feature: add OAI-PMH API

.github/workflows/tests.yml

@@ -31,21 +31,14 @@ jobs:
           python-version: [3.7, 3.8, 3.9]
           requirements-level: [pypi]
           db-service: [postgresql12]
-          search-service: [elasticsearch6, elasticsearch7]
+          search-service: [elasticsearch7]
           exclude:
           - python-version: 3.7
             search-service: elasticsearch7
-          - python-version: 3.8
-            search-service: elasticsearch6
-          - python-version: 3.9
-            search-service: elasticsearch6
           include:
           - db-service: postgresql12
             DB_EXTRAS: "postgresql"
 
-          - search-service: elasticsearch6
-            SEARCH_EXTRAS: "elasticsearch6"
-
           - search-service: elasticsearch7
             SEARCH_EXTRAS: "elasticsearch7"
 

invenio_rdm_records/ext.py

@@ -24,6 +24,14 @@
     SubjectsResourceConfig, SubjectsService, SubjectsServiceConfig
 from itsdangerous import SignatureExpired
 
+from invenio_rdm_records.oaiserver.resources.config import \
+    OAIPMHServerResourceConfig
+from invenio_rdm_records.oaiserver.resources.resources import \
+    OAIPMHServerResource
+from invenio_rdm_records.oaiserver.services.config import \
+    OAIPMHServerServiceConfig
+from invenio_rdm_records.oaiserver.services.services import OAIPMHServerService
+
 from . import config
 from .resources import RDMDraftFilesResourceConfig, \
     RDMParentRecordLinksResource, RDMParentRecordLinksResourceConfig, \
@@ -138,6 +146,7 @@ class ServiceConfigs:
             affiliations = AffiliationsServiceConfig
             names = NamesServiceConfig
             subjects = SubjectsServiceConfig
+            oaipmh_server = OAIPMHServerServiceConfig
 
         return ServiceConfigs
 
@@ -164,6 +173,10 @@ def init_services(self, app):
             config=service_configs.subjects
         )
 
+        self.oaipmh_server_service = OAIPMHServerService(
+            config=service_configs.oaipmh_server,
+        )
+
     def init_resource(self, app):
         """Initialize vocabulary resources."""
         self.records_resource = RDMRecordResource(
@@ -203,6 +216,12 @@ def init_resource(self, app):
             config=SubjectsResourceConfig,
         )
 
+        # OAI-PMH
+        self.oaipmh_server_resource = OAIPMHServerResource(
+            service=self.oaipmh_server_service,
+            config=OAIPMHServerResourceConfig,
+        )
+
     def fix_datacite_configs(self, app):
         """Make sure that the DataCite config items are strings."""
         datacite_config_items = [

invenio_rdm_records/oaiserver/__init__.py

@@ -0,0 +1,8 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2022 Graz University of Technology.
+#
+# 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.
+
+"""OAI-PMH API for InvenioRDM."""

invenio_rdm_records/oaiserver/resources/__init__.py

@@ -0,0 +1,8 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2022 Graz University of Technology.
+#
+# 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.
+
+"""OAI-PMH resources for InvenioRDM."""

