fix(metadata): find audiobook series through hardcover when audible leaves them untagged

static/app.js

@@ -360,7 +360,8 @@ async function openSeriesDialog(result, type = "audiobook", language = "") {
     seriesEpisodes = await api(
       `/api/search/series?series_id=${encodeURIComponent(result.series_id)}` +
         `&title=${encodeURIComponent(result.series || result.title)}` +
-        `&media_type=${type}&language=${encodeURIComponent(language)}`
+        `&media_type=${type}&language=${encodeURIComponent(language)}` +
+        `&source=${encodeURIComponent(result.source || "")}`
     );
   } catch (err) {
     $("#series-info").textContent = err.message;

tests/test_musicbrainz.py

@@ -82,3 +82,42 @@ def test_a_broken_fallback_never_breaks_the_search(monkeypatch):
     monkeypatch.setattr(metadata.musicbrainz, "search", boom)
     results = asyncio.run(metadata.audiobook_search("Toteninsel"))
     assert [r.title for r in results] == ["Da"]
+
+
+def test_audiobook_search_appends_hardcover_series_as_audiobooks(monkeypatch):
+    from wordarr.metadata.base import MetadataResult
+
+    async def fake_audible(query, language=""):
+        return [MetadataResult(media_type="audiobook", title="Die letzte Witwe")]
+
+    async def no_hits(query, language=""):
+        return []
+
+    async def fake_hardcover(query, language=""):
+        return [MetadataResult(media_type="ebook", source="hardcover",
+                               title="Die verstummte Frau", series="Will Trent",
+                               series_id="3670", volume=10)]
+
+    monkeypatch.setattr(metadata.audible, "search", fake_audible)
+    monkeypatch.setattr(metadata.musicbrainz, "search", no_hits)
+    monkeypatch.setattr(metadata.hardcover, "search", fake_hardcover)
+    results = asyncio.run(metadata.audiobook_search("Karin Slaughter", "german"))
+    assert [r.source for r in results] == ["audible", "hardcover"]
+    assert results[1].media_type == "audiobook"
+
+
+def test_audiobook_series_goes_to_the_provider_of_the_hit(monkeypatch):
+    from wordarr.metadata.base import MetadataResult
+
+    async def fake_series_books(series_id, language=""):
+        return [MetadataResult(media_type="ebook", source="hardcover",
+                               title="Triptych", volume=1)]
+
+    async def audible_must_not_run(series_id, title):
+        raise AssertionError("a hardcover id is no audible series asin")
+
+    monkeypatch.setattr(metadata.hardcover, "series_books", fake_series_books)
+    monkeypatch.setattr(metadata.audible, "series_episodes", audible_must_not_run)
+    lookup = metadata.SERIES_PROVIDERS["audiobook"]
+    books = asyncio.run(lookup("3670", "Will Trent", "german", "hardcover"))
+    assert [(b.title, b.media_type) for b in books] == [("Triptych", "audiobook")]

wordarr/api/search.py

@@ -19,15 +19,16 @@ async def search(media_type: str, q: str, language: str = "", fallback: bool = T
 
 @router.get("/series", response_model=list[MetadataResult])
 async def series(series_id: str, title: str = "", media_type: str = "audiobook",
-                 language: str = ""):
+                 language: str = "", source: str = ""):
     """All parts of a series, by number. Audible has no series endpoint, so
-    audiobooks are a keyword crawl that may miss episodes; Hardcover is exact."""
+    audiobooks are a keyword crawl that may miss episodes; Hardcover is exact.
+    The source picks the provider, a series id only means something to its own."""
     lookup = SERIES_PROVIDERS.get(media_type)
     if not lookup:
         raise HTTPException(400, f"no series lookup for media_type: {media_type}")
     if not series_id:
         raise HTTPException(400, "series_id required")
     try:
-        return await lookup(series_id, title, language)
+        return await lookup(series_id, title, language, source)
     except Exception as exc:
         raise HTTPException(502, f"metadata provider error: {exc}")

wordarr/metadata/__init__.py

@@ -58,38 +58,53 @@ async def ebook_search(query: str, language: str = "",
     return _merge(groups, _ebook_key)
 
 
+def _as_audiobook(hits: list[MetadataResult]) -> list[MetadataResult]:
+    return [r.model_copy(update={"media_type": "audiobook"}) for r in hits]
+
+
 async def audiobook_search(query: str, language: str = "",
                            fallback: bool = True) -> list[MetadataResult]:
-    """Audible first, MusicBrainz alongside it.
+    """Audible first, MusicBrainz and Hardcover alongside it.
 
     Audible answers almost any query with *something*, so "no good hit" cannot be
-    detected reliably - measuring it showed the fallback would never fire. Both
-    are therefore asked in parallel and the MusicBrainz hits are appended, marked
-    with their source. Callers that search automatically (the import dialog when
-    it opens) pass fallback=False to spare the rate limited service.
+    detected reliably - measuring it showed the fallback would never fire. All
+    are therefore asked in parallel and the extra hits are appended, marked with
+    their source. Hardcover is there for its series: Audible leaves most volumes
+    of a series untagged. Callers that search automatically (the import dialog
+    when it opens) pass fallback=False to spare the rate limited services.
     """
     if not fallback:
         return await audible.search(query, language)
 
-    hits, extra = await asyncio.gather(
+    hits, *extras = await asyncio.gather(
         audible.search(query, language),
         musicbrainz.search(query, language),
+        hardcover.search(query, language),
         return_exceptions=True,
     )
     if isinstance(hits, BaseException):
         raise hits
-    if isinstance(extra, BaseException):  # a slow fallback must not break search
-        log.warning("musicbrainz lookup failed for %r: %s", query, extra)
-        return hits
     seen = {(r.title.strip().lower(), r.volume) for r in hits}
-    return hits + [r for r in extra if (r.title.strip().lower(), r.volume) not in seen]
+    for module, extra in zip((musicbrainz, hardcover), extras):
+        if isinstance(extra, BaseException):  # a slow fallback must not break search
+            log.warning("%s lookup failed for %r: %s", module.__name__, query, extra)
+            continue
+        for r in _as_audiobook(extra):
+            if (key := (r.title.strip().lower(), r.volume)) not in seen:
+                seen.add(key)
+                hits.append(r)
+    return hits
 
 
-async def _audiobook_series(series_id: str, title: str, language: str = ""):
+async def _audiobook_series(series_id: str, title: str, language: str = "",
+                            source: str = ""):
+    if source == "hardcover":
+        return _as_audiobook(await hardcover.series_books(series_id, language))
     return await audible.series_episodes(series_id, title)
 
 
-async def _ebook_series(series_id: str, title: str = "", language: str = ""):
+async def _ebook_series(series_id: str, title: str = "", language: str = "",
+                        source: str = ""):
     return await hardcover.series_books(series_id, language)