moving scicrungh unders the service layer

pcrespov · pcrespov · commit 15dfa640c2fa · 2025-09-26T19:32:25.000+02:00
diff --git a/services/web/server/src/simcore_service_webserver/groups/_classifiers_rest.py b/services/web/server/src/simcore_service_webserver/groups/_classifiers_rest.py
@@ -8,9 +8,8 @@
 
 from .._meta import API_VTAG
 from ..login.decorators import login_required
-from ..scicrunch.models import ResearchResource, ResourceHit
+from ..scicrunch.models import ResourceHit
 from ..scicrunch.scicrunch_service import ScicrunchResourcesService
-from ..scicrunch.service_client import SciCrunch
 from ..security.decorators import permission_required
 from ..utils_aiohttp import envelope_json_response
 from ._classifiers_service import GroupClassifiersService
@@ -50,15 +49,9 @@ async def get_group_classifiers(request: web.Request):
 @handle_plugin_requests_exceptions
 async def get_scicrunch_resource(request: web.Request):
     rrid = request.match_info["rrid"]
-    rrid = SciCrunch.validate_identifier(rrid)
 
-    # check if in database first
     service = ScicrunchResourcesService(request.app)
-    resource: ResearchResource | None = await service.get_resource(rrid)
-    if not resource:
-        # otherwise, request to scicrunch service
-        scicrunch = SciCrunch.get_instance(request.app)
-        resource = await scicrunch.get_resource_fields(rrid)
+    resource = await service.get_or_fetch_resource(rrid)
 
     return envelope_json_response(resource.model_dump())
 
@@ -73,16 +66,8 @@ async def get_scicrunch_resource(request: web.Request):
 async def add_scicrunch_resource(request: web.Request):
     rrid = request.match_info["rrid"]
 
-    # check if exists
     service = ScicrunchResourcesService(request.app)
-    resource: ResearchResource | None = await service.get_resource(rrid)
-    if not resource:
-        # then request scicrunch service
-        scicrunch = SciCrunch.get_instance(request.app)
-        resource = await scicrunch.get_resource_fields(rrid)
-
-        # insert new or if exists, then update
-        await service.upsert_resource(resource)
+    resource = await service.add_resource(rrid)
 
     return envelope_json_response(resource.model_dump())
 
@@ -97,17 +82,7 @@ async def add_scicrunch_resource(request: web.Request):
 async def search_scicrunch_resources(request: web.Request):
     guess_name = str(request.query["guess_name"]).strip()
 
-    scicrunch = SciCrunch.get_instance(request.app)
-    hits: list[ResourceHit] = await scicrunch.search_resource(guess_name)
-
-    return envelope_json_response([hit.model_dump() for hit in hits])
-
-
-@handle_plugin_requests_exceptions
-async def search_scicrunch_resources(request: web.Request):
-    guess_name = str(request.query["guess_name"]).strip()
-
-    scicrunch = SciCrunch.get_instance(request.app)
-    hits: list[ResourceHit] = await scicrunch.search_resource(guess_name)
+    service = ScicrunchResourcesService(request.app)
+    hits: list[ResourceHit] = await service.search_resources(guess_name)
 
     return envelope_json_response([hit.model_dump() for hit in hits])
diff --git a/services/web/server/src/simcore_service_webserver/groups/_classifiers_service.py b/services/web/server/src/simcore_service_webserver/groups/_classifiers_service.py
@@ -25,7 +25,6 @@
 
 from ..scicrunch.errors import ScicrunchError
 from ..scicrunch.scicrunch_service import ScicrunchResourcesService
-from ..scicrunch.service_client import SciCrunch
 from ._classifiers_repository import GroupClassifierRepository
 
 _logger = logging.getLogger(__name__)
@@ -108,46 +107,41 @@ async def get_group_classifiers(
 
         # otherwise, build dynamic tree with RRIDs
         try:
-            return await build_rrids_tree_view(self.app, tree_view_mode=tree_view_mode)
+            return await self._build_rrids_tree_view(tree_view_mode=tree_view_mode)
         except ScicrunchError:
             # Return empty view on any error (including ScicrunchError)
             return {}
 