invenio_rdm_records/oaiserver/resources/config.py

@@ -0,0 +1,69 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2022 Graz University of Technology.
+#
+# 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.
+
+"""OAI-PMH resource configuration."""
+
+import marshmallow as ma
+from flask_babelex import lazy_gettext as _
+from flask_resources import HTTPJSONException, ResourceConfig, \
+    create_error_handler
+from invenio_records_resources.resources.errors import ErrorHandlersMixin
+from invenio_records_resources.resources.records.args import \
+    SearchRequestArgsSchema
+
+from ..services.errors import OAIPMHError, OAIPMHSetDoesNotExistError, \
+    OAIPMHSetIDDoesNotExistError
+
+oaipmh_error_handlers = {
+    **ErrorHandlersMixin.error_handlers,
+    OAIPMHSetDoesNotExistError: create_error_handler(
+        lambda e: HTTPJSONException(
+            code=404,
+            description=e.description,
+        )
+    ),
+    OAIPMHSetIDDoesNotExistError: create_error_handler(
+        lambda e: HTTPJSONException(
+            code=404,
+            description=e.description,
+        )
+    ),
+    OAIPMHError: create_error_handler(
+        lambda e: HTTPJSONException(
+            code=400,
+            description=e.description,
+        )
+    ),
+}
+
+
+class OAIPMHServerSearchRequestArgsSchema(SearchRequestArgsSchema):
+    """OAI-PMH request parameters."""
+
+    managed = ma.fields.Boolean()
+    sort_direction = ma.fields.Str()
+
+
+class OAIPMHServerResourceConfig(ResourceConfig):
+    """OAI-PMH resource config."""
+
+    # Blueprint configuration
+    blueprint_name = "oaipmh-server"
+    url_prefix = "/oaipmh"
+    routes = {
+        "set-prefix": "/sets",
+        "list": "",
+        "item": "/<id>",
+        "format-prefix": "/formats",
+    }
+
+    # Request parsing
+    request_read_args = {}
+    request_view_args = {"id": ma.fields.Int()}
+    request_search_args = OAIPMHServerSearchRequestArgsSchema
+
+    error_handlers = oaipmh_error_handlers

invenio_rdm_records/oaiserver/resources/resources.py

@@ -0,0 +1,114 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2022 Graz University of Technology.
+#
+# 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.
+
+"""OAI-PMH resource."""
+
+from flask import abort, g
+from flask.globals import request
+from flask_resources import Resource, resource_requestctx, response_handler, \
+    route
+from invenio_records_resources.resources.errors import ErrorHandlersMixin
+from invenio_records_resources.resources.records.resource import \
+    request_data, request_headers, request_search_args, request_view_args
+from invenio_records_resources.resources.records.utils import es_preference
+
+
+class OAIPMHServerResource(ErrorHandlersMixin, Resource):
+    """OAI-PMH server resource."""
+
+    def __init__(self, config, service):
+        """Constructor."""
+        super().__init__(config)
+        self.service = service
+
+    def create_url_rules(self):
+        """Create the URL rules for the OAI-PMH server resource."""
+        routes = self.config.routes
+        url_rules = [
+            route("GET", routes["set-prefix"] + routes["list"], self.search),
+            route("POST", routes["set-prefix"], self.create),
+            route("GET", routes["set-prefix"] + routes["item"], self.read),
+            route("PUT", routes["set-prefix"] + routes["item"], self.update),
+            route(
+                "DELETE", routes["set-prefix"] + routes["item"], self.delete
+            ),
+            route(
+                "GET",
+                routes["format-prefix"] + routes["list"],
+                self.read_formats,
+            ),
+        ]
+
+        return url_rules
+
+    #
+    # Primary Interface
+    #
+    @request_search_args
+    @response_handler(many=True)
+    def search(self):
+        """Perform a search over the items."""
+        identity = g.identity
+        hits = self.service.search(
+            identity=identity,
+            params=resource_requestctx.args,
+        )
+        return hits.to_dict(), 200
+
+    @request_data
+    @response_handler()
+    def create(self):
+        """Create an item."""
+        item = self.service.create(
+            g.identity,
+            resource_requestctx.data or {},
+        )
+        return item.to_dict(), 201
+
+    # @request_read_args
+    @request_view_args
+    @response_handler()
+    def read(self):
+        """Read an item."""
+        item = self.service.read(
+            g.identity,
+            resource_requestctx.view_args["id"],
+        )
+        return item.to_dict(), 200
+
+    @request_headers
+    @request_view_args
+    @request_data
+    @response_handler()
+    def update(self):
+        """Update an item."""
+        item = self.service.update(
+            g.identity,
+            resource_requestctx.view_args["id"],
+            resource_requestctx.data,
+        )
+        return item.to_dict(), 200
+
+    @request_headers
+    @request_view_args
+    def delete(self):
+        """Delete an item."""
+        self.service.delete(
+            g.identity,
+            resource_requestctx.view_args["id"],
+        )
+        return "", 204
+
+    @request_search_args
+    @response_handler(many=True)
+    def read_formats(self):
+        """Perform a search over the formats."""
+        identity = g.identity
+        hits = self.service.read_all_formats(
+            identity=identity,
+        )
+        return hits.to_dict(), 200

invenio_rdm_records/oaiserver/services/__init__.py

@@ -0,0 +1,8 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2022 Graz University of Technology.
+#
+# 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.
+
+"""OAI-PMH services for InvenioRDM."""

