fixes bauerj/paperless_app#23 and most of all other scanner apps out there.

src/paperless_tesseract/parsers.py

@@ -110,6 +110,24 @@ class RasterisedDocumentParser(DocumentParser):
                 f"Error while getting DPI from image {image}: {e}")
             return None
 
+    def calculate_a4_dpi(self, image):
+        try:
+            with Image.open(image) as im:
+                width, height = im.size
+                # divide image width by A4 width (210mm) in inches.
+                dpi = int(width / (21 / 2.54))
+                self.log(
+                    'debug',
+                    f"Estimated DPI {dpi} based on image width {width}"
+                )
+                return dpi
+
+        except Exception as e:
+            self.log(
+                'warning',
+                f"Error while calculating DPI for image {image}: {e}")
+            return None
+
     def parse(self, document_path, mime_type):
         mode = settings.OCR_MODE
 
@@ -162,6 +180,7 @@ class RasterisedDocumentParser(DocumentParser):
 
         if self.is_image(mime_type):
             dpi = self.get_dpi(document_path)
+            a4_dpi = self.calculate_a4_dpi(document_path)
             if dpi:
                 self.log(
                     "debug",
@@ -170,6 +189,8 @@ class RasterisedDocumentParser(DocumentParser):
                 ocr_args['image_dpi'] = dpi
             elif settings.OCR_IMAGE_DPI:
                 ocr_args['image_dpi'] = settings.OCR_IMAGE_DPI
+            elif a4_dpi:
+                ocr_args['image_dpi'] = a4_dpi
             else:
                 raise ParseError(
                     f"Cannot produce archive PDF for image {document_path}, "
@@ -241,6 +262,9 @@ def strip_excess_whitespace(text):
 
 def get_text_from_pdf(pdf_file):
 
+    if not os.path.isfile(pdf_file):
+        return None
+
     with open(pdf_file, "rb") as f:
         try:
             pdf = pdftotext.PDF(f)

src/paperless_tesseract/tests/test_parser.py

@@ -164,8 +164,21 @@ class TestParser(DirectoriesMixin, TestCase):
 
         self.assertRaises(ParseError, f)
 
+    @mock.patch("paperless_tesseract.parsers.ocrmypdf.ocr")
+    def test_image_calc_a4_dpi(self, m):
+        parser = RasterisedDocumentParser(None)
 
-    def test_image_no_dpi_fail(self):
+        parser.parse(os.path.join(self.SAMPLE_FILES, "simple-no-dpi.png"), "image/png")
+
+        m.assert_called_once()
+
+        args, kwargs = m.call_args
+
+        self.assertEqual(kwargs['image_dpi'], 62)
+
+    @mock.patch("paperless_tesseract.parsers.RasterisedDocumentParser.calculate_a4_dpi")
+    def test_image_dpi_fail(self, m):
+        m.return_value = None
         parser = RasterisedDocumentParser(None)
 
         def f():