fix: encoding in Device.get_text()

Author: dtopuzov

core/utils/image_utils.py

@@ -13,8 +13,6 @@
 import pytesseract
 from PIL import Image
 
-from core.settings import Settings
-
 
 class ImageUtils(object):
     @staticmethod
@@ -112,7 +110,4 @@ def get_text(image_path):
         row_text = pytesseract.image_to_string(image, lang='eng',
                                                config="-c tessedit_char_whitelist=%s_-." % char_whitelist).strip()
         text = "".join([s for s in row_text.splitlines(True) if s.strip()])
-        if Settings.PYTHON_VERSION < 3:
-            return str(text.decode('utf8').encode('utf8')).strip()
-        else:
-            return text.decode("utf-8").strip()
+        return text.encode(encoding='utf-8', errors='ignore')

core_tests/unit/utils/image_tests.py

@@ -22,6 +22,7 @@ class ImageUtilsTests(unittest.TestCase):
 
     app_image = os.path.join(current_folder, 'resources', 'app.png')
     iphone_image = os.path.join(current_folder, 'resources', 'screenshot.png')
+    unicode_image = os.path.join(current_folder, 'resources', 'unicode.png')
     blue = numpy.array([255, 188, 48])
     white = numpy.array([255, 255, 255])
 
@@ -54,6 +55,11 @@ def test_04_get_text(self):
         assert 'Reminders' in text
         assert 'Settings' in text
 
+        # Unicode text
+        text = ImageUtils.get_text(self.unicode_image)
+        assert 'Ter Stegen' in text
+        assert 'Neymar' in text
+
 
 if __name__ == '__main__':
     unittest.main()