In the late 1990’s, a wave of viruses spread through the Internet, delivered via email, using contact information culled from Microsoft Outlook. A virus would simply email copies of itself to each of the Outlook contacts that had an email address. This was possible because, at the time, Outlook did not take any steps to protect data from programs using the Outlook API, since that API was designed for ordinary developers, not virus authors.
Nowadays, many applications that hold onto contact data secure that data by requiring that a user explicitly grant rights for other programs to access the contact information. Those rights could be granted on a case-by-case basis or all at once at install time.
Android is no different, in that it requires permissions for applications to read or write contact data. Android’s permission system is useful well beyond contact data, and for content providers and services beyond those supplied by the Android framework.
You, as an Android developer, will frequently need to ensure your applications have the appropriate permissions to do what you want to do with other applications’ data. This chapter covers this topic, both the classic approach used for all permissions prior to Android 6.0 and the new runtime permission system used for certain permissions in Android 6.0+.
You may also elect to require permissions for other applications to use your data or services, if you make those available to other Android components. This will be discussed later in this book.
Permissions are occasionally a confusing topic in Android app development, more so now that Android 6.0 has arrived and has changed the permission system a fair bit. Here are some common questions about permissions to help get us started.
A permission is a way for Android (or, sometimes, a third-party app) to require an app developer to notify the user about something that the app will do that might raise concerns with the user. Only if an app holds a certain permission can the app do certain things that are defended by that permission.
Mechanically, permissions take the form of elements in the manifest.
Right now, we are focusing on requesting and holding permissions, and
so we will be working with the <uses-permission>
element.
Most permissions that you will deal with come from Android itself. Usually, the documentation will tell you when you need to request and hold one of these permissions.
However, occasionally the documentation has gaps.
If you are trying out some code and you crash with a SecurityException
the description of the exception may tell you that you need to hold
a certain permission — that means you need to add the corresponding
<uses-permission>
element to your manifest.
Third-party code, including Google’s own Play Services SDK, may define their own custom permissions. Once again, ideally, you find out that you need to request a permission through documentation, and otherwise you find out through crashing during testing.
There are dozens upon dozens of permissions in Android. Here are some of the permissions we will see in this book:
INTERNET
, if your application wishes to access the Internet through any
means from your own process, using anything from raw Java sockets through the WebView
widgetWRITE_EXTERNAL_STORAGE
, for writing data to external storageACCESS_COARSE_LOCATION
and ACCESS_FINE_LOCATION
, for determining where
the device isREAD_CONTACTS
, to get at personally-identifying information of arbitrary
contacts that the user has in their Contacts appIn this book and in casual conversation, we refer to the permissions using
the unique portion of their name (e.g., INTERNET
). Really, the full
name of the permission will usually have android.permission.
as a prefix
(e.g., android.permission.INTERNET
), for Android-defined permissions.
Custom permissions from third-party apps should use a different prefix.
You will need the full permission name, including the prefix, in your
manifest entries.
Put a <uses-permission>
element in your manifest, as a direct
child of the root <manifest>
element (i.e., as a peer element
of <application>
), with an android:name
attribute identifying the
permission that you are interested in.
For example, here is a sample manifest, with a request to hold
the WRITE_EXTERNAL_STORAGE
permission:
<?xml version="1.0"?>
<manifest package="com.commonsware.android.fileseditor"
xmlns:android="http://schemas.android.com/apk/res/android"
android:versionCode="1"
android:versionName="1.0">
<uses-permission android:name="android.permission.WRITE_EXTERNAL_STORAGE" />
<supports-screens
android:anyDensity="true"
android:largeScreens="true"
android:normalScreens="true"
android:smallScreens="true" />
<application
android:icon="@drawable/ic_launcher"
android:label="@string/app_name"
android:theme="@style/Theme.Apptheme">
<activity
android:name=".MainActivity"
android:label="@string/app_name">
<intent-filter>
<action android:name="android.intent.action.MAIN" />
<category android:name="android.intent.category.LAUNCHER" />
</intent-filter>
</activity>
</application>
</manifest>
This is sufficient for most permissions and most devices. Permissions
considered to be dangerous
need special attention on Android 6.0+, and
we will cover that in grand detail
later in this chapter.
