Human operator

.github/workflows/manual.yml

@@ -1,35 +1,140 @@
 name: Android Build
 
 on:
- # push:
-   # branches: [ main, develop ]
- # pull_request:
-  #  branches: [ main ]
+  push:
+    branches: [ human-operator, main ]
   workflow_dispatch:  # Ermöglicht manuelle Ausführung des Workflows
 
 jobs:
+  detect-changes:
+    runs-on: ubuntu-latest
+    outputs:
+      app_changed: ${{ steps.changes.outputs.app }}
+      humanoperator_changed: ${{ steps.changes.outputs.humanoperator }}
+      shared_changed: ${{ steps.changes.outputs.shared }}
+    steps:
+    - name: Checkout code
+      uses: actions/checkout@v4
+      with:
+        fetch-depth: 2  # Letzten 2 Commits holen für Diff
+
+    - name: Detect changed files
+      id: changes
+      run: |
+        # Bei workflow_dispatch immer alles bauen
+        if [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
+          echo "app=true" >> $GITHUB_OUTPUT
+          echo "humanoperator=true" >> $GITHUB_OUTPUT
+          echo "shared=true" >> $GITHUB_OUTPUT
+          echo "Manual dispatch - building all modules"
+          exit 0
+        fi
+
+        # Geänderte Dateien im letzten Commit ermitteln
+        CHANGED_FILES=$(git diff --name-only HEAD~1 HEAD 2>/dev/null || echo "")
+
+        # Falls kein vorheriger Commit existiert (erster Commit), alles bauen
+        if [ -z "$CHANGED_FILES" ]; then
+          echo "app=true" >> $GITHUB_OUTPUT
+          echo "humanoperator=true" >> $GITHUB_OUTPUT
+          echo "shared=true" >> $GITHUB_OUTPUT
+          echo "No previous commit found - building all modules"
+          exit 0
+        fi
+
+        echo "Changed files:"
+        echo "$CHANGED_FILES"
+
+        # Prüfen ob shared/root files geändert wurden (build.gradle, settings.gradle, etc.)
+        SHARED_CHANGED=false
+        if echo "$CHANGED_FILES" | grep -qE '^(build\.gradle|settings\.gradle|gradle\.properties|gradle/|buildSrc/)'; then
+          SHARED_CHANGED=true
+        fi
+
+        # Prüfen ob app/ Dateien geändert wurden
+        APP_CHANGED=false
+        if echo "$CHANGED_FILES" | grep -q '^app/'; then
+          APP_CHANGED=true
+        fi
+
+        # Prüfen ob humanoperator/ Dateien geändert wurden
+        HUMANOPERATOR_CHANGED=false
+        if echo "$CHANGED_FILES" | grep -q '^humanoperator/'; then
+          HUMANOPERATOR_CHANGED=true
+        fi
+
+        echo "app=$APP_CHANGED" >> $GITHUB_OUTPUT
+        echo "humanoperator=$HUMANOPERATOR_CHANGED" >> $GITHUB_OUTPUT
+        echo "shared=$SHARED_CHANGED" >> $GITHUB_OUTPUT
+
+        echo "Results: app=$APP_CHANGED, humanoperator=$HUMANOPERATOR_CHANGED, shared=$SHARED_CHANGED"
+
   build:
+    needs: detect-changes
     runs-on: ubuntu-latest
+    env:
+      BUILD_APP: ${{ needs.detect-changes.outputs.app_changed == 'true' || needs.detect-changes.outputs.shared_changed == 'true' }}
+      BUILD_HUMANOPERATOR: ${{ needs.detect-changes.outputs.humanoperator_changed == 'true' || needs.detect-changes.outputs.shared_changed == 'true' }}
 
     steps:
     - name: Checkout code
-      uses: actions/checkout@v3
+      uses: actions/checkout@v4
 
-    - name: Set up JDK
-      uses: actions/setup-java@v3
+    - name: Set up JDK 17
+      uses: actions/setup-java@v4
       with:
         java-version: '17'
         distribution: 'temurin'
         cache: gradle
 
