A content provider manages access to a central repository of data. You implement a provider as one or more classes in an Android application, along with elements in the manifest file. One of your classes implements a subclass {@link android.content.ContentProvider}, which is the interface between your provider and other applications. Although content providers are meant to make data available to other applications, you may of course have activities in your application that allow the user to query and modify the data managed by your provider.
The rest of this topic is a basic list of steps for building a content provider and a list of APIs to use.
Before you start building a provider, do the following:
You don't need a provider to use an SQLite database if the use is entirely within your own application.
Next, follow these steps to build your provider:
A content provider is the interface to data saved in a structured format. Before you create the interface, you must decide how to store the data. You can store the data in any form you like, and then design the interface to read and write the data as necessary.
These are some of the data storage technologies that are available in Android:
Remember that you don't have to use a database to implement your repository. A provider appears externally as a set of tables, similar to a relational database, but this is not a requirement for the provider's internal implementation.
Here are some tips for designing your provider's data structure:
_ID.
You can also use a BLOB to implement a schema-independent table. In this type of table, you define a primary key column, a MIME type column, and one or more generic columns as BLOB. The meaning of the data in the BLOB columns is indicated by the value in the MIME type column. This allows you to store different row types in the same table. The Contacts Provider's "data" table {@link android.provider.ContactsContract.Data} is an example of a schema-independent table.
A content URI is a URI that identifies data in a provider. Content URIs include the symbolic name of the entire provider (its authority) and a name that points to a table or file (a path). The optional id part points to an individual row in a table. Every data access method of {@link android.content.ContentProvider} has a content URI as an argument; this allows you to determine the table, row, or file to access.
The basics of content URIs are described in the topic Content Provider Basics.
A provider usually has a single authority, which serves as its Android-internal name. To
avoid conflicts with other providers, you should use Internet domain ownership (in reverse)
as the basis of your provider authority. Because this recommendation is also true for Android
package names, you can define your provider authority as an extension of the name
of the package containing the provider. For example, if your Android package name is
com.example.<appname>, you should give your provider the
authority com.example.<appname>.provider.
Developers usually create content URIs from the authority by appending paths that point to
individual tables. For example, if you have two tables table1 and
table2, you combine the authority from the previous example to yield the
content URIs
com.example.<appname>.provider/table1 and
com.example.<appname>.provider/table2. Paths aren't
limited to a single segment, and there doesn't have to be a table for each level of the path.
By convention, providers offer access to a single row in a table by accepting a content URI
with an ID value for the row at the end of the URI. Also by convention, providers match the
ID value to the table's _ID column, and perform the requested access against the
row that matches.
This convention facilitates a common design pattern for apps accessing a provider. The app
does a query against the provider and displays the resulting {@link android.database.Cursor}
in a {@link android.widget.ListView} using a {@link android.widget.CursorAdapter}.
The definition of {@link android.widget.CursorAdapter} requires one of the columns in the
{@link android.database.Cursor} to be _ID
The user then picks one of the displayed rows from the UI in order to look at or modify the
data. The app gets the corresponding row from the {@link android.database.Cursor} backing the
{@link android.widget.ListView}, gets the _ID value for this row, appends it to
the content URI, and sends the access request to the provider. The provider can then do the
query or modification against the exact row the user picked.
To help you choose which action to take for an incoming content URI, the provider API includes
the convenience class {@link android.content.UriMatcher}, which maps content URI "patterns" to
integer values. You can use the integer values in a switch statement that
chooses the desired action for the content URI or URIs that match a particular pattern.
A content URI pattern matches content URIs using wildcard characters:
*: Matches a string of any valid characters of any length.
#: Matches a string of numeric characters of any length.
As an example of designing and coding content URI handling, consider a provider with the
authority com.example.app.provider that recognizes the following content URIs
pointing to tables:
content://com.example.app.provider/table1: A table called table1.
content://com.example.app.provider/table2/dataset1: A table called
dataset1.
content://com.example.app.provider/table2/dataset2: A table called
dataset2.
content://com.example.app.provider/table3: A table called table3.
The provider also recognizes these content URIs if they have a row ID appended to them, as
for example content://com.example.app.provider/table3/1 for the row identified by
1 in table3.
The following content URI patterns would be possible:
content://com.example.app.provider/*
content://com.example.app.provider/table2/*:
dataset1
and dataset2, but doesn't match content URIs for table1 or
table3.
content://com.example.app.provider/table3/#: Matches a content URI
for single rows in table3, such as
content://com.example.app.provider/table3/6 for the row identified by
6.