Note that you are welcome to have zero, one, or several such
<uses-permission>
elements. Also note that some libraries that
you elect to use might add their own <uses-permission>
elements
to your manifest, through a process called “manifest merger”.
Well, that gets complicated. It depends on the permission, the version of Android the user is using, from where the user is installing the app, and the phase of the moon.
(well, OK, not really that last one)
Anyone who installs an app using Android Studio will not be prompted for permissions. The same holds true for anyone using anything else based on the Android SDK tools — while the app may request permissions, the user is not prompted for them, and the permissions are granted.
(Android 6.0+ and dangerous
permissions change this up a bit –
more on that later in this chapter)
If the user is running an Android 5.1 or older device, and the user
goes to install your app from the Play Store, the user will be presented
with a roster of permission groups that contain permissions that you
are requesting and that are considered to be dangerous
:
Figure 255: Permission Confirmation Screen, on Play Store Web Site
We will discuss more about permission groups and this dangerous
concept later in this chapter.
On Android 6.0 and higher, when the user installs your app from the
Play Store, what happens depends upon the value of targetSdkVersion
for
your app.
If your targetSdkVersion
is 22 or lower, you get the same behavior
as is described above, where the user sees the list of permission groups
which contain permissions that you are requesting and that are considered
to be dangerous
.
If your targetSdkVersion
is 23 or higher, the user is not prompted about
permissions at install time. Instead these prompts will occur
when the user runs your app and when you ask the user for the permissions,
as we will see later in this chapter.
If you install an app on Android 5.1 or older, by any means (e.g., downloading from a Web site), you will be prompted with a list of all requested permissions:
Figure 256: Permission Confirmation Screen, on Android 4.4
Note that this prompt will not appear until you actually have downloaded the app and have begun the installation process. Before then, the device cannot examine the manifest inside the APK file to find the permissions.
If your app’s targetSdkVersion
is 23 or higher, and you install the
app on an Android 6.0+ device by other means than the Play Store, you will
not be prompted about any permissions at install time:
Figure 257: Permission Confirmation Screen, on Android 6.0
Several bits of information make up a permission, and some of those affect app developers or users.
We have already seen that permissions have names, and you use them
in the android:name
attribute of the <uses-permission>
element to
identify a permission that you would like your app to hold.
Android framework-defined permissions will begin with android.permission
.
Permissions from libraries or third-party apps will have some other
prefix. Make sure that when you create your <uses-permission>
element that you are using the fully-qualified permission name,
including android.permission
or any other prefix.
Also note that Android is case-sensitive, so make sure you use the case
of the permission as documented (e.g., android.permission.INTERNET
).
Some versions of Android Studio had a bug where if you let the IDE
auto-complete a <uses-permission>
element for you, sometimes it would
have the android:name
value appear IN ALL CAPS. This is a bug that
has since been fixed, so hopefully it will not affect you in the future.
The definition of a permission, in the framework or in third-party
code, will have a “protection level”. This describes how the permission
itself should be validated. The two protection levels that you
will encounter most often are normal
and dangerous
.
A normal
permission is something that the user might care about, but
probably not. So, while we need to request the permission in the manifest
via <uses-permission>
, the user will not be bothered about this
permission at install time.
The classic example is the INTERNET
permission. Most Android apps wind
up requesting this permission, either for functionality written by the
developers or functionality pulled in from libraries (e.g., ad banners).
INTERNET
is considered normal
, so while we need to request the
INTERNET
permission in the manifest, the user is not informed about
this permission anymore at install time.
(the “anymore” note is because in the early days of Android, users were informed about all permissions, regardless of protection level)
Users can see normal
permissions, though, in other places:
A dangerous
permission is one that the definers of the permission
(e.g., Google) wants to ensure that the user is aware of and has agreed to.
Classically, this meant that the user would be prompted for this permission
at install time. On old versions of Android and the Play Store,
dangerous
permissions would be listed before normal
permissions.
With Android 6.0+, while dangerous
permissions are not displayed at
install time (for apps with a targetSdkVersion
of 23 or higher),
they will be displayed to the user while the app is running, before the
app tries doing something that requires one of those permissions.
This is a significant behavior change, so we will be covering it in depth
later in this chapter.
Permissions are collected into permission groups.
In the early days of Android, app developers were oblivious to this, as permission groups had no effect on app development, runtime behavior, or user experience.
