mapped objects.
This implementation relies on the list starting in the proper order, so be sure to put an
order_byon your relationship.
- ordering_attr¶ – Name of the attribute that stores the object’s order in the relationship.
- ordering_func¶ –
Optional. A function that maps the position in the Python list to a value to store in the
ordering_attr. Values returned are usually (but need not be!) integers.
ordering_funcis called with two positional parameters: the index of the element in the list, and the list itself.
If omitted, Python list indexes are used for the attribute values. Two basic pre-built numbering functions are provided in this module:
count_from_1. For more exotic examples like stepped numbering, alphabetical and Fibonacci numbering, see the unit tests.
- reorder_on_append¶ –
Default False. When appending an object with an existing (non-None) ordering value, that value will be left untouched unless
reorder_on_appendis true. This is an optimization to avoid a variety of dangerous unexpected database writes.
SQLAlchemy will add instances to the list via append() when your object loads. If for some reason the result set from the database skips a step in the ordering (say, row ‘1’ is missing but you get ‘2’, ‘3’, and ‘4’), reorder_on_append=True would immediately renumber the items to ‘1’, ‘2’, ‘3’. If you have multiple sessions making changes, any of whom happen to load this collection even in passing, all of the sessions would try to “clean up” the numbering in their commits, possibly causing all but one to fail with a concurrent modification error.
Recommend leaving this with the default of False, and just call
reorder()if you’re doing
append()operations with previously ordered instances or when doing some housekeeping after manual sql operations.
L.append(object) – append object to end
L.insert(index, object) – insert object before index
pop([index]) → item -- remove and return item at index (default last).¶
Raises IndexError if list is empty or index is out of range.
L.remove(value) – remove first occurrence of value. Raises ValueError if the value is not present.
Synchronize ordering for the entire collection.
Sweeps through the list and ensures that each object has accurate ordering information set.