The following code snippet shows how the methods in {@link android.content.UriMatcher} work.
This code handles URIs for an entire table differently from URIs for a
single row, by using the content URI pattern
content://<authority>/<path> for tables, and
content://<authority>/<path>/<id> for single rows.
The method {@link android.content.UriMatcher#addURI(String, String, int) addURI()} maps an
authority and path to an integer value. The method {@link android.content.UriMatcher#match(Uri)
match()} returns the integer value for a URI. A switch statement
chooses between querying the entire table, and querying for a single record:
public class ExampleProvider extends ContentProvider {
...
// Creates a UriMatcher object.
private static final UriMatcher sUriMatcher;
...
/*
* The calls to addURI() go here, for all of the content URI patterns that the provider
* should recognize. For this snippet, only the calls for table 3 are shown.
*/
...
/*
* Sets the integer value for multiple rows in table 3 to 1. Notice that no wildcard is used
* in the path
*/
sUriMatcher.addURI("com.example.app.provider", "table3", 1);
/*
* Sets the code for a single row to 2. In this case, the "#" wildcard is
* used. "content://com.example.app.provider/table3/3" matches, but
* "content://com.example.app.provider/table3 doesn't.
*/
sUriMatcher.addURI("com.example.app.provider", "table3/#", 2);
...
// Implements ContentProvider.query()
public Cursor query(
Uri uri,
String[] projection,
String selection,
String[] selectionArgs,
String sortOrder) {
...
/*
* Choose the table to query and a sort order based on the code returned for the incoming
* URI. Here, too, only the statements for table 3 are shown.
*/
switch (sUriMatcher.match(uri)) {
// If the incoming URI was for all of table3
case 1:
if (TextUtils.isEmpty(sortOrder)) sortOrder = "_ID ASC";
break;
// If the incoming URI was for a single row
case 2:
/*
* Because this URI was for a single row, the _ID value part is
* present. Get the last path segment from the URI; this is the _ID value.
* Then, append the value to the WHERE clause for the query
*/
selection = selection + "_ID = " uri.getLastPathSegment();
break;
default:
...
// If the URI is not recognized, you should do some error handling here.
}
// call the code to actually do the query
}
Another class, {@link android.content.ContentUris}, provides convenience methods for working
with the id part of content URIs. The classes {@link android.net.Uri} and
{@link android.net.Uri.Builder} include convenience methods for parsing existing
{@link android.net.Uri} objects and building new ones.
The {@link android.content.ContentProvider} instance manages access to a structured set of data by handling requests from other applications. All forms of access eventually call {@link android.content.ContentResolver}, which then calls a concrete method of {@link android.content.ContentProvider} to get access.
The abstract class {@link android.content.ContentProvider} defines six abstract methods that you must implement as part of your own concrete subclass. All of these methods except {@link android.content.ContentProvider#onCreate() onCreate()} are called by a client application that is attempting to access your content provider:
Notice that these methods have the same signature as the identically-named {@link android.content.ContentResolver} methods.
Your implementation of these methods should account for the following:
The
{@link android.content.ContentProvider#query(Uri, String[], String, String[], String)
ContentProvider.query()} method must return a {@link android.database.Cursor} object, or if it
fails, throw an {@link java.lang.Exception}. If you are using an SQLite database as your data
storage, you can simply return the {@link android.database.Cursor} returned by one of the
query() methods of the {@link android.database.sqlite.SQLiteDatabase} class.
If the query does not match any rows, you should return a {@link android.database.Cursor}
instance whose {@link android.database.Cursor#getCount()} method returns 0.
You should return null only if an internal error occurred during the query process.
If you aren't using an SQLite database as your data storage, use one of the concrete subclasses of {@link android.database.Cursor}. For example, the {@link android.database.MatrixCursor} class implements a cursor in which each row is an array of {@link java.lang.Object}. With this class, use {@link android.database.MatrixCursor#addRow(Object[]) addRow()} to add a new row.
Remember that the Android system must be able to communicate the {@link java.lang.Exception} across process boundaries. Android can do this for the following exceptions that may be useful in handling query errors:
The {@link android.content.ContentProvider#insert(Uri, ContentValues) insert()} method adds a new row to the appropriate table, using the values in the {@link android.content.ContentValues} argument. If a column name is not in the {@link android.content.ContentValues} argument, you may want to provide a default value for it either in your provider code or in your database schema.
