OCR

build.gradle

@@ -71,6 +71,8 @@ dependencies {
     compile("org.glassfish.jaxb:jaxb-runtime:2.3.2")
     compile("commons-io:commons-io:2.6")
     compile("org.apache.tika:tika-parsers:1.24")
+    compile("org.apache.pdfbox:pdfbox:2.0.1")
+    compile("net.sourceforge.tess4j:tess4j:3.4.0")
 
     // Remove devtools for release
     runtime('org.springframework.boot:spring-boot-devtools')

gradle/wrapper/gradle-wrapper.properties

@@ -1,5 +1,6 @@
+#Mon May 04 21:03:43 CEST 2020
+distributionUrl=https\://services.gradle.org/distributions/gradle-4.10.2-all.zip
 distributionBase=GRADLE_USER_HOME
 distributionPath=wrapper/dists
-distributionUrl=https\://services.gradle.org/distributions/gradle-4.10.2-bin.zip
-zipStoreBase=GRADLE_USER_HOME
 zipStorePath=wrapper/dists
+zipStoreBase=GRADLE_USER_HOME

src/main/kotlin/pl/edu/uj/ii/ksi/mordor/configuration/TesseractConfiguration.kt

@@ -0,0 +1,16 @@
+package pl.edu.uj.ii.ksi.mordor.configuration
+
+import net.sourceforge.tess4j.Tesseract
+import org.springframework.context.annotation.Bean
+import org.springframework.context.annotation.Configuration
+
+@Configuration
+class TesseractConfiguration {
+    @Bean
+    fun tesseract(): Tesseract {
+        val tesseract = Tesseract()
+        tesseract.setDatapath("./src/main/resources/")
+        tesseract.setLanguage("pol+eng")
+        return tesseract
+    }
+}

src/main/kotlin/pl/edu/uj/ii/ksi/mordor/services/AutoDetectTextExtractor.kt

@@ -0,0 +1,49 @@
+package pl.edu.uj.ii.ksi.mordor.services
+
+import java.io.File
+import java.io.IOException
+import net.sourceforge.tess4j.Tesseract
+import net.sourceforge.tess4j.TesseractException
+import org.apache.tika.Tika
+import org.slf4j.LoggerFactory
+import org.springframework.stereotype.Service
+import pl.edu.uj.ii.ksi.mordor.services.repository.RepositoryService
+
+@Service
+class AutoDetectTextExtractor(
+    private val tika: Tika,
+    private val tesseract: Tesseract
+) : FileTextExtractor {
+
+    companion object {
+        private val logger = LoggerFactory.getLogger(RepositoryService::class.java)
+    }
+
+    override fun extract(file: File, maxLength: Int): String? {
+        try {
+            val tikaContent = TikaFileTextExtractor(tika).extract(file)
+            if (!isScanned(tikaContent)) {
+                return tikaContent
+            }
+            val type = tika.detect(file)
+            if (type == "application/pdf") {
+                return PDFTextExtractor(tesseract).extract(file)
+            }
+            if (type.startsWith("image")) {
+                return ImageTextExtractor(tesseract).extract(file)
+            }
+        } catch (e: TesseractException) {
+            logger.error("Tesseract is unable do process OCR", e)
+        } catch (e: IOException) {
+            logger.error("File can not be read", e)
+        }
+        return null
+    }
+
+    private fun isScanned(content: String?): Boolean {
+        if (content == null) {
+            return true
+        }
+        return content.trim().isEmpty()
+    }
+}

src/main/kotlin/pl/edu/uj/ii/ksi/mordor/services/FileEntryCreator.kt

@@ -2,6 +2,7 @@ package pl.edu.uj.ii.ksi.mordor.services
 
 import java.io.File
 import javax.persistence.EntityManager
+import org.springframework.beans.factory.annotation.Qualifier
 import org.springframework.stereotype.Service
 import org.springframework.transaction.annotation.Transactional
 import pl.edu.uj.ii.ksi.mordor.persistence.entities.FileContent
@@ -14,7 +15,7 @@ import pl.edu.uj.ii.ksi.mordor.services.hash.FileHashProvider
 class FileEntryCreator(
     private val metadataExtractor: MetadataExtractor,
     private val entityManager: EntityManager,
-    private val fileTextExtractor: FileTextExtractor,
+    @Qualifier("autoDetectTextExtractor") private val fileTextExtractor: FileTextExtractor,
     private val hashProvider: FileHashProvider,
     private val metadataRepository: FileMetadataRepository
 ) {

