Compare commits

..
Author SHA1 Message Date
Kilian Schuettler 90b4869761 RED-6126: In the OCRService, OCR Text is not applied to Document
*removed init/terminate calls again
*added destroy()/close() at every opportunity
2023-02-24 13:07:42 +01:00
30 changed files with 150 additions and 511 deletions
@@ -1,4 +1,4 @@
FROM red/base-image:2.0.2
FROM red/base-image:2.0.0
COPY "libs/pdftron/OCRModuleLinux.tar.gz" .
RUN tar xvzf OCRModuleLinux.tar.gz
+20 -2
View File
@@ -14,7 +14,8 @@
<version>1.0-SNAPSHOT</version>
<properties>
<persistence-service.version>2.1.0</persistence-service.version>
<persistence-service.version>1.269.0</persistence-service.version>
<redaction-service.version>3.155.0</redaction-service.version>
<dsljson.version>1.9.9</dsljson.version>
</properties>
@@ -42,9 +43,26 @@
<dependency>
<groupId>com.iqser.red.service</groupId>
<artifactId>persistence-service-internal-api-v1</artifactId>
<artifactId>persistence-service-api-v1</artifactId>
<exclusions>
<exclusion>
<groupId>com.iqser.red.service</groupId>
<artifactId>redaction-service-api-v1</artifactId>
</exclusion>
</exclusions>
<version>${persistence-service.version}</version>
</dependency>
<dependency>
<groupId>com.iqser.red.service</groupId>
<artifactId>redaction-service-api-v1</artifactId>
<exclusions>
<exclusion>
<groupId>com.iqser.red.service</groupId>
<artifactId>persistence-service-api-v1</artifactId>
</exclusion>
</exclusions>
<version>${redaction-service.version}</version>
</dependency>
</dependencies>
</project>
@@ -1,17 +0,0 @@
package com.iqser.red.service.ocr.v1.api.model;
import static org.assertj.core.api.Assertions.assertThat;
import org.junit.jupiter.api.Test;
public class IdentityTest {
@Test
public void mockTest() {
int i = 1;
assertThat(i).isEqualTo(1);
}
}
@@ -10,11 +10,9 @@ import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Import;
import org.springframework.scheduling.annotation.EnableAsync;
import com.iqser.red.commons.spring.DefaultWebMvcConfiguration;
import com.iqser.red.service.ocr.v1.server.client.FileStatusProcessingUpdateClient;
import com.iqser.red.service.ocr.v1.server.configuration.MessagingConfiguration;
import com.iqser.red.service.ocr.v1.server.multitenancy.AsyncConfig;
import com.iqser.red.service.ocr.v1.server.multitenancy.MultiTenancyMessagingConfiguration;
import com.iqser.red.service.ocr.v1.server.multitenancy.MultiTenancyWebConfiguration;
import com.iqser.red.service.ocr.v1.server.settings.OcrServiceSettings;
import io.micrometer.core.aop.TimedAspect;
@@ -23,7 +21,7 @@ import io.micrometer.core.instrument.MeterRegistry;
@EnableAsync
@EnableConfigurationProperties(OcrServiceSettings.class)
@SpringBootApplication(exclude = {SecurityAutoConfiguration.class, ManagementWebSecurityAutoConfiguration.class})
@Import({MultiTenancyWebConfiguration.class, AsyncConfig.class, MessagingConfiguration.class, MultiTenancyMessagingConfiguration.class})
@Import({DefaultWebMvcConfiguration.class, MessagingConfiguration.class})
@EnableFeignClients(basePackageClasses = FileStatusProcessingUpdateClient.class)
public class Application {
@@ -2,7 +2,7 @@ package com.iqser.red.service.ocr.v1.server.client;
import org.springframework.cloud.openfeign.FeignClient;
import com.iqser.red.service.persistence.service.v1.api.internal.resources.FileStatusProcessingUpdateResource;
import com.iqser.red.service.persistence.service.v1.api.resources.FileStatusProcessingUpdateResource;
@FeignClient(name = "FileStatusProcessingUpdateResource", url = "${persistence-service.url}")
public interface FileStatusProcessingUpdateClient extends FileStatusProcessingUpdateResource {
@@ -1,10 +0,0 @@
package com.iqser.red.service.ocr.v1.server.client;
import org.springframework.cloud.openfeign.FeignClient;
import com.iqser.red.service.persistence.service.v1.api.internal.resources.TenantsResource;
@FeignClient(name = "TenantsResource", url = "${persistence-service.url}")
public interface TenantsClient extends TenantsResource {
}
@@ -1,7 +1,6 @@
package com.iqser.red.service.ocr.v1.server.model;
import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.Rectangle;
import com.iqser.red.service.redaction.v1.model.Rectangle;
import lombok.AllArgsConstructor;
import lombok.Data;
@@ -1,27 +0,0 @@
package com.iqser.red.service.ocr.v1.server.multitenancy;
