Non-Ascii characters support for the output file

PDFOC-91

Author: alexander.kischuk

pdfocr-tesseract4/src/test/java/com/itextpdf/pdfocr/IntegrationTestHelper.java

@@ -73,6 +73,7 @@ public class IntegrationTestHelper extends ExtendedITextTest {
     // directory with test files
     public static final String TEST_DIRECTORY = "./src/test/resources/com/itextpdf/pdfocr/";
     private static final String TARGET_FOLDER = "./target/test/resources/com/itextpdf/pdfocr/";
+    private static final String NON_ASCII_TARGET_DIRECTORY = "./target/test/resources/com/itextpdf/ñoñ-ascîî/";
 
     // directory with trained data for tests
     protected static final String LANG_TESS_DATA_DIRECTORY = TEST_DIRECTORY + "tessdata";
@@ -159,6 +160,16 @@ public static String getTargetDirectory() {
         return TARGET_FOLDER;
     }
 
+    /**
+     * Returns a non ascii target directory.
+     */
+    public static String getNonAsciiTargetDirectory() {
+        if (!Files.exists(java.nio.file.Paths.get(NON_ASCII_TARGET_DIRECTORY))) {
+            createDestinationFolder(NON_ASCII_TARGET_DIRECTORY);
+        }
+        return NON_ASCII_TARGET_DIRECTORY;
+    }
+
     protected static File getTessDataDirectory() {
         return new File(LANG_TESS_DATA_DIRECTORY);
     }

pdfocr-tesseract4/src/test/java/com/itextpdf/pdfocr/tessdata/TessDataIntegrationTest.java

@@ -623,6 +623,18 @@ public void testJapaneseScript() {
         Assert.assertEquals(expected, result);
     }
 
+    @Test
+    public void testTargetDirectoryWithNonAsciiPath() {
+        String imgPath = TEST_IMAGES_DIRECTORY + "german_01.jpg";
+        String expectedTxt = TEST_DOCUMENTS_DIRECTORY + "german_01" + testFileTypeName + ".txt";
+        List<String> languages = Collections.<String>singletonList("deu");
+        String resultTxtFile = getNonAsciiTargetDirectory() + getImageName(imgPath, languages) + ".txt";
+        doOcrAndSaveToTextFile(tesseractReader, imgPath, resultTxtFile, languages);
+
+        boolean result = compareTxtFiles(expectedTxt, resultTxtFile);
+        Assert.assertTrue(result);
+    }
+
     /**
      * Do OCR for given image and compare result text file with expected one.
      */