Content providers store and retrieve data and make it accessible to all applications. They're the only way to share data across applications; there's no common storage area that all Android packages can access.
Android ships with a number of content providers for common data types (audio, video, images, personal contact information, and so on). You can see some of them listed in the android.provider package. You can query these providers for the data they contain (although, for some, you must acquire the proper permission to read the data).
If you want to make your own data public, you have two options: You can create your own content provider (a ContentProvider subclass) or you can add the data to an existing provider — if there's one that controls the same type of data and you have permission to write to it.
This document is an introduction to using content providers. After a brief discussion of the fundamentals, it explores how to query a content provider, how to modify data controlled by a provider, and how to create a content provider of your own.
How a content provider actually stores its data under the covers is up to its designer. But all content providers implement a common interface for querying the provider and returning results — as well as for adding, altering, and deleting data.
It's an interface that clients use indirectly, most generally through
ContentResolver objects. You get a ContentResolver
by calling getContentResolver()
from within the implementation of an Activity
or other application component:
ContentResolver cr = getContentResolver();
You can then use the ContentResolver's methods to interact with whatever content providers you're interested in.
When a query is initiated, the Android system identifies the content provider that's the target of the query and makes sure that it is up and running. The system instantiates all ContentProvider objects; you never need to do it on your own. In fact, you never deal directly with ContentProvider objects at all. Typically, there's just a single instance of each type of ContentProvider. But it can communicate with multiple ContentResolver objects in different applications and processes. The interaction between processes is handled by the ContentResolver and ContentProvider classes.
Content providers expose their data as a simple table on a database model, where each row is a record and each column is data of a particular type and meaning. For example, information about people and their phone numbers might be exposed as follows:
_ID | NUMBER | NUMBER_KEY | LABEL | NAME | TYPE |
---|---|---|---|---|---|
13 | (425) 555 6677 | 425 555 6677 | Kirkland office | Bully Pulpit | TYPE_WORK |
44 | (212) 555-1234 | 212 555 1234 | NY apartment | Alan Vain | TYPE_HOME |
45 | (212) 555-6657 | 212 555 6657 | Downtown office | Alan Vain | TYPE_MOBILE |
53 | 201.555.4433 | 201 555 4433 | Love Nest | Rex Cars | TYPE_HOME |
Every record includes a numeric _ID
field that uniquely identifies
the record within the table. IDs can be used to match records in related
tables — for example, to find a person's phone number in one table
and pictures of that person in another.
A query returns a Cursor object that can move from record to record and column to column to read the contents of each field. It has specialized methods for reading each type of data. So, to read a field, you must know what type of data the field contains. (There's more on query results and Cursor objects later.)
Each content provider exposes a public URI (wrapped as a Uri
object) that uniquely identifies its data set. A content provider that controls
multiple data sets (multiple tables) exposes a separate URI for each one. All
URIs for providers begin with the string "content://
". The content:
scheme identifies the data as being controlled by a content provider.
If you're defining a content provider, it's a good idea to also define a
constant for its URI, to simplify client code and make future updates cleaner.
Android defines CONTENT_URI
constants for all the providers that come
with the platform. For example, the URI for the table that matches
phone numbers to people and the URI for the table that holds pictures of
people (both controlled by the Contacts content provider) are:
android.provider.Contacts.Phones.CONTENT_URI
android.provider.Contacts.Photos.CONTENT_URI
Similarly, the URIs for the table of recent phone calls and the table of calendar entries are:
android.provider.CallLog.Calls.CONTENT_URI
android.provider.Calendar.CONTENT_URI
The URI constant is used in all interactions with the content provider. Every ContentResolver method takes the URI as its first argument. It's what identifies which provider the ContentResolver should talk to and which table of the provider is being targeted.
You need three pieces of information to query a content provider:
If you're querying a particular record, you also need the ID for that record.
To query a content provider, you can use either the
ContentResolver.query()
method or the Activity.managedQuery()
method.
Both methods take the same set of arguments, and both return a
Cursor object. However, managedQuery()
causes the activity to manage the life cycle of the Cursor. A managed Cursor
handles all of the niceties, such as unloading itself when the activity pauses,
and requerying itself when the activity restarts. You can ask an Activity to
begin managing an unmanaged Cursor object for you by calling
Activity.startManagingCursor()
.