In the past few years, the “permission” prompts at install time have really been prompting about permission groups. The user is told that the app is requesting permissions from certain groups. Moreover, the blessing that the user gives — by virtue of continuing to install the app — is by group, not by permission. If some future update to the app would ask for a new permission, but one from a group that the user agreed to previously, the user would not be informed about this new permission request.
With Android 6.0, permission groups also extend to the runtime permission UX, as while we developers will still request individual permissions, the user will be asked to grant rights with respect to permission groups.
<uses-permission>
can have an android:maxSdkVersion
attribute.
This indicates the highest API level for which we need the permission.
If the app is running on newer versions of Android, skip the permission.
This is for cases where Android relaxes restrictions over time. We will
see an example of this, in the form of the WRITE_EXTERNAL_STORAGE
permission, in an upcoming chapter.
You might think that <uses-permission>
would have an
android:minSdkVersion
attribute to serve as the counterpart to
android:maxSdkVersion
. The minSdkVersion
would indicate the
lowest API level for which to request a permission; older devices
would skip the permission.
Alas, this is not available.
However, there is the awkwardly-named [uses-permission-sdk-23]
element.
This element functions identically to <uses-permission>
on Android 6.0+
devices. On older devices, it is ignored.
This element illustrates a problem with the permission system
in Android: you have to put all permissions that you want in the manifest.
Prior to the runtime permission system in Android 6.0, this would mean
that developers who need some controversial permission
(e.g., READ_CONTACTS
) for some fringe feature would need to request
the permission from everyone, not just those who use the feature. As
we will see, the runtime permission system lets us not bother the user
until they try using the secured feature. [uses-permission-sdk-23]
would allow us to not bother with the permission at all on older
devices, where its presence might scare away potential users.
Sometimes, Android introduces new permissions that govern behavior that
formerly did not require permissions. WRITE_EXTERNAL_STORAGE
is one example
– originally, applications could write to external storage without any
permission at all. Android 1.6 introduced WRITE_EXTERNAL_STORAGE
, required
before you can write to external storage. However, applications that were
written before Android 1.6 could not possibly request that permission, since it
did not exist at the time. Breaking those applications would seem to be a harsh
price for progress.
What Android does is “grandfather” in certain permissions for applications supporting earlier SDK versions.
For example, if your minSdkVersion
is 3 or lower, saying that you
support Android 1.5, your application will
automatically request WRITE_EXTERNAL_STORAGE
and READ_PHONE_STATE
, even if
you do not explicitly request those permissions. People installing your
application on an Android 1.5 device will see these requests.
Eventually, when you drop support for the older version (e.g., switch to
minSdkVersion
of 4 or higher), Android will no longer automatically
request those permissions. Hence, if your code really does need those
permissions, you will need to ask for them yourself.
In Android 6.0 and higher devices, permissions that are considered
to be dangerous
not only have to be requested via <uses-permission>
elements, but you also have to ask the user to grant you those permissions
at runtime. What you gain, though, is that users are not bothered with
these permissions at install time, and you can elect to delay asking
for certain permissions until such time as the user actually does
something that needs them.
This section will occasionally point out snippets of code from the
Permissions/PermissionMonger
sample project.
Let’s explore the runtime permissions system via a new series of questions.
There are nine permission groups that Android 6.0 manages as user-controllable permissions:
Permission Group | Permission |
---|---|
CALENDAR |
READ_CALENDAR , WRITE_CALENDAR
|
CAMERA |
CAMERA |
CONTACTS |
GET_ACCOUNTS , READ_CONTACTS , WRITE_CONTACTS
|
LOCATION |
ACCESS_COARSE_LOCATION , ACCESS_FINE_LOCATION
|
MICROPHONE |
RECORD_AUDIO |
PHONE |
ADD_VOICEMAIL , CALL_PHONE , PROCESS_OUTGOING_CALLS , READ_CALL_LOG , READ_PHONE_STATE , USE_SIP , WRITE_CALL_LOG
|
SENSORS |
BODY_SENSORS |
SMS |
READ_CELL_BROADCASTS , READ_SMS , RECEIVE_SMS , RECEIVE_MMS , RECEIVE_WAP_PUSH , SEND_SMS
|
STORAGE |
READ_EXTERNAL_STORAGE , WRITE_EXTERNAL_STORAGE
|
Users will be able to revoke permissions by group, through the Settings app. They can go into the page for your app, click on Permissions, and see a list of the permission groups for which you are requesting permissions:
Figure 258: Settings Screen for Permission Monger, Showing Permissions
The same <uses-permission>
elements as before. These declare the superset
of all possible permissions that you can have. If you do not have a
<uses-permission>
element for a particular permission, you cannot
ask for it at runtime, and the user cannot grant it to you.
