Merge pull request #1022 from dandi/asset-glob-param

Add support for glob/regex filtering on `NestedAsset` list endpoint

dandiapi/api/tests/test_asset.py

@@ -1240,3 +1240,80 @@ def test_asset_direct_info(api_client, asset):
         'created': TIMESTAMP_RE,
         'modified': TIMESTAMP_RE,
     }
+
+
+@pytest.mark.django_db
+@pytest.mark.parametrize(
+    'glob_pattern,expected_paths',
+    [
+        (
+            '*.txt',
+            ['a/b.txt', 'a/b/c.txt', 'a/b/c/d.txt', 'a/b/c/e.txt', 'a/b/d/e.txt'],
+        ),
+        (
+            'a/b/c/*',
+            ['a/b/c/d.txt', 'a/b/c/e.txt'],
+        ),
+        ('a/b/d/*', ['a/b/d/e.txt']),
+    ],
+)
+def test_asset_rest_glob(api_client, asset_factory, version, glob_pattern, expected_paths):
+    paths = ('a/b.txt', 'a/b/c.txt', 'a/b/c/d.txt', 'a/b/c/e.txt', 'a/b/d/e.txt')
+    for path in paths:
+        version.assets.add(asset_factory(path=path))
+
+    resp = api_client.get(
+        f'/api/dandisets/{version.dandiset.identifier}/versions/{version.version}/assets/',
+        {'glob': glob_pattern},
+    )
+
+    assert expected_paths == [asset['path'] for asset in resp.json()['results']]
+
+
+@pytest.mark.django_db
+@pytest.mark.parametrize(
+    'regex_pattern,expected_paths',
+    [
+        (
+            '[0-9].txt',
+            ['1.txt', '1/2/3.txt'],
+        ),
+        (
+            '[a-z].txt',
+            ['a/b/c/d.txt', 'a/b/c/e.txt', 'a/b/d/e.txt'],
+        ),
+    ],
+)
+def test_asset_rest_regex_valid(api_client, asset_factory, version, regex_pattern, expected_paths):
+    paths = ('1.txt', '1/2/3.txt', 'a/b/c/d.txt', 'a/b/c/e.txt', 'a/b/d/e.txt')
+    for path in paths:
+        version.assets.add(asset_factory(path=path))
+
+    resp = api_client.get(
+        f'/api/dandisets/{version.dandiset.identifier}/versions/{version.version}/assets/',
+        {'regex': regex_pattern},
+    )
+
+    assert expected_paths == [asset['path'] for asset in resp.json()['results']]
+
+
+@pytest.mark.django_db
+def test_asset_rest_regex_invalid(api_client, version):
+    resp = api_client.get(
+        f'/api/dandisets/{version.dandiset.identifier}/versions/{version.version}/assets/',
+        {'regex': '[[[[['},  # provide an invalid regex
+    )
+
+    assert resp.status_code == 400
+
+
+@pytest.mark.django_db
+def test_asset_rest_glob_regex_together(api_client, version):
+    """Test that including both a glob and regex returns a 400 error."""
+    resp = api_client.get(
+        f'/api/dandisets/{version.dandiset.identifier}/versions/{version.version}/assets/',
+        {'regex': '[0-9].txt', 'glob': '*.txt'},
+    )
+
+    assert resp.status_code == 400
+    assert resp.json() == {'glob': ['Cannot specify both glob and regex']}

dandiapi/api/views/asset.py

@@ -1,3 +1,7 @@
+from __future__ import annotations
+
+import re
+
 try:
     from storages.backends.s3boto3 import S3Boto3Storage
 except ImportError:
@@ -41,6 +45,7 @@
 )
 from dandiapi.api.views.serializers import (
     AssetDetailSerializer,
+    AssetListSerializer,
     AssetPathsQueryParameterSerializer,
     AssetPathsResponseSerializer,
     AssetSerializer,
@@ -505,6 +510,39 @@ def destroy(self, request, versions__dandiset__pk, versions__version, **kwargs):
 
         return Response(None, status=status.HTTP_204_NO_CONTENT)
 
+    @swagger_auto_schema(query_serializer=AssetListSerializer)
+    def list(self, request, *args, **kwargs):
+        serializer = AssetListSerializer(data=request.query_params)
+        serializer.is_valid(raise_exception=True)
+
+        queryset = self.filter_queryset(self.get_queryset())
+        glob_pattern: str | None = serializer.validated_data.get('glob')
+        regex_pattern: str | None = serializer.validated_data.get('regex')
+
+        if regex_pattern is not None:
+            try:
+                # Validate the regex by calling re.compile on it
+                re.compile(regex_pattern)
+                queryset = queryset.filter(path__iregex=regex_pattern)
+            except re.error:
+                return Response(
+                    data=f'{regex_pattern} is not a valid regex pattern.',
+                    status=status.HTTP_400_BAD_REQUEST,
+                )
+
+        if glob_pattern is not None:
+            queryset = queryset.filter(
+                path__iregex=glob_pattern.replace('*', '.*').replace('.', '\\.')
+            )
+
+        page = self.paginate_queryset(queryset)
+        if page is not None:
+            serializer = self.get_serializer(page, many=True)
+            return self.get_paginated_response(serializer.data)
+
+        serializer = self.get_serializer(queryset, many=True)
+        return Response(serializer.data)
+
     @swagger_auto_schema(
         manual_parameters=[PATH_PREFIX_PARAM],
         responses={200: AssetPathsResponseSerializer()},

dandiapi/api/views/serializers.py

@@ -197,3 +197,15 @@ class AssetPathsQueryParameterSerializer(serializers.Serializer):
     path_prefix = serializers.CharField(default='')
     page = serializers.IntegerField(default=1)
     page_size = serializers.IntegerField(default=DandiPagination.page_size)
+
+
+class AssetListSerializer(serializers.Serializer):
+    glob = serializers.CharField(required=False)
+    regex = serializers.CharField(required=False)
+
+    def validate(self, attrs):
+        if 'glob' in attrs and 'regex' in attrs:
+            raise serializers.ValidationError(
+                {'glob': 'Cannot specify both glob and regex'},
+            )
+        return super().validate(attrs)