import java.util.concurrent.Executor;
import org.springframework.context.annotation.Configuration;
import org.springframework.scheduling.annotation.AsyncConfigurerSupport;
import org.springframework.scheduling.concurrent.ThreadPoolTaskExecutor;
@Configuration
public class AsyncConfig extends AsyncConfigurerSupport {
@Override
public Executor getAsyncExecutor() {
ThreadPoolTaskExecutor executor = new ThreadPoolTaskExecutor();
executor.setCorePoolSize(7);
executor.setMaxPoolSize(42);
executor.setQueueCapacity(11);
executor.setThreadNamePrefix("TenantAwareTaskExecutor-");
executor.setTaskDecorator(new TenantAwareTaskDecorator());
executor.initialize();
return executor;
}
}
@@ -1,105 +0,0 @@
package com.iqser.red.service.ocr.v1.server.multitenancy;
import java.nio.ByteBuffer;
import java.nio.charset.StandardCharsets;
import java.security.SecureRandom;
import java.security.spec.KeySpec;
import java.util.Base64;
import javax.annotation.PostConstruct;
import javax.crypto.Cipher;
import javax.crypto.SecretKey;
import javax.crypto.SecretKeyFactory;
import javax.crypto.spec.GCMParameterSpec;
import javax.crypto.spec.PBEKeySpec;
import javax.crypto.spec.SecretKeySpec;
import org.springframework.beans.factory.annotation.Value;
import org.springframework.stereotype.Service;
import lombok.SneakyThrows;
@Service
public class EncryptionDecryptionService {
@Value("${ocr-service.crypto.key:redaction}")
private String key;
private SecretKey secretKey;
private byte[] iv;
@SneakyThrows
@PostConstruct
protected void postConstruct() {
SecureRandom secureRandom = new SecureRandom();
iv = new byte[12];
secureRandom.nextBytes(iv);
secretKey = generateSecretKey(key, iv);
}
@SneakyThrows
public String encrypt(String strToEncrypt) {
return Base64.getEncoder().encodeToString(encrypt(strToEncrypt.getBytes()));
}
@SneakyThrows
public String decrypt(String strToDecrypt) {
byte[] bytes = Base64.getDecoder().decode(strToDecrypt);
return new String(decrypt(bytes), StandardCharsets.UTF_8);
}
@SneakyThrows
public byte[] encrypt(byte[] data) {
Cipher cipher = Cipher.getInstance("AES/GCM/NoPadding");
GCMParameterSpec parameterSpec = new GCMParameterSpec(128, iv);
cipher.init(Cipher.ENCRYPT_MODE, secretKey, parameterSpec);
byte[] encryptedData = cipher.doFinal(data);
ByteBuffer byteBuffer = ByteBuffer.allocate(4 + iv.length + encryptedData.length);
byteBuffer.putInt(iv.length);
byteBuffer.put(iv);
byteBuffer.put(encryptedData);
return byteBuffer.array();
}
@SneakyThrows
public byte[] decrypt(byte[] encryptedData) {
ByteBuffer byteBuffer = ByteBuffer.wrap(encryptedData);
int noonceSize = byteBuffer.getInt();
if (noonceSize < 12 || noonceSize >= 16) {
throw new IllegalArgumentException("Nonce size is incorrect. Make sure that the incoming data is an AES encrypted file.");
}
byte[] iv = new byte[noonceSize];
byteBuffer.get(iv);
SecretKey secretKey = generateSecretKey(key, iv);
byte[] cipherBytes = new byte[byteBuffer.remaining()];
byteBuffer.get(cipherBytes);
Cipher cipher = Cipher.getInstance("AES/GCM/NoPadding");
GCMParameterSpec parameterSpec = new GCMParameterSpec(128, iv);
cipher.init(Cipher.DECRYPT_MODE, secretKey, parameterSpec);
return cipher.doFinal(cipherBytes);
}
@SneakyThrows
public SecretKey generateSecretKey(String password, byte[] iv) {
KeySpec spec = new PBEKeySpec(password.toCharArray(), iv, 65536, 128); // AES-128
SecretKeyFactory secretKeyFactory = SecretKeyFactory.getInstance("PBKDF2WithHmacSHA1");
byte[] key = secretKeyFactory.generateSecret(spec).getEncoded();
return new SecretKeySpec(key, "AES");
}
}
@@ -1,17 +0,0 @@
package com.iqser.red.service.ocr.v1.server.multitenancy;
import org.springframework.stereotype.Component;
import feign.RequestInterceptor;
import feign.RequestTemplate;
@Component
public class ForwardTenantInterceptor implements RequestInterceptor {
public static final String TENANT_HEADER_NAME = "X-TENANT-ID";
@Override
public void apply(RequestTemplate template) {
template.header(TENANT_HEADER_NAME, TenantContext.getTenantId());
}
}
@@ -1,48 +0,0 @@
package com.iqser.red.service.ocr.v1.server.multitenancy;
import static com.iqser.red.service.ocr.v1.server.multitenancy.ForwardTenantInterceptor.TENANT_HEADER_NAME;
import org.springframework.amqp.rabbit.config.AbstractRabbitListenerContainerFactory;