The first argument to either query()
or managedQuery()
is the provider URI
— the CONTENT_URI
constant that identifies a particular
ContentProvider and data set (see URIs earlier).
To restrict a query to just one record, you can append the _ID
value for
that record to the URI — that is, place a string matching the ID as the
last segment of the path part of the URI. For example, if the ID is 23,
the URI would be:
content://. . . ./23
There are some helper methods, particularly
ContentUris.withAppendedId()
and Uri.withAppendedPath()
,
that make it easy to append an ID to a URI. Both are static methods that return
a Uri object with the ID added. So, for example, if you were looking for record
23 in the database of people contacts, you might construct a query as follows:
import android.provider.Contacts.People; import android.content.ContentUris; import android.net.Uri; import android.database.Cursor; // Use the ContentUris method to produce the base URI for the contact with _ID == 23. Uri myPerson = ContentUris.withAppendedId(People.CONTENT_URI, 23); // Alternatively, use the Uri method to produce the base URI. // It takes a string rather than an integer. Uri myPerson = Uri.withAppendedPath(People.CONTENT_URI, "23"); // Then query for this specific record: Cursor cur = managedQuery(myPerson, null, null, null, null);
The other arguments to the query()
and managedQuery()
methods delimit
the query in more detail. They are:
null
value returns all columns. Otherwise, only columns that are listed by name
are returned. All the content providers that come with the platform define
constants for their columns. For example, the
android.provider.Contacts.Phones class
defines constants for the names of the columns in the phone table illustrated
earlier &mdash _ID
, NUMBER
, NUMBER_KEY
, NAME
,
and so on.A filter detailing which rows to return, formatted as an SQL WHERE
clause (excluding the WHERE
itself). A null
value returns
all rows (unless the URI limits the query to a single record).
Selection arguments.
A sorting order for the rows that are returned, formatted as an SQL
ORDER BY
clause (excluding the ORDER BY
itself). A null
value returns the records in the default order for the table, which may be
unordered.
Let's look at an example query to retrieve a list of contact names and their primary phone numbers:
import android.provider.Contacts.People; import android.database.Cursor; // Form an array specifying which columns to return. String[] projection = new String[] { People._ID, People._COUNT, People.NAME, People.NUMBER }; // Get the base URI for the People table in the Contacts content provider. Uri contacts = People.CONTENT_URI; // Make the query. Cursor managedCursor = managedQuery(contacts, projection, // Which columns to return null, // Which rows to return (all rows) null, // Selection arguments (none) // Put the results in ascending order by name People.NAME + " ASC");
This query retrieves data from the People table of the Contacts content
provider. It gets the name, primary phone number, and unique record ID for
each contact. It also reports the number of records that are returned as
the _COUNT
field of each record.
The constants for the names of the columns are defined in various interfaces
— _ID
and _COUNT
in
BaseColumns, NAME
in PeopleColumns, and NUMBER
in PhoneColumns. The
Contacts.People class implements
each of these interfaces, which is why the code example above could refer
to them using just the class name.
A query returns a set of zero or more database records. The names of the
columns, their default order, and their data types are specific to each
content provider.
But every provider has an _ID
column, which holds a unique numeric
ID for each record. Every provider can also report the number
of records returned as the _COUNT
column; its value
is the same for all rows.
Here is an example result set for the query in the previous section:
_ID | _COUNT | NAME | NUMBER |
---|---|---|---|
44 | 3 | Alan Vain | 212 555 1234 |
13 | 3 | Bully Pulpit | 425 555 6677 |
53 | 3 | Rex Cars | 201 555 4433 |
The retrieved data is exposed by a Cursor object that can be used to iterate backward or forward through the result set. You can use this object only to read the data. To add, modify, or delete data, you must use a ContentResolver object.
The Cursor object returned by a query provides access to a recordset of
results. If you have queried for a specific record by ID, this set will
contain only one value. Otherwise, it can contain multiple values.
(If there are no matches, it can also be empty.) You
can read data from specific fields in the record, but you must know the
data type of the field, because the Cursor object has a separate method
for reading each type of data — such as getString()
, getInt()
, and getFloat()
.
(However, for most types, if you call the method for reading strings,
the Cursor object will give you the String representation of the data.)
