Skip to main content

Data Storage Guide for Python

The LeanStorage Python SDK can be used to persist and query data in LeanCloud. The code below shows how you can create an object and store it into the cloud:

# Declare a class
Todo = leancloud.Object.extend('Todo')

# Create an object
todo = Todo()

# Set values of fields
todo.set('title', 'R&D Weekly Meeting')
todo.set('content', 'All team members, Tue 2pm')

# Save the object to the cloud

The SDK designed for each language interacts with the same REST API via HTTPS, offering fully functional interfaces for you to manipulate the data in the cloud.

Installing SDK

See Installing Python SDK.



The objects on the cloud are built around leancloud.Object. Each leancloud.Object contains key-value pairs of JSON-compatible data. This data is schema-free, which means that you don't need to specify ahead of time what keys exist on each leancloud.Object. Simply set whatever key-value pairs you want, and our backend will store them.

For example, the leancloud.Object storing a simple todo item may contain the following data:

title:      "Email Linda to Confirm Appointment",
isComplete: false,
priority: 2,
tags: ["work", "sales"]

Data Types

leancloud.Object supports a wide range of data types to be used for each field, including common ones like String, Number, Boolean, Object, Array, and Date. You can nest objects in JSON format to store more structured data within a single Object or Array field.

Special data types supported by leancloud.Object include Pointer and File, which are used to store a reference to another leancloud.Object and binary data respectively.

leancloud.Object also supports GeoPoint, a special data type you can use to store location-based data. See GeoPoints for more details.

Some examples:

// Basic types
from datetime import datetime

// Basic types
bool = True
number = 2018
string = 'Top Hit Songs'
date =
list = [string, number]
dictionary = {
'number': number,
'string': string

# Create an object
TestObject = leancloud.Object.extend('TestObject')
test_object = TestObject()
test_object.set('testString', string)
test_object.set('testNumber', number)
test_object.set('testBoolean', bool)
test_object.set('testList', list)
test_object.set('testDict', dictionary)
test_object.set('testDate', date)

We do not recommend storing large pieces of binary data like images or documents with leancloud.Object using byte[]. The size of each leancloud.Object should not exceed 128 KB. We recommend using leancloud.File for storing images, documents, and other types of files. To do so, create leancloud.File objects and assign them to fields of leancloud.Object. See Files for details.

Keep in mind that our backend stores dates in UTC format and the SDK will convert them to local times upon retrieval.

The date values displayed on

> Data are also converted to match your operating system's time zone. The only exception is that when you retrieve these date values through our REST API, they will remain in UTC format. You can manually convert them using appropriate time zones when necessary.

To learn about how you can protect the data stored on the cloud, see Data Security.

Creating Objects

The code below creates a new instance of leancloud.Object with class Todo:

# Create a new subclass of leancloud.Object
Todo = leancloud.Object.extend('Todo')

# Create a new instance of that class
todo = Todo()

The constructor takes a class name as a parameter so that the cloud knows the class you are using to create the object. A class is comparable to a table in a relational database. A class name starts with a letter and can only contain numbers, letters, and underscores.

Saving Objects

The following code saves a new object with class Todo to the cloud:

# Declare a class
Todo = leancloud.Object.extend('Todo')

# Create an object
todo = Todo()

# Set values of fields
todo.set('title', 'Sign up for Marathon')
todo.set('priority', 2)

# Save the object to the cloud

To make sure the object is successfully saved, take a look at

> Data > Todo in your app. You should see a new entry of data with something like this when you click on its objectId:

"title": "Sign up for Marathon",
"priority": 2,
"ACL": {
"*": {
"read": true,
"write": true
"objectId": "582570f38ac247004f39c24b",
"createdAt": "2017-11-11T07:19:15.549Z",
"updatedAt": "2017-11-11T07:19:15.549Z"

You don't have to create or set up a new class called Todo in

> Data before running the code above. If the class doesn't exist, it will be automatically created.