+    - name: Decode google-services.json (app)
+      env:
+        GOOGLE_SERVICES_JSON: ${{ secrets.GOOGLE_SERVICES_JSON_APP }}
+      run: printf '%s' "$GOOGLE_SERVICES_JSON" > app/google-services.json
+
+    - name: Decode google-services.json (humanoperator)
+      env:
+        GOOGLE_SERVICES_JSON: ${{ secrets.GOOGLE_SERVICES_JSON_HUMANOPERATOR }}
+      run: printf '%s' "$GOOGLE_SERVICES_JSON" > humanoperator/google-services.json
+
+    - name: Create local.properties
+      run: echo "sdk.dir=$ANDROID_HOME" > local.properties
+
+    - name: Fix gradle.properties for CI
+      run: |
+        sed -i '/org.gradle.java.home=/d' gradle.properties
+        sed -i 's/org.gradle.jvmargs=.*/org.gradle.jvmargs=-Xmx2048m -XX:MaxMetaspaceSize=512m/' gradle.properties
+        sed -i 's/kotlin.daemon.jvmargs=.*/kotlin.daemon.jvmargs=-Xmx1536m -XX:MaxMetaspaceSize=512m/' gradle.properties
+
     - name: Grant execute permission for gradlew
       run: chmod +x gradlew
 
-    - name: Build with Gradle
-      run: ./gradlew assembleRelease
+    - name: Build app module (debug)
+      if: env.BUILD_APP == 'true'
+      run: ./gradlew :app:assembleDebug
 
-    - name: Upload APK
+    - name: Build humanoperator module (debug)
+      if: env.BUILD_HUMANOPERATOR == 'true'
+      run: ./gradlew :humanoperator:assembleDebug
+
+    - name: Upload app APK
+      if: env.BUILD_APP == 'true'
       uses: actions/upload-artifact@v4
       with:
-        name: app-release
-        path: app/build/outputs/apk/release/app-release-unsigned.apk
+        name: app-debug
+        path: app/build/outputs/apk/debug/app-debug.apk
+
+    - name: Upload humanoperator APK
+      if: env.BUILD_HUMANOPERATOR == 'true'
+      uses: actions/upload-artifact@v4
+      with:
+        name: humanoperator-debug
+        path: humanoperator/build/outputs/apk/debug/humanoperator-debug.apk
+
+    - name: Build summary
+      run: |
+        echo "### Build Summary" >> $GITHUB_STEP_SUMMARY
+        echo "| Module | Built |" >> $GITHUB_STEP_SUMMARY
+        echo "|--------|-------|" >> $GITHUB_STEP_SUMMARY
+        echo "| app | ${{ env.BUILD_APP }} |" >> $GITHUB_STEP_SUMMARY
+        echo "| humanoperator | ${{ env.BUILD_HUMANOPERATOR }} |" >> $GITHUB_STEP_SUMMARY

.vscode/settings.json

@@ -0,0 +1,3 @@
+{
+    "editor.maxTokenizationLineLength": 20000
+}

app/build.gradle.kts

@@ -5,14 +5,20 @@ plugins {
     id("org.jetbrains.kotlin.plugin.serialization") version "1.9.20"
     id("com.google.android.libraries.mapsplatform.secrets-gradle-plugin")
     id("kotlin-parcelize")
+    id("com.google.gms.google-services")
+}
+
+// Redirect build output to C: drive (NTFS) to avoid corrupted ExFAT build cache
+if (System.getenv("CI") == null) {
+    layout.buildDirectory = file("C:/GradleBuild/app")
 }
 
 android {
     namespace = "com.google.ai.sample"
     compileSdk = 35
 
     defaultConfig {
-        applicationId = "com.google.ai.sample"
+        applicationId = "io.github.android_poweruser"
         minSdk = 26
         targetSdk = 35
         versionCode = 1
@@ -96,4 +102,14 @@ dependencies {
 
     // Camera Core to potentially fix missing JNI lib issue
     implementation("androidx.camera:camera-core:1.4.0")
+
+    // WebRTC
+    implementation("io.getstream:stream-webrtc-android:1.1.1")
+
+    // WebSocket for signaling
+    implementation("com.squareup.okhttp3:okhttp:4.12.0")
+
+    // Firebase
+    implementation(platform("com.google.firebase:firebase-bom:32.7.2"))
+    implementation("com.google.firebase:firebase-database")
 }

app/src/main/AndroidManifest.xml

@@ -1,6 +1,7 @@
 <?xml version="1.0" encoding="utf-8"?>
 <manifest xmlns:android="http://schemas.android.com/apk/res/android"
     xmlns:tools="http://schemas.android.com/tools">
+        <uses-permission android:name="android.permission.INTERNET" />
         <!-- Storage permissions -->
         <uses-permission android:name="android.permission.READ_EXTERNAL_STORAGE" android:maxSdkVersion="32" />
         <uses-permission android:name="android.permission.WRITE_EXTERNAL_STORAGE" android:maxSdkVersion="28" />