src/main/kotlin/pl/edu/uj/ii/ksi/mordor/services/ImageTextExtractor.kt

@@ -0,0 +1,40 @@
+package pl.edu.uj.ii.ksi.mordor.services
+
+import com.recognition.software.jdeskew.ImageDeskew
+import java.awt.image.BufferedImage
+import java.io.File
+import javax.imageio.ImageIO
+import net.sourceforge.tess4j.Tesseract
+import net.sourceforge.tess4j.TesseractException
+import net.sourceforge.tess4j.util.ImageHelper
+import org.slf4j.LoggerFactory
+import org.springframework.stereotype.Service
+
+@Service
+class ImageTextExtractor(private val tesseract: Tesseract) : FileTextExtractor {
+
+    val maxSkewAngle = 0.05
+
+    override fun extract(file: File, maxLength: Int): String? {
+        return extractTextFromBufferedImage(ImageIO.read(file))
+    }
+
+    private fun correctTwisted(image: BufferedImage?): BufferedImage? {
+        val imageSkewAngle = ImageDeskew(image).skewAngle
+        if (kotlin.math.abs(imageSkewAngle) > maxSkewAngle) {
+            return ImageHelper.rotateImage(ImageHelper.convertImageToGrayscale(image), -imageSkewAngle)
+        }
+        return ImageHelper.convertImageToGrayscale(image)
+    }
+
+    fun extractTextFromBufferedImage(image: BufferedImage?): String? {
+        try {
+            return tesseract.doOCR(correctTwisted(image))
+                    .replace("\\n{2,}", "\n")
+                    .trim { c -> c <= ' ' }
+        } catch (e: TesseractException) {
+            LoggerFactory.getLogger(this.javaClass).error("Tesseract is unable do process OCR on image", e)
+        }
+        return null
+    }
+}

src/main/kotlin/pl/edu/uj/ii/ksi/mordor/services/PDFTextExtractor.kt

@@ -0,0 +1,44 @@
+package pl.edu.uj.ii.ksi.mordor.services
+
+import java.awt.image.BufferedImage
+import java.io.File
+import java.util.LinkedList
+import net.sourceforge.tess4j.Tesseract
+import org.apache.pdfbox.pdmodel.PDDocument
+import org.apache.pdfbox.pdmodel.graphics.image.PDImageXObject
+import org.springframework.stereotype.Service
+
+@Service
+class PDFTextExtractor(private val tesseract: Tesseract) : FileTextExtractor {
+
+    override fun extract(file: File, maxLength: Int): String? {
+        val extracted = StringBuilder()
+        val bufferedImages = formatPDF(file)
+
+        if (!bufferedImages.isEmpty()) {
+            for (image in bufferedImages) {
+                val text: String? = ImageTextExtractor(tesseract).extractTextFromBufferedImage(image)
+                if (text != null) {
+                    extracted.append(text)
+                }
+            }
+        }
+        return extracted.toString()
+    }
+
+    private fun formatPDF(pdfFile: File): LinkedList<BufferedImage?> {
+        val bufferedImages = LinkedList<BufferedImage?>()
+        val doc: PDDocument = PDDocument.load(pdfFile)
+        for (page in doc.pages) {
+            val resources = page.resources
+            for (xObjectName in resources.xObjectNames) {
+                val xObject = resources.getXObject(xObjectName)
+                if (xObject is PDImageXObject) {
+                    bufferedImages.add(xObject.image)
+                }
+            }
+        }
+        doc.close()
+        return bufferedImages
+    }
+}