The Cursor lets you request the column name from the index of the column,
or the index number from the column name.
The following snippet demonstrates reading names and phone numbers from the query illustrated earlier:
import android.provider.Contacts.People; private void getColumnData(Cursor cur){ if (cur.moveToFirst()) { String name; String phoneNumber; int nameColumn = cur.getColumnIndex(People.NAME); int phoneColumn = cur.getColumnIndex(People.NUMBER); String imagePath; do { // Get the field values name = cur.getString(nameColumn); phoneNumber = cur.getString(phoneColumn); // Do something with the values. ... } while (cur.moveToNext()); } }
If a query can return binary data, such as an image or sound, the data
may be directly entered in the table or the table entry for that data may be
a string specifying a content:
URI that you can use to get the data.
In general, smaller amounts of data (say, from 20 to 50K or less) are most often
directly entered in the table and can be read by calling
Cursor.getBlob()
.
It returns a byte array.
If the table entry is a content:
URI, you should never try to open
and read the file directly (for one thing, permissions problems can make this
fail). Instead, you should call
ContentResolver.openInputStream()
to get an
InputStream object that you can use to read the data.
Data kept by a content provider can be modified by:
All data modification is accomplished using ContentResolver methods. Some content providers require a more restrictive permission for writing data than they do for reading it. If you don't have permission to write to a content provider, the ContentResolver methods will fail.
To add a new record to a content provider, first set up a map of key-value pairs
in a ContentValues object, where each key matches
the name of a column in the content provider and the value is the desired
value for the new record in that column. Then call ContentResolver.insert()
and pass
it the URI of the provider and the ContentValues map. This method returns
the full URI of the new record — that is, the provider's URI with
the appended ID for the new record. You can then use this URI to query and
get a Cursor over the new record, and to further modify the record.
Here's an example:
import android.provider.Contacts.People; import android.content.ContentResolver; import android.content.ContentValues; ContentValues values = new ContentValues(); // Add Abraham Lincoln to contacts and make him a favorite. values.put(People.NAME, "Abraham Lincoln"); // 1 = the new contact is added to favorites // 0 = the new contact is not added to favorites values.put(People.STARRED, 1); Uri uri = getContentResolver().insert(People.CONTENT_URI, values);
Once a record exists, you can add new information to it or modify existing information. For example, the next step in the example above would be to add contact information — like a phone number or an IM or e-mail address — to the new entry.
The best way to add to a record in the Contacts database is to append
the name of the table where the new data goes to the URI for the
record, then use the amended URI to add the new data values. Each
Contacts table exposes a name for this purpose as a CONTENT_DIRECTORY
constant. The following code continues the previous
example by adding a phone number and e-mail address for the record
just created:
Uri phoneUri = null; Uri emailUri = null; // Add a phone number for Abraham Lincoln. Begin with the URI for // the new record just returned by insert(); it ends with the _ID // of the new record, so we don't have to add the ID ourselves. // Then append the designation for the phone table to this URI, // and use the resulting URI to insert the phone number. phoneUri = Uri.withAppendedPath(uri, People.Phones.CONTENT_DIRECTORY); values.clear(); values.put(People.Phones.TYPE, People.Phones.TYPE_MOBILE); values.put(People.Phones.NUMBER, "1233214567"); getContentResolver().insert(phoneUri, values); // Now add an email address in the same way. emailUri = Uri.withAppendedPath(uri, People.ContactMethods.CONTENT_DIRECTORY); values.clear(); // ContactMethods.KIND is used to distinguish different kinds of // contact methods, such as email, IM, etc. values.put(People.ContactMethods.KIND, Contacts.KIND_EMAIL); values.put(People.ContactMethods.DATA, "test@example.com"); values.put(People.ContactMethods.TYPE, People.ContactMethods.TYPE_HOME); getContentResolver().insert(emailUri, values);
You can place small amounts of binary data into a table by calling
the version of ContentValues.put()
that takes a byte array.
That would work for a small icon-like image or a short audio clip, for example.
However, if you have a large amount of binary data to add, such as a photograph
or a complete song, put a content:
URI for the data in the table and call
ContentResolver.openOutputStream()
with the file's URI. (That causes the content provider to store the data
in a file and record the file path in a hidden field of the record.)
