alternative workaround for wagtail bug wagtail/wagtail#1326

CHANGELOG.md

@@ -1,3 +1,7 @@
+# 2.3.5
+
+- For truncated images, do not import instead of allow truncated image. Wagtail can't support them so it breaks later in Wagtail usage.
+
 # 2.3.4
 
 - Fix import of categories and tags

blog/management/commands/import_wordpress.py

@@ -1,25 +1,32 @@
-from PIL import ImageFile
 from django.core.management.base import BaseCommand
 from blog.wordpress_import import WordpressImport
 
 
-# https://stackoverflow.com/questions/12984426/python-pil-ioerror-image-file-truncated-with-big-images
-ImageFile.LOAD_TRUNCATED_IMAGES = True
-
-
 class Command(BaseCommand):
     def add_arguments(self, parser):
-        parser.add_argument('blog_index', help="Title of blog index page to attach blogs")
-        parser.add_argument('--url', help="Base url of wordpress instance")
-        parser.add_argument('--convert-images', default=False, type=bool, help="Find and convert images to Wagtail Images")
-        parser.add_argument('--create-users', default=False, type=bool, help="Create users out of found Authors")
+        parser.add_argument(
+            "blog_index", help="Title of blog index page to attach blogs"
+        )
+        parser.add_argument("--url", help="Base url of wordpress instance")
+        parser.add_argument(
+            "--convert-images",
+            default=False,
+            type=bool,
+            help="Find and convert images to Wagtail Images",
+        )
+        parser.add_argument(
+            "--create-users",
+            default=False,
+            type=bool,
+            help="Create users out of found Authors",
+        )
 
     def handle(self, *args, **options):
-        url = options.get('url')
-        url = url.rstrip('/')
+        url = options.get("url")
+        url = url.rstrip("/")
         wordpress_import = WordpressImport(
             url,
-            convert_images=options.get('convert_images'),
-            create_users=options.get('create_users')
+            convert_images=options.get("convert_images"),
+            create_users=options.get("create_users"),
         )
         wordpress_import.get_posts()

blog/wordpress_import.py

@@ -15,6 +15,8 @@
 
 User = get_user_model()
 
+logger = logging.getLogger(__name__)
+
 
 class WordpressImport:
     url = ""
@@ -66,8 +68,8 @@ def get_posts(self):
                     self.process_post(post)
 
     def process_post(self, post):
-        logging.debug(post["content"]["rendered"])
-        logging.info(".")
+        logger.debug(post["content"]["rendered"])
+        logger.info(".")
         try:
             page = BlogPage.objects.descendant_of(self.blog_index).get(
                 slug=post["slug"]
@@ -217,7 +219,12 @@ def create_images_from_urls_in_content(self, body):
             image.save()
             if img.has_attr("srcset"):
                 img["srcset"] = ""
-            new_url = image.get_rendition("original").url
-            img["src"] = new_url
+            try:
+                new_url = image.get_rendition("original").url
+                img["src"] = new_url
+            except OSError:
+                # Avoid https://github.com/wagtail/wagtail/issues/1326 by not importing it
+                logger.warning(f"image {image} is unable to be imported")
+                image.delete()
         soup.body.hidden = True
         return soup.body

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "wagtail_blog"
-version = "2.3.4"
+version = "2.3.5"
 description = "A wordpress like blog app implemented in wagtail"
 authors = ["David Burke"]
 license = "Apache License"