Merge pull request #16 from diging/develop

Story/geco 90 (#15)
diging · Feb 16, 2018 · 09fb31d · 09fb31d
2 parents 57e2a86 + cb06d86
commit 09fb31d
Show file tree

Hide file tree

Showing 6 changed files with 74 additions and 32 deletions.
diff --git a/cassiopeia/pom.xml b/cassiopeia/pom.xml
@@ -12,7 +12,7 @@
         <org.slf4j-version>1.7.5</org.slf4j-version>
         <org.springframework-version>4.3.1.RELEASE</org.springframework-version>
         <spring-security-version>4.1.3.RELEASE</spring-security-version>
-        <geco.requests.version>0.5</geco.requests.version>
+        <geco.requests.version>0.6</geco.requests.version>
         <geco.util.version>0.4.2</geco.util.version>
         <geco.september.util.version>0.2</geco.september.util.version>
 
@@ -35,6 +35,7 @@
         <tesseract.bin></tesseract.bin>
         <tesseract.data></tesseract.data>
         <tesseract.create.hocr>false</tesseract.create.hocr>
+        <tesseract.timeout>600</tesseract.timeout>
     </properties>
 
     <dependencies>

diff --git a/...n/java/edu/asu/diging/gilesecosystem/cassiopeia/core/service/impl/KafkaRequestSender.java b/...n/java/edu/asu/diging/gilesecosystem/cassiopeia/core/service/impl/KafkaRequestSender.java
@@ -30,59 +30,73 @@ public class KafkaRequestSender implements IKafkaRequestSender {
 
     @Autowired
     private IPropertiesManager propertyManager;
-    
+
     @Autowired
     private IRequestFactory<ICompletedOCRRequest, CompletedOCRRequest> requestFactory;
-    
+
     @Autowired
     private IRequestProducer requestProducer;
 
     @Autowired
     private ISystemMessageHandler messageHandler;
-    
+
     @PostConstruct
     public void init() {
         requestFactory.config(CompletedOCRRequest.class);
     }
 