Several built-in fields are provided by default which you don't need to specify in your code:

Built-in FieldTypeDescription
objectIdStringA unique identifier for each saved object.
ACLLCACLAccess Control List, a special object defining the read and write permissions of other people.
createdAtDateThe time the object was created.
updatedAtDateThe time the object was last modified.

Each of these fields is filled in by the cloud automatically and doesn't exist on the local leancloud.Object until a save operation has been completed.

Field names, or keys, can only contain letters, numbers, and underscores. A custom key can neither start with double underscores __, nor be identical to any system reserved words or built-in field names (ACL, className, createdAt, objectId, and updatedAt) regardless of letter cases.

Values can be strings, numbers, booleans, or even arrays and dictionaries — anything that can be JSON-encoded. See Data Types for more information.

We recommend that you adopt CamelCase naming convention to NameYourClassesLikeThis and nameYourKeysLikeThis, which keeps your code more readable.

Retrieving Objects

If an leancloud.Object is already in the cloud, you can retrieve it using its objectId with the following code:

Todo = leancloud.Object.extend('Todo')
query = Todo.query
todo = query.get('582570f38ac247004f39c24b')

# todo is the instance of the Todo object with objectId 582570f38ac247004f39c24b
title = todo.get('title')
priority = todo.get('priority')

# Acquire special properties.
object_id =
update_at = todo.updated_at
created_at = todo.created_at

After retrieving an object, you can use the get method to acquire the data stored in its fields. Be aware that objectId, updatedAt, and createdAt are 3 special properties that cannot be retrieved using the get method or modified with the set method. Each of these fields is filled in by the cloud only, so they don't exist on leancloud.Object until a save operation has been completed.

If you try to access a field or property that doesn't exist, the SDK will not raise an error. Instead, it will return None.

Refreshing Objects

If you need to refresh a local object with the latest version of it in the cloud, call the fetch method on it:

Todo = leancloud.Object.extend('Todo')
todo = Todo.create_without_data('582570f38ac247004f39c24b')

Updating Objects

To update an existing object, assign the new data to each field and call the save method. For example:

Todo = leancloud.Object.extend('Todo')
todo = Todo.create_without_data('582570f38ac247004f39c24b')
todo.set('content', 'Weekly meeting has been rescheduled to Wed 3pm for this week.')

The cloud automatically figures out which data has changed and only the fields with changes will be sent to the cloud. The fields you didn't update will remain intact.

Updating Data Conditionally

By passing in a query option when saving, you can specify conditions on the save operation so that the object can be updated atomically only when those conditions are met. If no object matches the conditions, the cloud will return error 305 to indicate that there was no update taking place.

For example, in the class Account there is a field called balance, and there are multiple incoming requests that want to modify this field. Since an account cannot have a negative balance, we can only allow a request to update the balance when the amount requested is lower than or equal to the balance:

Account = leancloud.Object.extend('Account')
account = Account.create_without_data('5745557f71cfe40068c6abe0')
# Atomically decrease balance by 100
amount = -100
account.increment('balance', amount)
# Add the condition
where = Account.query.greater_than_or_equal_to('balance', -amount)
# Return the latest data in the cloud upon completion.
# All the fields will be returned if the object is new,
# otherwise only fields with changes will be returned.
account.fetch_when_save = True
print('Balance: ', account.get('balance'))
except leancloud.LeanCloudError as e:
if e.code == 305:
print('Insufficient balance. Operation failed!')

The query option only works for existing objects. In other words, it does not affect objects that haven't been saved to the cloud yet.

The benefit of using the query option instead of combining AV.Query and leancloud.Object shows up when you have multiple clients trying to update the same field at the same time. The latter way is more cumbersome and may lead to potential inconsistencies.

Updating Counters

Take Twitter as an example, we need to keep track of how many Likes and Retweets a tweet has gained so far. Since a Like or Retweet action can be triggered simultaneously by multiple clients, saving objects with updated values directly can lead to inaccurate results. To make sure that the total number is stored correctly, you can atomically increase (or decrease) the value of a number field:

post.increment("likes", 1);

You can specify the amount of increment (or decrement) by providing an additional argument. If the argument is not provided, 1 is used by default.