import org.springframework.amqp.rabbit.core.RabbitTemplate;
import org.springframework.beans.BeansException;
import org.springframework.beans.factory.config.BeanPostProcessor;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
@Configuration
public class MultiTenancyMessagingConfiguration {
@Bean
public static BeanPostProcessor multitenancyBeanPostProcessor() {
return new BeanPostProcessor() {
@Override
public Object postProcessAfterInitialization(Object bean, String beanName) throws BeansException {
if (bean instanceof RabbitTemplate) {
((RabbitTemplate) bean).setBeforePublishPostProcessors(m -> {
m.getMessageProperties().setHeader(TENANT_HEADER_NAME, TenantContext.getTenantId());
return m;
});
} else if (bean instanceof AbstractRabbitListenerContainerFactory) {
((AbstractRabbitListenerContainerFactory<?>) bean).setAfterReceivePostProcessors(m -> {
String tenant = m.getMessageProperties().getHeader(TENANT_HEADER_NAME);
if (tenant != null) {
TenantContext.setTenantId(tenant);
} else {
throw new RuntimeException("No Tenant is set queue message");
}
return m;
});
}
return bean;
}
};
}
}
@@ -1,28 +0,0 @@
package com.iqser.red.service.ocr.v1.server.multitenancy;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.context.annotation.Configuration;
import org.springframework.web.servlet.config.annotation.InterceptorRegistry;
import com.iqser.red.commons.spring.DefaultWebMvcConfiguration;
@Configuration
public class MultiTenancyWebConfiguration extends DefaultWebMvcConfiguration {
private final TenantInterceptor tenantInterceptor;
@Autowired
public MultiTenancyWebConfiguration(TenantInterceptor tenantInterceptor) {
this.tenantInterceptor = tenantInterceptor;
}
@Override
public void addInterceptors(InterceptorRegistry registry) {
registry.addWebRequestInterceptor(tenantInterceptor);
}
}
@@ -1,45 +0,0 @@
package com.iqser.red.service.ocr.v1.server.multitenancy;
import org.springframework.stereotype.Service;
import com.iqser.red.service.ocr.v1.server.client.TenantsClient;
import com.iqser.red.storage.commons.model.AzureStorageConnection;
import com.iqser.red.storage.commons.model.S3StorageConnection;
import com.iqser.red.storage.commons.service.StorageConnectionProvider;
import lombok.RequiredArgsConstructor;
@Service
@RequiredArgsConstructor
public class StorageConnectionProviderImpl implements StorageConnectionProvider {
private final TenantsClient tenantsClient;
private final EncryptionDecryptionService encryptionDecryptionService;
@Override
public AzureStorageConnection getAzureStorageConnection(String tenantId) {
var tenant = tenantsClient.getTenant(tenantId);
return AzureStorageConnection.builder()
.connectionString(encryptionDecryptionService.decrypt(tenant.getAzureStorageConnection().getConnectionString()))
.containerName(tenant.getAzureStorageConnection().getContainerName())
.build();
}
@Override
public S3StorageConnection getS3StorageConnection(String tenantId) {
var tenant = tenantsClient.getTenant(tenantId);
return S3StorageConnection.builder()
.key(tenant.getS3StorageConnection().getKey())
.secret(encryptionDecryptionService.decrypt(tenant.getS3StorageConnection().getSecret()))
.signerType(tenant.getS3StorageConnection().getSignerType())
.bucketName(tenant.getS3StorageConnection().getBucketName())
.region(tenant.getS3StorageConnection().getRegion())
.endpoint(tenant.getS3StorageConnection().getEndpoint())
.build();
}
}
@@ -1,23 +0,0 @@
package com.iqser.red.service.ocr.v1.server.multitenancy;
import org.springframework.core.task.TaskDecorator;
import org.springframework.lang.NonNull;
public class TenantAwareTaskDecorator implements TaskDecorator {
@Override
@NonNull
public Runnable decorate(@NonNull Runnable runnable) {
String tenantId = TenantContext.getTenantId();
return () -> {
try {
TenantContext.setTenantId(tenantId);
runnable.run();
} finally {
TenantContext.setTenantId(null);
}
};
}
}
@@ -1,29 +0,0 @@
package com.iqser.red.service.ocr.v1.server.multitenancy;
import lombok.extern.slf4j.Slf4j;
@Slf4j
public final class TenantContext {
private static InheritableThreadLocal<String> currentTenant = new InheritableThreadLocal<>();
public static void setTenantId(String tenantId) {
log.debug("Setting tenantId to " + tenantId);
currentTenant.set(tenantId);
