The Query Class

Note: Developers building new applications are strongly encouraged to use the NDB Client Library, which has several benefits compared to this client library, such as automatic entity caching via the Memcache API. If you are currently using the older DB Client Library, read the DB to NDB Migration Guide

Class Query represents a query for retrieving entities from the App Engine Datastore. (See also the related class GqlQuery, which defines queries using GQL, a SQL-like query language.)

Query is defined in the module google.appengine.ext.db.

Note: The index-based query mechanism supports a wide range of queries and is suitable for most applications. However, it does not support some kinds of query common in other database technologies: in particular, joins and aggregate queries aren't supported within the Datastore query engine. See Datastore Queries page for limitations on Datastore queries.

Introduction

An application creates a query object for a given entity kind by calling either the Query constructor directly

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

q = db.Query(Song)

or the class method all() of the kind's model class:

q = Song.all()

Without further modification, the resulting instance of class Query will retrieve all existing entities of the specified kind. Method calls on the object can then be used to customize the query with additional filter criteria, ancestor conditions, and sort orders:

q.filter('title =', 'Imagine')
q.ancestor(ancestor_key)
q.order('-date')

For convenience, all of these methods return the query object itself so that they can cascaded in a single statement:

q.filter('title =', 'Imagine').ancestor(key).order('-date')

The application can then execute the query and access the results in any of the following ways:

  • Treat the query object as an iterable, to process matching entities one at a time:

    for song in q:
      print song.title

    This implicitly calls the query's run() method to generate the matching entities. It is thus equivalent to

    for song in q.run():
      print song.title

    You can set a limit on the number of results to process with the keyword argument limit:

    for song in q.run(limit=5):
      print song.title

    The iterator interface does not cache results, so creating a new iterator from the query object reiterates the same query from the beginning.

  • Call the query's get() method, to obtain the first single matching entity found in the Datastore:

    song = q.get()
    print song.title
  • Call the query's fetch() method, to obtain a list of all matching entities up to a specified number of results:

    results = q.fetch(limit=5)
    for song in results:
      print song.title

    As with run(), the query object does not cache results, so calling fetch() a second time reissues the same query.

    Note: You should rarely need to use this method; it is almost always better to use run() instead.

Constructor

The constructor for class Query is defined as follows:

class Query (model_class=None, keys_only=False, cursor=None, namespace=None, projection=None, distinct=False)

Creates an instance of class Query for retrieving entities from the App Engine Datastore.

Without further modification, the resulting query object will retrieve all existing entities of the kind specified by model_class. The instance methods filter(), ancestor(), and order() can then be used to customize the query with additional filter criteria, ancestor conditions, and sort orders.

Arguments

model_class
Model (or Expando) class representing the entity kind to which the query applies.
keys_only
If true, return only keys instead of complete entities. Keys-only queries are faster and cheaper than those that return complete entities.
cursor
Cursor position at which to resume query.
namespace
Namespace to use for query.
projection
List or tuple of names of properties to return. Only entities possessing the specified properties will be returned. If not specified, entire entities are returned by default. Projection queries are faster and cheaper than those that return complete entities.

Note: Specifying this parameter may change the query's index requirements.

distinct
In the case of Projection queries, distinct=True specifies that only completely unique results will be returned in a result set. This will only return the first result for entities which have the same values for the properties that are being projected.
True
Only return the first result for each distinct set of value for properties in the projection.
False
All results are returned.

Instance Methods

Instances of class Query have the following methods:

filter (property_operator, value)

Adds a property filter to the query. The query will return only entities whose properties satisfy all of its filters.

Arguments

property_operator
String consisting of a property name and an optional comparison operator (=, !=, <, <=, >, >=, IN), separated by a space: for example, 'age >'. If just a property name is specified without a comparison operator, the filter compares for equality (=) by default.
value
Value to compare with the property value. For example:

q.filter('height >', 42).filter('city =', 'Seattle')
q.filter('user =', users.get_current_user())

The comparison value specified should be of the same value type as that of the property being compared.

ancestor (ancestor)

Adds an ancestor filter to the query. The query will return only entities with the specified ancestor.

Argument

ancestor
Ancestor entity or key.
order (property)

Adds a sort order to the query. If more than one sort order is added, they will be applied in the order specified.

Argument

property
String giving the name of the property on which to sort, optionally preceded by a hyphen (-) to specify descending order. Omitting the hyphen specifies ascending order by default. For example:
# Order alphabetically by last name:
q.order('last_name')

# Order by height, tallest to shortest:
q.order('-height')
projection ()

Returns the tuple of properties in the projection or None.

is_keys_only ()

Returns a boolean value indicating whether the query is a keys-only query.

run (read_policy=STRONG_CONSISTENCY, deadline=60, offset=0, limit=None, batch_size=20, keys_only=False, projection=None, start_cursor=None, end_cursor=None)

Returns an iterable for looping over the results of the query. This allows you to specify the query's operation with parameter settings and access the results iteratively:

  1. Retrieves and discards the number of results specified by the offset argument.
  2. Retrieves and returns up to the maximum number of results specified by the limit argument.

