Migrate ckanext-fork to CKAN 2.11 + Python 3.10

.github/workflows/test.yml

@@ -1,18 +1,17 @@
 name: Tests
-on: [push, pull_request]
+on: [pull_request]
 jobs:
   test:
+    name: CKAN 2.11 - Focus on failing tests
     runs-on: ubuntu-latest
     container:
-      # The CKAN version tag of the Solr and Postgres containers should match
-      # the one of the container the tests run on.
-      # You can switch this base image with a custom image tailored to your project
-      image: openknowledge/ckan-dev:2.9
+      image: ckan/ckan-dev:2.11-py3.10
+      options: --user root
     services:
       solr:
-        image: ckan/ckan-solr-dev:2.9
+        image: ckan/ckan-solr:2.11-solr9
       postgres:
-        image: ckan/ckan-postgres-dev:2.9
+        image: ckan/ckan-postgres-dev:2.11
         env:
           POSTGRES_USER: postgres
           POSTGRES_PASSWORD: postgres
@@ -29,7 +28,7 @@ jobs:
       CKAN_REDIS_URL: redis://redis:6379/1
 
     steps:
-    - uses: actions/checkout@v2
+    - uses: actions/checkout@v6
     - name: Install requirements
       # Install any extra requirements your extension has here (dev requirements, other extensions etc)
       run: |
@@ -43,6 +42,9 @@ jobs:
         sed -i -e 's/use = config:.*/use = config:\/srv\/app\/src\/ckan\/test-core.ini/' test.ini
 
         ckan -c test.ini db init
+        ckan -c test.ini db upgrade
     - name: Run tests
-      run: pytest --ckan-ini=test.ini --cov=ckanext.fork --disable-warnings ckanext/fork
+      run: |
+        # Run ALL tests
+        pytest --ckan-ini=test.ini --disable-warnings ckanext/fork/tests/
 

ckanext/fork/actions.py

@@ -63,22 +63,51 @@ def resource_autocomplete(context, data_dict):
         datasets = _get_dataset_from_resource_uuid(context, q_lower)
 
     if not datasets:
-        datasets = toolkit.get_action('package_search')(context, {
-            "q": q,
-            "rows": 10,
+        # CKAN's Solr schema (ckan/config/solr/schema.xml) indexes `res_name` but
+        # NOT `res_id`, so package_search cannot find datasets by resource UUID
+        # substring (required by ADX-879). We fetch up to 100 datasets and filter
+        # resource IDs in Python. Capped by rows=100: instances with more datasets
+        # won't match resource IDs beyond the top 100 by default sort. Lifting
+        # this cap would require adding res_id to Solr (e.g. via
+        # IPackageController.before_index) so package_search can filter directly.
+        search_results = toolkit.get_action('package_search')(context, {
+            "q": "*:*",
+            "rows": 100,
             "include_private": True
-        })['results']
+        })
+        datasets = search_results['results']
+
+    # Split query into tokens for dataset-level matching (allows partial matches like "01")
+    query_tokens = q_lower.split()
 
     for dataset in datasets:
 
         if not dataset['resources']:
             continue
 
         resources = []
+        has_matching_resource = False
 
         for resource in dataset['resources']:
             last_modified = toolkit.h.time_ago_from_timestamp(resource['last_modified'])
-            match = q_lower in resource['name'].lower() or q_lower == resource['id']
+            # For resources: match based on number of matching tokens
+            # - Single token queries: require exact match (full string)
+            # - Multi-token queries: require at least 2 tokens to match
+            resource_lower = resource['name'].lower()
+            resource_id_lower = resource['id'].lower()
+
+            if len(query_tokens) == 1:
+                # Single token: use full string matching
+                match = q_lower in resource_lower or q_lower in resource_id_lower
+            else:
+                # Multi-token: require at least 2 tokens to match
+                match = sum(
+                    1 for token in query_tokens
+                    if token in resource_lower or token in resource_id_lower
+                ) >= 2
+
+            if match:
+                has_matching_resource = True
             resources.append({
                 'id': resource['id'],
                 'name': resource['name'],
@@ -89,15 +118,31 @@ def resource_autocomplete(context, data_dict):
             })
 
         organization_title = dataset.get('organization', {}).get('title', "")
-        match = q_lower in dataset['name'].lower() or q_lower in dataset['title'].lower()
-        pkg_list.append({
-            'id': dataset['id'],
-            'name': dataset['name'],
-            'title': dataset['title'],
-            'owner_org': organization_title,
-            'match': match,
-            'resources': resources
-        })
+        # For datasets: match any token from query (allows "01" in "Resource 01" to match "test-dataset-01")
+        dataset_match = any(
+            token in dataset['name'].lower() or token in dataset['title'].lower()
+            for token in query_tokens
+        )
+
+        # Only include dataset if it matches at dataset level OR has matching resources
+        if dataset_match or has_matching_resource:
+            pkg_list.append({
+                'id': dataset['id'],
+                'name': dataset['name'],
+                'title': dataset['title'],
+                'owner_org': organization_title,
+                'match': dataset_match,
+                'resources': resources
+            })
+
+    # Sort results: datasets with name/title matches first, then resource-only matches
+    # Within each group, maintain the order from package_search
+    for i, item in enumerate(pkg_list):
+        item['_original_index'] = i
+    pkg_list.sort(key=lambda x: (not x['match'], x['_original_index']))
+    # Remove the temporary index
+    for item in pkg_list:
+        del item['_original_index']
 
     return pkg_list
 
@@ -118,7 +163,7 @@ def _get_dataset_from_resource_uuid(context, uuid):
             {"id": resource['package_id']}
         )
         return [package]
-    except toolkit.NotFound:
+    except logic.NotFound:
         return []
 
 

ckanext/fork/tests/conftest.py

@@ -1,23 +1,32 @@
 import pytest
+from ckan import model
 from ckan.tests import factories
 from ckan.tests.helpers import call_action
 
 
 @pytest.fixture
 def forked_data():
+    """
+    Creates test data with fork metadata (requires activity plugin).
+
+    Note: Tests using this fixture must use:
+    @pytest.mark.usefixtures('clean_db_with_migrations', 'with_plugins')
+    """
     giftless_metadata = {
         "sha256": "dummysha",
         "size": 999,
         "lfs_prefix": "test/resource",
         "url_type": "upload"
     }
+    user = factories.User()
     organization = factories.Organization()
     forked_dataset = factories.Dataset(owner_org=organization['id'])
     forked_resource = factories.Resource(
         package_id=forked_dataset['id'],
         **giftless_metadata
     )
-    call_action('package_patch', id=forked_dataset['id'], notes='An activity')
+    # Trigger activity creation (factories don't create activities)
+    call_action('package_patch', context={'user': user['name']}, id=forked_dataset['id'], notes='An activity')
     forked_activity_id = call_action(
         'package_activity_list',
         id=forked_dataset['id']
@@ -27,3 +36,20 @@ def forked_data():
         'resource': forked_resource,
         'activity_id': forked_activity_id
     }
+
+
+@pytest.fixture
+def clean_db_with_migrations(clean_db):
+    """
+    Extends the standard clean_db fixture to add activity plugin schema changes.
+
+    In CKAN 2.11, the activity plugin adds a permission_labels column (text[])
+    to the activity table. The clean_db fixture rebuilds the database without
+    plugin-specific migrations, so we manually add the column here.
+    """
+    # Add the permission_labels column required by CKAN 2.11 activity plugin
+    model.Session.execute("""
+        ALTER TABLE activity
+        ADD COLUMN IF NOT EXISTS permission_labels text[];
+    """)
+    model.Session.commit()