services: updated file schema

* added access field to file schema
* updated metadata field to be nested with a new schema
* added tests for setting file metadata

invenio_rdm_records/records/mappings/os-v1/rdmrecords/drafts/draft-v6.0.0.json

@@ -1292,6 +1292,14 @@
               },
               "file_id": {
                 "enabled": false
+              },
+              "access": {
+                "type": "object",
+                "properties": {
+                  "hidden": {
+                    "type": "boolean"
+                  }
+                }
               }
             }
           }
@@ -1351,6 +1359,14 @@
               },
               "file_id": {
                 "enabled": false
+              },
+              "access": {
+                "type": "object",
+                "properties": {
+                  "hidden": {
+                    "type": "boolean"
+                  }
+                }
               }
             }
           }

invenio_rdm_records/records/mappings/os-v1/rdmrecords/records/record-v6.0.0.json

@@ -1243,6 +1243,14 @@
               },
               "file_id": {
                 "enabled": false
+              },
+              "access": {
+                "type": "object",
+                "properties": {
+                  "hidden": {
+                    "type": "boolean"
+                  }
+                }
               }
             }
           }
@@ -1302,6 +1310,14 @@
               },
               "file_id": {
                 "enabled": false
+              },
+              "access": {
+                "type": "object",
+                "properties": {
+                  "hidden": {
+                    "type": "boolean"
+                  }
+                }
               }
             }
           }

invenio_rdm_records/records/mappings/os-v2/rdmrecords/drafts/draft-v6.0.0.json

@@ -1292,6 +1292,14 @@
               },
               "file_id": {
                 "enabled": false
+              },
+              "access": {
+                "type": "object",
+                "properties": {
+                  "hidden": {
+                    "type": "boolean"
+                  }
+                }
               }
             }
           }
@@ -1351,6 +1359,14 @@
               },
               "file_id": {
                 "enabled": false
+              },
+              "access": {
+                "type": "object",
+                "properties": {
+                  "hidden": {
+                    "type": "boolean"
+                  }
+                }
               }
             }
           }

invenio_rdm_records/records/mappings/os-v2/rdmrecords/records/record-v6.0.0.json

@@ -1243,6 +1243,14 @@
               },
               "file_id": {
                 "enabled": false
+              },
+              "access": {
+                "type": "object",
+                "properties": {
+                  "hidden": {
+                    "type": "boolean"
+                  }
+                }
               }
             }
           }
@@ -1302,6 +1310,14 @@
               },
               "file_id": {
                 "enabled": false
+              },
+              "access": {
+                "type": "object",
+                "properties": {
+                  "hidden": {
+                    "type": "boolean"
+                  }
+                }
               }
             }
           }

invenio_rdm_records/records/mappings/v7/rdmrecords/drafts/draft-v6.0.0.json

@@ -1292,6 +1292,14 @@
               },
               "file_id": {
                 "enabled": false
+              },
+              "access": {
+                "type": "object",
+                "properties": {
+                  "hidden": {
+                    "type": "boolean"
+                  }
+                }
               }
             }
           }
@@ -1351,6 +1359,14 @@
               },
               "file_id": {
                 "enabled": false
+              },
+              "access": {
+                "type": "object",
+                "properties": {
+                  "hidden": {
+                    "type": "boolean"
+                  }
+                }
               }
             }
           }

invenio_rdm_records/records/mappings/v7/rdmrecords/records/record-v6.0.0.json

@@ -1243,6 +1243,14 @@
               },
               "file_id": {
                 "enabled": false
+              },
+              "access": {
+                "type": "object",
+                "properties": {
+                  "hidden": {
+                    "type": "boolean"
+                  }
+                }
               }
             }
           }
@@ -1302,6 +1310,14 @@
               },
               "file_id": {
                 "enabled": false
+              },
+              "access": {
+                "type": "object",
+                "properties": {
+                  "hidden": {
+                    "type": "boolean"
+                  }
+                }
               }
             }
           }

invenio_rdm_records/services/config.py

@@ -67,6 +67,7 @@
 from .result_items import GrantItem, GrantList, SecretLinkItem, SecretLinkList
 from .schemas import RDMParentSchema, RDMRecordSchema
 from .schemas.community_records import CommunityRecordsSchema
+from .schemas.files import FileSchema
 from .schemas.parent.access import AccessSettingsSchema
 from .schemas.parent.access import Grant as GrantSchema
 from .schemas.parent.access import Grants as GrantsSchema
@@ -263,6 +264,8 @@ class RDMFileRecordServiceConfig(FileServiceConfig, ConfiguratorMixin):
         ),
     }
 
+    file_schema = FileSchema
+
 
 class RDMRecordServiceConfig(RecordServiceConfig, ConfiguratorMixin):
     """RDM record draft service config."""
@@ -575,6 +578,8 @@ class RDMMediaFileRecordServiceConfig(FileServiceConfig, ConfiguratorMixin):
         "content": FileLink("{+api}/records/{id}/media-files/{key}/content"),
     }
 
+    file_schema = FileSchema
+
 
 class RDMFileDraftServiceConfig(FileServiceConfig, ConfiguratorMixin):
     """Configuration for draft files."""
@@ -617,6 +622,8 @@ class RDMFileDraftServiceConfig(FileServiceConfig, ConfiguratorMixin):
         ),
     }
 
