RED-4254: Added dirty hack in pdfbox classes to find words that contains uniqueCharacters with 2 chars like 'RA'

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/parsing/LegacyPDFStreamEngine.java

@@ -301,12 +301,24 @@ class LegacyPDFStreamEngine extends PDFStreamEngine
             nextY -= pageSize.getLowerLeftY();
         }
 
-        processTextPosition(new TextPosition(pageRotation, pageSize.getWidth(),
+        // This is a hack for unicode letter with 2 chars e.g. RA see unicodeProblem.pdf
-                pageSize.getHeight(), translatedTextRenderingMatrix, nextX, nextY,
+        if(unicodeMapping.length() == 2){
-                Math.abs(dyDisplay), dxDisplay,
+            processTextPosition(new TextPosition(pageRotation, pageSize.getWidth(),
-                Math.abs(spaceWidthDisplay), unicodeMapping, new int[] { code }, font,
+                    pageSize.getHeight(), translatedTextRenderingMatrix, nextX, nextY,
-                fontSize,
+                    Math.abs(dyDisplay), dxDisplay,
-                (int)(fontSize * textMatrix.getScalingFactorX())));
+                    Math.abs(spaceWidthDisplay), Character.toString(unicodeMapping.charAt(0)), new int[] { code }, font,
+                    fontSize,
+                    (int)(fontSize * textMatrix.getScalingFactorX())));
+            processTextPosition(new TextPosition(pageRotation, pageSize.getWidth(),
+                    pageSize.getHeight(), translatedTextRenderingMatrix, nextX, nextY,
+                    Math.abs(dyDisplay), dxDisplay,
+                    Math.abs(spaceWidthDisplay), Character.toString(unicodeMapping.charAt(1)), new int[] { code }, font,
+                    fontSize,
+                    (int)(fontSize * textMatrix.getScalingFactorX())));
+        } else {
+
+            processTextPosition(new TextPosition(pageRotation, pageSize.getWidth(), pageSize.getHeight(), translatedTextRenderingMatrix, nextX, nextY, Math.abs(dyDisplay), dxDisplay, Math.abs(spaceWidthDisplay), unicodeMapping, new int[]{code}, font, fontSize, (int) (fontSize * textMatrix.getScalingFactorX())));
+        }
     }
 
     /**

redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/RedactionIntegrationTest.java

@@ -807,6 +807,30 @@ public class RedactionIntegrationTest {
     }
 
 
+    @Test
+    public void testUnicodeProblem() throws IOException {
+
+        long start = System.currentTimeMillis();
+
+        AnalyzeRequest request = prepareStorage("files/new/unicodeProblem.pdf");
+        analyzeService.analyzeDocumentStructure(new StructureAnalyzeRequest(request.getDossierId(), request.getFileId()));
+        AnalyzeResult result = analyzeService.analyze(request);
+
+        AnnotateResponse annotateResponse = annotationService.annotate(AnnotateRequest.builder()
+                .dossierId(TEST_DOSSIER_ID)
+                .fileId(TEST_FILE_ID)
+                .build());
+
+        try (FileOutputStream fileOutputStream = new FileOutputStream(OsUtils.getTemporaryDirectory() + "/Annotated.pdf")) {
+            fileOutputStream.write(annotateResponse.getDocument());
+        }
+        long end = System.currentTimeMillis();
+
+        System.out.println("duration: " + (end - start));
+        System.out.println("numberOfPages: " + result.getNumberOfPages());
+    }
+
+
     @Test
     public void testRotations() throws IOException {
 

redaction-service-v1/redaction-service-server-v1/src/test/resources/dictionaries/PII.txt

@@ -12,4 +12,5 @@ Xinyi Y. Tao
 Dorn
 Prasher
 David
-annotation
+annotation
+J.B. RASCLE