fix(metadata): keep hardcover books whose editions carry no language

tests/test_ebooks.py

@@ -25,6 +25,7 @@ GERMAN_EDITION = {
     "isbn_10": None,
     "release_date": "1957-03-01",
     "cached_image": "https://example.invalid/hobbit-de.jpg",
+    "language": {"code2": "de"},
 }
 
 
@@ -83,6 +84,38 @@ def test_hardcover_drops_books_without_an_edition_in_that_language(monkeypatch):
     assert asyncio.run(hardcover.search("hobbit", "german")) == []
 
 
+def test_hardcover_keeps_a_book_whose_editions_carry_no_language(monkeypatch):
+    """German translations are often their own book row with untagged editions."""
+    def handler(request: httpx.Request) -> httpx.Response:
+        import json
+        if "editions" in json.loads(request.content)["query"]:
+            return httpx.Response(200, json={"data": {
+                "editions": [{"book_id": 42, "title": "Der Hobbit", "language": None}],
+                "other": [],
+            }})
+        return httpx.Response(200, json={"data": {"search": {"results": {"hits": [HARDCOVER_HIT]}}}})
+
+    monkeypatch.setattr(hardcover, "TOKEN", "test-token")
+    mock_client(monkeypatch, hardcover, handler)
+    hit = asyncio.run(hardcover.search("hobbit", "german"))[0]
+    assert (hit.title, hit.language) == ("Der Hobbit", "")
+
+
+def test_hardcover_drops_an_untagged_edition_of_a_foreign_book(monkeypatch):
+    def handler(request: httpx.Request) -> httpx.Response:
+        import json
+        if "editions" in json.loads(request.content)["query"]:
+            return httpx.Response(200, json={"data": {
+                "editions": [{"book_id": 42, "title": "The Hobbit", "language": None}],
+                "other": [{"book_id": 42}],
+            }})
+        return httpx.Response(200, json={"data": {"search": {"results": {"hits": [HARDCOVER_HIT]}}}})
+
+    monkeypatch.setattr(hardcover, "TOKEN", "test-token")
+    mock_client(monkeypatch, hardcover, handler)
+    assert asyncio.run(hardcover.search("hobbit", "german")) == []
+
+
 SERIES_ROWS = {
     "name": "Die drei Fragezeichen",
     "book_series": [
@@ -129,7 +162,8 @@ def test_hardcover_series_sorts_by_volume_and_keeps_the_first_row(monkeypatch):
 def test_hardcover_series_prefers_the_row_with_an_edition_in_the_language(monkeypatch):
     calls = series_client(monkeypatch, [
         {"book_id": 3, "title": "Der Phantomsee", "isbn_13": "9783440032251",
-         "release_date": "1969-01-01", "cached_image": None},
+         "release_date": "1969-01-01", "cached_image": None,
+         "language": {"code2": "de"}},
     ])
     books = asyncio.run(hardcover.series_books("10146", "german"))
     assert calls[1] == {"ids": [1, 2, 3, 4], "lang": "de"}
@@ -280,3 +314,22 @@ def test_automatic_search_asks_only_the_best_provider(monkeypatch):
     monkeypatch.setattr(openlibrary, "search", lambda q, lang="", fb=True: record("ol", q, lang))
     asyncio.run(metadata.ebook_search("hobbit", "german", fallback=False))
     assert called == ["gb"]  # no token, so Google Books stands in for Hardcover
+
+
+def test_hardcover_puts_hits_of_unknown_language_last(monkeypatch):
+    untagged = {"document": {**HARDCOVER_HIT["document"], "id": "7", "title": "Bundle"}}
+
+    def handler(request: httpx.Request) -> httpx.Response:
+        import json
+        if "editions" in json.loads(request.content)["query"]:
+            return httpx.Response(200, json={"data": {"editions": [
+                {"book_id": 7, "title": "Bundle", "language": None},
+                GERMAN_EDITION,
+            ]}})
+        return httpx.Response(200, json={"data": {"search": {"results": {
+            "hits": [untagged, HARDCOVER_HIT]}}}})
+
+    monkeypatch.setattr(hardcover, "TOKEN", "test-token")
+    mock_client(monkeypatch, hardcover, handler)
+    hits = asyncio.run(hardcover.search("hobbit", "german"))
+    assert [h.title for h in hits] == ["Der Hobbit", "Bundle"]