}
public static String getTenantId() {
return currentTenant.get();
}
public static void clear() {
currentTenant.remove();
}
}
@@ -1,35 +0,0 @@
package com.iqser.red.service.ocr.v1.server.multitenancy;
import org.springframework.stereotype.Component;
import org.springframework.ui.ModelMap;
import org.springframework.web.context.request.WebRequest;
import org.springframework.web.context.request.WebRequestInterceptor;
@Component
public class TenantInterceptor implements WebRequestInterceptor {
public static final String TENANT_HEADER_NAME = "X-TENANT-ID";
@Override
public void preHandle(WebRequest request) {
if (request.getHeader(TENANT_HEADER_NAME) != null) {
TenantContext.setTenantId(request.getHeader(TENANT_HEADER_NAME));
}
}
@Override
public void postHandle(WebRequest request, ModelMap model) {
TenantContext.clear();
}
@Override
public void afterCompletion(WebRequest request, Exception ex) {
}
}
@@ -0,0 +1,2 @@
package com.iqser.red.service.ocr.v1.server.service;public class DataExtractionService {
}
@@ -7,8 +7,7 @@ import org.apache.commons.io.IOUtils;
import org.springframework.stereotype.Service;
import com.iqser.red.service.ocr.v1.server.model.image.ImageServiceResponse;
import com.iqser.red.service.ocr.v1.server.multitenancy.TenantContext;
import com.iqser.red.service.persistence.service.v1.api.shared.model.dossiertemplate.dossier.file.FileType;
import com.iqser.red.service.persistence.service.v1.api.model.dossiertemplate.dossier.file.FileType;
import com.iqser.red.storage.commons.service.StorageService;
import lombok.RequiredArgsConstructor;
@@ -32,39 +31,39 @@ public class FileStorageService {
@SneakyThrows
public byte[] getOriginalFile(String dossierId, String fileId) {
return IOUtils.toByteArray(storageService.getObject(TenantContext.getTenantId(), getStorageId(dossierId, fileId, FileType.ORIGIN)).getInputStream());
return IOUtils.toByteArray(storageService.getObject(getStorageId(dossierId, fileId, FileType.ORIGIN)).getInputStream());
}
@SneakyThrows
public InputStream getOriginalFileAsStream(String dossierId, String fileId) {
return storageService.getObject(TenantContext.getTenantId(), getStorageId(dossierId, fileId, FileType.ORIGIN)).getInputStream();
return storageService.getObject(getStorageId(dossierId, fileId, FileType.ORIGIN)).getInputStream();
}
public void storeOriginalFile(String dossierId, String fileId, InputStream stream) {
storageService.storeObject(TenantContext.getTenantId(), getStorageId(dossierId, fileId, FileType.ORIGIN), stream);
storageService.storeObject(getStorageId(dossierId, fileId, FileType.ORIGIN), stream);
}
public boolean untouchedFileExists(String dossierId, String fileId) {
return storageService.objectExists(TenantContext.getTenantId(), getStorageId(dossierId, fileId, FileType.UNTOUCHED));
return storageService.objectExists(getStorageId(dossierId, fileId, FileType.UNTOUCHED));
}
public void storeUntouchedFile(String dossierId, String fileId, byte[] data) {
storageService.storeObject(TenantContext.getTenantId(), getStorageId(dossierId, fileId, FileType.UNTOUCHED), new ByteArrayInputStream(data));
storageService.storeObject(getStorageId(dossierId, fileId, FileType.UNTOUCHED), new ByteArrayInputStream(data));
}
@SneakyThrows
public ImageServiceResponse getImageServiceResponse(String dossierId, String fileId) {
return storageService.readJSONObject(TenantContext.getTenantId(), getStorageId(dossierId, fileId, FileType.IMAGE_INFO), ImageServiceResponse.class);
return storageService.readJSONObject(getStorageId(dossierId, fileId, FileType.IMAGE_INFO), ImageServiceResponse.class);
}
}
@@ -87,13 +87,17 @@ public class InvisibleElementRemovalService {
.visitedXObjIds(visitedXObjIds)
.build();
removeClippedElementsAndInvisibleTextAndRememberOverlappedElements(page, writer, context);
context.visitedXObjIds().clear();
removeOverlappedElements(page, writer, context);
reader.end();
writer.end();
}
try {
pdfDoc.save(out, SDFDoc.SaveMode.LINEARIZED, null);
} catch (Exception e) {
@@ -226,7 +230,6 @@ public class InvisibleElementRemovalService {
formWriter.setDefaultGState(context.reader());
processElements(formWriter, context);
formWriter.end();
formWriter.destroy();
context.reader().end();
}
@@ -234,7 +237,6 @@ public class InvisibleElementRemovalService {
