feature/pdct-1777-support-multiple-geos-in-get

app/model/family.py

@@ -11,6 +11,7 @@ class FamilyReadDTO(BaseModel):
     title: str
     summary: str
     geography: str
+    geographies: list[str]
     category: str
     status: str
     metadata: dict[str, list[str]]
@@ -40,6 +41,7 @@ class FamilyWriteDTO(BaseModel):
     title: str
     summary: str
     geography: str
+    geographies: list[str]
     category: str
     metadata: dict[str, list[str]]
     collections: list[str]

app/repository/family.py

@@ -22,10 +22,11 @@
 from db_client.models.organisation.users import Organisation
 from sqlalchemy import Column, and_
 from sqlalchemy import delete as db_delete
-from sqlalchemy import desc, or_
+from sqlalchemy import desc, func, or_
 from sqlalchemy import update as db_update
 from sqlalchemy.exc import NoResultFound, OperationalError
-from sqlalchemy.orm import Query, Session
+from sqlalchemy.orm import Query, Session, lazyload
+from sqlalchemy.sql import Subquery
 from sqlalchemy_utils import escape_like
 
 from app.errors import RepositoryError
@@ -34,13 +35,22 @@
 
 _LOGGER = logging.getLogger(__name__)
 
-FamilyGeoMetaOrg = Tuple[Family, Geography, FamilyMetadata, Corpus, Organisation]
+FamilyGeoMetaOrg = Tuple[
+    Family,
+    list[str],
+    FamilyMetadata,
+    Corpus,
+    Organisation,
+]
 
 
-def _get_query(db: Session) -> Query:
+def _get_query_search_endpoint(db: Session) -> Query:
     # NOTE: SqlAlchemy will make a complete hash of query generation
     #       if columns are used in the query() call. Therefore, entire
     #       objects are returned.
+    # NOTE: Whilst we are updating the queries to handle multiple geographies, we are keeping
+    #       the returned results for search as they are currently to enable backwards
+    #       compatibility, and as to not extend beyond this pr.
     return (
         db.query(Family, Geography, FamilyMetadata, Corpus, Organisation)
         .join(FamilyGeography, FamilyGeography.family_import_id == Family.import_id)
@@ -55,9 +65,19 @@ def _get_query(db: Session) -> Query:
     )
 
 
-def _family_to_dto(
-    db: Session, fam_geo_meta_corp_org: FamilyGeoMetaOrg
+def _family_to_dto_search_endpoint(
+    db: Session,
+    fam_geo_meta_corp_org: Tuple[
+        Family,
+        Geography,
+        FamilyMetadata,
+        Corpus,
+        Organisation,
+    ],
 ) -> FamilyReadDTO:
+    # NOTE: Whilst we are updating the queries to handle multiple geographies, we are keeping
+    #       the returned results for search as they are currently to enable backwards
+    #       compatibility, and as to not extend beyond this pr.
     fam, geo_value, meta, corpus, org = fam_geo_meta_corp_org
     metadata = cast(dict, meta.value)
     org = cast(str, org.name)
@@ -67,6 +87,111 @@ def _family_to_dto(
         title=str(fam.title),
         summary=str(fam.description),
         geography=str(geo_value.value),
+        geographies=[str(geo_value.value)],
+        category=str(fam.family_category),
+        status=str(fam.family_status),
+        metadata=metadata,
+        slug=str(fam.slugs[0].name if len(fam.slugs) > 0 else ""),
+        events=[str(e.import_id) for e in fam.events],
+        published_date=fam.published_date,
+        last_updated_date=fam.last_updated_date,
+        documents=[str(d.import_id) for d in fam.family_documents],
+        collections=[
+            c.collection_import_id
+            for c in db.query(CollectionFamily).filter(
+                fam.import_id == CollectionFamily.family_import_id
+            )
+        ],
+        organisation=org,
+        corpus_import_id=cast(str, corpus.import_id),
+        corpus_title=cast(str, corpus.title),
+        corpus_type=cast(str, corpus.corpus_type_name),
+        created=cast(datetime, fam.created),
+        last_modified=cast(datetime, fam.last_modified),
+    )
+
+
+def get_family_geography_subquery(db: Session) -> Subquery:
+    """
+    Creates a subquery to aggregate geography values for families, accomodating
+    those with multiple associated geographies.
+
+    :param db Session: the database connection
+    :return Query: A subquery containing family import IDs and their associated geography values
+    """
+    return (
+        db.query(
+            FamilyGeography.family_import_id,
+            func.array_agg(Geography.value).label("geography_values"),
+        )
+        .join(Geography, Geography.id == FamilyGeography.geography_id)
+        .group_by(FamilyGeography.family_import_id)
+        .subquery()
+    )
+
+
+def _get_query(db: Session) -> Query:
+    # NOTE: SqlAlchemy will make a complete hash of query generation
+    #       if columns are used in the query() call. Therefore, entire
+    #       objects are returned.
+
+    geography_subquery = get_family_geography_subquery(db)
+
+    query = (
+        db.query(
+            Family,
+            geography_subquery.c.geography_values,
+            FamilyMetadata,
+            Corpus,
+            Organisation,
+        )
+        .join(
+            geography_subquery,
+            geography_subquery.c.family_import_id == Family.import_id,
+        )
+        .join(FamilyMetadata, FamilyMetadata.family_import_id == Family.import_id)
+        .join(FamilyCorpus, FamilyCorpus.family_import_id == Family.import_id)
+        .join(Corpus, Corpus.import_id == FamilyCorpus.corpus_import_id)
+        .join(Organisation, Corpus.organisation_id == Organisation.id)
+        .group_by(
+            Family.import_id,
+            Family.title,
+            geography_subquery.c.geography_values,
+            FamilyMetadata.family_import_id,
+            Corpus.import_id,
+            Organisation,
+        )
+        .options(
+            # Disable any default eager loading as this was causing multiplicity due to
+            # implicit joins in relationships on the selected models.
+            lazyload("*")
+        )
+    )
+    _LOGGER.error(query)
+
+    return query
+
+
+def _family_to_dto(
+    db: Session, fam_geo_meta_corp_org: FamilyGeoMetaOrg
+) -> FamilyReadDTO:
+    (
+        fam,
+        geo_values,
+        meta,
+        corpus,
+        org,
+    ) = fam_geo_meta_corp_org
+
+    metadata = cast(dict, meta.value)
+    org = cast(str, org.name)
+
+    return FamilyReadDTO(
+        import_id=str(fam.import_id),
+        title=str(fam.title),
+        summary=str(fam.description),
+        geography=str(geo_values[0]),
+        geographies=[str(value) for value in geo_values],
         category=str(fam.family_category),
         status=str(fam.family_status),
         metadata=metadata,
@@ -209,21 +334,24 @@ def search(
         search.append(Family.family_status == term.capitalize())
 