+    file_schema = FileSchema
+
 
 class RDMMediaFileDraftServiceConfig(FileServiceConfig, ConfiguratorMixin):
     """Configuration for draft media files."""
@@ -644,3 +651,5 @@ class RDMMediaFileDraftServiceConfig(FileServiceConfig, ConfiguratorMixin):
         "content": FileLink("{+api}/records/{id}/draft/media-files/{key}/content"),
         "commit": FileLink("{+api}/records/{id}/draft/media-files/{key}/commit"),
     }
+
+    file_schema = FileSchema

invenio_rdm_records/services/schemas/files.py

@@ -11,11 +11,27 @@
 from invenio_vocabularies.services.schema import (
     VocabularyRelationSchema as VocabularySchema,
 )
-from marshmallow import Schema, fields, pre_load
+from marshmallow import RAISE, Schema, fields, pre_load
 from marshmallow_utils.fields import NestedAttribute, SanitizedUnicode
 from marshmallow_utils.permissions import FieldPermissionsMixin
 
 
+class FileMetadataSchema(Schema):
+    """Schema for file metadata."""
+
+    class Meta:
+        """Meta."""
+
+        unknown = RAISE
+
+    page = fields.Integer()
+    type = fields.String()
+    language = fields.String()
+    encoding = fields.String()
+    charset = fields.String()
+    previewer = fields.String()
+
+
 class FileSchema(Schema):
     """File schema."""
 
@@ -28,7 +44,8 @@ class FileSchema(Schema):
 
     # FileRecord fields
     key = SanitizedUnicode()
-    metadata = fields.Dict()
+    metadata = fields.Nested(FileMetadataSchema)
+    access = fields.Dict(attribute="file.access")
 
 
 class FilesSchema(Schema, FieldPermissionsMixin):

tests/resources/test_iiif_image_api.py

@@ -9,6 +9,7 @@
 
 from PIL import Image
 from werkzeug.utils import secure_filename
+import dictdiffer
 
 
 def publish_record_with_images(
@@ -112,15 +113,20 @@ def test_iiif_info(
     recid = publish_record_with_images(client, file_id, minimal_record, headers)
     response = client.get(f"/iiif/record:{recid}:{file_id}/info.json")
     assert response.status_code == 200
-    assert response.json == {
-        "@context": "http://iiif.io/api/image/2/context.json",
-        "profile": ["http://iiif.io/api/image/2/level2.json"],
-        "protocol": "http://iiif.io/api/image",
-        "@id": f"https://127.0.0.1:5000/api/iiif/record:{recid}:{file_id}",
-        "tiles": [{"width": 256, "scaleFactors": [1, 2, 4, 8, 16, 32, 64]}],
-        "width": 1280,
-        "height": 1024,
-    }
+    assert not list(
+        dictdiffer.diff(
+            response.json,
+            {
+                "@context": "http://iiif.io/api/image/2/context.json",
+                "profile": ["http://iiif.io/api/image/2/level2.json"],
+                "protocol": "http://iiif.io/api/image",
+                "@id": f"https://127.0.0.1:5000/api/iiif/record:{recid}:{file_id}",
+                "tiles": [{"width": 256, "scaleFactors": [1, 2, 4, 8, 16, 32, 64]}],
+                "width": 1280,
+                "height": 1024,
+            },
+        )
+    )
 
     ## Testing with filename with a slash ##
 

tests/services/files/test_metadata_files.py

@@ -0,0 +1,76 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2024 CERN.
+#
+# Invenio-RDM-Records is free software; you can redistribute it and/or modify
+# it under the terms of the MIT License; see LICENSE file for more details.
+"""Test RDM records files (metadata)."""
+
+from io import BytesIO
+
+import pytest
+from marshmallow.exceptions import ValidationError
+
+from invenio_rdm_records.proxies import current_rdm_records_service
+
+
+def test_valid_metadata_set(running_app, db, location, minimal_record, identity_simple):
+    """Test setting file metadata."""
+    data = minimal_record.copy()
+    data["files"] = {"enabled": True}
+    service = current_rdm_records_service
+
+    file_service = service.files
+
+    # Create
+    draft = service.create(identity_simple, data)
+
+    # Initialize files and add valid metadata
+    metadata = {
+        "page": 1,
+        "encoding": "utf-8",
+        "language": "en",
+        "charset": "utf-8",
+        "type": "pdf",
+    }
+    service.draft_files.init_files(
+        identity_simple, draft.id, data=[{"key": "test.pdf", "metadata": metadata}]
+    )
+    service.draft_files.set_file_content(
+        identity_simple, draft.id, "test.pdf", BytesIO(b"test file")
+    )
+    service.draft_files.commit_file(identity_simple, draft.id, "test.pdf")
+
+    # Publish the record
+    record = service.publish(identity_simple, draft.id)
+
+    # Get file metadata
+    result = file_service.list_files(identity_simple, draft.id)
+    assert result.to_dict()["entries"][0]["metadata"] == metadata
+    assert result.to_dict()["entries"][0]["access"]["hidden"] is False  # default value
+
+
+def test_invalid_metadata_set(
+    running_app, db, location, minimal_record, identity_simple
+):
+    """Test setting file metadata with invalid data."""
+    data = minimal_record.copy()
+    data["files"] = {"enabled": True}
+    service = current_rdm_records_service
+
+    # Create
+    draft = service.create(identity_simple, data)
+
+    # Initialize files and add valid metadata
+    metadata = {
+        "page": 1,
+        "encoding": "utf-8",
+        "language": "en",
+        "charset": "utf-8",
+        "type": "pdf",
+        "invalid": "invalid",
+    }
+    with pytest.raises(ValidationError):
+        service.draft_files.init_files(
+            identity_simple, draft.id, data=[{"key": "test.pdf", "metadata": metadata}]
+        )