Updating Arrays

There are several operations that can be used to atomically update an array associated with a given key:

  • add() appends the given object to the end of an array.
  • add_unique() adds the given object into an array only if it is not in it. The object will be inserted at a random position.
  • remove() removes all instances of the given object from an array.

For example, Todo has a field named alarms for keeping track of the times at which a user wants to be alerted. The following code adds the times to the alarms field:

from datetime import datetime

alarm1 = datetime(2018, 4, 30, 7, 10, 00)
alarm2 = datetime(2018, 4, 30, 7, 20, 00)
alarm3 = datetime(2018, 4, 30, 7, 30, 00)

Todo = leancloud.Object.extend('Todo')
todo = Todo()
todo.add_unique('alarms', alarm1)
todo.add_unique('alarms', alarm2)
todo.add_unique('alarms', alarm3)

Deleting Objects

The following code deletes a Todo object from the cloud:

Todo = leancloud.Object.extend('Todo')
todo = Todo.create_without_data('582570f38ac247004f39c24b')

Removing data from the cloud should always be dealt with great caution as it may lead to non-recoverable data loss. We strongly advise that you read ACL Guide to understand the risks thoroughly. You should also consider implementing class-level, object-level, and field-level permissions for your classes in the cloud to guard against unauthorized data operations.

Batch Processing

# Batch create and update

# Batch delete

The following code sets isComplete of all Todo objects to be true:

Todo = leancloud.Object.extend('Todo')
# Get a collection of todos to work on
todo1 = Todo()
todo2 = Todo()
todo3 = Todo()
# Update value
todo1.set('isComplete', True)
todo2.set('isComplete', True)
todo3.set('isComplete', True)
# Save all at once
Todo.save_all([todo1, todo2, todo3])

Although each function call sends multiple operations in one single network request, saving operations and fetching operations are billed as separate API calls for each object in the collection, while deleting operations are billed as a single API call.

Data Models

Objects may have relationships with other objects. For example, in a blogging application, a Post object may have relationships with many Comment objects. The Data Storage service supports three kinds of relationships, including one-to-one, one-to-many, and many-to-many.

One-to-One and One-to-Many Relationships

One-to-one and one-to-many relationships are modeled by saving leancloud.Object as a value in the other object. For example, each Comment in a blogging app might correspond to one Post.

The following code creates a new Post with a single Comment:

# Create a post
Post = leancloud.Object.extend('Post')
post = Post()
post.set('title', 'I am starving!')
post.set('content', 'Hmmm, where should I go for lunch?')

# Create a comment
Comment = leancloud.Object.extend('Comment')
comment = Comment()
comment.set('content', 'KFC is the best!')

# Add the post as a property of the comment
comment.set('parent', post)

# This will save both post and comment

Internally, the backend will store the referred-to object with the Pointer type in just one place in order to maintain consistency. You can also link objects using their objectIds like this:

Post = leancloud.Object.extend('Post')
post = Post.create_without_data('57328ca079bc44005c2472d0')
comment.set('post', post)

See Relational Queries for instructions on how to query relational data.

Many-to-Many Relationships

The easiest way to model many-to-many relationships is to use arrays. In most cases, using arrays helps you reduce the number of queries you need to make and leads to better performance. However, if additional properties need to be attached to the relationships between two classes, using join tables would be a better choice. Keep in mind that the additional properties are used to describe the relationships between classes rather than any single class.

We recommend you to use join tables if the total number of objects of any class exceeds 100.


We've already seen how you can retrieve a single object from the cloud with leancloud.Object, but it doesn't seem to be powerful enough when you need to retrieve multiple objects that match certain conditions at once. In such a situation, AV.Query would be a more efficient tool you can use.

Basic Queries

The general steps of performing a basic query include:

  1. Creating AV.Query.
  2. Putting conditions on it.
  3. Retrieving an array of objects matching the conditions.

The code below retrieves all Student objects whose lastName is Smith:

query = leancloud.Query('Student')
query.equal_to('lastName', 'Smith')
student_list = query.find()

Query Constraints

There are several ways to put constraints on the objects found by leancloud.Object.