-    /* (non-Javadoc)
-     * @see edu.asu.diging.gilesecosystem.cassiopeia.core.service.impl.IKafkaRequestSender#sendRequest(java.lang.String, java.lang.String, java.lang.String, java.lang.String, edu.asu.diging.gilesecosystem.cassiopeia.core.service.impl.RequestInfo)
+    /*
+     * (non-Javadoc)
+     * 
+     * @see edu.asu.diging.gilesecosystem.cassiopeia.core.service.impl.
+     * IKafkaRequestSender#sendRequest(java.lang.String, java.lang.String,
+     * java.lang.String, java.lang.String,
+     * edu.asu.diging.gilesecosystem.cassiopeia.core.service.impl.RequestInfo)
      */
     @Override
     public void sendRequest(String requestId, String documentId, RequestInfo info) {
         String restEndpoint = propertyManager.getProperty(Properties.BASE_URL);
         if (restEndpoint.endsWith("/")) {
-            restEndpoint = restEndpoint.substring(0, restEndpoint.length()-1);
+            restEndpoint = restEndpoint.substring(0, restEndpoint.length() - 1);
+        }
+
+        String fileEndpoint = null;
+
+        if (info.getStatus() == RequestStatus.COMPLETE) {
+            fileEndpoint = restEndpoint + DownloadFileController.GET_FILE_URL
+                    .replace(DownloadFileController.REQUEST_ID_PLACEHOLDER, requestId)
+                    .replace(DownloadFileController.DOCUMENT_ID_PLACEHOLDER, documentId)
+                    .replace(DownloadFileController.FILENAME_PLACEHOLDER,
+                            info.getFilename());
         }
-
-        String fileEndpoint = restEndpoint + DownloadFileController.GET_FILE_URL
-                .replace(DownloadFileController.REQUEST_ID_PLACEHOLDER, requestId)
-                .replace(DownloadFileController.DOCUMENT_ID_PLACEHOLDER, documentId)
-                .replace(DownloadFileController.FILENAME_PLACEHOLDER, info.getFilename());
-
+
         ICompletedOCRRequest completedRequest = null;
         try {
-            completedRequest = requestFactory.createRequest(requestId, info.getUploadId());
+            completedRequest = requestFactory.createRequest(requestId,
+                    info.getUploadId());
         } catch (InstantiationException | IllegalAccessException e) {
-            messageHandler.handleMessage("Could not create request.", e, MessageType.ERROR);
+            messageHandler.handleMessage("Could not create request.", e,
+                    MessageType.ERROR);
             // this should never happen if used correctly
         }
-        
+
         completedRequest.setDocumentId(documentId);
         completedRequest.setDownloadPath(info.getPath());
         completedRequest.setSize(info.getSize());
         completedRequest.setDownloadUrl(fileEndpoint);
         completedRequest.setFilename(info.getImageFilename());
         completedRequest.setFileId(info.getFileId());
-        completedRequest.setStatus(RequestStatus.COMPLETE);
+        completedRequest.setStatus(info.getStatus());
+        completedRequest.setErrorMsg(info.getErrorMsg());
         completedRequest.setOcrDate(OffsetDateTime.now(ZoneId.of("UTC")).toString());
         completedRequest.setTextFilename(info.getFilename());
-        
+
         try {
-            requestProducer.sendRequest(completedRequest, propertyManager.getProperty(Properties.KAFKA_TOPIC_OCR_COMPLETE));
+            requestProducer.sendRequest(completedRequest,
+                    propertyManager.getProperty(Properties.KAFKA_TOPIC_OCR_COMPLETE));
         } catch (MessageCreationException e) {
             messageHandler.handleMessage("Could not send message.", e, MessageType.ERROR);
         }
     }
-    
+
 }
diff --git a/.../src/main/java/edu/asu/diging/gilesecosystem/cassiopeia/core/service/impl/OCRManager.java b/.../src/main/java/edu/asu/diging/gilesecosystem/cassiopeia/core/service/impl/OCRManager.java
@@ -33,6 +33,7 @@
 import edu.asu.diging.gilesecosystem.cassiopeia.core.service.IKafkaRequestSender;
 import edu.asu.diging.gilesecosystem.cassiopeia.core.service.IOCRManager;
 import edu.asu.diging.gilesecosystem.requests.IOCRRequest;
+import edu.asu.diging.gilesecosystem.requests.RequestStatus;
 import edu.asu.diging.gilesecosystem.septemberutil.properties.MessageType;
 import edu.asu.diging.gilesecosystem.septemberutil.service.ISystemMessageHandler;
 import edu.asu.diging.gilesecosystem.util.files.IFileStorageManager;
@@ -82,18 +83,25 @@ public void processOCRRequest(IOCRRequest request) {
         Metadata metadata = new Metadata();
         BodyContentHandler handler = new BodyContentHandler();
 
-        String ocrResult = null;
+        RequestInfo info = null;
         try (InputStream stream = new ByteArrayInputStream(image)) {
             ocrParser.parse(stream, handler, metadata, parseContext);
-            ocrResult = handler.toString();
+            String ocrResult = handler.toString();
+            info = saveTextToFile(request.getRequestId(), request.getDocumentId(), ocrResult, request.getFilename(), ".txt");
+            info.setUploadId(request.getUploadId());
+            info.setFileId(request.getFileId());
+            info.setStatus(RequestStatus.COMPLETE);
+            info.setImageFilename(request.getFilename());
         } catch (SAXException | TikaException | IOException e) {
             messageHandler.handleMessage("Error during ocr.", e, MessageType.ERROR);
-            // FIXME: send to monitoring app
+            info = new RequestInfo(null, 0, null);
+            info.setUploadId(request.getUploadId());
+            info.setFileId(request.getFileId());
+            info.setStatus(RequestStatus.FAILED);
+            info.setErrorMsg(e.getMessage());
+            info.setImageFilename(request.getFilename());
         }
 