On Android 6.0+, you can call a checkSelfPermission()
method, available
on any Context
(e.g., your Activity
). This will return either
PERMISSION_GRANTED
or PERMISSION_DENIED
, depending on whether or
not the user granted you permission or you were automatically given
permission (e.g., for normal
permissions).
For a simpler boolean
check to see if you have the permission, you could
have your own hasPermission()
method:
private boolean hasPermission(String perm) {
return(PackageManager.PERMISSION_GRANTED==checkSelfPermission(perm));
}
Then you can use that hasPermission()
call where you need it.
For example, the PermissionMonger app requests five permissions in the manifest:
<uses-permission android:name="android.permission.ACCESS_FINE_LOCATION"/>
<uses-permission android:name="android.permission.CAMERA"/>
<uses-permission android:name="android.permission.INTERNET"/>
<uses-permission android:name="android.permission.READ_CONTACTS"/>
<uses-permission android:name="android.permission.WRITE_EXTERNAL_STORAGE"/>
The UI is then a table showing the current status of those five permissions:
private void updateTable() {
location.setText(String.valueOf(canAccessLocation()));
camera.setText(String.valueOf(canAccessCamera()));
internet.setText(String.valueOf(hasPermission(Manifest.permission.INTERNET)));
contacts.setText(String.valueOf(canAccessContacts()));
storage.setText(String.valueOf(hasPermission(Manifest.permission.WRITE_EXTERNAL_STORAGE)));
}
private boolean canAccessLocation() {
return(hasPermission(Manifest.permission.ACCESS_FINE_LOCATION));
}
private boolean canAccessCamera() {
return(hasPermission(Manifest.permission.CAMERA));
}
private boolean canAccessContacts() {
return(hasPermission(Manifest.permission.READ_CONTACTS));
}
At the outset, we only have the one “normal” permission: INTERNET
:
Figure 259: Permission Monger, Showing Initial Permissions
The checkSelfPermission()
method on Context
is only available on
API Level 23. You can, if you wish, wrap your call to checkSelfPermission()
in a check of the API level of the device you are running on:
if (Build.VERSION.SDK_INT>=Build.VERSION_CODES.M) {
if (checkSelfPermission(Manifest.permission.WRITE_EXTERNAL_STORAGE)==
PackageManager.PERMISSION_GRANTED) {
// do something cool
}
}
A simpler approach is to use ContextCompat
, from the support-v4
library. This has a static
implementation of checkSelfPermission()
that takes a Context
and your permission string as parameters. It
returns the same value (e.g., PackageManager.PERMISSION_GRANTED
)
as does the checkSelfPermission()
that ships with Android 6.0.
But, if you are running on an older device, it checks the version for
you and returns PackageManager.PERMISSION_GRANTED
for older devices.
So, the above code snippet turns into:
if (ContextCompat.checkSelfPermission(this,
Manifest.permission.WRITE_EXTERNAL_STORAGE)==
PackageManager.PERMISSION_GRANTED) {
// do something cool
}
(assuming that this
is a subclass of Context
, like Activity
)
If the user grants you access to some permission group, the only way the user can revoke that is via the Settings app. If the user does revoke access to a permission group, your process is terminated.
Hence, while your code is running, you will have all permissions that you started with, plus any new ones that the user grants on the fly based upon your request. There should be no circumstance where your process is running yet you lose a permission.
That being said, your app is not specifically notified about losing
the permission. You should be calling checkSelfPermission()
to determine
what you can and cannot do, at least for every process invocation. And,
since the call appears to be reasonably cheap, you should just call it
whenever you need to know whether you can perform a particular operation.
Note that usually your app will be in the background, if it is running at all, at the time when a runtime permission is revoked. Particularly in an Android 7.0+ multi-window environment, though, it is possible that you will still be visible when the user revokes a runtime permission. Your process is terminated in any case, and if you were visible, your UI is removed from the screen.