This method should return the content URI for the new row. To construct this, append the new
row's _ID (or other primary key) value to the table's content URI, using
{@link android.content.ContentUris#withAppendedId(Uri, long) withAppendedId()}.
The {@link android.content.ContentProvider#delete(Uri, String, String[]) delete()} method does not have to physically delete rows from your data storage. If you are using a sync adapter with your provider, you should consider marking a deleted row with a "delete" flag rather than removing the row entirely. The sync adapter can check for deleted rows and remove them from the server before deleting them from the provider.
The {@link android.content.ContentProvider#update(Uri, ContentValues, String, String[])
update()} method takes the same {@link android.content.ContentValues} argument used by
{@link android.content.ContentProvider#insert(Uri, ContentValues) insert()}, and the
same selection and selectionArgs arguments used by
{@link android.content.ContentProvider#delete(Uri, String, String[]) delete()} and
{@link android.content.ContentProvider#query(Uri, String[], String, String[], String)
ContentProvider.query()}. This may allow you to re-use code between these methods.
The Android system calls {@link android.content.ContentProvider#onCreate() onCreate()} when it starts up the provider. You should perform only fast-running initialization tasks in this method, and defer database creation and data loading until the provider actually receives a request for the data. If you do lengthy tasks in {@link android.content.ContentProvider#onCreate() onCreate()}, you will slow down your provider's startup. In turn, this will slow down the response from the provider to other applications.
For example, if you are using an SQLite database you can create a new {@link android.database.sqlite.SQLiteOpenHelper} object in {@link android.content.ContentProvider#onCreate() ContentProvider.onCreate()}, and then create the SQL tables the first time you open the database. To facilitate this, the first time you call {@link android.database.sqlite.SQLiteOpenHelper#getWritableDatabase getWritableDatabase()}, it automatically calls the {@link android.database.sqlite.SQLiteOpenHelper#onCreate(SQLiteDatabase) SQLiteOpenHelper.onCreate()} method.
The following two snippets demonstrate the interaction between {@link android.content.ContentProvider#onCreate() ContentProvider.onCreate()} and {@link android.database.sqlite.SQLiteOpenHelper#onCreate(SQLiteDatabase) SQLiteOpenHelper.onCreate()}. The first snippet is the implementation of {@link android.content.ContentProvider#onCreate() ContentProvider.onCreate()}:
public class ExampleProvider extends ContentProvider
/*
* Defines a handle to the database helper object. The MainDatabaseHelper class is defined
* in a following snippet.
*/
private MainDatabaseHelper mOpenHelper;
// Defines the database name
private static final String DBNAME = "mydb";
// Holds the database object
private SQLiteDatabase db;
public boolean onCreate() {
/*
* Creates a new helper object. This method always returns quickly.
* Notice that the database itself isn't created or opened
* until SQLiteOpenHelper.getWritableDatabase is called
*/
mOpenHelper = new MainDatabaseHelper(
getContext(), // the application context
DBNAME, // the name of the database)
null, // uses the default SQLite cursor
1 // the version number
);
return true;
}
...
// Implements the provider's insert method
public Cursor insert(Uri uri, ContentValues values) {
// Insert code here to determine which table to open, handle error-checking, and so forth
...
/*
* Gets a writeable database. This will trigger its creation if it doesn't already exist.
*
*/
db = mOpenHelper.getWritableDatabase();
}
}
The next snippet is the implementation of {@link android.database.sqlite.SQLiteOpenHelper#onCreate(SQLiteDatabase) SQLiteOpenHelper.onCreate()}, including a helper class:
...
// A string that defines the SQL statement for creating a table
private static final String SQL_CREATE_MAIN = "CREATE TABLE " +
"main " + // Table's name
"(" + // The columns in the table
" _ID INTEGER PRIMARY KEY, " +
" WORD TEXT"
" FREQUENCY INTEGER " +
" LOCALE TEXT )";
...
/**
* Helper class that actually creates and manages the provider's underlying data repository.
*/
protected static final class MainDatabaseHelper extends SQLiteOpenHelper {
/*
* Instantiates an open helper for the provider's SQLite data repository
* Do not do database creation and upgrade here.
*/
MainDatabaseHelper(Context context) {
super(context, DBNAME, null, 1);
}
/*
* Creates the data repository. This is called when the provider attempts to open the
* repository and SQLite reports that it doesn't exist.