private void processPath(Element pathElement, ElementWriter writer, InvisibleElementRemovalContext context) throws PDFNetException {
PathData pathData = pathElement.getPathData();
if (pathData.getOperators().length == 0 && pathData.getPoints().length == 0) {
@@ -364,7 +366,6 @@ public class InvisibleElementRemovalService {
formWriter.setDefaultGState(context.reader());
processOverlappedElements(formWriter, context);
formWriter.end();
formWriter.destroy();
context.reader().end();
}
@@ -389,7 +390,8 @@ public class InvisibleElementRemovalService {
switch (operator) {
case PathData.e_moveto -> linePath.moveTo(points.next(), points.next());
case PathData.e_lineto -> linePath.lineTo(points.next(), points.next());
case PathData.e_cubicto -> linePath.curveTo(points.next(), points.next(), points.next(), points.next(), points.next(), points.next());
case PathData.e_cubicto ->
linePath.curveTo(points.next(), points.next(), points.next(), points.next(), points.next(), points.next());
case PathData.e_closepath -> linePath.closePath();
case PathData.e_rect -> {
double x = points.next();
@@ -1,35 +1,23 @@
package com.iqser.red.service.ocr.v1.server.service;
import static java.lang.String.format;
import java.io.ByteArrayInputStream;
import java.io.ByteArrayOutputStream;
import java.io.IOException;
import java.io.InputStream;
import java.io.OutputStream;
import java.util.HashMap;
import java.util.Map;
import org.springframework.amqp.rabbit.core.RabbitTemplate;
import org.springframework.stereotype.Service;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.iqser.red.service.ocr.v1.api.model.OCRStatusUpdateResponse;
import com.iqser.red.service.ocr.v1.server.configuration.MessagingConfiguration;
import com.iqser.red.service.ocr.v1.server.settings.OcrServiceSettings;
import com.pdftron.common.PDFNetException;
import com.pdftron.pdf.OCRModule;
import com.pdftron.pdf.OCROptions;
import com.pdftron.pdf.Optimizer;
import com.pdftron.pdf.PDFDoc;
import com.pdftron.pdf.Page;
import com.pdftron.pdf.RectCollection;
import com.pdftron.pdf.*;
import com.pdftron.sdf.SDFDoc;
import io.micrometer.core.annotation.Timed;
import lombok.RequiredArgsConstructor;
import lombok.SneakyThrows;
import lombok.extern.slf4j.Slf4j;
import org.springframework.amqp.rabbit.core.RabbitTemplate;
import org.springframework.stereotype.Service;
import java.io.*;
import java.util.Map;
import static java.lang.String.format;
@Slf4j
@Service
@@ -72,10 +60,7 @@ public class OCRService {
log.debug("Start invisible element removal for file with dossierId {} and fileId {}", dossierId, fileId);
invisibleElementRemovalService.removeInvisibleElements(fileStream, transferOutputStream, false);
long removalEnd = System.currentTimeMillis();
log.debug("Invisible element removal successful for file with dossierId {} and fileId {}, took {}s",
dossierId,
fileId,
format("%.1f", (removalEnd - removalStart) / 1000.0));
log.debug("Invisible element removal successful for file with dossierId {} and fileId {}, took {}s", dossierId, fileId, format("%.1f", (removalEnd - removalStart) / 1000.0));
}
try (InputStream transferInputStream = new ByteArrayInputStream(transferOutputStream.toByteArray())) {
long ocrStart = System.currentTimeMillis();
@@ -93,22 +78,18 @@ public class OCRService {
PDFDoc pdfDoc = new PDFDoc(fileStream);
Map<Integer, RectCollection> pageIdToRectCollection = imagePositionRetrievalService.getImagePositionPerPage(pdfDoc, true);
rabbitTemplate.convertAndSend(MessagingConfiguration.OCR_STATUS_UPDATE_RESPONSE_QUEUE,
objectMapper.writeValueAsString(OCRStatusUpdateResponse.builder().fileId(fileId).numberOfPagesToOCR(pageIdToRectCollection.size()).build()));
// Optimization:
// When a page does not have a TextZone, PDFTron whites out the page. But, PDFTron scans it anyway, resulting in a longer runtime.
// So, we need to remove pages without images.
// Furthermore, creating a new document is *much* faster than reusing the same document and adding/removing pages one by one.
// Therefore, we create a new Document with a single page for every page that contains an image.
// For some reason, if we insert the OCR singlePageDoc into the original PDFDoc inside the loop, for some documents, the RAM Usage increases exponentially with every page.
