add(metadata): hardcover and google books as ebook providers

tests/test_ebooks.py

@@ -0,0 +1,168 @@
+"""The three ebook providers and the search that merges them."""
+import asyncio
+
+import httpx
+import pytest
+
+from wordarr import metadata
+from wordarr.metadata import googlebooks, hardcover, openlibrary
+
+HARDCOVER_HIT = {
+    "document": {
+        "id": "42",
+        "title": "The Hobbit",
+        "author_names": ["J. R. R. Tolkien"],
+        "release_year": 1937,
+        "isbns": ["9780261102217"],
+        "image": {"url": "https://example.invalid/hobbit.jpg"},
+        "series_names": [],
+    }
+}
+GERMAN_EDITION = {
+    "book_id": 42,
+    "title": "Der Hobbit",
+    "isbn_13": "9783608939828",
+    "isbn_10": None,
+    "release_date": "1957-03-01",
+    "cached_image": "https://example.invalid/hobbit-de.jpg",
+}
+
+
+def mock_client(monkeypatch, module, handler):
+    real = httpx.AsyncClient
+    monkeypatch.setattr(module.httpx, "AsyncClient",
+                        lambda **kw: real(transport=httpx.MockTransport(handler), **kw))
+
+
+@pytest.fixture
+def hardcover_calls(monkeypatch):
+    seen = []
+
+    def handler(request: httpx.Request) -> httpx.Response:
+        import json
+        body = json.loads(request.content)
+        seen.append((body["variables"], request.headers.get("authorization")))
+        if "editions" in body["query"]:
+            return httpx.Response(200, json={"data": {"editions": [GERMAN_EDITION]}})
+        return httpx.Response(200, json={"data": {"search": {"results": {"hits": [HARDCOVER_HIT]}}}})
+
+    monkeypatch.setattr(hardcover, "TOKEN", "test-token")
+    mock_client(monkeypatch, hardcover, handler)
+    return seen
+
+
+def test_hardcover_needs_no_network_without_a_token(monkeypatch):
+    monkeypatch.setattr(hardcover, "TOKEN", "")
+    assert asyncio.run(hardcover.search("hobbit")) == []
+
+
+def test_hardcover_parses_a_hit(hardcover_calls):
+    hit = asyncio.run(hardcover.search("hobbit"))[0]
+    assert (hit.title, hit.authors, hit.year) == ("The Hobbit", "J. R. R. Tolkien", 1937)
+    assert hit.external_id == "9780261102217"
+    assert hit.source == "hardcover"
+    assert hardcover_calls[0][1] == "Bearer test-token"
+
+
+def test_hardcover_swaps_in_the_german_edition(hardcover_calls):
+    hit = asyncio.run(hardcover.search("hobbit", "german"))[0]
+    assert (hit.title, hit.year, hit.language) == ("Der Hobbit", 1957, "german")
+    assert hit.external_id == "9783608939828"
+    assert hardcover_calls[1][0] == {"ids": [42], "lang": "de"}
+
+
+def test_hardcover_drops_books_without_an_edition_in_that_language(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": []}})
+        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")) == []
+
+
+def test_google_books_restricts_the_language(monkeypatch):
+    seen = []
+
+    def handler(request: httpx.Request) -> httpx.Response:
+        seen.append(dict(request.url.params))
+        return httpx.Response(200, json={"items": [{"volumeInfo": {
+            "title": "Der Hobbit",
+            "authors": ["J. R. R. Tolkien"],
+            "publishedDate": "1957",
+            "language": "de",
+            "industryIdentifiers": [{"type": "ISBN_13", "identifier": "9783608939828"}],
+            "imageLinks": {"thumbnail": "http://example.invalid/de.jpg"},
+        }}]})
+
+    mock_client(monkeypatch, googlebooks, handler)
+    hit = asyncio.run(googlebooks.search("hobbit", "german"))[0]
+    assert seen[0]["langRestrict"] == "de"
+    assert (hit.title, hit.language, hit.year) == ("Der Hobbit", "german", 1957)
+    assert hit.cover_url.startswith("https://")  # the API still hands out http
+
+
+def test_google_books_searches_an_isbn_as_an_isbn(monkeypatch):
+    seen = []
+
+    def handler(request: httpx.Request) -> httpx.Response:
+        seen.append(dict(request.url.params))
+        return httpx.Response(200, json={"items": []})
+
+    mock_client(monkeypatch, googlebooks, handler)
+    asyncio.run(googlebooks.search("978-3-608-93982-8"))
+    assert seen[0]["q"] == "isbn:9783608939828"
+
+
+def test_open_library_filters_and_labels_the_language(monkeypatch):
+    seen = []
+
+    def handler(request: httpx.Request) -> httpx.Response:
+        seen.append(dict(request.url.params))
+        return httpx.Response(200, json={"docs": [{
+            "title": "Der Hobbit",
+            "author_name": ["J. R. R. Tolkien"],
+            "first_publish_year": 1957,
+            "language": ["ger", "eng"],
+            "isbn": ["9783608939828"],
+        }]})
+
+    mock_client(monkeypatch, openlibrary, handler)
+    hit = asyncio.run(openlibrary.search("hobbit", "german"))[0]
+    assert seen[0]["q"] == "hobbit language:ger"
+    assert hit.language == "german"
+    assert hit.source == "openlibrary"
+
+
+def test_ebook_search_merges_the_providers_and_survives_a_broken_one(monkeypatch):
+    async def hits(source, **fields):
+        return [metadata.MetadataResult(media_type="ebook", source=source, **fields)]
+
+    monkeypatch.setattr(hardcover, "TOKEN", "test-token")
+    monkeypatch.setattr(hardcover, "search", lambda q, lang="", fb=True: hits(
+        "hardcover", title="Der Hobbit", external_id="9783608939828"))
+    monkeypatch.setattr(googlebooks, "search", lambda q, lang="", fb=True: hits(
+        "googlebooks", title="Der Hobbit", external_id="9783608939828"))  # same edition
+
+    async def boom(q, lang="", fb=True):
+        raise httpx.ConnectError("open library is down")
+
+    monkeypatch.setattr(openlibrary, "search", boom)
+    results = asyncio.run(metadata.ebook_search("hobbit", "german"))
+    assert [r.source for r in results] == ["hardcover"]
+
+
+def test_automatic_search_asks_only_the_best_provider(monkeypatch):
+    called = []
+
+    async def record(name, q, lang):
+        called.append(name)
+        return []
+
+    monkeypatch.setattr(hardcover, "TOKEN", "")
+    monkeypatch.setattr(googlebooks, "search", lambda q, lang="", fb=True: record("gb", q, lang))
+    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