*/
public void onCreate(SQLiteDatabase db) {
// Creates the main table
db.execSQL(SQL_CREATE_MAIN);
}
}
The {@link android.content.ContentProvider} class has two methods for returning MIME types:
The {@link android.content.ContentProvider#getType(Uri) getType()} method returns a {@link java.lang.String} in MIME format that describes the type of data returned by the content URI argument. The {@link android.net.Uri} argument can be a pattern rather than a specific URI; in this case, you should return the type of data associated with content URIs that match the pattern.
For common types of data such as as text, HTML, or JPEG, {@link android.content.ContentProvider#getType(Uri) getType()} should return the standard MIME type for that data. A full list of these standard types is available on the IANA MIME Media Types website.
For content URIs that point to a row or rows of table data, {@link android.content.ContentProvider#getType(Uri) getType()} should return a MIME type in Android's vendor-specific MIME format:
vnd
android.cursor.item/
android.cursor.dir/
vnd.<name>.<type>
You supply the <name> and <type>.
The <name> value should be globally unique,
and the <type> value should be unique to the corresponding URI
pattern. A good choice for <name> is your company's name or
some part of your application's Android package name. A good choice for the
<type> is a string that identifies the table associated with the
URI.
For example, if a provider's authority is
com.example.app.provider, and it exposes a table named
table1, the MIME type for multiple rows in table1 is:
vnd.android.cursor.dir/vnd.com.example.provider.table1
For a single row of table1, the MIME type is:
vnd.android.cursor.item/vnd.com.example.provider.table1
If your provider offers files, implement {@link android.content.ContentProvider#getStreamTypes(Uri, String) getStreamTypes()}. The method returns a {@link java.lang.String} array of MIME types for the files your provider can return for a given content URI. You should filter the MIME types you offer by the MIME type filter argument, so that you return only those MIME types that the client wants to handle.
For example, consider a provider that offers photo images as files in .jpg,
.png, and .gif format.
If an application calls {@link android.content.ContentResolver#getStreamTypes(Uri, String)
ContentResolver.getStreamTypes()} with the filter string image/* (something that
is an "image"),
then the {@link android.content.ContentProvider#getStreamTypes(Uri, String)
ContentProvider.getStreamTypes()} method should return the array:
{ "image/jpeg", "image/png", "image/gif"}
If the app is only interested in .jpg files, then it can call
{@link android.content.ContentResolver#getStreamTypes(Uri, String)
ContentResolver.getStreamTypes()} with the filter string *\/jpeg, and
{@link android.content.ContentProvider#getStreamTypes(Uri, String)
ContentProvider.getStreamTypes()} should return:
{"image/jpeg"}
If your provider doesn't offer any of the MIME types requested in the filter string,
{@link android.content.ContentProvider#getStreamTypes(Uri, String) getStreamTypes()}
should return null.
A contract class is a public final class that contains constant definitions for the
URIs, column names, MIME types, and other meta-data that pertain to the provider. The class
establishes a contract between the provider and other applications by ensuring that the provider
can be correctly accessed even if there are changes to the actual values of URIs, column names,
and so forth.
A contract class also helps developers because it usually has mnemonic names for its constants, so developers are less likely to use incorrect values for column names or URIs. Since it's a class, it can contain Javadoc documentation. Integrated development environments such as Eclipse can auto-complete constant names from the contract class and display Javadoc for the constants.
Developers can't access the contract class's class file from your application, but they can
statically compile it into their application from a .jar file you provide.
The {@link android.provider.ContactsContract} class and its nested classes are examples of contract classes.
Permissions and access for all aspects of the Android system are described in detail in the topic Security and Permissions. The topic Data Storage also described the security and permissions in effect for various types of storage. In brief, the important points are:
If you want to use content provider permissions to control access to your data, then you should store your data in internal files, SQLite databases, or the "cloud" (for example, on a remote server), and you should keep files and databases private to your application.
All applications can read from or write to your provider, even if the underlying data is
private, because by default your provider does not have permissions set. To change this,
set permissions for your provider in your manifest file, using attributes or child
elements of the
<provider> element. You can set permissions that apply to the entire provider,
or to certain tables, or even to certain records, or all three.
You define permissions for your provider with one or more
<permission> elements in your manifest file. To make the
permission unique to your provider, use Java-style scoping for the
android:name attribute. For example, name the read permission
com.example.app.provider.permission.READ_PROVIDER.
