check for doubles

main.py

@@ -13,6 +13,7 @@ import db
 import sftp as sftp_module
 import sync
 import uploader
+from uploader import delete_book, fetch_all_books, find_duplicate_groups
 
 logging.basicConfig(level=logging.INFO, format="%(asctime)s %(levelname)s %(name)s — %(message)s")
 log = logging.getLogger(__name__)
@@ -185,6 +186,34 @@ async def test_calibre():
     return {"ok": ok, "message": message}
 
 
+# --- Duplicates ---
+
+@app.get("/duplicates", response_class=HTMLResponse)
+async def duplicates_page(request: Request):
+    cfg = config.load()
+    error = None
+    groups: list = []
+    total_books = 0
+    try:
+        books = fetch_all_books(cfg.calibre)
+        total_books = len(books)
+        groups = find_duplicate_groups(books)
+    except Exception as e:
+        error = str(e)
+    return templates.TemplateResponse(request, "duplicates.html", {
+        "groups": groups,
+        "total_books": total_books,
+        "error": error,
+    })
+
+
+@app.post("/api/delete_book/{book_id}")
+async def delete_book_api(book_id: int):
+    cfg = config.load()
+    ok, message = delete_book(cfg.calibre, book_id)
+    return {"ok": ok, "message": message}
+
+
 # --- Data reset ---
 
 @app.post("/settings/reset-sync-data")

templates/base.html

@@ -11,6 +11,7 @@
     <a href="/" class="brand">CalibreSync</a>
     <a href="/">Dashboard</a>
     <a href="/books">Books</a>
+    <a href="/duplicates">Duplicates</a>
     <a href="/settings">Settings</a>
   </nav>
   <main>

templates/duplicates.html

@@ -0,0 +1,81 @@
+{% extends "base.html" %}
+{% block title %}Duplicates — CalibreSync{% endblock %}
+
+{% block content %}
+<div class="page-header">
+  <h1>Duplicate books in Calibre-Web</h1>
+</div>
+
+{% if error %}
+  <div class="alert alert-warning">Could not fetch books from Calibre-Web: {{ error }}</div>
+{% else %}
+  <p class="muted small" style="margin-bottom:1.5rem">
+    Scanned <strong>{{ total_books }}</strong> book(s) —
+    {% if groups %}
+      found <strong>{{ groups|length }}</strong> duplicate group(s).
+      Books are grouped by normalised title. Keep the one you want and delete the rest.
+    {% else %}
+      no duplicates found.
+    {% endif %}
+  </p>
+
+  {% for group in groups %}
+  <div class="form-section" style="margin-bottom:1rem">
+    <h3 style="margin-top:0">{{ group[0].title }}</h3>
+    <table>
+      <thead>
+        <tr>
+          <th>ID</th>
+          <th>Title</th>
+          <th>Authors</th>
+          <th>Format</th>
+          <th></th>
+        </tr>
+      </thead>
+      <tbody>
+        {% for book in group %}
+        <tr id="row-{{ book.id }}">
+          <td class="mono muted">{{ book.id }}</td>
+          <td>{{ book.title }}</td>
+          <td>{{ book.authors }}</td>
+          <td>{{ book.format or "—" }}</td>
+          <td>
+            <button class="btn btn-danger" style="padding:0.2rem 0.7rem;font-size:0.85rem"
+                    onclick="deleteBook({{ book.id }}, this)">Delete</button>
+            <span id="status-{{ book.id }}" class="muted small" style="margin-left:0.5rem"></span>
+          </td>
+        </tr>
+        {% endfor %}
+      </tbody>
+    </table>
+  </div>
+  {% endfor %}
+{% endif %}
+
+<script>
+async function deleteBook(id, btn) {
+  if (!confirm('Delete book ID ' + id + ' from Calibre-Web?')) return;
+  btn.disabled = true;
+  btn.textContent = 'Deleting…';
+  const status = document.getElementById('status-' + id);
+  try {
+    const r = await fetch('/api/delete_book/' + id, {method: 'POST'});
+    const data = await r.json();
+    if (data.ok) {
+      document.getElementById('row-' + id).style.opacity = '0.35';
+      btn.textContent = 'Deleted';
+      status.textContent = '✓';
+    } else {
+      btn.disabled = false;
+      btn.textContent = 'Delete';
+      status.textContent = 'Failed: ' + data.message;
+      status.style.color = 'var(--error, #f87171)';
+    }
+  } catch (e) {
+    btn.disabled = false;
+    btn.textContent = 'Delete';
+    status.textContent = 'Error: ' + e;
+  }
+}
+</script>
+{% endblock %}

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)