The code below filters out objects with Jack as firstName:

query.not_equal_to("firstName", 'Jack')

For sortable types like numbers and strings, you can use comparisons in queries:

# Restricts to age < 18
query.less_than('age', 18)

# Restricts to age <= 18
query.less_than_or_equal_to('age', 18)

# Restricts to age > 18
query.greater_than('age', 18)

# Restricts to age >= 18
query.greater_than_or_equal_to('age', 18)

You can apply multiple constraints to a single query, and objects will only be in the results if they match all of the constraints. In other words, it's like concatenating constraints with AND:

query.equal_to("firstName", 'Jack')
query.greater_than('age', 18)

You can limit the number of results by setting limit (defaults to 100):

# Limit to at most 10 results

For performance reasons, the maximum value allowed for limit is 1000, meaning that the cloud would only return 1,000 results even if it is set to be greater than 1000.

If you need exactly one result, you may use first for convenience:

Todo = leancloud.Object.extend('Todo')
query = Todo.query
query.equal_to('priority', 2)
todo = query.first()

You can skip a certain number of results by setting skip:

// Skip the first 20 results

You can implement pagination in your app by using skip together with limit:

Todo = leancloud.Object.extend('Todo')
query = Todo.query
query.equal_to('priority', 2)

Keep in mind that the higher the skip goes, the slower the query will run. You may consider using createdAt or updatedAt (which are indexed) to set range boundaries for large datasets to make queries more efficient. You may also use the last value returned from an auto-increment field along with limit for pagination.

For sortable types, you can control the order in which results are returned:

# Sorts the results in ascending order by the createdAt property

# Sorts the results in descending order by the createdAt property

You can even attach multiple sorting rules to a single query:


To retrieve objects that have or do not have particular fields:

# Finds objects that have the 'images' field

# Finds objects that don't have the 'images' field

You can restrict the fields returned by providing a list of keys with selectKeys. The code below retrieves todos with only the title and content fields (and also special built-in fields including objectId, createdAt, and updatedAt):

// Finds objects that have the 'images' field

// Finds objects that don't have the 'images' field

You can restrict the fields returned by providing a list of keys with select. The code below retrieves todos with only the title and content fields (and also special built-in fields such as objectId, createdAt, and updatedAt):

Todo = leancloud.Object.extend('Todo')
query = Todo.query'title', 'content')
todo = query.first()

title = todo.get('title') # √
content = todo.get('content') # √
notes = todo.get('notes') # None

You can add a minus prefix to the attribute name for inverted selection. For example, if you do not care about the post author, use -author. The inverted selection also applies to preserved attributes and can be used with dot notations, e.g., -pubUser.createdAt.

The unselected fields can be fetched later with fetchInBackground. See Refreshing Objects.

Queries on String Values

Use startsWith to restrict to string values that start with a particular string. Similar to a LIKE operator in SQL, it is indexed so it is efficient for large datasets:

Todo = leancloud.Object.extend('Todo')
query = Todo.query
# SQL equivalent: title LIKE 'lunch%'
query.startswith("title", "lunch")

Use contains to restrict to string values that contain a particular string:

Todo = leancloud.Object.extend('Todo')
query = Todo.query
# SQL equivalent: title LIKE '%lunch%'
query.contains("title", "lunch")

Unlike startsWith, contains can't take advantage of indexes, so it is not encouraged to be used for large datasets.

Please note that both startsWith and contains perform case-sensitive matching, so the examples above will not look for string values containing Lunch, LUNCH, etc.

If you are looking for string values that do not contain a particular string, use whereMatches with regular expressions:

Todo = leancloud.Object.extend('Todo')
query = Todo.query
# 'title' without 'ticket' (case-insensitive)
query.matched('title', '^((?!ticket).)*$', ignore_case=True)

However, performing queries with regular expressions as constraints can be very expensive, especially for classes with over 100,000 records. The reason behind this is that queries like this can't take advantage of indexes and will lead to exhaustive scanning of the whole dataset to find the matching objects. We recommend that you take a look at our In-App Searching feature, a full-text search solution we provide to improve your app's searching ability and user experience.