     condition = and_(*search) if len(search) > 1 else search[0]
+
     try:
-        query = _get_query(db).filter(condition)
+        query = _get_query_search_endpoint(db).filter(condition)
         if org_id is not None:
             query = query.filter(Organisation.id == org_id)
+
         found = (
             query.order_by(desc(Family.last_modified))
             .limit(search_params["max_results"])
             .all()
         )
+
     except OperationalError as e:
         if "canceling statement due to statement timeout" in str(e):
             raise TimeoutError
         raise RepositoryError(e)
 
-    return [_family_to_dto(db, f) for f in found]
+    return [_family_to_dto_search_endpoint(db, f) for f in found]
 
 
 def update(db: Session, import_id: str, family: FamilyWriteDTO, geo_id: int) -> bool:

docker-compose-test.yml

@@ -1,7 +1,7 @@
 name: navigator-admin-backend-test
 services:
   webapp:
-    command: poetry run pytest --disable-warnings ${TEST}
+    command: poetry run pytest --disable-warnings -vv ${TEST}
     build:
       context: ./
       dockerfile: Dockerfile.webapp

tests/helpers/family.py

@@ -11,6 +11,7 @@ def create_family_read_dto(
     title: str = "title",
     summary: str = "summary",
     geography: str = "CHN",
+    geographies: list[str] = ["CHN", "BRB", "BHS"],
     category: str = FamilyCategory.LEGISLATIVE.value,
     metadata: Optional[dict] = None,
     slug: str = "slug",
@@ -34,6 +35,7 @@ def create_family_read_dto(
         title=title,
         summary=summary,
         geography=geography,
+        geographies=geographies,
         category=category,
         status="status",
         metadata=metadata,
@@ -87,6 +89,7 @@ def create_family_write_dto(
     title: str = "title",
     summary: str = "summary",
     geography: str = "CHN",
+    geographies: list[str] = ["CHN", "BRB", "BHS"],
     category: str = FamilyCategory.LEGISLATIVE.value,
     metadata: Optional[dict] = None,
     collections: Optional[list[str]] = None,
@@ -106,6 +109,7 @@ def create_family_write_dto(
         title=title,
         summary=summary,
         geography=geography,
+        geographies=geographies,
         category=category,
         metadata=metadata,
         collections=collections,

tests/integration_tests/family/test_search.py

@@ -19,6 +19,7 @@ def test_search_geographies(
                 "title": "title",
                 "summary": "gregarious magazine rub",
                 "geography": "ALB",
+                "geographies": ["ALB"],
                 "category": "UNFCCC",
                 "status": "Created",
                 "metadata": {"author": ["CPR"], "author_type": ["Party"]},
@@ -38,6 +39,7 @@ def test_search_geographies(
                 "title": "title",
                 "summary": "flour umbrella established",
                 "geography": "ZMB",
+                "geographies": ["ZMB"],
                 "category": "UNFCCC",
                 "status": "Created",
                 "metadata": {"author": ["CPR"], "author_type": ["Party"]},
@@ -74,6 +76,71 @@ def test_search_geographies(
         assert ids == expected_ids
 
