chore: add doi and publisher name to search

Author: olevski

components/renku_data_services/search/db.py

@@ -80,6 +80,8 @@ def _dataconnector_to_entity_doc(dc: DataConnector | GlobalDataConnector) -> Dat
         description=dc.description,
         keywords=dc.keywords if dc.keywords is not None else [],
         version=DocVersions.off(),
+        doi=dc.doi if hasattr(dc, "doi") else None,
+        publisherName=dc.publisher_name if hasattr(dc, "publisher_name") else None,
     )
 
 

components/renku_data_services/search/solr_user_query.py

@@ -18,13 +18,15 @@
     Created,
     CreatedByIs,
     DirectMemberIs,
+    DoiIs,
     IdIs,
     InheritedMemberIs,
     KeywordIs,
     NameIs,
     NamespaceIs,
     Order,
     OrderBy,
+    PublisherNameIs,
     RoleIs,
     SlugIs,
     SortableField,
@@ -344,6 +346,14 @@ async def visit_slug_is(self, ft: SlugIs) -> None:
         """Process slug-is segment."""
         self.__append(st.field_is_any(Fields.slug, ft.values.map(st.from_str)))
 
+    async def visit_doi_is(self, ft: DoiIs) -> None:
+        """Process doi-is segment."""
+        self.__append(st.field_is_any(Fields.doi, ft.values.map(st.from_str)))
+
+    async def visit_publisher_name_is(self, ft: PublisherNameIs) -> None:
+        """Process publisher_name-is segment."""
+        self.__append(st.field_is_any(Fields.publisher_name, ft.values.map(st.from_str)))
+
     async def visit_visibility_is(self, ft: VisibilityIs) -> None:
         """Process visibility-is segment."""
         self.__append(st.field_is_any(Fields.visibility, ft.values.map(st.from_visibility)))

components/renku_data_services/search/user_query.py

@@ -51,6 +51,8 @@ class Field(StrEnum):
     namespace = "namespace"
     direct_member = "direct_member"
     inherited_member = "inherited_member"
+    doi = "doi"
+    publisher_name = "publisher_name"
 
 
 class Comparison(StrEnum):
@@ -631,6 +633,54 @@ async def accept(self, visitor: SegmentVisitior) -> None:
         return await visitor.visit_role_is(self)
 
 
+@dataclass
+class DoiIs(FieldComparison):
+    """Compare the doi against a list of values."""
+
+    values: Nel[str]
+
+    @property
+    def field(self) -> Field:
+        """The field name."""
+        return Field.doi
+
+    @property
+    def cmp(self) -> Comparison:
+        """The comparison operation."""
+        return Comparison.is_equal
+
+    def _render_value(self) -> str:
+        return self.values.mk_string(",", Helper.quote)
+
+    async def accept(self, visitor: SegmentVisitior) -> None:
+        """Apply this to the visitor."""
+        return await visitor.visit_doi_is(self)
+
+
+@dataclass
+class PublisherNameIs(FieldComparison):
+    """Compare the publisher name against a list of values."""
+
+    values: Nel[str]
+
+    @property
+    def field(self) -> Field:
+        """The field name."""
+        return Field.publisher_name
+
+    @property
+    def cmp(self) -> Comparison:
+        """The comparison operation."""
+        return Comparison.is_equal
+
+    def _render_value(self) -> str:
+        return self.values.mk_string(",", Helper.quote)
+
+    async def accept(self, visitor: SegmentVisitior) -> None:
+        """Apply this to the visitor."""
+        return await visitor.visit_publisher_name_is(self)
+
+
 @dataclass
 class Text(SegmentBase):
     """A query part that is not corresponding to a specific field."""
@@ -712,6 +762,8 @@ async def accept(self, visitor: SegmentVisitior) -> None:
     | RoleIs
     | InheritedMemberIs
     | DirectMemberIs
+    | DoiIs
+    | PublisherNameIs
 )
 
 
@@ -812,6 +864,16 @@ def role_is(cls, role: Role, *args: Role) -> Segment:
         """Return role-is query segment."""
         return RoleIs(Nel(role, list(args)))
 
+    @classmethod
+    def doi_is(cls, doi: str, *args: str) -> Segment:
+        """Return slug-is query segment."""
+        return DoiIs(Nel(doi, list(args)))
+
+    @classmethod
+    def publisher_name_is(cls, publisher_name: str, *args: str) -> Segment:
+        """Return slug-is query segment."""
+        return PublisherNameIs(Nel(publisher_name, list(args)))
+
 
 @dataclass
 class UserQuery:
@@ -920,6 +982,16 @@ async def visit_inherited_member_is(self, ft: InheritedMemberIs) -> None:
         """Visit inherited-member-is."""
         ...
 
