Open
Conversation
ba5f9c1 to
f49f919
Compare
The idea is as such: create a custom version of SharedPreferences that treats preference keys prefixed with `encrypted` differently, saving them to an instance of EncryptedSharedPreferences. To actually use this MultiSharedPreferences in a PreferenceFragmentCompat, we have to employ a little hack of writing its private field, as for some reason the fragment does not allow customizing it's shared preferences. # Conflicts: # relay/build.gradle.kts
If we have a bad master key, or the encrypted file can't be decrypted using it, completely reset encrypted shared preferences. It is quite possible that we will never see this in action. # Conflicts: # app/src/main/java/com/ubergeek42/WeechatAndroid/utils/MultiSharedPreferences.kt
This moves `applicationContext` to the `utils` package. Additionally, it is mirrored in the tests to allow late initialization. This will be used in the following commit. This is highly experimental. I have no idea what I am doing.
Roboelectric is used to simplify testing with application context and
platform APIs. It provides these things without running on a real device.
We are using JUnit 5. Roboelectric needs JUnit 4; to make this work,
we use org.junit.vintage:junit-vintage-engine.
For in-memory implementation of shared preferences,
we use io.github.ivanshafran:shared-preferences-mock.
Please ignore this:
testImplementation("androidx.test:runner:1.5.2")
testImplementation("org.mockito:mockito-core:5.8.0")
I would prefer to only use the `include` rules, but this is quite hard when it comes to shared preferences. The problem is that the default shared preferences file name contains the application id. While it is possible to use application id placeholder in the manifest, it is not currently possible in the resource files. It is also apparently not possible to use string resources, it just does not work. We could also use a Gradle plugin that gives us placeholder replacements, but Android Stem seems to be only parsing string resources, which is neat but doesn't fit our use. The only feasible way here would be duplicating the resources for every build variant. Rules for API ≤ 30 can be tested with the following Bash script, which is a modified version of the script suitable for > 30 found here: at https://developer.android.com/guide/topics/data/testingbackup #!/bin/bash -eu : "${1?"Usage: $0 package name"}" adb shell bmgr enable true # $ adb shell bmgr list transports # android/com.android.internal.backup.LocalTransport # com.google.android.gms/.backup.BackupTransportService adb shell bmgr transport android/com.android.internal.backup.LocalTransport | grep -q "Selected transport" || (echo "Error: error selecting local transport"; exit 1) adb shell settings put secure backup_local_transport_parameters 'is_encrypted=true' adb shell bmgr backupnow "$1" | grep -F "Package $1 with result: Success" || (echo "Backup failed"; exit 1) apk_path_list=$(adb shell pm path "$1") OIFS=$IFS IFS=$'\n' apk_number=0 for apk_line in $apk_path_list do (( ++apk_number )) apk_path=${apk_line:8:1000} # MSYS_NO_PATHCONV=1 allows running this script in Git Bash # Copy before pulling to avoid “remote object ... does not exist” errors # See this answer josemigallas https://stackoverflow.com/questions/41150038/#43157127 MSYS_NO_PATHCONV=1 adb shell cp "$apk_path" "/storage/emulated/0/Download/temp_base.apk" MSYS_NO_PATHCONV=1 adb pull "/storage/emulated/0/Download/temp_base.apk" "myapk${apk_number}.apk" done IFS=$OIFS adb shell pm uninstall --user 0 "$1" apks=$(seq -f 'myapk%.f.apk' 1 $apk_number) adb install -t "$apks" adb shell bmgr transport com.google.android.gms/.backup.BackupTransportService rm $apks echo "Done"
f49f919 to
9a9bf5c
Compare
Collaborator
Author
|
I still want to do this despite https://www.reddit.ecom/r/androiddev/comments/1cj3krj/jetsec_crypto_is_now_deprecated/, and, more importantly, despite https://issuetracker.google.com/issues/176215143. As for the deprecation, ESP is a tiny wrapper around SP and tink, and we can just continue updating tink. As for the latter, it seems that the issue has 3 major sources:
Finally, there just seems to be no alternative. We would still want to go the same basic way, using some sort of a secret in AKS to encrypt data. |
Collaborator
|
That's… a lot. 🙃 Would it possible to have the version bumps and build cleanups as a separate PR (without the weird merge commit) so this code is just about the encryption? |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #574. This is based on top of #575. See individual commit messages. Suggest reviewing commit by commit.
TODO: Add include rules for backups. I suppose we only want to backup preferences, custom fonts and themes.https://developer.android.com/guide/topics/data/autobackup#IncludingFileshttps://developer.android.com/guide/topics/data/testingbackup