If you are facing performance issues with queries, please refer to Optimizing Performance for possible workarounds and best practices.

Queries on Array Values

The code below looks for all the objects with work as an element of its array field tags:

query.equal_to('tags', 'work')

To look for objects whose array field tags contains three elements:

query.size_equal_to('tags', 3)

You can also look for objects whose array field tags contains work, sales, and appointment:

query.contains_all('tags', ['work', 'sales', 'appointment'])

To retrieve objects whose field matches any one of the values in a given list, you can use contained_in instead of performing multiple queries. The code below constructs a query that retrieves todo items with priority to be 1 or 2:

# Single query
Todo = leancloud.Object.extend('Todo')
priority_one_or_two = Todo.query
priority_one_or_two.contained_in('priority', [1, 2])
# Mission completed :)

# ---------------
# vs.
# ---------------

# Multiple queries
Todo = leancloud.Object.extend('Todo')

priority_one = Todo.query
priority_one.equal_to('priority', 1)

priority_two = Todo.query
priority_two.equal_to('priority', 2)

priority_one_or_two = leancloud.Query.or_(priority_one, priority_two)
# Kind of verbose :(

Conversely, you can use not_contained_in if you want to retrieve objects that do not match any of the values in a list.

Relational Queries

There are several ways to perform queries for relational data. To retrieve objects whose given field matches a particular leancloud.Object, you can use equal_to just like how you use it for other data types. For example, if each Comment has a Post object in its post field, you can fetch all the comments for a particular Post with the following code:

Post = leancloud.Object.extend('Post')
post = Post.create_without_data('57328ca079bc44005c2472d0')
query = leancloud.Query('Comment')
query.equal_to('post', post)
comment_list = query.find()

To retrieve objects whose given field contains an leancloud.Object that matches a different query, you can use matches_query. The code below constructs a query that looks for all the comments for posts with images:

inner_query = leancloud.Query('Post')

query = leancloud.Query('Comment')
query.matches_query('post', inner_query)

To retrieve objects whose given field does not contain an leancloud.Object that matches a different query, use does_not_match_query instead.

Sometimes you may need to look for related objects from different classes without extra queries. In such situations, you can use include on the same query. The following code retrieves the last 10 comments together with the posts related to them:

query = leancloud.Query('Comment')

# Retrieve the most recent ones

# Only retrieve the last 10

# Include the related post together with each comment

comment_list = query.find()
for comment in comment_list:
# This does not require a network access
post = comment.get('post')

Caveats about Inner Queries

The Data Storage service is not built on relational databases, which makes it impossible to join tables while querying. For the relational queries mentioned above, what we would do is to perform an inner query first (with 100 as the default limit and 1000 as the maximum) and then insert the result from this query into the outer query. If the number of records matching the inner query exceeds the limit and the outer query contains other constraints, the amount of the records returned in the end could be zero or less than your expectation since only the records within the limit would be inserted into the outer query.

The following actions can be taken to solve the problem:

  • Make sure the number of records in the result of the inner query is no more than 100. If it is between 100 and 1,000, set 1000 as the limit of the inner query.
  • Create redundancy for the fields being queried by the inner query on the table for the outer query.
  • Repeat the same query with different skip values until all the records are gone through (performance issue could occur if the value of skip gets too big).

Counting Objects

If you just need to count how many objects match a query but do not need to retrieve the actual objects, use count instead of find. For example, to count how many todos have been completed:

Todo = leancloud.Object.extend('Todo')
query = Todo.query
query.equal_to('isComplete', True)
count = query.count()

Compound Queries

Compound queries can be used if complex query conditions need to be specified. A compound query is a logical combination (OR or AND) of subqueries.