The following list describes the scope of provider permissions, starting with the permissions that apply to the entire provider and then becoming more fine-grained. More fine-grained permissions take precedence over ones with larger scope:
android:permission attribute of the
<provider> element.
android:readPermission and
android:writePermission attributes of the
<provider> element. They take precedence over the permission required by
android:permission.
<path-permission> child element of the
<provider> element. For each content URI you specify, you can specify a
read/write permission, a read permission, or a write permission, or all three. The read and
write permissions take precedence over the read/write permission. Also, path-level
permission takes precedence over provider-level permissions.
Consider the permissions you need to implement an email provider and app, when you want to allow an outside image viewer application to display photo attachments from your provider. To give the image viewer the necessary access without requiring permissions, set up temporary permissions for content URIs for photos. Design your email app so that when the user wants to display a photo, the app sends an intent containing the photo's content URI and permission flags to the image viewer. The image viewer can then query your email provider to retrieve the photo, even though the viewer doesn't have the normal read permission for your provider.
To turn on temporary permissions, either set the
android:grantUriPermissions attribute of the
<provider> element, or add one or more
<grant-uri-permission> child elements to your
<provider> element. If you use temporary permissions, you have to call
{@link android.content.Context#revokeUriPermission(Uri, int)
Context.revokeUriPermission()} whenever you remove support for a content URI from your
provider, and the content URI is associated with a temporary permission.
The attribute's value determines how much of your provider is made accessible.
If the attribute is set to true, then the system will grant temporary
permission to your entire provider, overriding any other permissions that are required
by your provider-level or path-level permissions.
If this flag is set to false, then you must add
<grant-uri-permission> child elements to your
<provider> element. Each child element specifies the content URI or
URIs for which temporary access is granted.
To delegate temporary access to an application, an intent must contain the {@link android.content.Intent#FLAG_GRANT_READ_URI_PERMISSION} or the {@link android.content.Intent#FLAG_GRANT_WRITE_URI_PERMISSION} flags, or both. These are set with the {@link android.content.Intent#setFlags(int) setFlags()} method.
If the
android:grantUriPermissions attribute is not present, it's assumed to be
false.
Like {@link android.app.Activity} and {@link android.app.Service} components,
a subclass of {@link android.content.ContentProvider}
must be defined in the manifest file for its application, using the
<provider> element. The Android system gets the following information from
the element:
android:name
)
android:grantUriPermssions: Temporary permission flag.
android:permission: Single provider-wide read/write permission.
android:readPermission: Provider-wide read permission.
android:writePermission: Provider-wide write permission.
Permissions and their corresponding attributes are described in more detail in the section Implementing Content Provider Permissions.
android:enabled: Flag allowing the system to start the provider.
android:exported: Flag allowing other applications to use this provider.
android:initOrder: The order in which this provider should be started,
relative to other providers in the same process.
android:multiProcess: Flag allowing the system to start the provider
in the same process as the calling client.
android:process: The name of the process in which the provider should
run.
android:syncable: Flag indicating that the provider's data is to be
sync'ed with data on a server.
The attributes are fully documented in the dev guide topic for the
<provider>
element.
android:icon: A drawable resource containing an icon for the provider.
The icon appears next to the provider's label in the list of apps in
Settings > Apps > All.
android:label: An informational label describing the provider or its
data, or both. The label appears in the list of apps in
Settings > Apps > All.
The attributes are fully documented in the dev guide topic for the
<provider> element.
Applications can access a content provider indirectly with an {@link android.content.Intent}. The application does not call any of the methods of {@link android.content.ContentResolver} or {@link android.content.ContentProvider}. Instead, it sends an intent that starts an activity, which is often part of the provider's own application. The destination activity is in charge of retrieving and displaying the data in its UI. Depending on the action in the intent, the destination activity may also prompt the user to make modifications to the provider's data. An intent may also contain "extras" data that the destination activity displays in the UI; the user then has the option of changing this data before using it to modify the data in the provider.
You may want to use intent access to help ensure data integrity. Your provider may depend on having data inserted, updated, and deleted according to strictly defined business logic. If this is the case, allowing other applications to directly modify your data may lead to invalid data. If you want developers to use intent access, be sure to document it thoroughly. Explain to them why intent access using your own application's UI is better than trying to modify the data with their code.
Handling an incoming intent that wishes to modify your provider's data is no different from handling other intents. You can learn more about using intents by reading the topic Intents and Intent Filters.