The loop's performance thus scales linearly with the sum of offset + limit. If you know how many results you want to retrieve, you should always set an explicit limit value.

This method uses asynchronous prefetching to improve performance. By default, it retrieves its results from the Datastore in small batches, allowing the application to stop the iteration and avoid retrieving more results than are needed.

Tip: To retrieve all available results when their number is unknown, set batch_size to a large value, such as 1000.

Tip: If you don't need to change the default argument values, you can just use the query object directly as an iterable to control the loop. This implicitly calls run() with default arguments.

Arguments

read_policy
Read policy specifying desired level of data consistency:
STRONG_CONSISTENCY
Guarantees the freshest results, but limited to a single entity group.
EVENTUAL_CONSISTENCY
Can span multiple entity groups, but may occasionally return stale results. In general, eventually consistent queries run faster than strongly consistent queries, but there is no guarantee.

Note: Global (non-ancestor) queries ignore this argument.

deadline
Maximum time, in seconds, to wait for Datastore to return a result before aborting with an error. Accepts either an integer or a floating-point value. Cannot be set higher than the default value (60 seconds), but can be adjusted downward to ensure that a particular operation fails quickly (for instance, to return a faster response to the user, retry the operation, try a different operation, or add the operation to a task queue).
offset
Number of results to skip before returning the first one.
limit
Maximum number of results to return. If omitted or set to None, all available results will be retrieved by default.
batch_size
Number of results to attempt to retrieve per batch. If limit is set, defaults to the specified limit; otherwise defaults to 20.
keys_only
If true, return only keys instead of complete entities. Keys-only queries are faster and cheaper than those that return complete entities.
projection
List or tuple of names of properties to return. Only entities possessing the specified properties will be returned. If not specified, entire entities are returned by default. Projection queries are faster and cheaper than those that return complete entities.

Note: Specifying this parameter may change the query's index requirements.

start_cursor
Cursor position at which to start query.
end_cursor
Cursor position at which to end query.
get (read_policy=STRONG_CONSISTENCY, deadline=60, offset=0, keys_only=False, projection=None, start_cursor=None, end_cursor=None)

Executes the query and returns the first result, or None if no results are found.

Arguments

read_policy
Read policy specifying desired level of data consistency:
STRONG_CONSISTENCY
Guarantees the freshest results, but limited to a single entity group.
EVENTUAL_CONSISTENCY
Can span multiple entity groups, but may occasionally return stale results. In general, eventually consistent queries run faster than strongly consistent queries, but there is no guarantee.

Note: Global (non-ancestor) queries ignore this argument.

deadline
Maximum time, in seconds, to wait for Datastore to return a result before aborting with an error. Accepts either an integer or a floating-point value. Cannot be set higher than the default value (60 seconds), but can be adjusted downward to ensure that a particular operation fails quickly (for instance, to return a faster response to the user, retry the operation, try a different operation, or add the operation to a task queue).
offset
Number of results to skip before returning the first one.
keys_only
If true, return only keys instead of complete entities. Keys-only queries are faster and cheaper than those that return complete entities.
projection
List or tuple of names of properties to return. Only entities possessing the specified properties will be returned. If not specified, entire entities are returned by default. Projection queries are faster and cheaper than those that return complete entities.

Note: Specifying this parameter may change the query's index requirements.

start_cursor
Cursor position at which to start query.
end_cursor
Cursor position at which to end query.
fetch (limit, read_policy=STRONG_CONSISTENCY, deadline=60, offset=0, keys_only=False, projection=None, start_cursor=None, end_cursor=None)

Executes the query and returns a (possibly empty) list of results:

  1. Retrieves and discards the number of results specified by the offset argument.
  2. Retrieves and returns up to the maximum number of results specified by the limit argument.

The method's performance thus scales linearly with the sum of offset + limit.

Note: This method is merely a thin wrapper around the run() method, and is less efficient and more memory-intensive than using run() directly. You should rarely need to use fetch(); it is provided mainly for convenience in cases where you need to retrieve a full in-memory list of query results.

Tip: The fetch() method is designed to retrieve only the number of results specified by the limit argument. To retrieve all available results of a query when their number is unknown, use run() with a large batch size, such as run(batch_size=1000), instead of fetch().

Arguments

limit
Maximum number of results to return. If set to None, all available results will be retrieved.
read_policy
Read policy specifying desired level of data consistency:
STRONG_CONSISTENCY
Guarantees the freshest results, but limited to a single entity group.
EVENTUAL_CONSISTENCY
Can span multiple entity groups, but may occasionally return stale results. In general, eventually consistent queries run faster than strongly consistent queries, but there is no guarantee.

Note: Global (non-ancestor) queries ignore this argument.

deadline
Maximum time, in seconds, to wait for Datastore to return a result before aborting with an error. Accepts either an integer or a floating-point value. Cannot be set higher than the default value (60 seconds), but can be adjusted downward to ensure that a particular operation fails quickly (for instance, to return a faster response to the user, retry the operation, try a different operation, or add the operation to a task queue).
offset
Number of results to skip before returning the first one.
keys_only
If true, return only keys instead of complete entities. Keys-only queries are faster and cheaper than those that return complete entities.
projection
List or tuple of names of properties to return. Only entities possessing the specified properties will be returned. If not specified, entire entities are returned by default. Projection queries are faster and cheaper than those that return complete entities.

Note: Specifying this parameter may change the query's index requirements.

start_cursor
Cursor position at which to start query.
end_cursor
Cursor position at which to end query.
count (read_policy=STRONG_CONSISTENCY, deadline=60, offset=0, limit=1000, start_cursor=None, end_cursor=None)

Returns the number of results matching the query. This is faster by a constant factor than actually retrieving all of the results, but the running time still scales linearly with the sum of offset + limit. Unless the result count is expected to be small, it is best to specify a limit argument; otherwise the method will continue until it finishes counting or times out.

Arguments

read_policy
Read policy specifying desired level of data consistency:
STRONG_CONSISTENCY
Guarantees the freshest results, but limited to a single entity group.
EVENTUAL_CONSISTENCY
Can span multiple entity groups, but may occasionally return stale results. In general, eventually consistent queries run faster than strongly consistent queries, but there is no guarantee.

Note: Global (non-ancestor) queries ignore this argument.

deadline
Maximum time, in seconds, to wait for Datastore to return a result before aborting with an error. Accepts either an integer or a floating-point value. Cannot be set higher than the default value (60 seconds), but can be adjusted downward to ensure that a particular operation fails quickly (for instance, to return a faster response to the user, retry the operation, try a different operation, or add the operation to a task queue).
offset
Number of results to skip before counting the first one.
limit
Maximum number of results to count.
start_cursor
Cursor position at which to start query.
end_cursor
Cursor position at which to end query.
index_list ()

Returns a list of indexes used by an executed query, including primary, composite, kind, and single-property indexes.

Caution: Invoking this method on a query that has not yet been executed will raise an AssertionError exception.

Note: This feature is not fully supported on the development server. When used with the development server, the result is either the empty list or a list containing exactly one composite index.

For example, the following code prints various information about the indexes used by a query:

# other imports ...
import webapp2
from google.appengine.api import users
from google.appengine.ext import db

class Greeting(db.Model):
  author = db.StringProperty()
  content = db.StringProperty(multiline=True)
  date = db.DateTimeProperty(auto_now_add=True)

class MainPage(webapp2.RequestHandler):
  def get(self):
    user = users.get_current_user()
    q = db.Query(Greeting)
    q.filter("author =", user.user_id())
    q.order("-date")
    q.fetch(100)
    index_list = q.index_list()
    for ix in index_list:
      self.response.out.write("Kind: %s" % ix.kind())
      self.response.out.write("<br />")
      self.response.out.write("Has ancestor? %s" % ix.has_ancestor())
      self.response.out.write("<br />")
      for name, direction in ix.properties():
        self.response.out.write("Property name: "+name)
        self.response.out.write("<br />")
        if direction == db.Index.DESCENDING:
          self.response.out.write("Sort direction: DESCENDING")
        else:
          self.response.out.write("Sort direction: ASCENDING")
        self.response.out.write("<br />")

This produces output like the following for each index:

Kind: Greeting
Has ancestor? False
Property name: author
Sort direction: ASCENDING
Property name: date
Sort direction: DESCENDING
cursor ()

Returns a base64-encoded cursor string denoting the position in the query's result set following the last result retrieved. The cursor string is safe to use in HTTP GET and POST parameters, and can also be stored in the Datastore or Memcache. A future invocation of the same query can provide this string via the start_cursor parameter or the with_cursor() method to resume retrieving results from this position.

Caution: Invoking this method on a query that has not yet been executed will raise an AssertionError exception.

Note: Not all queries are compatible with cursors; see the Datastore Queries page for more information.

with_cursor (start_cursor, end_cursor=None)

Specifies the starting and (optionally) ending positions within a query's result set from which to retrieve results. The cursor strings denoting the starting and ending positions can be obtained by calling cursor() after a previous invocation of the query. The current query must be identical to that earlier invocation, including the entity kind, property filters, ancestor filters, and sort orders.

Arguments

start_cursor
Base64-encoded cursor string specifying where to start the query.
end_cursor
Base64-encoded cursor string specifying where to end the query.