-        RequestInfo info = saveTextToFile(request.getRequestId(), request.getDocumentId(), ocrResult, request.getFilename(), ".txt");
-        info.setUploadId(request.getUploadId());
-        info.setFileId(request.getFileId());
 
         kafkaRequestSender.sendRequest(request.getRequestId(), request.getDocumentId(), info);
     }
@@ -127,7 +135,6 @@ protected RequestInfo saveTextToFile(String requestId,
         if (!fileExtentions.startsWith(".")) {
             fileExtentions = "." + fileExtentions;
         }
-        String imageFilename = filename;
         filename = filename + fileExtentions;
 
         String filePath = docFolder + File.separator + filename;
@@ -151,6 +158,6 @@ protected RequestInfo saveTextToFile(String requestId,
         }
 
         String relativePath = storageManager.getFileFolderPathInBaseFolder(requestId, documentId, null);
-        return new RequestInfo(relativePath + File.separator + filename, fileObject.length(), imageFilename, filename);
+        return new RequestInfo(relativePath + File.separator + filename, fileObject.length(), filename);
     }
 }
diff --git a/...src/main/java/edu/asu/diging/gilesecosystem/cassiopeia/core/service/impl/RequestInfo.java b/...src/main/java/edu/asu/diging/gilesecosystem/cassiopeia/core/service/impl/RequestInfo.java
@@ -1,5 +1,7 @@
 package edu.asu.diging.gilesecosystem.cassiopeia.core.service.impl;
 
+import edu.asu.diging.gilesecosystem.requests.RequestStatus;
+
 public class RequestInfo {
 
     private String path;
@@ -8,12 +10,13 @@ public class RequestInfo {
     private String imageFilename;
     private String uploadId;
     private String fileId;
+    private RequestStatus status;
+    private String errorMsg;
 
-    public RequestInfo(String path, long size, String imageFilename, String filename) {
+    public RequestInfo(String path, long size, String filename) {
         this.path = path;
         this.size = size;
         this.filename = filename;
-        this.imageFilename = imageFilename;
     }
 
     public String getPath() {
@@ -63,6 +66,22 @@ public String getFilename() {
     public void setFilename(String filename) {
         this.filename = filename;
     }
+
+    public RequestStatus getStatus() {
+        return status;
+    }
+
+    public void setStatus(RequestStatus status) {
+        this.status = status;
+    }
+
+    public String getErrorMsg() {
+        return errorMsg;
+    }
+
+    public void setErrorMsg(String errorMsg) {
+        this.errorMsg = errorMsg;
+    }
 
 
 }
diff --git a/...main/java/edu/asu/diging/gilesecosystem/cassiopeia/core/service/impl/RequestResender.java b/...main/java/edu/asu/diging/gilesecosystem/cassiopeia/core/service/impl/RequestResender.java
@@ -56,7 +56,8 @@ public boolean accept(File dir, String name) {
                     String imageFilename = textFile.getName();
                     imageFilename = imageFilename.substring(0, imageFilename.lastIndexOf("."));
 
-                    RequestInfo info = new RequestInfo(relativePath + File.separator + textFile.getName(), textFile.length(), imageFilename, textFile.getName());
+                    RequestInfo info = new RequestInfo(relativePath + File.separator + textFile.getName(), textFile.length(), textFile.getName());
+                    info.setImageFilename(imageFilename);
                     kafkaRequestSender.sendRequest(requestId, docId, info);
                     requestCounter++;
                 }

diff --git a/cassiopeia/src/main/resources/config.properties b/cassiopeia/src/main/resources/config.properties
@@ -16,7 +16,7 @@ tesseract_bin_folder=${tesseract.bin}
 tesseract_data_folder=${tesseract.data}
 tesseract_create_hocr=${tesseract.create.hocr}
 # how much time to give to tesseract before timing out (in sec)
-tesseract_timeout=240
+tesseract_timeout=${tesseract.timeout}
 
 #OCR Type
 ocr_plainText=Plain Text