To ask the user for one of the runtime permissions, call requestPermissions()
on your Activity
. This takes a String
array of the permissions that you are
requesting and a locally-unique integer to identify this request from any
other similar requests that you may be making. This int
serves in
much the same role as does the int
passed into startActivityForResult()
,
though you should keep the value to 8 bits (0
to 255
) for
maximum compatibility.
For example, PermissionMonger will check in onCreate()
to see if
we can access locations or access contacts, and if not, it will request
access to those two permissions:
if (!canAccessLocation() || !canAccessContacts()) {
requestPermissions(INITIAL_PERMS, INITIAL_REQUEST);
}
INITIAL_PERMS
and INITIAL_REQUEST
are just static final
data members:
private static final String[] INITIAL_PERMS={
Manifest.permission.ACCESS_FINE_LOCATION,
Manifest.permission.READ_CONTACTS
};
private static final int INITIAL_REQUEST=1337;
When the app is first launched, dialogs will appear, one per permission that you requested, asking the user if they would be so kind as to allow your app to do the things that you requested:
Figure 260: Permission Monger, Requesting READ_CONTACTS Permission
Figure 261: Permission Monger, Requesting ACCESS_FINE_LOCATION Permission
When the user has proceeded through the dialogs, you will be called
with onRequestPermissionsResult()
. You are passed three parameters:
requestPermissions()
call,
to identify which requestPermissions()
call this is the result forString
array of the requested permissionsint
array of the corresponding results (PERMISSION_GRANTED
or
PERMISSION_DENIED
)Whether you use those latter two parameters or simply call checkSelfPermission()
again is up to you. Regardless, at this point, you should determine
what you got, so you know how to react, such as disabling things that the
user cannot use given the lack of permission.
Just as ContextCompat
offers a backwards-compatible implementation
of checkSelfPermission()
, ActivityCompat
offers a backwards-compatible
implementation of requestPermissions()
that you can use. Otherwise,
you will want to take other steps to ensure that you only call
requestPermissions()
on API Level 23+ devices.
That depends a bit on the nature of the permission.
In an ideal world, your app can function without any of the revocable permissions granted to you, albeit perhaps in a limited fashion. In that case, you might ask for permission only when the user tries to do something (e.g., taps on an action bar item) for which you definitely need the permission.
However, sometimes you will need permission to be at all useful to the user. In that case, you will need to ask for permission when the app opens.
In either case, though, bear in mind that while the user will see the dialog asking for permission, the user may not understand why you are asking for this permission. You need to make sure that the user understands the cost/benefit trade-off in granting the permission — in other words, what does the user get out of the deal?
For permissions that you are requesting based on user input, you might
pop your own dialog or other UI explaining what you want and why you want
it, before calling requestPermissions()
. For permissions that you
would want to ask for when the app starts up, make sure that you
clearly explain the need for the permissions and what the user gets in
exchange as part of a one-time introductory tutorial,
one that might also be accessed via an overflow item or nav drawer
entry as part of your app’s help facility.
One limitation with the requestPermissions()
implementation is that
it is oblivious to configuration changes.
For example, suppose that in onCreate()
of your activity, you check
to see if you have been granted a runtime permission (via
checkSelfPermission()
), and if you have not, you call requestPermissions()
to request it from the user. This displays the dialog. Now the user rotates
the screen. If the user denies the permission, by default, the user will
immediately see the permission dialog again… because your activity
will have been destroyed and recreated, and your onCreate()
will see
that you do not have the permission, and so you ask for it again.
In cases like this, you will need to track whether you are in the
permission-request flow (e.g., via a boolean
saved in the instance
state Bundle
) and skip requesting the permission if you have been
recreated in the middle of that flow.
If you were requesting permission as a direct response to some bit of user input (e.g., user tapped on an action bar item), and the user rejects the permission you need to do the work, obviously you cannot do the work. Depending on overall flow, showing a dialog or something to explain why you cannot do what the user asked for may be needed. In some cases, you may deem it to be obvious, by virtue of the fact that the user saw the permission-request dialog and said “deny”.
If you were requesting permission pre-emptively, such as when the activity starts, you will need to decide whether that decision needs to be reflected in the current UI (e.g., “no data available” messages, disabled action bar items).
