Merge pull request #193 from chinapandaman/PPF-190

PPF-190: any image to jpg

Author: chinapandaman

PyPDFForm/__init__.py

@@ -4,4 +4,4 @@
 
 PyPDFForm = Wrapper
 
-__version__ = "0.3.2"
+__version__ = "0.3.3"

PyPDFForm/core/image.py

@@ -48,3 +48,26 @@ def rotate_image(image_stream: bytes, rotation: Union[float, int]) -> bytes:
         rotated_buff.close()
 
         return result
+
+    @staticmethod
+    def any_image_to_jpg(image_stream: bytes) -> bytes:
+        """Converts an image of any type to jpg."""
+
+        buff = BytesIO()
+        buff.write(image_stream)
+        buff.seek(0)
+
+        image = Img.open(buff)
+
+        if image.format == "JPEG":
+            buff.close()
+            return image_stream
+
+        rgb_image = image.convert("RGB")
+        with BytesIO() as f:
+            rgb_image.save(f, format="JPEG")
+            f.seek(0)
+            result = f.read()
+
+        buff.close()
+        return result

PyPDFForm/middleware/wrapper.py

@@ -120,6 +120,9 @@ def _fill(
                 self.elements[key].validate_constants()
                 self.elements[key].validate_value()
                 self.elements[key].validate_text_attributes()
+                if self.elements[key].type == ElementType.image:
+                    value = ImageCore().any_image_to_jpg(value)
+                    self.elements[key].value = value
 
         self.stream = FillerCore().fill(self.stream, self.elements)
 
@@ -151,6 +154,7 @@ def _simple_fill(
             if isinstance(value, bytes):
                 if not ImageCore().is_image(value):
                     raise InvalidFormDataError
+                data[key] = ImageCore().any_image_to_jpg(value)
 
         if not isinstance(editable, bool):
             raise InvalidEditableParameterError
@@ -243,6 +247,7 @@ def draw_image(
         if not ImageCore().is_image(image):
             raise InvalidImageError
 
+        image = ImageCore().any_image_to_jpg(image)
         image = ImageCore().rotate_image(image, rotation)
 
         if not isinstance(page_number, int):

image_samples/after_converted.jpg

@@ -132,3 +132,23 @@ def test_draw_image_on_one_page_f_obj_param(
         assert obj.stream == expected
     else:
         assert obj.stream[:32767] == expected[:32767]
+
+
+def test_draw_png_image_on_one_page(template_stream, image_samples, pdf_samples):
+    with open(os.path.join(pdf_samples, "sample_pdf_with_png_image.pdf"), "rb+") as f:
+        obj = PyPDFForm(template_stream).draw_image(
+            os.path.join(image_samples, "before_converted.png"),
+            2,
+            100,
+            100,
+            400,
+            225,
+        )
+
+        expected = f.read()
+
+        if os.name == "nt":
+            assert len(obj.stream) == len(expected)
+            assert obj.stream == expected
+        else:
+            assert obj.stream[:32767] == expected[:32767]

image_samples/before_converted.png

@@ -522,6 +522,51 @@ def test_fill_images_f_obj_params_explicitly_setting_elements(
         assert obj.stream[:32767] == expected[:32767]
 
 
+def test_fill_png_images(pdf_samples, image_samples):
+    with open(os.path.join(pdf_samples, "sample_filled_png_images.pdf"), "rb+") as f:
+        expected = f.read()
+
+    obj = PyPDFForm(
+        os.path.join(pdf_samples, "sample_template_with_image_field.pdf"),
+        simple_mode=False,
+    ).fill(
+        {
+            "image_1": os.path.join(image_samples, "before_converted.png"),
+            "image_2": os.path.join(image_samples, "before_converted.png"),
+            "image_3": os.path.join(image_samples, "before_converted.png"),
+        }
+    )
+
+    if os.name == "nt":
+        assert len(obj.stream) == len(expected)
+        assert obj.stream == expected
+    else:
+        assert obj.stream[:16383] == expected[:16383]
+
+
+def test_simple_fill_png_images(pdf_samples, image_samples):
+    with open(
+        os.path.join(pdf_samples, "sample_filled_png_images_simple_mode.pdf"), "rb+"
+    ) as f:
+        expected = f.read()
+
+    obj = PyPDFForm(
+        os.path.join(pdf_samples, "sample_template_with_image_field.pdf"),
+    ).fill(
+        {
+            "image_1": os.path.join(image_samples, "before_converted.png"),
+            "image_2": os.path.join(image_samples, "before_converted.png"),
+            "image_3": os.path.join(image_samples, "before_converted.png"),
+        }
+    )
+
+    if os.name == "nt":
+        assert len(obj.stream) == len(expected)
+        assert obj.stream == expected
+    else:
+        assert obj.stream[:16383] == expected[:16383]
+
+
 def test_simple_fill_radiobutton(pdf_samples, template_with_radiobutton_stream):
     with open(
         os.path.join(pdf_samples, "sample_filled_radiobutton_simple.pdf"), "rb+"

pdf_samples/v2/sample_filled_png_images.pdf

@@ -25,6 +25,18 @@ def image_stream(image_samples):
         return f.read()
 
 
+@pytest.fixture
+def before_converted(image_samples):
+    with open(os.path.join(image_samples, "before_converted.png"), "rb+") as f:
+        return f.read()
+
+
+@pytest.fixture
+def after_converted(image_samples):
+    with open(os.path.join(image_samples, "after_converted.jpg"), "rb+") as f:
+        return f.read()
+
+
 def test_is_image(image_stream):
     assert not ImageCore().is_image(b"bad_stream")
     assert ImageCore().is_image(image_stream)
@@ -47,3 +59,13 @@ def test_rotate_image(image_stream):
 
     buff.close()
     rotated_buff.close()
+
+
+def test_any_image_to_jpg(before_converted, after_converted):
+    if os.name == "nt":
+        assert ImageCore().any_image_to_jpg(before_converted) == after_converted
+    else:
+        assert (
+            ImageCore().any_image_to_jpg(before_converted)[:511]
+            == after_converted[:511]
+        )