In this regard, the MediaStore content
provider, the main provider that dispenses image, audio, and video
data, employs a special convention: The same URI that is used with
query()
or managedQuery()
to get meta-information
about the binary data (such as, the caption of a photograph or the
date it was taken) is used with openInputStream()
to get the data itself. Similarly, the same URI that is used with
insert()
to put meta-information into a MediaStore record
is used with openOutputStream()
to place the binary data there.
The following code snippet illustrates this convention:
import android.provider.MediaStore.Images.Media; import android.content.ContentValues; import java.io.OutputStream; // Save the name and description of an image in a ContentValues map. ContentValues values = new ContentValues(3); values.put(Media.DISPLAY_NAME, "road_trip_1"); values.put(Media.DESCRIPTION, "Day 1, trip to Los Angeles"); values.put(Media.MIME_TYPE, "image/jpeg"); // Add a new record without the bitmap, but with the values just set. // insert() returns the URI of the new record. Uri uri = getContentResolver().insert(Media.EXTERNAL_CONTENT_URI, values); // Now get a handle to the file for that record, and save the data into it. // Here, sourceBitmap is a Bitmap object representing the file to save to the database. try { OutputStream outStream = getContentResolver().openOutputStream(uri); sourceBitmap.compress(Bitmap.CompressFormat.JPEG, 50, outStream); outStream.close(); } catch (Exception e) { Log.e(TAG, "exception while writing image", e); }
To batch update a group of records (for example, to change "NY" to "New York"
in all fields), call the ContentResolver.update()
method with the columns and values to change.
To delete a single record, call {ContentResolver.delete()
with the URI of a specific row.
To delete multiple rows, call ContentResolver.delete()
with the URI of the type of record to delete (for example, android.provider.Contacts.People.CONTENT_URI
) and an SQL WHERE
clause defining which rows to delete. (Caution:
Be sure to include a valid WHERE
clause if you're deleting a general
type, or you risk deleting more records than you intended!).
To create a content provider, you must:
Extend the ContentProvider class to provide access to the data.
Declare the content provider in the manifest file for your application (AndroidManifest.xml).
The following sections have notes on the last two of these tasks.
You define a ContentProvider subclass to expose your data to others using the conventions expected by ContentResolver and Cursor objects. Principally, this means implementing six abstract methods declared in the ContentProvider class:
query()
insert()
update()
delete()
getType()
onCreate()
The query()
method must return a Cursor object
that can iterate over the requested data. Cursor itself is an interface, but
Android provides some ready-made Cursor objects that you can use. For example,
SQLiteCursor can iterate over data stored in
an SQLite database. You get the Cursor object by calling any of the SQLiteDatabase class's query()
methods. There are other Cursor implementations — such as MatrixCursor — for data not stored in a database.
Because these ContentProvider methods can be called from various ContentResolver objects in different processes and threads, they must be implemented in a thread-safe manner.
As a courtesy, you might also want to call ContentResolver.notifyChange()
to notify listeners when there are
modifications to the data.
Beyond defining the subclass itself, there are other steps you should take to simplify the work of clients and make the class more accessible:
public static final
Uri
named CONTENT_URI
. This is the string that represents the full
content:
URI that your content provider handles. You must define a
unique string for this value. The best solution is to use the fully-qualified
class name of the content provider (made lowercase). So, for example, the
URI for a TransportationProvider class could be defined as follows:
public static final Uri CONTENT_URI = Uri.parse("content://com.example.codelab.transporationprovider");
If the provider has subtables, also define CONTENT_URI
constants for
each of the subtables. These URIs should all have the same authority (since
that identifies the content provider), and be distinguished only by their paths.
For example:
content://com.example.codelab.transporationprovider/train
content://com.example.codelab.transporationprovider/air/domestic
content://com.example.codelab.transporationprovider/air/international
For an overview of content:
URIs, see the Content URI
Summary at the end of this document.
Define the column names that the content provider will return to clients.
If you are using an underlying database, these column names are typically
identical to the SQL database column names they represent. Also define
public static
String constants that clients can use to specify
the columns in queries and other instructions.
