Add corpus type to config (#252)

* Add corpus types to config

* Bump to 2.17.15

* Check uniqueness

* Don't include taxonomy in ct on config

* Revert "Don't include taxonomy in ct on config"

This reverts commit b967c7a.

app/model/config.py

@@ -27,5 +27,6 @@ class ConfigReadDTO(BaseModel):
 
     geographies: Sequence[dict]
     corpora: Sequence[CorpusData]
+    corpus_types: Sequence[dict]
     languages: Mapping[str, str]
     document: DocumentConfig

app/repository/config.py

@@ -90,6 +90,8 @@ def get(db: Session, user: UserContext) -> ConfigReadDTO:
     corpora = get_corpora(db, user)
     languages = {lang.language_code: lang.name for lang in db.query(Language).all()}
 
+    corpus_types = get_unique_corpus_types(corpora)
+
     # Now Document config
     doc_config = DocumentConfig(
         variants=[
@@ -101,6 +103,31 @@ def get(db: Session, user: UserContext) -> ConfigReadDTO:
     return ConfigReadDTO(
         geographies=geographies,
         corpora=corpora,
+        corpus_types=corpus_types,
         languages=languages,
         document=doc_config,
     )
+
+
+def get_unique_corpus_types(corpora):
+    """Generate a list of unique corpus type dictionaries.
+
+    :param List[Corpus] corpora: List of corpus items.
+    :return List[Dict[str, Any]]: List of unique corpus type
+        dictionaries.
+    """
+    seen = set()
+    unique_corpus_types = []
+    for c in corpora:
+        # Create a tuple of name and description for uniqueness check
+        unique_key = (c.corpus_type, c.corpus_type_description)
+        if unique_key not in seen:
+            seen.add(unique_key)
+            unique_corpus_types.append(
+                {
+                    "name": c.corpus_type,
+                    "description": c.corpus_type_description,
+                    "taxonomy": c.taxonomy,
+                }
+            )
+    return unique_corpus_types

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "admin_backend"
-version = "2.17.14"
+version = "2.17.15"
 description = ""
 authors = ["CPR-dev-team <[email protected]>"]
 packages = [{ include = "app" }, { include = "tests" }]

tests/integration_tests/config/test_config.py

@@ -4,7 +4,13 @@
 
 from tests.integration_tests.setup_db import setup_db
 
-EXPECTED_CONFIG_KEYS = {"geographies", "corpora", "languages", "document"}
+EXPECTED_CONFIG_KEYS = {
+    "geographies",
+    "corpora",
+    "corpus_types",
+    "languages",
+    "document",
+}
 EXPECTED_CORPORA_CONFIG_KEYS = {
     "corpus_import_id",
     "title",

tests/mocks/repos/config_repo.py

@@ -19,6 +19,7 @@ def mock_get(_) -> Optional[ConfigReadDTO]:
         return ConfigReadDTO(
             geographies=[],
             corpora=[],
+            corpus_types=[],
             languages={},
             document=DocumentConfig(variants=[]),
         )

tests/mocks/services/config_service.py

@@ -16,6 +16,7 @@ def mock_get_config(_) -> ConfigReadDTO:
         return ConfigReadDTO(
             geographies=[],
             corpora=[],
+            corpus_types=[],
             languages={},
             document=DocumentConfig(variants=[]),
         )

tests/unit_tests/routers/config/test_get_config.py

@@ -9,6 +9,7 @@ def test_get_when_ok(client: TestClient, user_header_token, config_service_mock)
     keys = data.keys()
     assert "geographies" in keys
     assert "corpora" in keys
+    assert "corpus_types" in keys
     assert "languages" in keys
     assert "document" in keys
     assert config_service_mock.get.call_count == 1