// This is why, we replace the OCRed Pages outside the main loop.
int numProcessedPages = 0;
Map<Integer, PDFDoc> pageIdToSinglePagePdfDoc = new HashMap<>();
for (Integer pageId : pageIdToRectCollection.keySet()) {
try {
PDFDoc singlePagePdfDoc = extractSinglePagePdfDoc(pdfDoc, pageId);
processOcr(pageIdToRectCollection, pageId, singlePagePdfDoc);
++numProcessedPages;
log.info("{}/{} Page {} done, OCR regions {}",
numProcessedPages,
@@ -116,8 +97,9 @@ public class OCRService {
pageId,
getAllOcrTextZonesAsString(pageIdToRectCollection, pageId));
pageIdToSinglePagePdfDoc.put(pageId, singlePagePdfDoc);
++numProcessedPages;
replaceOriginalPageWithOcrPage(pdfDoc, pageId, singlePagePdfDoc);
singlePagePdfDoc.close();
rabbitTemplate.convertAndSend(MessagingConfiguration.OCR_STATUS_UPDATE_RESPONSE_QUEUE,
objectMapper.writeValueAsString(OCRStatusUpdateResponse.builder()
@@ -131,9 +113,6 @@ public class OCRService {
throw new RuntimeException(e);
}
}
log.info("Copying {} OCRed Pages into original Document", pageIdToSinglePagePdfDoc.size());
pageIdToSinglePagePdfDoc.forEach((pageId, singlePagePdfDoc) -> replaceOriginalPageWithOcrPage(pdfDoc, pageId, singlePagePdfDoc));
Optimizer.optimize(pdfDoc);
rabbitTemplate.convertAndSend(MessagingConfiguration.OCR_STATUS_UPDATE_RESPONSE_QUEUE,
objectMapper.writeValueAsString(OCRStatusUpdateResponse.builder()
@@ -143,6 +122,7 @@ public class OCRService {
.ocrFinished(true)
.build()));
Optimizer.optimize(pdfDoc);
try {
pdfDoc.save(out, SDFDoc.SaveMode.LINEARIZED, null);
} catch (Exception e) {
@@ -174,13 +154,11 @@ public class OCRService {
}
@SneakyThrows
private static void replaceOriginalPageWithOcrPage(PDFDoc pdfDoc, Integer pageId, PDFDoc singlePagePdfDoc) {
private static void replaceOriginalPageWithOcrPage(PDFDoc pdfDoc, Integer pageId, PDFDoc ocrPageDoc) throws PDFNetException {
Page ocrPage = singlePagePdfDoc.getPage(1);
Page ocrPage = ocrPageDoc.getPage(1);
pdfDoc.pageInsert(pdfDoc.getPageIterator(pageId), ocrPage);
pdfDoc.pageRemove(pdfDoc.getPageIterator(pageId + 1));
singlePagePdfDoc.close();
}
@@ -40,6 +40,10 @@ management:
storage:
signer-type: 'AWSS3V4SignerType'
bucket-name: 'redaction'
region: 'us-east-1'
endpoint: 'https://s3.amazonaws.com'
backend: 's3'
pdftron.license: ${PDFTRON_LICENSE}
@@ -1,7 +1,7 @@
package com.iqser.red.service.ocr.v1.server;
import static com.iqser.red.service.ocr.v1.server.utils.OsUtils.getTemporaryDirectory;
import static com.iqser.red.service.ocr.v1.server.utils.PdfTextExtraction.extractAllTextFromDocument;
import static com.iqser.red.service.ocr.v1.server.utils.PdfUtils.extractAllTextFromDocument;
import static org.assertj.core.api.Assertions.assertThat;
import java.io.FileInputStream;
@@ -16,19 +16,16 @@ import org.springframework.boot.test.context.SpringBootTest;
import org.springframework.core.io.ClassPathResource;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.iqser.red.service.ocr.v1.server.multitenancy.TenantContext;
import com.iqser.red.service.ocr.v1.server.service.FileStorageService;
import com.iqser.red.service.ocr.v1.server.service.OCRService;
import com.iqser.red.service.persistence.service.v1.api.shared.model.dossiertemplate.dossier.file.FileType;
import com.iqser.red.service.persistence.service.v1.api.model.dossiertemplate.dossier.file.FileType;
import com.pdftron.pdf.OCRModule;
import io.micrometer.prometheus.PrometheusMeterRegistry;
import io.micrometer.prometheus.PrometheusTimer;
import lombok.SneakyThrows;
@Disabled // OCRModule is not available on build server. If you want to run the test set the property at the top.