app/src/main/kotlin/com/google/ai/sample/ApiKeyDialog.kt

@@ -88,7 +88,7 @@ fun ApiKeyDialog(
                             ApiProvider.GOOGLE -> "https://makersuite.google.com/app/apikey"
                             ApiProvider.CEREBRAS -> "https://cloud.cerebras.ai/"
                             ApiProvider.VERCEL -> "https://vercel.com/ai-gateway"
-                            ApiProvider.HUMAN_EXPERT -> return@Button // No API key needed
+                            ApiProvider.HUMAN_EXPERT -> return@Button
                         }
                         val intent = Intent(Intent.ACTION_VIEW, Uri.parse(url))
                         context.startActivity(intent)

app/src/main/kotlin/com/google/ai/sample/GenerativeAiViewModelFactory.kt

@@ -9,6 +9,7 @@ import com.google.ai.client.generativeai.GenerativeModel
 import com.google.ai.client.generativeai.type.generationConfig
 import com.google.ai.sample.feature.live.LiveApiManager
 import com.google.ai.sample.feature.multimodal.PhotoReasoningViewModel
+import com.google.ai.sample.util.GenerationSettingsPreferences
 
 // Model options
 enum class ApiProvider {
@@ -44,24 +45,32 @@ enum class ModelOption(
         "https://huggingface.co/na5h13/gemma-3n-E4B-it-litert-lm/resolve/main/gemma-3n-E4B-it-int4.litertlm?download=true",
         "4.92 GB"
     ),
-    HUMAN_EXPERT("Human Expert", "human-expert", ApiProvider.HUMAN_EXPERT)
+    HUMAN_EXPERT("Human Expert", "human-expert", ApiProvider.HUMAN_EXPERT);
+
+    /** Whether this model supports TopK/TopP/Temperature settings */
+    val supportsGenerationSettings: Boolean
+        get() = this != HUMAN_EXPERT
 }
 
 val GenerativeViewModelFactory = object : ViewModelProvider.Factory {
     override fun <T : ViewModel> create(
         viewModelClass: Class<T>,
         extras: CreationExtras
     ): T {
-        val config = generationConfig {
-            temperature = 0.0f
-        }
-
         // Get the application context from extras
         val application = checkNotNull(extras[ViewModelProvider.AndroidViewModelFactory.APPLICATION_KEY])
+        val currentModel = GenerativeAiViewModelFactory.getCurrentModel()
+
+        // Load per-model generation settings
+        val genSettings = GenerationSettingsPreferences.loadSettings(application.applicationContext, currentModel.modelName)
+        val config = generationConfig {
+            temperature = genSettings.temperature
+            topP = genSettings.topP
+            topK = genSettings.topK
+        }
 
         // Get the API key from MainActivity
         val mainActivity = MainActivity.getInstance()
-        val currentModel = GenerativeAiViewModelFactory.getCurrentModel()
         val apiKey = if (currentModel == ModelOption.GEMMA_3N_E4B_IT || currentModel == ModelOption.HUMAN_EXPERT) {
             "offline-no-key-needed" // Dummy key for offline/human expert models
         } else {
@@ -75,8 +84,6 @@ val GenerativeViewModelFactory = object : ViewModelProvider.Factory {
         return with(viewModelClass) {
             when {
                 isAssignableFrom(PhotoReasoningViewModel::class.java) -> {
-                    val currentModel = GenerativeAiViewModelFactory.getCurrentModel()
-
                     if (currentModel.modelName.contains("live")) {
                         // Live API models
                         val liveApiManager = LiveApiManager(apiKey, currentModel.modelName)

app/src/main/kotlin/com/google/ai/sample/MainActivity.kt

@@ -120,12 +120,14 @@ class MainActivity : ComponentActivity() {
     // MediaProjection
     private lateinit var mediaProjectionManager: MediaProjectionManager
     private lateinit var mediaProjectionLauncher: ActivityResultLauncher<Intent>
+    private lateinit var webRtcMediaProjectionLauncher: ActivityResultLauncher<Intent>
 
     private var currentScreenInfoForScreenshot: String? = null
 
     private lateinit var navController: NavHostController
     private var isProcessingExplicitScreenshotRequest: Boolean = false
     private var onMediaProjectionPermissionGranted: (() -> Unit)? = null
+    private var onWebRtcMediaProjectionResult: ((Int, Intent) -> Unit)? = null
 
     private val screenshotRequestHandler = object : BroadcastReceiver() {
         override fun onReceive(context: Context?, intent: Intent?) {
@@ -187,15 +189,28 @@ class MainActivity : ComponentActivity() {
         // This should be guaranteed by its placement in onCreate.
         if (!::mediaProjectionManager.isInitialized) {
             Log.e(TAG, "requestMediaProjectionPermission: mediaProjectionManager not initialized!")
-            // Optionally, initialize it here as a fallback, though it indicates an issue with onCreate ordering
-            // mediaProjectionManager = getSystemService(Context.MEDIA_PROJECTION_SERVICE) as MediaProjectionManager
-            // Toast.makeText(this, "Error: Projection manager not ready. Please try again.", Toast.LENGTH_SHORT).show()
             return
         }
         val intent = mediaProjectionManager.createScreenCaptureIntent()
         mediaProjectionLauncher.launch(intent)
     }
 
