feat(ebook-search): implement async file path resolution and EPUB discovery

tests/ebook_search/test_core.py

@@ -27,7 +27,13 @@ from python.ebook_search.bm25_corpus import (
 )
 from python.ebook_search.config import EbookSearchConfig, RerankConfig, load_config
 from python.ebook_search.embeddings import MODEL_DIMENSIONS, ensure_embedding_models
-from python.ebook_search.ingest import chunk_text, find_existing_source
+from python.ebook_search.ingest import (
+    chunk_text,
+    find_existing_source,
+    find_library_epubs,
+    resolve_ingest_path,
+    sha256_file,
+)
 from python.ebook_search.search import (
     SearchResponse,
     SearchResult,
@@ -106,6 +112,33 @@ async def test_find_existing_source_matches_path_or_hash() -> None:
         assert await find_existing_source(session, Path("/new/book.epub"), "a" * 64) == source
 
 
+async def test_async_ingest_path_resolution_and_hashing(tmp_path: Path) -> None:
+    """Ingest file metadata uses AnyIO's asynchronous path operations."""
+    path = tmp_path / "book.epub"
+    path.write_bytes(b"epub content")
+
+    resolved_path = await resolve_ingest_path(path)
+
+    assert resolved_path == await resolved_path.resolve()
+    assert await sha256_file(resolved_path) == "4b71fc6fc452ae1ff7832bfe374ae56ab4d0649acf6af8ab4d068d858ea60448"
+
+
+async def test_find_library_epubs_uses_async_path_walk(tmp_path: Path) -> None:
+    """Library discovery finds EPUB files through AnyIO's async path API."""
+    (tmp_path / "nested").mkdir()
+    first_epub = tmp_path / "first.epub"
+    second_epub = tmp_path / "nested" / "second.epub"
+    first_epub.write_bytes(b"first")
+    second_epub.write_bytes(b"second")
+    (tmp_path / "not-a-book.txt").write_text("ignore", encoding="utf-8")
+
+    path, epub_paths = await find_library_epubs(str(tmp_path))
+
+    assert str(path) == str(tmp_path)
+    assert epub_paths is not None
+    assert [str(epub_path) for epub_path in epub_paths] == [str(first_epub), str(second_epub)]
+
+
 async def test_bm25_corpus_uses_existing_search_text_without_duplicate_metadata() -> None:
     engine = await build_async_engine()
     async with AsyncSession(engine, expire_on_commit=False) as session: