Rely on PDFBox-Android to reduce the size of generated PDFs

app/build.gradle.kts

@@ -60,6 +60,7 @@ dependencies {
     implementation(libs.litert.support)
     implementation(libs.litert.metadata)
     implementation(libs.opencv)
+    implementation(libs.pdfbox)
 
     testImplementation(libs.junit)
     testImplementation(libs.assertj)

app/src/main/java/org/mydomain/myscan/MainActivity.kt

@@ -2,7 +2,6 @@ package org.mydomain.myscan
 
 import android.content.Context
 import android.content.Intent
-import android.graphics.Bitmap
 import android.media.MediaScannerConnection
 import android.os.Bundle
 import android.os.Environment
@@ -32,7 +31,7 @@ class MainActivity : ComponentActivity() {
 
     override fun onCreate(savedInstanceState: Bundle?) {
         super.onCreate(savedInstanceState)
-        initOpenCV()
+        initLibraries()
         val viewModel: MainViewModel by viewModels { MainViewModel.getFactory(this) }
         enableEdgeToEdge()
         setContent {
@@ -68,19 +67,15 @@ class MainActivity : ComponentActivity() {
         viewModel: MainViewModel,
         context: Context
     ): () -> Unit = {
-        val document = viewModel.createPdf()
         val outputDir = File(cacheDir, "pdfs").apply { mkdirs() }
         val outputFile = File(outputDir, "scan_${System.currentTimeMillis()}.pdf")
         var success = true
         try {
-            FileOutputStream(outputFile).use { outputStream ->
+            val fileOutputStream = FileOutputStream(outputFile)
-                document.writeTo(outputStream)
+            viewModel.createPdf(fileOutputStream)
-            }
         } catch (_: IOException) {
             Toast.makeText(context, "Failed to share PDF", Toast.LENGTH_SHORT).show()
             success = false
-        } finally {
-            document.close()
         }
         if (success) {
             val uri = FileProvider.getUriForFile(
@@ -101,13 +96,12 @@ class MainActivity : ComponentActivity() {
         viewModel: MainViewModel,
         context: Context
     ): () -> Unit = {
-        val document = viewModel.createPdf()
         try {
             val downloadsDir = Environment.getExternalStoragePublicDirectory(Environment.DIRECTORY_DOWNLOADS)
             if (!downloadsDir.exists()) downloadsDir.mkdirs()
             val file = File(downloadsDir, "scan_${System.currentTimeMillis()}.pdf")
             val outputStream = FileOutputStream(file)
-            document.writeTo(outputStream)
+            viewModel.createPdf(outputStream)
             outputStream.flush()
             outputStream.close()
 
@@ -119,12 +113,12 @@ class MainActivity : ComponentActivity() {
         } catch (e: Exception) {
             Log.e("MyScan", "Failed to save PDF", e)
             Toast.makeText(context, "Failed to save PDF", Toast.LENGTH_SHORT).show()
-        } finally {
-            document.close()
         }
     }
 
-    private fun initOpenCV() {
+    private fun initLibraries() {
+        com.tom_roush.pdfbox.android.PDFBoxResourceLoader.init(applicationContext)
+
         if (!OpenCVLoader.initLocal()) {
             Log.e("OpenCV", "Initialization failed")
         } else {

app/src/main/java/org/mydomain/myscan/MainViewModel.kt

@@ -19,6 +19,7 @@ import kotlinx.coroutines.flow.map
 import kotlinx.coroutines.launch
 import kotlinx.coroutines.withContext
 import java.io.ByteArrayOutputStream
+import java.io.OutputStream
 
 class MainViewModel(
     private val imageSegmentationService: ImageSegmentationService,
@@ -128,9 +129,9 @@ class MainViewModel(
         return BitmapFactory.decodeByteArray(bytes, 0, bytes.size)
     }
 
-    fun createPdf(): PdfDocument {
+    fun createPdf(outputStream: OutputStream) {
         val jpegs = imageRepository.imageIds().asSequence()
             .map { id -> imageRepository.getContent(id) }
-        return createPdfFromJpegs(jpegs)
+        writePdfFromJpegs(jpegs, outputStream)
     }
 }

app/src/main/java/org/mydomain/myscan/PdfGeneration.kt

@@ -1,21 +1,28 @@
 package org.mydomain.myscan
 
 import android.graphics.Bitmap
-import android.graphics.BitmapFactory
-import android.graphics.pdf.PdfDocument
 import androidx.core.graphics.scale
+import com.tom_roush.pdfbox.pdmodel.PDDocument
+import com.tom_roush.pdfbox.pdmodel.PDPage
+import com.tom_roush.pdfbox.pdmodel.PDPageContentStream
+import com.tom_roush.pdfbox.pdmodel.PDPageContentStream.AppendMode
+import com.tom_roush.pdfbox.pdmodel.common.PDRectangle
+import com.tom_roush.pdfbox.pdmodel.graphics.image.JPEGFactory
+import java.io.OutputStream
 import kotlin.math.max
 
-fun createPdfFromJpegs (jpegs: Sequence<ByteArray>): PdfDocument {
+fun writePdfFromJpegs(jpegs: Sequence<ByteArray>, outputStream: OutputStream) {
-    val document = PdfDocument()
+    PDDocument().use { document ->
-    for ((index, jpegBytes) in jpegs.withIndex()) {
+        for (jpegBytes in jpegs) {
-        val bitmap = BitmapFactory.decodeByteArray(jpegBytes, 0, jpegBytes.size)
+            val image = JPEGFactory.createFromByteArray(document, jpegBytes)
-        val pageInfo = PdfDocument.PageInfo.Builder(bitmap.width, bitmap.height, index + 1).create()
+            val page = PDPage(PDRectangle(image.width.toFloat(), image.height.toFloat()))
-        val page = document.startPage(pageInfo)
+            document.addPage(page)
-        page.canvas.drawBitmap(bitmap, 0f, 0f, null)
+            val contentStream = PDPageContentStream(document, page, AppendMode.OVERWRITE, false)
-        document.finishPage(page)
+            contentStream.drawImage(image, 0f, 0f)
+            contentStream.close()
+        }
+        document.save(outputStream)
     }
-    return document
 }
 
 fun resizeImage(original: Bitmap): Bitmap {

gradle/libs.versions.toml

@@ -13,6 +13,7 @@ litert = "1.2.0"
 opencv = "4.11.0"
 flowlayout =  "0.36.0"
 assertj = "3.27.3"
+pdfbox = "2.0.27.0"
 
 [libraries]
 androidx-core-ktx = { group = "androidx.core", name = "core-ktx", version.ref = "coreKtx" }
@@ -41,6 +42,7 @@ litert = { group = "com.google.ai.edge.litert", name = "litert", version.ref = "
 litert-support = { group = "com.google.ai.edge.litert", name = "litert-support", version.ref = "litert" }
 litert-metadata = { group = "com.google.ai.edge.litert", name = "litert-metadata", version.ref = "litert" }
 opencv = { group="org.opencv", name="opencv", version.ref = "opencv" }
+pdfbox = { group = "com.tom-roush", name = "pdfbox-android", version.ref = "pdfbox" }
 
 assertj =  { group="org.assertj", name="assertj-core", version.ref = "assertj" }