invenio_rdm_records/oaiserver/services/config.py

@@ -0,0 +1,108 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2022 Graz University of Technology.
+#
+# 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.
+
+"""OAI-PMH service API configuration."""
+
+from flask_babelex import gettext as _
+from invenio_oaiserver.models import OAISet
+from invenio_records_resources.services import ServiceConfig
+from invenio_records_resources.services.base import Link
+from invenio_records_resources.services.records.links import pagination_links
+from sqlalchemy import asc, desc
+
+from invenio_rdm_records.oaiserver.services.schema import \
+    OAIPMHMetadataFormat, OAIPMHSetSchema
+
+from ..services.links import OAIPMHSetLink
+from ..services.permissions import OAIPMHServerPermissionPolicy
+from ..services.results import OAIMetadataFormatItem, OAIMetadataFormatList, \
+    OAISetItem, OAISetList
+
+
+class SearchOptions:
+    """Search options."""
+
+    sort_default = 'created'
+    sort_direction_default = 'asc'
+
+    sort_direction_options = {
+        "asc": dict(
+            title=_('Ascending'),
+            fn=asc,
+        ),
+        "desc": dict(
+            title=_('Descending'),
+            fn=desc,
+        ),
+    }
+
+    sort_options = {
+        "name": dict(
+            title=_('Name'),
+            fields=['name'],
+        ),
+        "spec": dict(
+            title=_('Spec'),
+            fields=['spec'],
+        ),
+        "created": dict(
+            title=_('Created'),
+            fields=['created'],
+        ),
+        "updated": dict(
+            title=_('Updated'),
+            fields=['updated'],
+        ),
+    }
+    pagination_options = {
+        "default_results_per_page": 25,
+    }
+
+
+class OAIPMHServerServiceConfig(ServiceConfig):
+    """Service factory configuration."""
+
+    # Common configuration
+    permission_policy_cls = OAIPMHServerPermissionPolicy
+    result_item_cls = OAISetItem
+    result_list_cls = OAISetList
+
+    metadata_format_result_item_cls = OAIMetadataFormatItem
+    metadata_format_result_list_cls = OAIMetadataFormatList
+
+    # Record specific configuration
+    record_cls = OAISet
+
+    # Search configuration
+    search = SearchOptions
+
+    # Service schema
+    schema = OAIPMHSetSchema
+
+    metadata_format_schema = OAIPMHMetadataFormat
+
+    links_item = {
+        "self": OAIPMHSetLink("{+api}/oaipmh/sets/{id}"),
+        "oai-listrecords": OAIPMHSetLink(
+            "{+ui}/oai2d?verb=ListRecords&metadataPrefix=oai_dc&set={spec}"
+        ),
+        "oai-listidentifiers": OAIPMHSetLink(
+            "{+ui}/oai2d?verb=ListIdentifiers&metadataPrefix=oai_dc&set={spec}"
+        ),
+    }
+
+    links_search = {
+        **pagination_links("{+api}/oaipmh/sets{?args*}"),
+        "oai-listsets": Link("{+ui}/oai2d?verb=ListSets"),
+        "oai-listrecords": Link(
+            "{+ui}/oai2d?verb=ListRecords&metadataPrefix=oai_dc"
+        ),
+        "oai-listidentifiers": Link(
+            "{+ui}/oai2d?verb=ListIdentifiers&metadataPrefix=oai_dc"
+        ),
+        "oai-identify": Link("{+ui}/oai2d?verb=Identify"),
+    }