transfer: Upload thumbnails for image files.

Fixes: #35029.

zerver/lib/transfer.py

@@ -1,6 +1,7 @@
 import logging
 import os
 from concurrent.futures import ProcessPoolExecutor, as_completed
+from glob import glob
 
 import bmemcached
 import magic
@@ -83,6 +84,35 @@ def _transfer_message_files_to_s3(attachment: Attachment) -> None:
                 storage_class=settings.S3_UPLOADS_STORAGE_CLASS,
             )
             logging.info("Uploaded message file in path %s", file_path)
+        thumbnail_dir = os.path.join(settings.LOCAL_FILES_DIR, "thumbnail", attachment.path_id)
+        if os.path.isdir(thumbnail_dir):
+            thumbnails = 0
+            for thumbnail_path in glob(os.path.join(thumbnail_dir, "*")):
+                with open(thumbnail_path, "rb") as f:
+                    # This relies on the thumbnails having guessable
+                    # content-type from their path, in order to avoid
+                    # having to fetch the ImageAttachment inside the
+                    # ProcessPoolExecutor.  We also have no clean way
+                    # to prefetch those rows via select_related in the
+                    # outer query, as they match on `path_id`, which
+                    # is not supported as a foreign key.
+                    guessed_type = guess_type(thumbnail_path)[0]
+                    upload_content_to_s3(
+                        s3backend.uploads_bucket,
+                        os.path.join(
+                            "thumbnail", attachment.path_id, os.path.basename(thumbnail_path)
+                        ),
+                        guessed_type,
+                        None,
+                        f.read(),
+                        storage_class=settings.S3_UPLOADS_STORAGE_CLASS,
+                    )
+                thumbnails += 1
+            logging.info(
+                "Uploaded %d thumbnails into %s",
+                thumbnails,
+                os.path.join("thumbnail", attachment.path_id),
+            )
     except FileNotFoundError:  # nocoverage
         pass
 

zerver/tests/test_transfer.py

@@ -13,7 +13,7 @@ from zerver.lib.test_helpers import (
     get_test_image_file,
     read_test_image_file,
 )
-from zerver.lib.thumbnail import resize_emoji
+from zerver.lib.thumbnail import ThumbnailFormat, resize_emoji
 from zerver.lib.transfer import (
     transfer_avatars_to_s3,
     transfer_emoji_to_s3,
@@ -69,13 +69,27 @@ class TransferUploadsToS3Test(ZulipTestCase):
 
         upload_message_attachment("dummy1.txt", "text/plain", b"zulip1!", hamlet)
         upload_message_attachment("dummy2.txt", "text/plain", b"zulip2!", othello)
+        with (
+            self.thumbnail_formats(ThumbnailFormat("webp", 100, 75, animated=False)),
+            self.captureOnCommitCallbacks(execute=True),
+        ):
+            access_path, _ = upload_message_attachment(
+                "img.png", "image/png", read_test_image_file("img.png"), hamlet
+            )
+        self.assertTrue(access_path.startswith("/user_uploads/"))
+        image_path_id = access_path.removeprefix("/user_uploads/")
+        assert settings.LOCAL_FILES_DIR is not None
+        thumbnail_path = os.path.join(
+            settings.LOCAL_FILES_DIR, "thumbnail", image_path_id, "100x75.webp"
+        )
+        self.assertTrue(os.path.exists(thumbnail_path))
 
         with self.assertLogs(level="INFO"):
             transfer_message_files_to_s3(1)
 
         attachments = Attachment.objects.all().order_by("id")
 
-        self.assert_length(list(bucket.objects.all()), 2)
+        self.assert_length(list(bucket.objects.all()), 4)
 
         s3_dummy1 = bucket.Object(attachments[0].path_id).get()
         self.assertEqual(s3_dummy1["Body"].read(), b"zulip1!")
@@ -91,6 +105,26 @@ class TransferUploadsToS3Test(ZulipTestCase):
             {"realm_id": str(attachments[1].realm_id), "user_profile_id": str(othello.id)},
         )
 
+        s3_image = bucket.Object(attachments[2].path_id).get()
+        self.assertEqual(
+            s3_image["Body"].read(),
+            read_test_image_file("img.png"),
+        )
+        self.assertEqual(
+            s3_image["Metadata"],
+            {"realm_id": str(attachments[2].realm_id), "user_profile_id": str(hamlet.id)},
+        )
+
+        s3_image_thumbnail = bucket.Object(
+            os.path.join("thumbnail", attachments[2].path_id, "100x75.webp")
+        ).get()
+        self.assertEqual(s3_image_thumbnail["Metadata"], {})
+        with open(thumbnail_path, "rb") as thumbnail_file:
+            self.assertEqual(
+                s3_image_thumbnail["Body"].read(),
+                thumbnail_file.read(),
+            )
+
     @mock_aws
     def test_transfer_emoji_to_s3(self) -> None:
         bucket = create_s3_buckets(settings.S3_AVATAR_BUCKET)[0]