fix: Accept SVG as avatar_url (#2083)

Make download_avatar task to accept .svg up to 3MB

Issue: AAH-2836

Author: rochacbruno

CHANGES/2836.bugfix

@@ -0,0 +1 @@
+Support SVG avatar image on namespaces

galaxy_ng/app/management/commands/download-namespace-logos.py

@@ -24,15 +24,51 @@ class Command(BaseCommand):
 
     def add_arguments(self, parser):
         parser.add_argument("--namespace", help="find and sync only this namespace name")
+        parser.add_argument(
+            "--sha-report",
+            default=False,
+            action="store_true",
+            required=False,
+            help="report the number of namespaces with avatar_url but missing avatar_sha256",
+            dest="sha_report",
+        )
+        parser.add_argument(
+            "--only-missing-sha",
+            default=False,
+            action="store_true",
+            required=False,
+            help="When set it will limit the logo download only to those namespaces missing sha",
+            dest="only_missing_sha",
+        )
 
     def echo(self, message, style=None):
         style = style or self.style.SUCCESS
         self.stdout.write(style(message))
 
     def handle(self, *args, **options):
+        # all namespaces having avatar_url must have avatar_sha256 set
+        # query for namespaces missing avatar_sha256
+        ns_missing_avatar_sha = Namespace.objects.filter(
+            _avatar_url__isnull=False,
+            last_created_pulp_metadata__avatar_sha256__isnull=True
+        )
+        if ns_missing_avatar_sha:
+            self.echo(
+                f"{ns_missing_avatar_sha.count()} Namespaces missing avatar_sha256",
+                self.style.ERROR
+            )
+            self.echo(", ".join(ns_missing_avatar_sha.values_list("name", flat=True)))
+        else:
+            self.echo("There are no namespaces missing avatar_sha256!")
+
+        if options["sha_report"]:  # --sha-report indicated only report was requested
+            return
+
+        self.echo("Proceeding with namespace logo downloads")
 
         kwargs = {
             'namespace_name': options['namespace'],
+            'only_missing_sha': options['only_missing_sha'],
         }
 
         task = dispatch(
@@ -52,11 +88,23 @@ def handle(self, *args, **options):
             sys.exit(1)
 
 
-def download_all_logos(namespace_name=None):
+def download_all_logos(namespace_name=None, only_missing_sha=False):
+    """Force logo downloads.
+    namespace: limit to specified namespace (or list of namespaces)
+    only_missing_sha: Limit to namespaces having avatar_url but missing avatar_sha256
+    """
     if namespace_name:
-        qs = Namespace.objects.filter(name=namespace_name)
+        namespaces = namespace_name.split(",")
+        qs = Namespace.objects.filter(name__in=namespaces)
     else:
         qs = Namespace.objects.all()
+
+    if only_missing_sha:
+        qs = qs.filter(
+            _avatar_url__isnull=False,
+            last_created_pulp_metadata__avatar_sha256__isnull=True
+        )
+
     for namespace in qs:
         download_logo = False
         if namespace._avatar_url:

galaxy_ng/app/tasks/namespaces.py

@@ -1,5 +1,7 @@
 import aiohttp
 import asyncio
+import contextlib
+import xml.etree.cElementTree as et
 
 from django.db import transaction
 from django.forms.fields import ImageField
@@ -15,6 +17,9 @@
 from galaxy_ng.app.models import Namespace
 
 
+MAX_AVATAR_SIZE = 3 * 1024 * 1024  # 3MB
+
+
 def dispatch_create_pulp_namespace_metadata(galaxy_ns, download_logo):
 
     dispatch(
@@ -26,11 +31,16 @@ def dispatch_create_pulp_namespace_metadata(galaxy_ns, download_logo):
     )
 
 
-def _download_avatar(url):
+def _download_avatar(url, namespace_name):
+    # User-Agent needs to be added to avoid timing out on throtled servers.
+    headers = {
+        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:71.0)'  # +
+        ' Gecko/20100101 Firefox/71.0'
+    }
     timeout = aiohttp.ClientTimeout(total=None, sock_connect=600, sock_read=600)
     conn = aiohttp.TCPConnector(force_close=True)
     session = aiohttp.ClientSession(
-        connector=conn, timeout=timeout, headers=None, requote_redirect_url=False
+        connector=conn, timeout=timeout, headers=headers, requote_redirect_url=False
     )
 
     try:
@@ -41,19 +51,29 @@ def _download_avatar(url):
     finally:
         asyncio.get_event_loop().run_until_complete(session.close())
 
-    try:
+    # Limit size of the avatar to avoid memory issues when validating it
+    if img.artifact_attributes["size"] > MAX_AVATAR_SIZE:
+        raise ValidationError(
+            f"Avatar for {namespace_name} on {url} larger than {MAX_AVATAR_SIZE / 1024 / 1024}MB"
+        )
+
+    with contextlib.suppress(Artifact.DoesNotExist):
         return Artifact.objects.get(sha256=img.artifact_attributes["sha256"])
-    except Artifact.DoesNotExist:
-        pass
 
     with open(img.path, "rb") as f:
         tf = PulpTemporaryUploadedFile.from_file(f)
-
         try:
             ImageField().to_python(tf)
         except ValidationError:
-            print("file is not an image")
-            return
+            # Not a PIL valid image lets handle SVG case
+            tag = None
+            with contextlib.suppress(et.ParseError):
+                f.seek(0)
+                tag = et.parse(f).find(".").tag
+            if tag != '{http://www.w3.org/2000/svg}svg':
+                raise ValidationError(
+                    f"Provided avatar_url for {namespace_name} on {url} is not a valid image"
+                )
 
         # the artifact has to be saved before the file is closed, or s3transfer
         # will throw an error.
@@ -71,7 +91,7 @@ def _create_pulp_namespace(galaxy_ns_pk, download_logo):
     avatar_artifact = None
 
     if download_logo:
-        avatar_artifact = _download_avatar(galaxy_ns._avatar_url)
+        avatar_artifact = _download_avatar(galaxy_ns._avatar_url, galaxy_ns.name)
 
     avatar_sha = None
     if avatar_artifact:

galaxy_ng/tests/integration/api/test_namespace_management.py

@@ -177,6 +177,7 @@ def test_namespace_edit_logo(galaxy_client):
     wait_for_all_tasks_gk(gc)
     updated_again_namespace = gc.get(f"_ui/v1/my-namespaces/{name}/")
     assert updated_namespace["avatar_url"] != updated_again_namespace["avatar_url"]
+    assert updated_namespace["avatar_sha256"] is not None
 
     # verify no additional namespaces are created
     resp = gc.get("_ui/v1/my-namespaces/")