@SpringBootTest(properties = {"pdftron.ocrmodule.path=/YourOCRModulePath"})
@SpringBootTest(properties = {"pdftron.ocrmodule.path=/home/kschuettler/iqser/PDFTron/ocr/Lib/"})
public class OcrServiceIntegrationTest extends AbstractTest {
@Autowired
@@ -42,12 +39,13 @@ public class OcrServiceIntegrationTest extends AbstractTest {
@BeforeEach
@SneakyThrows
@Disabled
public void assertOCRModuleIsLoaded() {
assert OCRModule.isModuleAvailable();
}
@Test
@Disabled // OCRModule is not available on build server. If you want to run the test set the property at the top.
public void testOCRMetrics() {
testOCR("Watermark");
@@ -63,13 +61,38 @@ public class OcrServiceIntegrationTest extends AbstractTest {
@Test
@Disabled // OCRModule is not available on build server. If you want to run the test set the property at the top.
public void testOcr() {
// check visually for most of the images containing text, the resulting text is kind of nonsense, just ensure it is there
String text = testOCR("StitchedImagesMultiPage");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
testOCR("131 IDD0000261725");
}
@Test
@Disabled // OCRModule is not available on build server. If you want to run the test set the property at the top.
public void testManyRotatedImages() {
// check visually for most of the images containing text, the resulting text is kind of nonsense, just ensure it is there
String text = testOCR("manyRotatedImages");
@@ -78,6 +101,7 @@ public class OcrServiceIntegrationTest extends AbstractTest {
@Test
@Disabled // OCRModule is not available on build server. If you want to run the test set the property at the top.
public void testMergeImages() {
// check visually for most of the images containing text, the resulting text is kind of nonsense, just ensure it is there
String text = testOCR("merge_images");
@@ -98,6 +122,7 @@ public class OcrServiceIntegrationTest extends AbstractTest {
@Test
@Disabled // OCRModule is not available on build server. If you want to run the test set the property at the top.
public void testOCRWatermark() {
assertThat(testOCR("Watermark")).contains("syngenta");
@@ -105,6 +130,7 @@ public class OcrServiceIntegrationTest extends AbstractTest {
@Test
@Disabled // OCRModule is not available on build server. If you want to run the test set the property at the top.
public void testOCRInvisibleText() {
String text = testOCR("InvisibleText");
@@ -119,7 +145,7 @@ public class OcrServiceIntegrationTest extends AbstractTest {
ClassPathResource pdfFileResource = new ClassPathResource("files/" + fileName + ".pdf");
var originId = FileStorageService.getStorageId("dossier", "file", FileType.ORIGIN);
try (var fileStream = pdfFileResource.getInputStream()) {
storageService.storeObject(TenantContext.getTenantId(), originId, fileStream);
storageService.storeObject(originId, fileStream);
}
try (var out = new FileOutputStream(getTemporaryDirectory() + "/" + fileName + ".pdf")) {
ocrService.runOcrOnDocument("dossier", "file", out);
@@ -0,0 +1,14 @@
package com.iqser.red.service.ocr.v1.server.service;
import static org.junit.jupiter.api.Assertions.*;
import org.junit.jupiter.api.Test;
class DataExtractionServiceTest {
@Test
void extractData() {
}
}
@@ -1,8 +1,8 @@
package com.iqser.red.service.ocr.v1.server.service;
import static com.iqser.red.service.ocr.v1.server.utils.OsUtils.getTemporaryDirectory;
import static com.iqser.red.service.ocr.v1.server.utils.PdfDraw.drawGrid;
import static com.iqser.red.service.ocr.v1.server.utils.PdfDraw.drawRectCollection;
import static com.iqser.red.service.ocr.v1.server.utils.PdfUtils.drawGrid;
import static com.iqser.red.service.ocr.v1.server.utils.PdfUtils.drawRectCollection;
import static org.assertj.core.api.Assertions.assertThat;
import java.io.FileInputStream;
@@ -1,7 +1,7 @@
package com.iqser.red.service.ocr.v1.server.service;
import static com.iqser.red.service.ocr.v1.server.utils.OsUtils.getTemporaryDirectory;
import static com.iqser.red.service.ocr.v1.server.utils.PdfTextExtraction.extractAllTextFromDocument;
import static com.iqser.red.service.ocr.v1.server.utils.PdfUtils.extractAllTextFromDocument;
import static org.assertj.core.api.AssertionsForClassTypes.assertThat;
import java.io.FileInputStream;
@@ -32,7 +32,7 @@ public class FileSystemBackedStorageService implements StorageService {
@SneakyThrows
@Override
public InputStreamResource getObject(String tenantId, String objectId) {
