check for doubles

uploader.py

@@ -157,6 +157,58 @@ class CalibreClient:
             return "error"
 
 
+def fetch_all_books(cfg: CalibreConfig) -> list[dict]:
+    """Fetch every book from Calibre-Web via /ajax/listbooks. Returns raw row dicts."""
+    client = CalibreClient(cfg)
+    client._ensure_auth()
+    all_books: list[dict] = []
+    page_size = 100
+    start = 0
+    while True:
+        resp = client._session.get(
+            f"{cfg.url}/ajax/listbooks",
+            params={"start": start, "length": page_size, "sort": "title", "order": "asc"},
+            timeout=30,
+        )
+        resp.raise_for_status()
+        data = resp.json()
+        rows = data.get("rows", [])
+        all_books.extend(rows)
+        if start + page_size >= data.get("total_count", 0):
+            break
+        start += page_size
+    return all_books
+
+
+def delete_book(cfg: CalibreConfig, book_id: int) -> tuple[bool, str]:
+    """Delete a book from Calibre-Web by ID."""
+    client = CalibreClient(cfg)
+    client._ensure_auth()
+    resp = client._session.post(
+        f"{cfg.url}/delete/{book_id}",
+        data={"csrf_token": client._upload_csrf} if client._upload_csrf else {},
+        timeout=30,
+    )
+    if resp.ok:
+        return True, "Deleted"
+    return False, f"HTTP {resp.status_code}"
+
+
+def find_duplicate_groups(books: list[dict]) -> list[list[dict]]:
+    """Group books by normalised title; return only groups with 2+ entries."""
+    from collections import defaultdict
+    groups: dict[str, list[dict]] = defaultdict(list)
+    for book in books:
+        words = _normalize_words(book.get("title", ""))
+        key = " ".join(sorted(words))
+        if key:
+            groups[key].append(book)
+    return sorted(
+        [g for g in groups.values() if len(g) > 1],
+        key=lambda g: g[0].get("title", "").lower(),
+    )
+
+
 def test_connection(cfg: CalibreConfig) -> tuple[bool, str]:
     try:
         client = CalibreClient(cfg)