+    @abstractmethod
+    async def visit_doi_is(self, ft: DoiIs) -> None:
+        """Visit doi-is."""
+        ...
+
+    @abstractmethod
+    async def visit_publisher_name_is(self, ft: PublisherNameIs) -> None:
+        """Visit doi-is."""
+        ...
+
 
 class UserQueryVisitor[T](SegmentVisitior):
     """A visitor to transform user queries."""
@@ -1001,6 +1073,16 @@ async def visit_visibility_is(self, ft: VisibilityIs) -> None:
         """Forwards to `visit_field_term`."""
         return await self.visit_field_term(ft)
 
+    @override
+    async def visit_doi_is(self, ft: DoiIs) -> None:
+        """Forwards to `visit_field_term`."""
+        return await self.visit_field_term(ft)
+
+    @override
+    async def visit_publisher_name_is(self, ft: PublisherNameIs) -> None:
+        """Forwards to `visit_field_term`."""
+        return await self.visit_field_term(ft)
+
 
 class EmptyUserQueryVisitor[T](UserQueryFieldTermVisitor[T]):
     """A visitor with every method doing nothing.

components/renku_data_services/search/user_query_parser.py

@@ -28,6 +28,7 @@
     CreatedByIs,
     DateTimeCalc,
     DirectMemberIs,
+    DoiIs,
     Field,
     Helper,
     IdIs,
@@ -40,6 +41,7 @@
     PartialDate,
     PartialDateTime,
     PartialTime,
+    PublisherNameIs,
     RelativeDate,
     RoleIs,
     SlugIs,
@@ -115,6 +117,10 @@ def _make_field_term(args: tuple[str, Nel[str]]) -> Parser:
             return success(NamespaceIs(values))
         case Field.created_by:
             return success(CreatedByIs(values))
+        case Field.doi:
+            return success(DoiIs(values))
+        case Field.publisher_name:
+            return success(PublisherNameIs(values))
         case _:
             return fail(f"Invalid field name: {field}")
 

components/renku_data_services/solr/entity_documents.py

@@ -237,6 +237,8 @@ class DataConnector(EntityDoc, frozen=True):
     isNamespace: Annotated[Literal[False], BeforeValidator(lambda e: False)] = False
     namespaceDetails: ResponseBody | None = None
     creatorDetails: ResponseBody | None = None
+    publisherName: str | None = None
+    doi: str | None = None
 
     @property
     def entity_type(self) -> EntityType:
@@ -287,7 +289,7 @@ def _add_tzinfo(cls, v: datetime) -> datetime:
 
     @classmethod
     def from_dict(cls, d: dict[str, Any]) -> DataConnector:
-        """Create a Project from a dictionary."""
+        """Create a data connector from a dictionary."""
         return DataConnector.model_validate(d)
 
 

components/renku_data_services/solr/entity_schema.py

@@ -50,6 +50,10 @@ class Fields:
     creator_details: Final[FieldName] = FieldName("creatorDetails")
     namespace_details: Final[FieldName] = FieldName("namespaceDetails")
 
+    # data connector fields
+    doi: Final[FieldName] = FieldName("doi")
+    publisher_name: Final[FieldName] = FieldName("publisherName")
+
 
 class Analyzers:
     """A collection of analyzers."""
@@ -155,4 +159,14 @@ class FieldTypes:
         ],
         requires_reindex=True,
     ),
+    SchemaMigration(
+        version=13,
+        commands=[
+            AddCommand(Field.of(Fields.doi, FieldTypes.string)),
+            AddCommand(CopyFieldRule(source=Fields.doi, dest=Fields.content_all)),
+            AddCommand(Field.of(Fields.publisher_name, FieldTypes.string)),
+            AddCommand(CopyFieldRule(source=Fields.publisher_name, dest=Fields.content_all)),
+        ],
+        requires_reindex=False,
+    ),
 ]

components/renku_data_services/solr/solr_schema.py

@@ -110,7 +110,9 @@ class FieldTypeClasses:
     type_float = FieldTypeClass("FloatPointField")
     type_double = FieldTypeClass("DoublePointField")
     type_text = FieldTypeClass("TextField")
+    """TextField gets tokenized in Solr by default in our deployment."""
     type_str = FieldTypeClass("StrField")
+    """StrField does not get tokenized in Solr by default in our deployment."""
     type_uuid = FieldTypeClass("UUIDField")
     type_rank = FieldTypeClass("RankField")
     type_date_point = FieldTypeClass("DatePointField")