Merge branch 'dev'

urieli · Sep 20, 2023 · 259f213 · 259f213
2 parents beb165b + cde7ff6
commit 259f213
Show file tree

Hide file tree

Showing 10 changed files with 28 additions and 16 deletions.
diff --git a/jochre_core/pom.xml b/jochre_core/pom.xml
@@ -6,7 +6,7 @@
   <parent>
     <groupId>com.joliciel.jochre</groupId>
     <artifactId>jochre-parent</artifactId>
-    <version>2.6.3</version>
+    <version>2.6.4</version>
     <relativePath>../jochre_parent</relativePath>
   </parent>
   <build>

diff --git a/jochre_core/src/main/java/com/joliciel/jochre/Jochre.java b/jochre_core/src/main/java/com/joliciel/jochre/Jochre.java
@@ -1078,7 +1078,12 @@ public void doCommandTransform(CorpusSelectionCriteria criteria, List<DocumentOb
   public void doCommandAnalyse(File sourceFile, MostLikelyWordChooser wordChooser, Set<Integer> pages,
       List<DocumentObserver> observers, List<PdfImageObserver> imageObservers) throws IOException {
 
-    JochreDocumentGenerator documentGenerator = this.getDocumentGenerator(sourceFile.getName(), wordChooser, pages, observers, imageObservers);
+    boolean imageFile = sourceFile.getName().toLowerCase().endsWith(".png")
+        || sourceFile.getName().toLowerCase().endsWith(".jpg") || sourceFile.getName().toLowerCase().endsWith(".jpeg")
+        || sourceFile.getName().toLowerCase().endsWith(".gif") || sourceFile.getName().toLowerCase().endsWith(".tif")
+        || sourceFile.getName().toLowerCase().endsWith(".tiff");
+
+    JochreDocumentGenerator documentGenerator = this.getDocumentGenerator(sourceFile.getName(), wordChooser, pages, observers, imageObservers, imageFile);
 
     if (!sourceFile.exists())
       throw new JochreException("The file " + sourceFile.getPath() + " does not exist");
@@ -1089,10 +1094,7 @@ public void doCommandAnalyse(File sourceFile, MostLikelyWordChooser wordChooser,
         pdfDocumentProcessor.addImageObserver(imageObserver);
       }
       pdfDocumentProcessor.process();
-    } else if (sourceFile.getName().toLowerCase().endsWith(".png")
-        || sourceFile.getName().toLowerCase().endsWith(".jpg") || sourceFile.getName().toLowerCase().endsWith(".jpeg")
-        || sourceFile.getName().toLowerCase().endsWith(".gif") || sourceFile.getName().toLowerCase().endsWith(".tif")
-        || sourceFile.getName().toLowerCase().endsWith(".tiff")) {
+    } else if (imageFile) {
       ImageFileDocumentExtractor extractor = new ImageFileDocumentExtractor(sourceFile, documentGenerator);
       extractor.extractDocument();
     } else if (sourceFile.isDirectory()) {
@@ -1104,7 +1106,7 @@ public void doCommandAnalyse(File sourceFile, MostLikelyWordChooser wordChooser,
   }
 
   private JochreDocumentGenerator getDocumentGenerator(String fileName, MostLikelyWordChooser wordChooser, Set<Integer> pages,
-                                List<DocumentObserver> observers, List<PdfImageObserver> imageObservers) throws IOException {
+                                List<DocumentObserver> observers, List<PdfImageObserver> imageObservers, boolean exitOnError) throws IOException {
 
     ClassificationModel letterModel = jochreSession.getLetterModel();
 
@@ -1134,7 +1136,7 @@ private JochreDocumentGenerator getDocumentGenerator(String fileName, MostLikely
     ImageAnalyser analyser = new BeamSearchImageAnalyser(boundaryDetector, letterGuesser, wordChooser, jochreSession);
     analyser.addObserver(letterGuessObserver);
 
-    JochreDocumentGenerator documentGenerator = new JochreDocumentGenerator(fileName, "", jochreSession);
+    JochreDocumentGenerator documentGenerator = new JochreDocumentGenerator(fileName, "", jochreSession, exitOnError);
     documentGenerator.addDocumentObserver(analyser);
 
     for (DocumentObserver observer : observers)
@@ -1546,7 +1548,7 @@ public void imageToAlto4(BufferedImage image, String fileName, Writer writer) th
     List<DocumentObserver> documentObservers = new ArrayList<>();
     AltoXMLExporter altoXMLExporter = new AltoXMLExporter(writer, 4);
     documentObservers.add(altoXMLExporter);
-    JochreDocumentGenerator documentGenerator = this.getDocumentGenerator(fileName, wordChooser, myPages, documentObservers, new ArrayList<>());
+    JochreDocumentGenerator documentGenerator = this.getDocumentGenerator(fileName, wordChooser, myPages, documentObservers, new ArrayList<>(), true);
     ImageDocumentExtractor documentExtractor = new ImageDocumentExtractor(image, fileName, documentGenerator);
     documentExtractor.extractDocument();
   }

diff --git a/jochre_core/src/main/java/com/joliciel/jochre/doc/JochreDocumentGenerator.java b/jochre_core/src/main/java/com/joliciel/jochre/doc/JochreDocumentGenerator.java
@@ -79,6 +79,7 @@ public class JochreDocumentGenerator implements SourceFileProcessor, Monitorable
   private List<DocumentObserver> documentObservers = new ArrayList<>();
 
   private final JochreSession jochreSession;
+  private final boolean exitOnError;
 