Note that we do not support GeoPoint or non-filtering constraints (e.g. near, withinGeoBox, limit, skip, ascending, descending, include) in the subqueries of a compound query.

OR-ed Query Constraints

An object will be returned as long as it fulfills any one of the subqueries. The code below constructs a query that looks for all the todos that either have priorities higher than or equal to 3, or are already completed:

Todo = leancloud.Object.extend('Todo')

priority_query = Todo.query
priority_query.greater_than_or_equal_to('priority', 3)

is_complete_query = Todo.query
is_complete_query.equal_to('isComplete', True)

query = leancloud.Query.or_(priority_query, is_complete_query)

Queries regarding GeoPoint cannot be present among OR-ed queries.

AND-ed Query Constraints

The effect of using AND-ed query is the same as adding constraints to AV.Query. The code below constructs a query that looks for all the todos that are created between 2016-11-13 and 2016-12-02:

from datetime import datetime

Todo = leancloud.Object.extend('Todo')

start_date_query = Todo.query
start_date_query.greater_than_or_equal_to('createdAt', datetime(2016, 11, 13))

end_date_query = Todo.query
end_date_query.less_than('createdAt', datetime(2016, 12, 3))

query = leancloud.Query.and_(start_date_query, end_date_query)

While using an AND-ed query by itself doesn't bring anything new compared to a basic query, to combine two or more OR-ed queries, you have to use AND-ed queries:

from datetime import datetime

Todo = leancloud.Object.extend('Todo')

created_at_query = Todo.query
created_at_query.greater_than_or_equal_to('createdAt', datetime(2018, 4, 30))
created_at_query.less_than('createdAt', datetime(2018, 5, 1))

location_query = Todo.query

priority_2_query = Todo.query
priority_2_query.equal_to('priority', 2)

priority_3_query = Todo.query
priority_3_query.equal_to('priority', 3)

priority_query = leancloud.Query.or_(priority_2_query, priority_3_query)
time_location_query = leancloud.Query.or_(created_at_query, location_query)
query = leancloud.Query.and_(priority_query, time_location_query)

Optimizing Performance

There are several factors that could lead to potential performance issues when you conduct a query, especially when more than 100,000 records are returned at a time. We are listing some common ones here so you can design your apps accordingly to avoid them:

  • Querying with "not equal to" or "not include" (index will not work)
  • Querying on strings with a wildcard at the beginning of the pattern (index will not work)
  • Using count with conditions (all the entries will be gone through)
  • Using skip for a large number of entries (all the entries that need to be skipped will be gone through)
  • Sorting without an index (querying and sorting cannot share a composite index unless the conditions used on them are both covered by the same one)
  • Querying without an index (the conditions used on the query cannot share a composite index unless all of them are covered by the same one; additional time will be consumed if excessive data falls under the uncovered conditions)


leancloud.File allows you to store application files in the cloud that would otherwise be too large or cumbersome to fit into a regular leancloud.Object. The most common use case is storing images, but you can also use it for documents, videos, music, and any other binary data.

Creating Files

You can create a file from a string:

from StringIO import StringIO

data = StringIO('LeanCloud')
# resume.txt is the file name
file = leancloud.File('resume.txt', data)

You can also create a file from byte values with buffer:

data = buffer('\x4c\x65\x61\x6e\x43\x6c\x6f\x75\x64')
file = leancloud.File('resume.txt', data)

You can also create a file from a URL:

file = leancloud.File.create_with_url('logo.png', '')

When creating files from URLs, the SDK will not upload the actual files into the cloud but will store the addresses of the files as strings. This will not lead to actual traffic for uploading files, as opposed to creating files in other ways by doing which the files will be actually stored into the cloud.

