From 41494ee689aaf8cd9fbdd6cb6ad1b8c4d403b85d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Martin=20M=C3=BCller?= <mail@mueller-martin.net>
Date: Mon, 21 Feb 2022 22:06:43 +0100
Subject: [PATCH 1/4] Remove alpha layer from PNG files for img2pdf

Fixes issue #1254
---
 src/paperless_tesseract/parsers.py | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)

diff --git a/src/paperless_tesseract/parsers.py b/src/paperless_tesseract/parsers.py
index d53879542..edb3190ac 100644
--- a/src/paperless_tesseract/parsers.py
+++ b/src/paperless_tesseract/parsers.py
@@ -66,6 +66,16 @@ class RasterisedDocumentParser(DocumentParser):
             "image/gif",
         ]
 
+    def has_alpha(self, image):
+        try:
+            with Image.open(image) as im:
+                return im.mode in ('RGBA', 'LA')
+        except Exception as e:
+            self.log(
+                'warning',
+                f"Error while check for alpha channel in image {image}: {e}")
+            return None
+
     def get_dpi(self, image):
         try:
             with Image.open(image) as im:
@@ -182,6 +192,18 @@ class RasterisedDocumentParser(DocumentParser):
         if self.is_image(mime_type):
             dpi = self.get_dpi(input_file)
             a4_dpi = self.calculate_a4_dpi(input_file)
+
+            if self.has_alpha(input_file):
+                self.log(
+                    "info",
+                    f"Removing alpha layer from {input_file} for compatibility with img2pdf"
+                )
+                with Image.open(input_file) as im:
+                    background = Image.new('RGBA', im.size, (255, 255, 255))
+                    background.alpha_composite(im)
+                    background = background.convert('RGB')
+                    background.save(input_file, format=im.format)
+
             if dpi:
                 self.log(
                     "debug",

From 2a47b3f1a1e2fa7cc7499fe39abcc167a2c75a57 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Martin=20M=C3=BCller?= <mail@mueller-martin.net>
Date: Mon, 21 Feb 2022 22:34:34 +0100
Subject: [PATCH 2/4] Fix code style (line too long)

---
 src/paperless_tesseract/parsers.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/src/paperless_tesseract/parsers.py b/src/paperless_tesseract/parsers.py
index edb3190ac..02de914ce 100644
--- a/src/paperless_tesseract/parsers.py
+++ b/src/paperless_tesseract/parsers.py
@@ -196,7 +196,8 @@ class RasterisedDocumentParser(DocumentParser):
             if self.has_alpha(input_file):
                 self.log(
                     "info",
-                    f"Removing alpha layer from {input_file} for compatibility with img2pdf"
+                    f"Removing alpha layer from {input_file} "
+                    "for compatibility with img2pdf"
                 )
                 with Image.open(input_file) as im:
                     background = Image.new('RGBA', im.size, (255, 255, 255))

From 73a8569d21113e6e16d2b133eed7c0b48d173156 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Martin=20M=C3=BCller?= <mail@mueller-martin.net>
Date: Mon, 21 Feb 2022 22:38:25 +0100
Subject: [PATCH 3/4] Modify test for PNG image with alpha

---
 src/paperless_tesseract/tests/test_parser.py | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/src/paperless_tesseract/tests/test_parser.py b/src/paperless_tesseract/tests/test_parser.py
index e39f87017..1ee295edf 100644
--- a/src/paperless_tesseract/tests/test_parser.py
+++ b/src/paperless_tesseract/tests/test_parser.py
@@ -181,13 +181,14 @@ class TestParser(DirectoriesMixin, TestCase):
 
         self.assertContainsStrings(parser.get_text(), ["This is a test document."])
 
-    def test_image_simple_alpha_fail(self):
+    def test_image_simple_alpha(self):
         parser = RasterisedDocumentParser(None)
 
-        def f():
-            parser.parse(os.path.join(self.SAMPLE_FILES, "simple-alpha.png"), "image/png")
+        parser.parse(os.path.join(self.SAMPLE_FILES, "simple-alpha.png"), "image/png")
 
-        self.assertRaises(ParseError, f)
+        self.assertTrue(os.path.isfile(parser.archive_path))
+
+        self.assertContainsStrings(parser.get_text(), ["This is a test document."])
 
     def test_image_calc_a4_dpi(self):
         parser = RasterisedDocumentParser(None)

From 1e288100a96315ff1fa4de2be7872fa42e68502c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Martin=20M=C3=BCller?= <mail@mueller-martin.net>
Date: Mon, 21 Feb 2022 22:58:19 +0100
Subject: [PATCH 4/4] Remove unneded exception handler from has_alpha()

---
 src/paperless_tesseract/parsers.py | 10 ++--------
 1 file changed, 2 insertions(+), 8 deletions(-)

diff --git a/src/paperless_tesseract/parsers.py b/src/paperless_tesseract/parsers.py
index 02de914ce..c1eddcefe 100644
--- a/src/paperless_tesseract/parsers.py
+++ b/src/paperless_tesseract/parsers.py
@@ -67,14 +67,8 @@ class RasterisedDocumentParser(DocumentParser):
         ]
 
     def has_alpha(self, image):
-        try:
-            with Image.open(image) as im:
-                return im.mode in ('RGBA', 'LA')
-        except Exception as e:
-            self.log(
-                'warning',
-                f"Error while check for alpha channel in image {image}: {e}")
-            return None
+        with Image.open(image) as im:
+            return im.mode in ('RGBA', 'LA')
 
     def get_dpi(self, image):
         try: