PDCT-1057 step1 - add corpus info to config

app/api/api_v1/schemas/metadata.py

@@ -5,10 +5,22 @@
 TaxonomyData = Mapping[str, Mapping[str, Union[bool, Sequence[str]]]]
 
 
+class CorpusData(BaseModel):
+    """Contains the Corpus and CorpusType info"""
+
+    corpus_import_id: str
+    title: str
+    description: str
+    corpus_type: str
+    corpus_type_description: str
+    taxonomy: TaxonomyData
+
+
 class OrganisationConfig(BaseModel):
     """Definition of stats used on homepage"""
 
-    taxonomy: TaxonomyData
+    taxonomy: TaxonomyData  # TODO: Remove this in subsequent deploy see PDCT-1057
+    copora: Sequence[CorpusData]
     total: int
     count_by_category: Mapping[str, int]
 

app/core/organisation.py

@@ -1,5 +1,5 @@
 from dataclasses import asdict
-from typing import cast
+from typing import Sequence, cast
 
 from db_client.models.dfce.family import (
     Corpus,
@@ -13,7 +13,7 @@
 from sqlalchemy import func
 from sqlalchemy.orm import Session
 
-from app.api.api_v1.schemas.metadata import OrganisationConfig, TaxonomyData
+from app.api.api_v1.schemas.metadata import CorpusData, OrganisationConfig, TaxonomyData
 
 
 def get_organisation_taxonomy(db: Session, org_id: int) -> Taxonomy:
@@ -38,6 +38,7 @@ def get_organisation_taxonomy(db: Session, org_id: int) -> Taxonomy:
     return {k: TaxonomyEntry(**v) for k, v in taxonomy[0].items()}
 
 
+# TODO: Remove this function as part of PDCT-1067
 def get_organisation_taxonomy_by_name(db: Session, org_name: str) -> TaxonomyData:
     """
     Returns the TaxonomyConfig for the named organisation
@@ -71,6 +72,48 @@ def get_organisation_taxonomy_by_name(db: Session, org_name: str) -> TaxonomyDat
     }
 
 
+def _to_corpus_data(row, event_types) -> CorpusData:
+    return CorpusData(
+        corpus_import_id=row.corpus_import_id,
+        title=row.title,
+        description=row.description,
+        corpus_type=row.corpus_type,
+        corpus_type_description=row.corpus_type_description,
+        taxonomy={
+            **row.taxonomy,
+            "event_types": asdict(event_types),
+        },
+    )
+
+
+def get_copora_for_org(db: Session, org_name: str) -> Sequence[CorpusData]:
+    copora = (
+        db.query(
+            Corpus.import_id.label("corpus_import_id"),
+            Corpus.title.label("title"),
+            Corpus.description.label("description"),
+            Corpus.corpus_type_name.label("corpus_type"),
+            CorpusType.description.label("corpus_type_description"),
+            CorpusType.valid_metadata.label("taxonomy"),
+        )
+        .join(
+            Corpus,
+            Corpus.corpus_type_name == CorpusType.name,
+        )
+        .join(Organisation, Organisation.id == Corpus.organisation_id)
+        .filter(Organisation.name == org_name)
+        .all()
+    )
+
+    event_types = db.query(FamilyEventType).all()
+    entry = TaxonomyEntry(
+        allow_blanks=False,
+        allowed_values=[r.name for r in event_types],
+        allow_any=False,
+    )
+    return [_to_corpus_data(row, entry) for row in copora]
+
+
 def get_organisation_config(db: Session, org: Organisation) -> OrganisationConfig:
     total = (
         db.query(Family)
@@ -98,8 +141,10 @@ def get_organisation_config(db: Session, org: Organisation) -> OrganisationConfi
         else:
             count_by_category[category] = 0
 
+    org_name = cast(str, org.name)
     return OrganisationConfig(
         total=total,
         count_by_category=count_by_category,
-        taxonomy=get_organisation_taxonomy_by_name(db, cast(str, org.name)),
+        taxonomy=get_organisation_taxonomy_by_name(db, org_name),
+        copora=get_copora_for_org(db, org_name),
     )

tests/routes/test_config.py

@@ -9,6 +9,39 @@
 from app.core.util import tree_table_to_json
 from app.db.session import SessionLocal
 
+LEN_ORG_CONFIG = 4
+EXPECTED_CCLW_TAXONOMY = {
+    "instrument",
+    "keyword",
+    "sector",
+    "topic",
+    "framework",
+    "hazard",
+    "event_types",
+}
+EXPECTED_CCLW_EVENTS = [
+    "Amended",
+    "Appealed",
+    "Closed",
+    "Declaration Of Climate Emergency",
+    "Dismissed",
+    "Entered Into Force",
+    "Filing",
+    "Granted",
+    "Implementation Details",
+    "International Agreement",
+    "Net Zero Pledge",
+    "Other",
+    "Passed/Approved",
+    "Repealed/Replaced",
+    "Set",
+    "Settled",
+    "Updated",
+]
+
+
+EXPECTED_UNFCCC_TAXONOMY = {"author", "author_type", "event_types"}
+
 
 def _add_family(test_db, import_id: str, cat: FamilyCategory, corpus_import_id):
     test_db.add(
@@ -42,50 +75,6 @@ def test_config_endpoint_content(data_client, data_db):
 
     assert "geographies" in response_json
     assert len(response_json["geographies"]) == 8
-
-    assert "organisations" in response_json
-
-    assert "CCLW" in response_json["organisations"]
-    cclw_taxonomy = response_json["organisations"]["CCLW"]["taxonomy"]
-    assert set(cclw_taxonomy) == {
-        "instrument",
-        "keyword",
-        "sector",
-        "topic",
-        "framework",
-        "hazard",
-        "event_types",
-    }
-    cclw_taxonomy_event_types = cclw_taxonomy["event_types"]["allowed_values"]
-    cclw_expected_event_types = [
-        "Amended",
-        "Appealed",
-        "Closed",
-        "Declaration Of Climate Emergency",
-        "Dismissed",
-        "Entered Into Force",
-        "Filing",
-        "Granted",
-        "Implementation Details",
-        "International Agreement",
-        "Net Zero Pledge",
-        "Other",
-        "Passed/Approved",
-        "Repealed/Replaced",
-        "Set",
-        "Settled",
-        "Updated",
-    ]
-    assert set(cclw_taxonomy_event_types) ^ set(cclw_expected_event_types) == set()
-
-    assert "UNFCCC" in response_json["organisations"]
-    unfccc_taxonomy = response_json["organisations"]["UNFCCC"]["taxonomy"]
-    assert set(unfccc_taxonomy) == {"author", "author_type", "event_types"}
-    assert set(unfccc_taxonomy["author_type"]["allowed_values"]) == {
-        "Party",
-        "Non-Party",
-    }
-
     assert "languages" in response_json
     assert len(response_json["languages"]) == 7893
     assert "fra" in response_json["languages"]
@@ -100,16 +89,48 @@ def test_config_endpoint_content(data_client, data_db):
     assert len(response_json["document_variants"]) == 2
     assert "Original Language" in response_json["document_variants"]
 
-    org_config = response_json["organisations"]["CCLW"]
-    assert len(org_config) == 3
-    assert "taxonomy" in org_config
-    assert org_config["total"] == 0
-    assert org_config["count_by_category"] == {
+    # Now test organisations
+    assert "organisations" in response_json
+
+    assert "CCLW" in response_json["organisations"]
+    cclw_org = response_json["organisations"]["CCLW"]
+    assert len(cclw_org) == LEN_ORG_CONFIG
+
+    # Test the counts are there (just CCLW)
+    assert cclw_org["total"] == 0
+    assert cclw_org["count_by_category"] == {
         "Executive": 0,
         "Legislative": 0,
         "UNFCCC": 0,
     }
 
+    assert "UNFCCC" in response_json["organisations"]
+    unfccc_org = response_json["organisations"]["UNFCCC"]
+    assert len(unfccc_org) == LEN_ORG_CONFIG
+
+    # Old tests - to be removed in PDCT-1057
+    cclw_taxonomy = cclw_org["taxonomy"]
+    assert set(cclw_taxonomy) == EXPECTED_CCLW_TAXONOMY
+    cclw_taxonomy_event_types = cclw_taxonomy["event_types"]["allowed_values"]
+    assert set(cclw_taxonomy_event_types) ^ set(EXPECTED_CCLW_EVENTS) == set()
+
+    unfccc_taxonomy = unfccc_org["taxonomy"]
+    assert set(unfccc_taxonomy) == EXPECTED_UNFCCC_TAXONOMY
+    assert set(unfccc_taxonomy["author_type"]["allowed_values"]) == {
+        "Party",
+        "Non-Party",
+    }
+
+    # New taxonomy tests
+    cclw_copora = cclw_org["copora"]
+    assert len(cclw_copora) == 1
+    assert cclw_copora[0]["corpus_import_id"] == "CCLW.corpus.i00000001.n0000"
+    assert cclw_copora[0]["corpus_type"] == "Laws and Policies"
+    assert cclw_copora[0]["corpus_type_description"] == "Laws and policies"
+    assert cclw_copora[0]["description"] == "CCLW national policies"
+    assert cclw_copora[0]["title"] == "CCLW national policies"
+    assert set(cclw_copora[0]["taxonomy"]) ^ EXPECTED_CCLW_TAXONOMY == set()
+
 
 def test_config_endpoint_cclw_stats(data_client, data_db):
     url_under_test = "/api/v1/config"
@@ -144,7 +165,7 @@ def test_config_endpoint_cclw_stats(data_client, data_db):
     response_json = response.json()
 
     org_config = response_json["organisations"]["CCLW"]
-    assert len(org_config) == 3
+    assert len(org_config) == LEN_ORG_CONFIG
     assert org_config["total"] == 6
 
     laws = org_config["count_by_category"]["Legislative"]