Be sure to include an integer column named "_id
"
(with the constant _ID
) for
the IDs of the records. You should have this field whether or not you have
another field (such as a URL) that is also unique among all records. If
you're using the SQLite database, the _ID
field should be the
following type:
INTEGER PRIMARY KEY AUTOINCREMENT
The AUTOINCREMENT
descriptor is optional. But without it, SQLite
increments an ID counter field to the next number above the largest
existing number in the column. If you delete the last row, the next row added
will have the same ID as the deleted row. AUTOINCREMENT
avoids this
by having SQLite increment to the next largest value whether deleted or not.
Carefully document the data type of each column. Clients need this information to read the data.
If you are handling a new data type, you must define a new MIME type
to return in your implementation of ContentProvider.getType()
.
The type depends in part on whether or not the content:
URI submitted
to getType()
limits the request to a specific record. There's one
form of the MIME type for a single record and another for multiple records.
Use the Uri methods to help determine what is being
requested. Here is the general format for each type:
For a single record: vnd.android.cursor.item/vnd.yourcompanyname.contenttype
For example, a request for train record 122, like this URI,
content://com.example.transportationprovider/trains/122
might return this MIME type:
vnd.android.cursor.item/vnd.example.rail
For multiple records: vnd.android.cursor.dir/vnd.yourcompanyname.contenttype
For example, a request for all train records, like the following URI,
content://com.example.transportationprovider/trains
might return this MIME type:
vnd.android.cursor.dir/vnd.example.rail
If you are exposing byte data that's too big to put in the table itself
— such as a large bitmap file — the field that exposes the
data to clients should actually contain a content:
URI string.
This is the field that gives clients access to the data file. The record
should also have another field, named "_data
" that lists the exact file
path on the device for that file. This field is not intended to be read by
the client, but by the ContentResolver. The client will call ContentResolver.openInputStream()
on the user-facing field holding the URI for the item. The ContentResolver
will request the "_data
" field for that record, and because
it has higher permissions than a client, it should be able to access
that file directly and return a read wrapper for the file to the client.
For an example of a private content provider implementation, see the NodePadProvider class in the Notepad sample application that ships with the SDK.
To let the Android system know about the content provider you've developed,
declare it with a <provider>
element in the application's
AndroidManifest.xml file. Content providers that are not declared in the
manifest are not visible to the Android system
The name
attribute is the fully qualified name of the ContentProvider
subclass. The authorities
attribute is the authority part of the
content:
URI that identifies the provider.
For example if the ContentProvider subclass is AutoInfoProvider, the
<provider>
element might look like this:
<provider name="com.example.autos.AutoInfoProvider" authorities="com.example.autos.autoinfoprovider" . . . /> </provider>
Note that the authorities
attribute omits the path part of a
content:
URI. For example, if AutoInfoProvider controlled subtables
for different types of autos or different manufacturers,
content://com.example.autos.autoinfoprovider/honda
content://com.example.autos.autoinfoprovider/gm/compact
content://com.example.autos.autoinfoprovider/gm/suv
those paths would not be declared in the manifest. The authority is what identifies the provider, not the path; your provider can interpret the path part of the URI in any way you choose.
Other <provider>
attributes can set permissions to read and
write data, provide for an icon and text that can be displayed to users,
enable and disable the provider, and so on. Set the multiprocess
attribute to "true
" if data does not need to be synchronized between
multiple running versions of the content provider. This permits an instance
of the provider to be created in each client process, eliminating the need
to perform IPC.
Here is a recap of the important parts of a content URI:
The authority part of the URI; it identifies the content provider.
For third-party applications, this should be a fully-qualified class name
(reduced to lowercase) to ensure uniqueness. The authority is declared in
the <provider>
element's authorities
attribute:
<provider name=".TransportationProvider" authorities="com.example.transportationprovider" . . . >
The path that the content provider uses to determine what kind of data is
being requested. This can be zero or more segments long. If the content provider
exposes only one type of data (only trains, for example), it can be absent.
If the provider exposes several types, including subtypes, it can be several
segments long — for example, "land/bus
", "land/train
",
"sea/ship
", and "sea/submarine
" to give four possibilities.
The ID of the specific record being requested, if any. This is the
_ID
value of the requested record. If the request is not limited to
a single record, this segment and the trailing slash are omitted:
content://com.example.transportationprovider/trains