assemblyline-core 4.7.5.dev45__tar.gz → 4.7.5.dev46__tar.gz

{assemblyline_core-4.7.5.dev45 → assemblyline_core-4.7.5.dev46}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: assemblyline-core
-Version: 4.7.5.dev45
+Version: 4.7.5.dev46
 Summary: Assemblyline 4 - Core components
 Home-page: https://github.com/CybercentreCanada/assemblyline-core/
 Author: CCCS Assemblyline development team

assemblyline_core-4.7.5.dev46/assemblyline_core/VERSION

@@ -0,0 +1 @@
+4.7.5.dev46

{assemblyline_core-4.7.5.dev45 → assemblyline_core-4.7.5.dev46}/assemblyline_core/updater/helper.py

@@ -70,7 +70,13 @@ class DockerRegistry(ContainerRegistry):
 class AzureContainerRegistry(ContainerRegistry):
     def _get_proprietary_registry_tags(self, server, image_name, auth, verify, proxies=None, token_server=None):
         # Find latest tag for each types
-        url = f"https://{server}/v2/{image_name}/tags/list"
+        # Explicitly request the maximum page size (1000) supported by ACR. Without the 'n' parameter,
+        # ACR defaults to only 100 tags per page, alphabetically sorted (not by push date/time).
+        # Repositories with more tags than the page size (e.g. long-running services with nightly
+        # builds) would silently have their most recent tags dropped if pagination isn't followed,
+        # since lexicographic ordering of variable-length build numbers doesn't match numeric order
+        # (e.g. "100" sorts before "99").
+        url = f"https://{server}/v2/{image_name}/tags/list?n=1000"
 
         # Get tag list
         headers = {}
@@ -94,12 +100,25 @@ class AzureContainerRegistry(ContainerRegistry):
                 resp = self._perform_request(url, headers, verify, proxies)
 
         # At this point, we should have a response from the API
-        if resp and resp.ok:
-            # Test for positive list of tags
+        tags = []
+        while resp and resp.ok:
             resp_data = resp.json()
-            return resp_data['tags'] or []
+            tags.extend(resp_data.get('tags') or [])
 
-        return []
+            # Link header format: <https://{server}/v2/{image_name}/tags/list?n=1000&last=...>; rel="next"
+            next_link = resp.links.get("next")
+            if not next_link:
+                # If there's no next link, we've reached the last page of results
+                break
+
+            # Prepare the next URL for the subsequent request. The 'next' link may be relative or absolute.
+            next_url = next_link["url"]
+            if next_url.startswith('/'):
+                # Resolve relative URL to absolute URL if necessary
+                next_url = f"https://{server}{next_url}"
+            resp = self._perform_request(next_url, headers, verify, proxies)
+
+        return tags
 
 class GitHubContainerRegistry(ContainerRegistry):
     def _get_proprietary_registry_tags(self, server, image_name, auth, verify, proxies=None, token_server=None):
@@ -300,6 +319,8 @@ def get_latest_tag_for_service(service_config: ServiceConfig, system_config: Sys
         tags = registry._get_proprietary_registry_tags(server, image_name, auth,
                                                        not system_config.services.allow_insecure_registry,
                                                        proxies, token_server)
+        logger.info(f"{prefix}Fetched {len(tags)} raw tag(s) for service {service_name} - {image_name} "
+                f"from {server} before filtering.")
     # Pre-filter tags to only consider 'compatible' tags relative to the running system
     tags = [tag for tag in tags
             if re.match(f"({FRAMEWORK_VERSION})\\.({SYSTEM_VERSION})\\.\\d+\\.({update_channel})\\d+", tag)]

{assemblyline_core-4.7.5.dev45 → assemblyline_core-4.7.5.dev46}/assemblyline_core.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: assemblyline-core
-Version: 4.7.5.dev45
+Version: 4.7.5.dev46
 Summary: Assemblyline 4 - Core components
 Home-page: https://github.com/CybercentreCanada/assemblyline-core/
 Author: CCCS Assemblyline development team

{assemblyline_core-4.7.5.dev45 → assemblyline_core-4.7.5.dev46}/assemblyline_core.egg-info/SOURCES.txt

@@ -74,5 +74,6 @@ test/test_scaler.py
 test/test_scheduler.py
 test/test_signature_client.py
 test/test_tasking_client.py
+test/test_updater_helper.py
 test/test_vacuum.py
 test/test_workflow.py

assemblyline_core-4.7.5.dev46/test/test_updater_helper.py

@@ -0,0 +1,47 @@
+from unittest.mock import MagicMock, patch
+
+from assemblyline_core.updater.helper import AzureContainerRegistry
+
+SERVER = "registry.example"
+IMAGE_NAME = "namespace/service"
+FIRST_PAGE_URL = f"https://{SERVER}/v2/{IMAGE_NAME}/tags/list?n=1000"
+NEXT_PAGE_URL = f"https://{SERVER}/v2/{IMAGE_NAME}/tags/list?n=1000&last=tag-b"
+
+
+def _make_response(tags, links=None):
+    resp = MagicMock()
+    resp.ok = True
+    resp.json.return_value = {"tags": tags}
+    resp.links = links or {}
+    return resp
+
+
+def test_azure_registry_follows_pagination_links():
+    first_page = _make_response(
+        ["tag-a", "tag-b"],
+        links= {
+            'next': {
+                'url': NEXT_PAGE_URL,
+                'rel': 'next'
+            }
+        }
+    )
+    second_page = _make_response(["tag-c"], links=None)
+    requested_urls = []
+
+    def fake_get(url, headers=None, verify=None, proxies=None):
+        requested_urls.append(url)
+        if url == FIRST_PAGE_URL:
+            return first_page
+        if url == NEXT_PAGE_URL:
+            return second_page
+        raise AssertionError(f"Unexpected URL requested: {url}")
+
+    registry = AzureContainerRegistry()
+    with patch("assemblyline_core.updater.helper.requests.get", side_effect=fake_get):
+        tags = registry._get_proprietary_registry_tags(
+            server=SERVER, image_name=IMAGE_NAME, auth="Bearer token", verify=True
+        )
+
+    assert tags == ["tag-a", "tag-b", "tag-c"]
+    assert requested_urls == [FIRST_PAGE_URL, NEXT_PAGE_URL]

assemblyline_core-4.7.5.dev45/assemblyline_core/VERSION

@@ -1 +0,0 @@
-4.7.5.dev45