-
-# HELPERS FOR API HANDLERS --------------
-
-
-async def build_rrids_tree_view(
-    app: web.Application, tree_view_mode: Literal["std"] = "std"
-) -> dict[str, Any]:
-    if tree_view_mode != "std":
-        raise web.HTTPNotImplemented(
-            text="Currently only 'std' option for the classifiers tree view is implemented"
+    async def _build_rrids_tree_view(
+        self, tree_view_mode: Literal["std"] = "std"
+    ) -> dict[str, Any]:
+        if tree_view_mode != "std":
+            msg = "Currently only 'std' option for the classifiers tree view is implemented"
+            raise NotImplementedError(msg)
+
+        service = ScicrunchResourcesService(self.app)
+
+        flat_tree_view: dict[TreePath, ClassifierItem] = {}
+        for resource in await service.list_resources(include_url=True):
+            try:
+                validated_item = ClassifierItem(
+                    classifier=resource.rrid,
+                    display_name=resource.name.title(),
+                    short_description=resource.description,
+                    url=resource.url,
+                )
+
+                node = TypeAdapter(TreePath).validate_python(
+                    validated_item.display_name.replace(":", " ")
+                )
+                flat_tree_view[node] = validated_item
+
+            except ValidationError as err:
+                _logger.warning(
+                    "Cannot convert RRID into a classifier item. Skipping. Details: %s",
+                    err,
+                )
+
+        return Classifiers.model_construct(classifiers=flat_tree_view).model_dump(
+            exclude_unset=True
         )
-
-    scicrunch = SciCrunch.get_instance(app)
-    service = ScicrunchResourcesService(app)
-
-    flat_tree_view: dict[TreePath, ClassifierItem] = {}
-    for resource in await service.list_resources():
-        try:
-            validated_item = ClassifierItem(
-                classifier=resource.rrid,
-                display_name=resource.name.title(),
-                short_description=resource.description,
-                url=scicrunch.get_resolver_web_url(resource.rrid),
-            )
-
-            node = TypeAdapter(TreePath).validate_python(
-                validated_item.display_name.replace(":", " ")
-            )
-            flat_tree_view[node] = validated_item
-
-        except ValidationError as err:
-            _logger.warning(
-                "Cannot convert RRID into a classifier item. Skipping. Details: %s", err
-            )
-
-    return Classifiers.model_construct(classifiers=flat_tree_view).model_dump(
-        exclude_unset=True
-    )
diff --git a/services/web/server/src/simcore_service_webserver/scicrunch/_service.py b/services/web/server/src/simcore_service_webserver/scicrunch/_service.py
@@ -9,7 +9,8 @@
 from pydantic import ValidationError
 
 from ._repository import ScicrunchResourcesRepository
-from .models import ResearchResource, ResearchResourceAtdB
+from .models import ResearchResource, ResearchResourceAtdB, ResourceHit
+from .service_client import SciCrunch
 
 _logger = logging.getLogger(__name__)
 
@@ -20,8 +21,9 @@ class ScicrunchResourcesService:
     def __init__(self, app: web.Application):
         self.app = app
         self._repo = ScicrunchResourcesRepository.create_from_app(app)
+        self._scicrunch = SciCrunch.get_instance(self.app)
 
-    async def list_resources(self) -> list[ResearchResource]:
+    async def list_resources(self, include_url: bool = False) -> list[ResearchResource]:
         """List all research resources as domain models."""
         rows = await self._repo.list_all_resources()
         if not rows:
@@ -30,7 +32,15 @@ async def list_resources(self) -> list[ResearchResource]:
         resources = []
         for row in rows:
             try:
-                resource = ResearchResource.model_validate(dict(row))
+                resource_data = dict(row)
+
+                # Add resolver URL if requested
+                if include_url:
+                    resource_data["url"] = self._scicrunch.get_resolver_web_url(
+                        row.rrid
+                    )
+
+                resource = ResearchResource.model_validate(resource_data)
                 resources.append(resource)
             except ValidationError as err:
                 _logger.warning(
@@ -88,3 +98,37 @@ async def upsert_resource(self, resource: ResearchResource) -> ResearchResource:
         values = resource.model_dump(exclude_unset=True)
         row = await self._repo.upsert_resource(values)
         return ResearchResource.model_validate(dict(row))
+
+    async def search_resources(self, guess_name: str) -> list[ResourceHit]:
+        """Search for research resources using SciCrunch API."""
+        guess_name = guess_name.strip()
+        if not guess_name:
+            return []
+
+        return await self._scicrunch.search_resource(guess_name)
+
+    async def add_resource(self, rrid: str) -> ResearchResource:
+        """Add a research resource by RRID, fetching from SciCrunch if not in database."""
+        # Check if exists in database first
+        resource = await self.get_resource(rrid)
+        if resource:
+            return resource
+
+        # If not found, request from scicrunch service
+        resource = await self._scicrunch.get_resource_fields(rrid)
+
+        # Insert new or update if exists
+        return await self.upsert_resource(resource)
+
+    async def get_or_fetch_resource(self, rrid: str) -> ResearchResource:
+        """Get resource from database first, fetch from SciCrunch API if not found."""
+        # Validate the RRID format first
+        validated_rrid = SciCrunch.validate_identifier(rrid)
+
+        # Check if in database first
+        resource = await self.get_resource(validated_rrid)
+        if resource:
+            return resource
+
+        # Otherwise, request from scicrunch service
+        return await self._scicrunch.get_resource_fields(validated_rrid)
diff --git a/services/web/server/src/simcore_service_webserver/scicrunch/models.py b/services/web/server/src/simcore_service_webserver/scicrunch/models.py
@@ -1,12 +1,12 @@
 """
-   Domain models at every interface: scicrunch API, pg database and webserver API
+Domain models at every interface: scicrunch API, pg database and webserver API
 """
 
 import logging
 import re
 from datetime import datetime
 
-from pydantic import field_validator, ConfigDict, BaseModel, Field
+from pydantic import BaseModel, ConfigDict, Field, HttpUrl, field_validator
 
 logger = logging.getLogger(__name__)
 
@@ -62,11 +62,13 @@ class ResearchResource(BaseModel):
     )
     name: str
     description: str
+    url: HttpUrl | None = None
 
     @field_validator("rrid", mode="before")
     @classmethod
-    def format_rrid(cls, v):
+    def _format_rrid(cls, v):
         return normalize_rrid_tags(v, with_prefix=True)
+
     model_config = ConfigDict(from_attributes=True, str_strip_whitespace=True)
 
 
