Python 3.9 is now generally available.

google.appengine.ext.db.Query

A Query instance queries over instances of Models.

Inherits From: expected_type

You construct a query with a model class, like this:

class Story(db.Model): title = db.StringProperty() date = db.DateTimeProperty()

query = Query(Story)

You modify a query with filters and orders like this:

query.filter('title =', 'Foo') query.order('-date') query.ancestor(key_or_model_instance)

Every query can return an iterator, so you access the results of a query by iterating over it:

for story in query: print story.title

For convenience, all of the filtering and ordering methods return "self", so the easiest way to use the query interface is to cascade all filters and orders in the iterator line like this:

for story in Query(story).filter('title =', 'Foo').order('-date'): print story.title

model_class Model class to build query for.
keys_only Whether the query should return full entities or only keys.
projection A tuple of strings representing the property names to include in the projection this query should produce or None. Setting a projection is similar to specifying SELECT prop1, prop2, ... in SQL. See _BaseQuery.projection for details on projection queries.
distinct A boolean, true if the projection should be distinct. See _BaseQuery.is_distinct for details on distinct queries.
cursor A compiled query from which to resume.
namespace The namespace to use for this query.

Methods

ancestor

View source

Sets an ancestor for this query.

This restricts the query to only return results that descend from a given model instance. In other words, all of the results will have the ancestor as their parent, or parent's parent, etc. The ancestor itself is also a possible result!

Args
ancestor Model or Key (that has already been saved)

Returns
Self to support method chaining.

Raises
TypeError if the argument isn't a Key or Model; NotSavedError if it is, but isn't saved yet.

count

View source

Number of entities this query fetches.

Beware: count() ignores the LIMIT clause on GQL queries.

Args
limit A number. If there are more results than this, stop short and just return this number. Providing this argument makes the count operation more efficient.
kwargs Any keyword arguments accepted by datastore_query.QueryOptions().

Returns
Number of entities this query fetches.

cursor

View source

Get a serialized cursor for an already executed query.

The returned cursor effectively lets a future invocation of a similar query to begin fetching results immediately after the last returned result from this query invocation.

Returns
A base64-encoded serialized cursor.

Raises
AssertionError If the query has not been executed.

fetch

View source

Return a list of items selected using SQL-like limit and offset.

Always use run(limit=...) instead of fetch() when iterating over a query.

Beware: offset must read and discard all skipped entities. Use cursor()/with_cursor() instead.

Args
limit Maximum number of results to return.
offset Optional number of results to skip first; default zero.
kwargs Any keyword arguments accepted by datastore_query.QueryOptions().

Returns
A list of db.Model instances. There may be fewer than 'limit' results if there aren't enough results to satisfy the request.

filter

View source

Add filter to query.

Args
property_operator string with the property and operator to filter by.
value the filter value.

Returns
Self to support method chaining.

Raises
PropertyError if invalid property is provided.

get

View source

Get first result from this.

Beware: get() ignores the LIMIT clause on GQL queries.

Args
kwargs Any keyword arguments accepted by datastore_query.QueryOptions().

Returns
First result from running the query if there are any, else None.

index_list

View source

Get the index list for an already executed query.

Returns
A list of indexes used by the query.

Raises
AssertionError If the query has not been executed.

is_distinct

View source

Returns true if the projection query should be distinct.

This is equivalent to the SQL syntax: SELECT DISTINCT. It is only available for projection queries, it is not valid to specify distinct without also specifying projection properties.

Distinct projection queries on entities with multi-valued properties will return the same entity multiple times, once for each unique combination of properties included in the projection.

Returns
True if this projection query is distinct.

is_keys_only

View source

Returns whether this query is keys only.

Returns
True if this query returns keys, False if it returns entities.

order

View source

Set order of query result.

To use descending order, prepend - (minus) to the property name, e.g., -date rather than date.

Args
property Property to sort on.

Returns
Self to support method chaining.

Raises
PropertyError if invalid property is provided.

projection

View source

Returns the tuple of properties in the projection or None.

Projected results differ from normal results in multiple ways:

  • they only contain a portion of the original entity and cannot be put;
  • properties defined on the model, but not included in the projections will have a value of None, even if the property is required or has a default value;
  • multi-valued properties (such as a ListProperty) will only contain a single value.
  • dynamic properties not included in the projection will not appear on the model instance.
  • dynamic properties included in the projection are deserialized into their indexed type. Specifically one of str, bool, long, float, GeoPt, Key or User. If the original type is known, it can be restored using datastore_types.RestoreFromIndexValue.

However, projection queries are significantly faster than normal queries.

Projection queries on entities with multi-valued properties will return the same entity multiple times, once for each unique combination of values for properties included in the order, an inequality property, or the projected properties.

Returns
The list of properties in the projection, or None if no projection is set on this query.

run

View source

Iterator for this query.

If you know the number of results you need, use run(limit=...) instead, or use a GQL query with a LIMIT clause. It's more efficient. If you want all results use run(batch_size=<large number>).

Args
kwargs Any keyword arguments accepted by datastore_query.QueryOptions().

Returns
Iterator for this query.

with_cursor

View source

Set the start and end of this query using serialized cursors.

Conceptually cursors point to the position between the last result returned and the next result so running a query with each of the following cursors combinations will return all results in four chunks with no duplicate results:

query.with_cursor(end_cursor=cursor1) query.with_cursors(cursor1, cursor2) query.with_cursors(cursor2, cursor3) query.with_cursors(start_cursor=cursor3)

For example if the cursors pointed to: cursor: 1 2 3 result: a b c d e f g h

The results returned by these queries would be [a, b], [c, d], [e, f], [g, h] respectively.

Cursors are pinned to the position just after the previous result (last result, exclusive), so if results are inserted or deleted between the time the cursor was made and these queries are executed, the cursors stay pinned to these positions. For example:

delete(b, f, g, h) put(a1, b1, c1, d1) cursor: 1(b) 2(d) 3(f) result: a a1 b1 c c1 d d1 e

The results returned by these queries would now be: [a, a1], [b1, c, c1, d], [d1, e], [] respectively.

Args
start_cursor The cursor position at which to start or None
end_cursor The cursor position at which to end or None

Returns
This Query instance, for chaining.

Raises
BadValueError when cursor is not valid.

__getitem__

View source

Support for query[index] and query[start:stop].

Beware: this ignores the LIMIT clause on GQL queries.

Args
arg Either a single integer, corresponding to the query[index] syntax, or a Python slice object, corresponding to the query[start:stop] or query[start:stop:step] syntax.

Returns
A single Model instance when the argument is a single integer. A list of Model instances when the argument is a slice.

__iter__

View source

Iterator for this query.

If you know the number of results you need, consider fetch() instead, or use a GQL query with a LIMIT clause. It's more efficient.