   /**
    * Constructor for existing documents.
@@ -89,6 +90,7 @@ public class JochreDocumentGenerator implements SourceFileProcessor, Monitorable
   public JochreDocumentGenerator(JochreDocument jochreDocument, JochreSession jochreSession) {
     this.jochreSession = jochreSession;
     this.doc = jochreDocument;
+    this.exitOnError = false;
   }
 
   /**
@@ -100,9 +102,14 @@ public JochreDocumentGenerator(JochreDocument jochreDocument, JochreSession joch
    *            user-friendly name for the document (required if saving)
    */
   public JochreDocumentGenerator(String filename, String userFriendlyName, JochreSession jochreSession) {
+    this(filename, userFriendlyName, jochreSession, false);
+  }
+
+  public JochreDocumentGenerator(String filename, String userFriendlyName, JochreSession jochreSession, boolean exitOnError) {
     this.jochreSession = jochreSession;
     this.filename = filename;
     this.userFriendlyName = userFriendlyName;
+    this.exitOnError = exitOnError;
   }
 
   @Override
@@ -214,6 +221,9 @@ public JochreImage onImageFound(JochrePage jochrePage, BufferedImage image, Stri
       } catch (SegmentationException se) {
         LOG.error("segmentation failed", se);
         sourceImage.clearSegmentation();
+        if (exitOnError) {
+          throw se;
+        }
       }
 
       if (currentMonitor != null)

diff --git a/jochre_distribution/pom.xml b/jochre_distribution/pom.xml
@@ -6,7 +6,7 @@
   <parent>
     <groupId>com.joliciel.jochre</groupId>
     <artifactId>jochre-parent</artifactId>
-    <version>2.6.3</version>
+    <version>2.6.4</version>
     <relativePath>../jochre_parent</relativePath>
   </parent>
   <build>

diff --git a/jochre_parent/pom.xml b/jochre_parent/pom.xml
@@ -4,7 +4,7 @@
   <modelVersion>4.0.0</modelVersion>
   <groupId>com.joliciel.jochre</groupId>
   <artifactId>jochre-parent</artifactId>
-  <version>2.6.3</version>
+  <version>2.6.4</version>
   <packaging>pom</packaging>
   <name>Jochre</name>
   <description>Java Optical CHaracter Recognition</description>

diff --git a/jochre_search/pom.xml b/jochre_search/pom.xml
@@ -6,7 +6,7 @@
   <parent>
     <groupId>com.joliciel.jochre</groupId>
     <artifactId>jochre-parent</artifactId>
-    <version>2.6.3</version>
+    <version>2.6.4</version>
     <relativePath>../jochre_parent</relativePath>
   </parent>
   <packaging>jar</packaging>

diff --git a/jochre_search_webapp/pom.xml b/jochre_search_webapp/pom.xml
@@ -5,7 +5,7 @@
   <parent>
     <groupId>com.joliciel.jochre</groupId>
     <artifactId>jochre-parent</artifactId>
-    <version>2.6.3</version>
+    <version>2.6.4</version>
     <relativePath>../jochre_parent</relativePath>
   </parent>
   <packaging>war</packaging>

diff --git a/jochre_utils/pom.xml b/jochre_utils/pom.xml
@@ -5,7 +5,7 @@
   <parent>
     <groupId>com.joliciel.jochre</groupId>
     <artifactId>jochre-parent</artifactId>
-    <version>2.6.3</version>
+    <version>2.6.4</version>
     <relativePath>../jochre_parent</relativePath>
   </parent>
   <build>

diff --git a/jochre_web/pom.xml b/jochre_web/pom.xml
@@ -5,7 +5,7 @@
   <parent>
     <groupId>com.joliciel.jochre</groupId>
     <artifactId>jochre-parent</artifactId>
-    <version>2.6.3</version>
+    <version>2.6.4</version>
     <relativePath>../jochre_parent</relativePath>
   </parent>
   <packaging>war</packaging>

diff --git a/jochre_yiddish/pom.xml b/jochre_yiddish/pom.xml
@@ -6,7 +6,7 @@
   <parent>
     <groupId>com.joliciel.jochre</groupId>
     <artifactId>jochre-parent</artifactId>
-    <version>2.6.3</version>
+    <version>2.6.4</version>
     <relativePath>../jochre_parent</relativePath>
   </parent>
   <packaging>jar</packaging>