One thing you can do to help here is to detect when this has
occurred before you request permissions again. Before you call
requestPermissions()
, you can call shouldShowRequestPermissionRationale()
,
supplying the name of a permission. This will return true
if the
user had previously declined to grant you permission, in cases
where Android thinks that the user might benefit from learning a bit
more about why you need the permission. You can use this to determine
whether you should show some explanatory UI of your own first, before
continuing with the permission request, or if you should just go
ahead and call requestPermissions()
.
Note that ActivityCompat
also has a backwards-compatible
implementation of shouldShowRequestPermissionRationale()
,
so you can avoid your own API level checks.
The second time you ask a user for a particular runtime permission, the user will have a “Never ask again” checkbox:
Figure 262: Permission Monger, Requesting ACCESS_FINE_LOCATION Permission (Again)
If the user checks that and clicks the Deny button, not only will
you not get the runtime permission now, but all future requests will
immediately call onRequestPermissionsResult()
indicating that
your request for permission was denied. The only way the user can now
grant you this permission is via the Settings app.
You need to handle this situation with grace and aplomb.
Choices include:
For permissions that, when denied, leave your app in a completely useless state, you may wind up just displaying a screen on app startup that says “sorry, but this app is useless to you”, with options for the user to uninstall the app or grant you the desired permissions.
Note that shouldShowRequestPermissionRationale()
returns false
if
the user declined the permission and checked the checkbox to ask you
to stop pestering the user.
Older devices behave as they always have. Since you still list the
permissions in the manifest, those permissions will be granted to you
if the user installs the app, and the user will be notified about those
permissions as part of the installation process. If you are checking
the API level yourself, or you are using ContextCompat
and ActivityCompat
as described above, your code should just work.
Apps with a targetSdkVersion
below 23, on the surface, behave
on Android 6.0+ as they
would on an older device: the user is prompted for all permissions, and the
app is granted those permissions if the app is installed.
However, the user will still be able to go into Settings and revoke permissions from these apps, for any permissions the app requests that are in one of the runtime permission groups.
Generally, you will wind up ignoring the issue. All your calls to methods protected by
permissions that the user revoke will still “work”, insofar as they will
not throw a SecurityException
. However, you just will not get any results
back or have the intended effects. So, for example, if you try to query()
the ContactsContract
ContentProvider
, and the user revoked your access
to contact-related permissions, the query()
will return an empty
Cursor
. This is a completely valid response, even ignoring the permission
issue, as it is entirely possible that the user has no contacts. Your app
should be handling these cases gracefully anyway. Hence, in theory, even
if you do nothing special regarding the lost permissions, your app should
survive, albeit with reduced functionality for the user. Dave Smith
outlines the expected results
for legacy apps calling methods sans permission.
However, all else being equal, you should set your targetSdkVersion
to at least 23 and opt into the runtime permission system. After all,
the Play Store requires a higher targetSdkVersion
,
and other app distribution channels might follow suit.
If the user clears your app’s data through the Settings app, the runtime
permissions are cleared as well. Behavior at this point will be as if
your app had been just installed — checkSelfPermission()
will return
PERMISSION_DENIED
, and you will need to request the permissions.
While the runtime permission system provides a reasonable user-facing UI, having to deal with that UI constantly as a developer can be a significant pain. For testing and debugging purposes, there are some command-line options for granting and revoking permissions that you can use.
checkSelfPermission()
on ContextCompat
always returns PERMISSION_GRANTED
if either:
targetSdkVersion
below 23, orPERMISSION_DENIED
will be returned only if you have opted into
the new runtime permission system (targetSdkVersion
of 23 or higher),
you are running on Android 6.0 or higher, and the user either never granted
the permission or revoked it through the Settings app.
The key is that even if you are running on Android 6.0, with an older
targetSdkVersion
(so all permissions are requested at install time and
are granted to you automatically), checkSelfPermission()
still returns
PERMISSION_GRANTED
even if the user revoked the permission in
Settings.
The Android Support libraries — specifically support-v4
— added
a PermissionChecker
class with a checkPermission()
static method.
If you are running on Android 6.0+ with an older targetSdkVersion
,
checkPermission()
will return PERMISSION_DENIED_APP_OP
if the
user revoked the permission in Settings.
Hence, PermissionChecker
is useful in cases where you have a really
large code base, and you want to try to better handle cases where
users revoke permissions, but you are not in position to do a complete
implementation of the runtime permissions system. However, it is merely a
stopgap — your long-term plan should be to raise your targetSdkVersion
to 23 or higher and implement the runtime permissions properly.
Occasionally, our need for runtime permissions is fairly straightforward: we need them to do anything meaningful in the app, and so we need to request them up front and exit the app if the user declines to grant one or more of them. Some of the samples in this book fit this model, in part because many of the samples in this book are fairly small apps.
Some of those apps will use a variation of an AbstractPermissionActivity
to
hide all the runtime permission work, allowing the sample’s “real” activity to
focus on demonstrating the portion of the Android SDK that the sample is tied to.
One such AbstractPermissionActivity
can be found in the
Files/FilesEditor
sample project. Later in the book, we will look at this sample to
see how to do file I/O on Android. Here, though, let’s take a look at
AbstractPermissionActivity
.
The idea is that AbstractPermissionActivity
will handle the runtime permissions,
in its onCreate()
implementation. Subclasses will need to implement three
methods to make this work:
getDesiredPermissions()
, which returns the names of the permissions that the
app wantsonReady()
, which will be called once permission is granted by the user, and
serves as an onCreate()
substitute for the subclassonPermissionDenied()
, which will be called if the user declines granting
the permission, so the subclass can do something (e.g., show a Toast
, then
finish()
and go away)onCreate()
will see if we have the desired permissions, and if not, it will
call requestPermissions()
to ask for those that we do not already hold:
@Override
protected void onCreate(Bundle savedInstanceState) {
super.onCreate(savedInstanceState);
this.state=savedInstanceState;
if (state!=null) {
isInPermission=state.getBoolean(STATE_IN_PERMISSION, false);
}
if (hasAllPermissions(getDesiredPermissions())) {
onReady(state);
}
else if (!isInPermission) {
isInPermission=true;
ActivityCompat
.requestPermissions(this,
netPermissions(getDesiredPermissions()),
REQUEST_PERMISSION);
}
}
hasAllPermissions()
iterates over the permission array from getDesiredPermissions()
and returns true
if we hold them all, false
otherwise:
private boolean hasAllPermissions(String[] perms) {
for (String perm : perms) {
if (!hasPermission(perm)) {
return(false);
}
}
return(true);
}
protected boolean hasPermission(String perm) {
return(ContextCompat.checkSelfPermission(this, perm)==
PackageManager.PERMISSION_GRANTED);
}
If we hold all of the permissions, we go ahead and call onReady()
, so the
activity can start its real work. Otherwise, we call requestPermissions()
on ActivityCompat
, using a netPermissions()
method to identify those
permissions that we do not already hold:
private String[] netPermissions(String[] wanted) {
ArrayList<String> result=new ArrayList<String>();
for (String perm : wanted) {
if (!hasPermission(perm)) {
result.add(perm);
}
}
return(result.toArray(new String[result.size()]));
}
In onRequestPermissionResult()
, depending on whether we now hold all the
desired permissions, we call onReady()
or onPermissionDenied()
:
@Override
public void onRequestPermissionsResult(int requestCode,
String[] permissions,
int[] grantResults) {
isInPermission=false;
if (requestCode==REQUEST_PERMISSION) {
if (hasAllPermissions(getDesiredPermissions())) {
onReady(state);
}
else {
onPermissionDenied();
}
}
}
It is possible that the user will rotate the device or otherwise trigger a configuration change while our permission dialog is in the foreground. Since our activity is still visible behind that dialog, we get destroyed and recreated… but we do not want to re-raise the permission dialog again.
That is why we have a boolean
, named isInPermission
, that tracks whether or
not we are in the middle of requesting permissions. We hold onto that value
in onSaveInstanceState()
:
@Override
protected void onSaveInstanceState(Bundle outState) {
super.onSaveInstanceState(outState);
outState.putBoolean(STATE_IN_PERMISSION, isInPermission);
}
We restore it in onCreate()
. If we do not hold all of the desired permissions,
but isInPermission
is true
, we skip requesting the permissions, since we
are in the middle of doing so already.
As noted earlier, later in the book, we will look at how a subclass
of AbstractPermissionActivity
works, and you will see AbstractPermissionActivity
used in other sample apps as well.