Add books languages endpoint

backend/src/cms_backend/api/routes/books.py

@@ -13,10 +13,11 @@
 from cms_backend.db.book import get_book as db_get_book
 from cms_backend.db.book import move_book as db_move_book
 from cms_backend.db.book import recover_book as db_recover_book
+from cms_backend.db.books import get_book_languages as db_get_book_languages
 from cms_backend.db.books import get_books as db_get_books
 from cms_backend.db.books import get_zim_urls as db_get_zim_urls
 from cms_backend.schemas import BaseModel
-from cms_backend.schemas.models import ZimUrlsSchema
+from cms_backend.schemas.models import BookLanguagesSchema, ZimUrlsSchema
 from cms_backend.schemas.orms import (
     BookFullSchema,
     BookLightSchema,
@@ -80,6 +81,13 @@ def get_zim_urls(
     return db_get_zim_urls(session, zim_ids)
 
 
+@router.get("/languages")
+def get_book_languages(
+    session: Annotated[OrmSession, Depends(gen_dbsession)],
+) -> BookLanguagesSchema:
+    return db_get_book_languages(session)
+
+
 @router.get("/{book_id}")
 def get_book(
     book_id: Annotated[UUID, Path()],

backend/src/cms_backend/db/books.py

@@ -6,7 +6,7 @@
 
 from cms_backend.db import count_from_stmt
 from cms_backend.db.models import Book, BookLocation, Collection, CollectionTitle, Title
-from cms_backend.schemas.models import ZimUrlSchema, ZimUrlsSchema
+from cms_backend.schemas.models import BookLanguagesSchema, ZimUrlSchema, ZimUrlsSchema
 from cms_backend.schemas.orms import BookLightSchema, ListResult
 from cms_backend.utils.filename import construct_download_url
 
@@ -197,3 +197,20 @@ def get_zim_urls(session: OrmSession, zim_ids: list[UUID]) -> ZimUrlsSchema:
                 )
 
     return result
+
+
+def get_book_languages(session: OrmSession) -> BookLanguagesSchema:
+    """Get the sorted list of language codes used by production books."""
+    stmt = select(Book.zim_metadata).where(Book.location_kind == "prod")
+
+    languages: set[str] = set()
+    for zim_metadata in session.scalars(stmt):
+        language_value = zim_metadata.get("Language")
+        if not isinstance(language_value, str):
+            continue
+
+        for language in language_value.split(","):
+            if normalized_language := language.strip():
+                languages.add(normalized_language)
+
+    return BookLanguagesSchema(languages=sorted(languages))

backend/src/cms_backend/schemas/models.py

@@ -16,3 +16,7 @@ class ZimUrlSchema(BaseModel):
 
 class ZimUrlsSchema(BaseModel):
     urls: dict[UUID, list[ZimUrlSchema]]
+
+
+class BookLanguagesSchema(BaseModel):
+    languages: list[str]

backend/tests/api/routes/test_books.py

@@ -3,6 +3,7 @@
 
 import pytest
 from fastapi.testclient import TestClient
+from sqlalchemy.orm import Session as OrmSession
 
 from cms_backend.db.models import Book, Title
 
@@ -258,6 +259,35 @@ def test_get_books_filter_by_id(
     assert response_doc["items"][0]["id"] == str(book1.id)
 
 
+def test_get_book_languages(
+    client: TestClient,
+    dbsession: OrmSession,
+    create_book: Callable[..., Book],
+):
+    """Test books languages endpoint returns sorted distinct production languages."""
+    prod_book = create_book(zim_metadata={"Language": "eng, fra"})
+    prod_book.location_kind = "prod"
+
+    other_prod_book = create_book(zim_metadata={"Language": " deu ,eng, ,spa "})
+    other_prod_book.location_kind = "prod"
+
+    create_book(zim_metadata={"Language": "hin"})
+    staging_book = create_book(zim_metadata={"Language": "ita"})
+    staging_book.location_kind = "staging"
+
+    blank_language_book = create_book(zim_metadata={"Language": " , "})
+    blank_language_book.location_kind = "prod"
+
+    missing_language_book = create_book(zim_metadata={"Name": "no-language"})
+    missing_language_book.location_kind = "prod"
+
+    dbsession.flush()
+
+    response = client.get("/v1/books/languages")
+    assert response.status_code == HTTPStatus.OK
+    assert response.json() == {"languages": ["deu", "eng", "fra", "spa"]}
+
+
 def test_get_book_by_id(
     client: TestClient,
     book: Book,

backend/tests/db/test_books.py

@@ -14,7 +14,7 @@
     move_book,
     recover_book,
 )
-from cms_backend.db.books import get_books, get_zim_urls
+from cms_backend.db.books import get_book_languages, get_books, get_zim_urls
 from cms_backend.db.exceptions import RecordDoesNotExistError
 from cms_backend.db.models import (
     Book,
@@ -375,6 +375,34 @@ def test_get_zim_urls(
     assert view_url.collection == collection.name
 
 
+def test_get_book_languages(
+    dbsession: OrmSession,
+    create_book: Callable[..., Book],
+):
+    """Return the sorted distinct language codes from production books only."""
+    prod_book = create_book(zim_metadata={"Language": "eng, fra"})
+    prod_book.location_kind = "prod"
+
+    other_prod_book = create_book(zim_metadata={"Language": " deu ,eng, ,spa "})
+    other_prod_book.location_kind = "prod"
+
+    create_book(zim_metadata={"Language": "hin"})
+    staging_book = create_book(zim_metadata={"Language": "ita"})
+    staging_book.location_kind = "staging"
+
+    blank_language_book = create_book(zim_metadata={"Language": " , "})
+    blank_language_book.location_kind = "prod"
+
+    missing_language_book = create_book(zim_metadata={"Name": "no-language"})
+    missing_language_book.location_kind = "prod"
+
+    dbsession.flush()
+
+    result = get_book_languages(dbsession)
+
+    assert result.languages == ["deu", "eng", "fra", "spa"]
+
+
 def test_get_zim_urls_book_with_subpath(
     dbsession: OrmSession,
     create_book: Callable[..., Book],