Accroding to python-openxml/python-docx#187

python-docx can't parse some jpeg files correctly, so we use PIL
to format all jpeg files to workaround this issue when generate docx
file.

backend/api/admin/actions/export_docx.py

@@ -16,6 +16,9 @@
 
 from docxcompose.composer import Composer
 
+import PIL
+import time
+
 LOGGER = logging.getLogger('django')
 
 CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))
@@ -332,7 +335,16 @@ def _attchments(self):
             generator.new(
                 self.document, f"附件 {ConvertToLowerCaseNumbers(index)}", 12)
             data = urlopen(image.image_path).read()
-            self.document.add_picture(BytesIO(data), width=Mm(150))
+
+            image_data = PIL.Image.open(BytesIO(data))
+            if image_data.format == "JPEG":
+                # Use PIL to save all jpeg files again to workaround python-docx bug
+                # https://github.com/python-openxml/python-docx/issues/187
+                tmp_image_data = BytesIO()
+                image_data.save(tmp_image_data, format="jpeg")
+                self.document.add_picture(tmp_image_data, width=Mm(150))
+            else:
+                self.document.add_picture(BytesIO(data), width=Mm(150))
 
     def _page_break(self):
         self.document.add_page_break()

backend/api/admin/test/test_actions.py

@@ -57,7 +57,12 @@ def has_perm(self, perm):
         "factory_type": "2-1",
         "images": [
             "https://i.imgur.com/3XPyVuF.png",
+            # python-docx can't parse this image correctly
+            # https://github.com/python-openxml/python-docx/issues/187
+            "https://i.imgur.com/Q3fplPG.jpg",
             "https://i.imgur.com/eHQ8uWo.jpg",
+            "https://i.imgur.com/4AiXzf8.jpg",
+            "https://i.imgur.com/Jvh1OQm.jpg",
         ],
         "lat": 23.123,
         "lng": 120.2,
@@ -118,3 +123,20 @@ def test_export_doc_action(self):
         assert response[
             'Content-Type'
         ] == 'application/vnd.openxmlformats-officedocument.wordprocessingml.document'
+
+    def test_python_docx_workaround(self):
+        # python-docx can't parse some jpeg image correctly
+        # https://github.com/python-openxml/python-docx/issues/187
+        # So we use PIL to format the JPEG file to workaround this.
+        data = {
+            'action': 'export_as_docx',
+            'select_across': 0,
+            'index': 0,
+            '_selected_action': self.factories[1].id
+        }
+        response = self.client.post('/admin/api/factory/', data)
+
+        assert response.status_code == 200, f"status_code should be 200 but {response.status_code}"
+        assert response[
+                   'Content-Type'
+               ] == 'application/vnd.openxmlformats-officedocument.wordprocessingml.document'