问题
I am sure that Firebase is counting all my development work too in its analytics. I open my app like hundred times a day to debug and test on a few devices, it's really skewing up my readings.
I have used a function to get me a somewhat unique ID to represent my devices and ignored all its analytics through code.
public static String getPsuedoID() {
String m_szDevIDShort = "35" + (Build.BOARD.length() % 10)
+ (Build.BRAND.length() % 10) + (Build.VERSION.SDK_INT % 10)
+ (Build.DEVICE.length() % 10) + (Build.DISPLAY.length() % 10)
+ (Build.MODEL.length() % 10) + (Build.PRODUCT.length() % 10);
String serial;
try {
serial = android.os.Build.class.getField("SERIAL").get(null).toString();
return new UUID(m_szDevIDShort.hashCode(), serial.hashCode()).toString();
} catch (Exception exception) {
serial = "getUniquePsuedoIDfailed";
}
return new UUID(m_szDevIDShort.hashCode(), serial.hashCode()).toString();
}
But I just figured out it is not as unique as I assumed. Apparently, my ID was same as around a few (very few) users.
Is there a foolproof method to do just this?
回答1:
You can control analytics collection using manifest metadata with the setting defined by a manifestPlaceholder:
<application
android:name="MyApplication"
//... >
<meta-data
android:name="firebase_analytics_collection_deactivated"
android:value="${analytics_deactivated}" />
//...
</application>
Then define the placeholder value in the build variant blocks of your build.gradle
file:
buildTypes {
debug {
manifestPlaceholders = [analytics_deactivated: "true"]
//...
}
release {
manifestPlaceholders = [analytics_deactivated: "false"]
//...
}
回答2:
A solution using the ANDROID_ID
Android has a provision of a unique device ID called the ANDROID_ID, a 64-bit number that is randomly generated when the user first sets up the device and remains constant thereafter. By leveraging this ANDROID_ID we can uniquely identify the testing devices.
Note: Unlike Firebase's provision of controlling analytics collection, this method will also work on Release builds/APKs.
1) Get ANDROID_ID of all your testing devices:
private static String getDeviceID(Context c) {
return Settings.Secure.getString(c.getContentResolver(), Settings.Secure.ANDROID_ID);
}
2) Once done, add the IDs to an array:
private static String testingDeviceIDs[] = {"8ab5946d3d65e893", "ada1247bfb6cfa5d", ...};
3) Check if the device is one of the testing devices.
private static boolean isDeviceForTesting(Context c) {
for (String testingID : testingDeviceIDs)
if (getDeviceID(c).equals(testingID))
return true;
return false;
}
4) Finally, log Firebase events only if the device is not a testing device.
static void logFirebaseEvent(Context c, String name) {
if (!isDeviceForTesting(c))
FirebaseAnalytics.getInstance(c).logEvent(name, null);
}
回答3:
Should be able to do something like following:
if (BuildConfig.DEBUG) {
FirebaseAnalytics.getInstance(getApplicationContext()).setAnalyticsCollectionEnabled(false);
}
回答4:
If what you were looking for, it was to disable the Firebase Analytics / tracking for the test devices that run the Firebase tests or the Play Console Pre-launch reports (which are actually Firebase tests), then you need to disable them based on a system string:
String testLabSetting = Settings.System.getString(context.getContentResolver(), "firebase.test.lab");
if ("true".equals(testLabSetting)) {
FirebaseAnalytics.getInstance(context).setAnalyticsCollectionEnabled(false);
}
So basically, it's a combination of checking if your device is a Firebase Test device, and then you disable the FirebaseAnalytics for it.
However, it won't entirely stop Firebase Analytics, and the event session_start
will keep showing up.
回答5:
The firebase docs only describe that it can be disabled from within the apk.
AndroidManifest.xml
<meta-data android:name="firebase_analytics_collection_enabled" android:value="false" />
Activity.java
setAnalyticsCollectionEnabled(false);
source: https://firebase.google.com/support/guides/disable-analytics
来源:https://stackoverflow.com/questions/40959903/exclude-testing-device-from-firebase-analytics-logging