 
+def test_search_retrieves_families_with_multiple_geographies(
+    client: TestClient, data_db: Session, superuser_header_token
+):
+    setup_db(data_db)
+    add_data(
+        data_db,
+        [
+            {
+                "import_id": "A.0.0.4",
+                "title": "title",
+                "summary": "gregarious magazine rub",
+                "geography": "ALB",
+                "geographies": ["ALB", "USA", "BRB"],
+                "category": "UNFCCC",
+                "status": "Created",
+                "metadata": {"author": ["CPR"], "author_type": ["Party"]},
+                "organisation": "UNFCCC",
+                "corpus_import_id": "UNFCCC.corpus.i00000001.n0000",
+                "corpus_title": "UNFCCC Submissions",
+                "corpus_type": "Intl. agreements",
+                "slug": "Slug4",
+                "events": ["E.0.0.3"],
+                "published_date": "2018-12-24T04:59:33Z",
+                "last_updated_date": "2018-12-24T04:59:33Z",
+                "documents": ["D.0.0.1", "D.0.0.2"],
+                "collections": ["C.0.0.4"],
+            },
+            {
+                "import_id": "A.0.0.5",
+                "title": "title",
+                "summary": "flour umbrella established",
+                "geography": "AGO",
+                "geographies": ["AGO", "ALB"],
+                "category": "UNFCCC",
+                "status": "Created",
+                "metadata": {"author": ["CPR"], "author_type": ["Party"]},
+                "organisation": "UNFCCC",
+                "corpus_import_id": "UNFCCC.corpus.i00000001.n0000",
+                "corpus_title": "UNFCCC Submissions",
+                "corpus_type": "Intl. agreements",
+                "slug": "Slug5",
+                "events": ["E.0.0.3"],
+                "published_date": "2018-12-24T04:59:33Z",
+                "last_updated_date": "2018-12-24T04:59:33Z",
+                "documents": ["D.0.0.1", "D.0.0.2"],
+                "collections": ["C.0.0.4"],
+            },
+        ],
+    )
+
+    test_geography = {"display_name": "albania", "iso_code": "ALB"}
+
+    geographies_query = f"&geography={test_geography['display_name']}"
+    response = client.get(
+        f"/api/v1/families/?{geographies_query}",
+        headers=superuser_header_token,
+    )
+    assert response.status_code == status.HTTP_200_OK
+    data = response.json()
+    for item in data:
+        assert "geographies" in item
+        assert isinstance(item["geographies"], list)
+        assert test_geography["iso_code"] in item["geographies"]
+
+
 def test_search_family_super(
     client: TestClient, data_db: Session, superuser_header_token
 ):

tests/integration_tests/setup_db.py

@@ -32,6 +32,7 @@ class DBEntry(TypedDict):
     title: str
     summary: str
     geography: str
+    geographies: list[str]
     category: str
     status: str
     metadata: dict
@@ -53,6 +54,7 @@ class DBEntry(TypedDict):
         "title": "apple",
         "summary": "",
         "geography": "AFG",
+        "geographies": ["AFG"],
         "category": "UNFCCC",
         "status": "Created",
         "metadata": {
@@ -79,6 +81,7 @@ class DBEntry(TypedDict):
         "title": "apple orange banana",
         "summary": "apple",
         "geography": "ZWE",
+        "geographies": ["ZWE"],
         "category": "UNFCCC",
         "status": "Created",
         "metadata": {
@@ -105,6 +108,7 @@ class DBEntry(TypedDict):
         "title": "title",
         "summary": "orange peas",
         "geography": "AFG",
+        "geographies": ["AFG"],
         "category": "UNFCCC",
         "status": "Created",
         "metadata": {"author": ["CPR"], "author_type": ["Party"]},

tests/mocks/services/family_service.py

@@ -59,6 +59,7 @@ def mock_update_family(
                 data.title,
                 data.summary,
                 data.geography,
+                data.geographies,
                 data.category,
                 data.metadata,
                 "slug",