LeanCloud will auto-detect the type of the file you are uploading based on the file extension, but you can also specify the Content-Type (commonly referred to as MIME type):

file = leancloud.File('resume.txt', StringIO('{"company":"LeanCloud"}'), 'application/json')

But the most common method for creating files is to upload them from local paths:

with open('/tmp/avatar.jpg', 'rb') as f:
file = leancloud.File('avatar.jpg', f)

You can specify the path of the uploaded file via the key property. For example, use a robots.txt to restrict search engines from crawling uploaded files:

with open('/tmp/robots.txt', 'rb') as f:
file = leancloud.File('robots.txt', f)
file.key = 'robots.txt'

Saving Files

By saving a file, you store it into the cloud and get a permanent URL pointing to it:

A file successfully uploaded can be found in

> Files and cannot be modified later. If you need to change the file, you have to upload the modified file again and a new objectId and URL will be generated.

You can associate a file with leancloud.Object after it has been saved:

Todo = leancloud.Object.extend('Todo')
todo = Todo()
todo.set('title', 'Get Cakes')
# attachments is an Array field
todo.add('attachments', file)

You can also construct an AV.Query to query files:

query = leancloud.Query('_File')

Note that the url field of internal files (files uploaded to the file service) is dynamically generated by the cloud, which will switch custom domain names automatically. Therefore, querying files by the url field is only applicable to external files (files created by saving the external URL directly to the _File table). Query internal files by the key field (path in URL) instead.

File Metadata

When uploading a file, you can attach additional properties to it with metaData. A file's metaData cannot be updated once the file is stored to the cloud.

# Set metadata
file.metadata['author'] = 'LeanCloud'

# Get all metadata
metadata = file.metadata
# Get author
author = metadata['author']
# Get file name
file_name =
# Get size (not available for files created from base64-encoded strings or URLs)
size = file.size

Deleting Files

The code below deletes a file from the cloud:

file = leancloud.File.create_without_data('552e0a27e4b0643b709e891e')

By default, a file is not allowed to be deleted. We recommend you delete files by accessing our REST API with the Master Key. You can also allow certain users and roles to delete files by going to

> Files > Permission and select Others > Permission settings > delete..


You can associate real-world latitude and longitude coordinates with an object by adding an LCGeoPoint to the leancloud.Object. By doing so, queries on the proximity of an object to a given point can be performed, allowing you to implement functions like looking for users or places nearby easily.

To associate a point with an object, you need to create the point first. The code below creates an LCGeoPoint with 39.9 as latitude and 116.4 as longitude:

point = leancloud.GeoPoint(39.9, 116.4)

Now you can store the point into an object as a regular field:

todo.set('location', point);

Geo Queries

With a number of existing objects with spatial coordinates, you can find out which of them are closest to a given point, or are contained within a particular area. This can be done by adding another restriction to AV.Query using whereNear. The code below returns a list of Todo objects with location closest to a given point:

query = leancloud.Query('Todo')
point = leancloud.GeoPoint(39.9, 116.4)
query.near('location', point)

# Limit to 10 results
todo_list = query.find()

Additional sorting conditions like orderByAscending and orderByDescending will gain higher priorities than the default order by distance.

To have the results limited within a certain distance, check out within_kilometers, within_miles, and within_radians in our API docs.

You can also query for the set of objects that are contained within a rectangular bounding box with within_geo_box:


query = leancloud.Query('Todo')
southwest = leancloud.GeoPoint(30, 115)
northeast = leancloud.GeoPoint(40, 118)
query.within_geo_box('location', southwest, northeast)

Caveats about GeoPoints

Points should not exceed the extreme ends of the ranges. Latitude should be between -90.0 and 90.0. Longitude should be between -180.0 and 180.0. Attempting to set latitude or longitude out of bounds will cause an error. Also, each leancloud.Object can only have one field for LCGeoPoint.


See TDS Authentication Guide.


As your app grows in scope and user base, you may find yourself needing more coarse-grained control over access to pieces of your data than user-linked ACLs can provide. To address this requirement, we support a form of role-based access control. Check the detailed ACL Guide to learn how to set it up for your objects.

Full-Text Search offers a better way to search through the information contained within your app. It's built with search engine capabilities that you can easily tap into your app. Effective and useful searching functionality in your app is crucial for helping users find what they need. For more details, see Full-Text Search Guide.