fix OCR in apx

android/app/build.gradle.kts

@@ -34,6 +34,10 @@ android {
             // TODO: Add your own signing config for the release build.
             // Signing with the debug keys for now, so `flutter run --release` works.
             signingConfig = signingConfigs.getByName("debug")
+            // Flutter turns R8 on for release. That strips ML Kit classes
+            // discovered via manifest metadata / JNI, so OCR fails in the APK.
+            isMinifyEnabled = false
+            isShrinkResources = false
             proguardFiles(
                 getDefaultProguardFile("proguard-android-optimize.txt"),
                 "proguard-rules.pro",

android/app/proguard-rules.pro

@@ -1,10 +1,15 @@
-# ML Kit text recognition ships Latin only; the plugin still references
-# optional script packs that are not in this app.
--dontwarn com.google.mlkit.vision.text.chinese.ChineseTextRecognizerOptions$Builder
--dontwarn com.google.mlkit.vision.text.chinese.ChineseTextRecognizerOptions
--dontwarn com.google.mlkit.vision.text.devanagari.DevanagariTextRecognizerOptions$Builder
--dontwarn com.google.mlkit.vision.text.devanagari.DevanagariTextRecognizerOptions
--dontwarn com.google.mlkit.vision.text.japanese.JapaneseTextRecognizerOptions$Builder
--dontwarn com.google.mlkit.vision.text.japanese.JapaneseTextRecognizerOptions
--dontwarn com.google.mlkit.vision.text.korean.KoreanTextRecognizerOptions$Builder
--dontwarn com.google.mlkit.vision.text.korean.KoreanTextRecognizerOptions
+# ML Kit text recognition is wired up through manifest metadata, reflection,
+# and JNI. R8 otherwise strips those classes and OCR fails in release APKs.
+-keep class com.google.mlkit.** { *; }
+-keep class com.google.android.gms.internal.mlkit_** { *; }
+-keep class com.google.android.gms.internal.mlkit_vision_** { *; }
+-keep class com.google.android.libraries.vision.** { *; }
+-keep class com.google_mlkit_commons.** { *; }
+-keep class com.google_mlkit_text_recognition.** { *; }
+-keep class * implements com.google.firebase.components.ComponentRegistrar { *; }
+
+# The Flutter plugin references optional script packs that this app does not ship.
+-dontwarn com.google.mlkit.vision.text.chinese.**
+-dontwarn com.google.mlkit.vision.text.devanagari.**
+-dontwarn com.google.mlkit.vision.text.japanese.**
+-dontwarn com.google.mlkit.vision.text.korean.**

android/gradle.properties

@@ -1,5 +1,7 @@
 org.gradle.jvmargs=-Xmx8G -XX:MaxMetaspaceSize=4G -XX:ReservedCodeCacheSize=512m -XX:+HeapDumpOnOutOfMemoryError
 android.useAndroidX=true
+# Flutter enables R8 for release by default. That breaks on-device ML Kit OCR.
+shrink=false
 # This newDsl flag was added by the Flutter template
 android.newDsl=false
 # This builtInKotlin flag was added by the Flutter template

lib/scan_screen.dart

@@ -103,10 +103,17 @@ class _ScanScreenState extends State<ScanScreen> {
           _error = 'No text found in this image.';
         }
       });
-    } catch (error) {
+    } on PlatformException catch (error) {
+      debugPrint('OCR failed: ${error.code} ${error.message}');
       if (!mounted) return;
       setState(() {
-        _error = 'Could not read text from this image.';
+        _error = _ocrFailureMessage(error.message);
+      });
+    } catch (error) {
+      debugPrint('OCR failed: $error');
+      if (!mounted) return;
+      setState(() {
+        _error = _ocrFailureMessage(error.toString());
       });
     } finally {
       await recognizer.close();
@@ -120,6 +127,12 @@ class _ScanScreenState extends State<ScanScreen> {
     }
   }
 
+  String _ocrFailureMessage(String? detail) {
+    const fallback = 'Could not read text from this image.';
+    if (detail == null || detail.trim().isEmpty) return fallback;
+    return '$fallback\n$detail';
+  }
+
   Future<void> _openReview(String rawText) async {
     final items = ReceiptParser.parse(rawText);
     if (!mounted) return;