+    /**
+     * Request a fresh MediaProjection permission specifically for WebRTC (Human Expert).
+     * This does NOT start ScreenCaptureService - the result is passed directly to the callback.
+     */
+    fun requestMediaProjectionForWebRTC(onResult: (Int, Intent) -> Unit) {
+        Log.d(TAG, "Requesting MediaProjection permission for WebRTC")
+        onWebRtcMediaProjectionResult = onResult
+
+        if (!::mediaProjectionManager.isInitialized) {
+            Log.e(TAG, "requestMediaProjectionForWebRTC: mediaProjectionManager not initialized!")
+            return
+        }
+        val intent = mediaProjectionManager.createScreenCaptureIntent()
+        webRtcMediaProjectionLauncher.launch(intent)
+    }
+
     fun takeAdditionalScreenshot() {
         if (ScreenCaptureService.isRunning()) {
             Log.d(TAG, "MainActivity: Instructing ScreenCaptureService to take an additional screenshot.")
@@ -286,7 +301,7 @@ class MainActivity : ComponentActivity() {
 
         when (currentTrialState) {
             TrialManager.TrialState.EXPIRED_INTERNET_TIME_CONFIRMED -> {
-                trialInfoMessage = "Your 30-minute trial period has ended. Please subscribe to the app to continue using it."
+                trialInfoMessage = "Please support the development of the app so that you can continue using it \uD83C\uDF89"
                 showTrialInfoDialog = true
                 Log.d(TAG, "updateTrialState: Set message to \'$trialInfoMessage\', showTrialInfoDialog = true (EXPIRED)")
             }
@@ -444,6 +459,10 @@ class MainActivity : ComponentActivity() {
                 if (result.resultCode == Activity.RESULT_OK && result.data != null) {
                     val shouldTakeScreenshotOnThisStart = this@MainActivity.isProcessingExplicitScreenshotRequest
                     Log.i(TAG, "MediaProjection permission granted. Starting ScreenCaptureService. Explicit request: $shouldTakeScreenshotOnThisStart")
+
+                    // Notify ViewModel about the permission grant (for Human Expert WebRTC)
+                    photoReasoningViewModel?.onMediaProjectionPermissionGranted(result.resultCode, result.data!!)
+
                     val serviceIntent = Intent(this, ScreenCaptureService::class.java).apply {
                         action = ScreenCaptureService.ACTION_START_CAPTURE
                         putExtra(ScreenCaptureService.EXTRA_RESULT_CODE, result.resultCode)
@@ -487,6 +506,21 @@ class MainActivity : ComponentActivity() {
                 }
             }
 
+            // Separate WebRTC MediaProjection launcher - does NOT start ScreenCaptureService
+            webRtcMediaProjectionLauncher = registerForActivityResult(
+                ActivityResultContracts.StartActivityForResult()
+            ) { result ->
+                if (result.resultCode == Activity.RESULT_OK && result.data != null) {
+                    Log.i(TAG, "WebRTC MediaProjection permission granted.")
+                    onWebRtcMediaProjectionResult?.invoke(result.resultCode, result.data!!)
+                    onWebRtcMediaProjectionResult = null
+                } else {
+                    Log.w(TAG, "WebRTC MediaProjection permission denied.")
+                    Toast.makeText(this, "Screen capture permission denied", Toast.LENGTH_SHORT).show()
+                    onWebRtcMediaProjectionResult = null
+                }
+            }
+
         // Keyboard visibility listener
         val rootView = findViewById<View>(android.R.id.content)
         onGlobalLayoutListener = ViewTreeObserver.OnGlobalLayoutListener {
@@ -1222,7 +1256,7 @@ fun TrialExpiredDialog(
                 )
                 Spacer(modifier = Modifier.height(16.dp))
                 Text(
-                    text = "Your 7-day trial period has ended. Please subscribe to the app to continue using it.",
+                    text = "Please support the development of the app so that you can continue using it \uD83C\uDF89",
                     style = MaterialTheme.typography.bodyMedium,
                     modifier = Modifier.align(Alignment.CenterHorizontally)
                 )