public InputStreamResource getObject(String objectId) {
var res = dataMap.get(objectId);
if (res == null) {
@@ -44,22 +44,28 @@ public class FileSystemBackedStorageService implements StorageService {
@Override
public void deleteObject(String tenantId, String objectId) {
public void deleteObject(String objectId) {
dataMap.remove(objectId);
}
@Override
public boolean objectExists(String tenantId, String objectId) {
public boolean objectExists(String objectId) {
return dataMap.containsKey(objectId);
}
@Override
public void init() {
}
@Override
@SneakyThrows
public <T> void storeJSONObject(String tenantId, String objectId, T any) {
public <T> void storeJSONObject(String objectId, T any) {
File tempFile = File.createTempFile("test", ".tmp");
getMapper().writeValue(new FileOutputStream(tempFile), any);
@@ -75,7 +81,7 @@ public class FileSystemBackedStorageService implements StorageService {
@Override
@SneakyThrows
public <T> T readJSONObject(String tenantId, String objectId, Class<T> clazz) {
public <T> T readJSONObject(String objectId, Class<T> clazz) {
if (dataMap.get(objectId) == null || !dataMap.get(objectId).exists()) {
throw new StorageObjectDoesNotExist("Stored object not found");
@@ -98,7 +104,7 @@ public class FileSystemBackedStorageService implements StorageService {
@Override
@SneakyThrows
public void storeObject(String tenantId, String objectId, InputStream stream) {
public void storeObject(String objectId, InputStream stream) {
File tempFile = File.createTempFile("test", ".tmp");
@@ -1,35 +0,0 @@
package com.iqser.red.service.ocr.v1.server.utils;
import java.io.IOException;
import java.io.InputStream;
import java.util.ArrayList;
import java.util.List;
import com.pdftron.common.PDFNetException;
import com.pdftron.pdf.PDFDoc;
import com.pdftron.pdf.Page;
import com.pdftron.pdf.PageIterator;
import com.pdftron.pdf.TextExtractor;
public class PdfTextExtraction {
public static String extractAllTextFromDocument(InputStream fileStream) throws IOException, PDFNetException {
PDFDoc pdfDoc = new PDFDoc(fileStream);
TextExtractor extractor = new TextExtractor();
List<String> texts = new ArrayList<>();
PageIterator iterator = pdfDoc.getPageIterator();
while (iterator.hasNext()) {
Page page = iterator.next();
extractor.begin(page);
texts.add(extractor.getAsText());
}
extractor.destroy();
pdfDoc.close();
return String.join("\n", texts);
}
}
@@ -1,18 +1,32 @@
package com.iqser.red.service.ocr.v1.server.utils;
import com.pdftron.pdf.ColorPt;
import com.pdftron.pdf.ColorSpace;
import com.pdftron.pdf.Element;
import com.pdftron.pdf.ElementBuilder;
import com.pdftron.pdf.ElementWriter;
import com.pdftron.pdf.Page;
import com.pdftron.pdf.Rect;
import com.pdftron.pdf.RectCollection;
import com.pdftron.common.PDFNetException;
import com.pdftron.pdf.*;
import lombok.SneakyThrows;
import java.io.IOException;
import java.io.InputStream;
import java.util.ArrayList;
import java.util.List;
public class PdfDraw {
public class PdfUtils {
public static String extractAllTextFromDocument(InputStream fileStream) throws IOException, PDFNetException {
PDFDoc pdfDoc = new PDFDoc(fileStream);
TextExtractor extractor = new TextExtractor();
List<String> texts = new ArrayList<>();
PageIterator iterator = pdfDoc.getPageIterator();
while (iterator.hasNext()) {
Page page = iterator.next();
extractor.begin(page);
texts.add(extractor.getAsText());
}
extractor.destroy();
pdfDoc.close();
return String.join("\n", texts);
}
@SneakyThrows
public static void drawGrid(ElementWriter writer, Page page) {
@@ -45,7 +59,6 @@ public class PdfDraw {
eb.destroy();
}
@SneakyThrows
public static void drawRectCollection(ElementWriter writer, RectCollection rectCollection) {
@@ -70,5 +83,4 @@ public class PdfDraw {
colorPt.destroy();
eb.destroy();
}
}
+2 -2
View File
@@ -7,7 +7,7 @@
<parent>
<groupId>com.iqser.red</groupId>
<artifactId>platform-dependency</artifactId>
<version>1.17.0</version>
<version>1.16.0</version>
<relativePath/>
</parent>
@@ -27,7 +27,7 @@
<dependency>
<groupId>com.iqser.red</groupId>
<artifactId>platform-commons-dependency</artifactId>
<version>1.22.0</version>
<version>1.20.0</version>
<scope>import</scope>
<type>pom</type>
</dependency>
+1 -1
View File
@@ -6,7 +6,7 @@
<groupId>com.iqser.red.service</groupId>
<artifactId>ocr-service</artifactId>
<version>2.0-SNAPSHOT</version>
<version>1.0-SNAPSHOT</version>
<packaging>pom</packaging>