feat: Phase 5 – PDF/Bild-Beleg-Import (lokal, Tesseract-OCR)

Backend:
- services/ocr: PDF-Text via PyMuPDF, OCR (Tesseract) für Scans/Bilder,
  heuristischer Parser (Datum + Betrag je Zeile, CH/DE/US-Beträge).
- routers/documents: /documents/preview (Upload -> Text -> Kandidaten ->
  gemeinsame Prüf-/Buchungslogik des CSV-Imports). csv_import refaktoriert:
  detect_rows() von CSV und Beleg gemeinsam genutzt.
- requirements: PyMuPDF, pytesseract, pillow; install.sh installiert tesseract-ocr
  (+ deu). pytest: Heuristik-Parser und PDF-Vorschau.

Frontend:
- Import-Seite: CSV/PDF-Umschalter; PDF-Modus mit Beleg-Auslesen, extrahiertem
  Text zur Kontrolle und Best-Effort-Hinweis; gleiche Prüfansicht + Buchen.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

backend/app/services/csv_import.py

@@ -153,16 +153,45 @@ def _apply_rules(rules: list[CategorizationRule], payee: str, purpose: str) -> i
     return None
 
 
-def build_preview(db: Session, user: User, account: Account, template: ImportTemplate, data: bytes) -> list[dict]:
+def parse_file_to_candidates(data: bytes, template: ImportTemplate) -> list[dict]:
+    """Zerlegt die Datei anhand des Templates in rohe Kandidaten-Zeilen."""
     cmap = json.loads(template.column_map or "{}")
 
     def col(field: str) -> int | None:
         idx = cmap.get(field)
         return idx if isinstance(idx, int) and idx >= 0 else None
 
-    rows = read_rows(data, template)
+    out: list[dict] = []
+    for r in read_rows(data, template):
+        def cell(field: str) -> str:
+            c = col(field)
+            return r[c].strip() if c is not None and c < len(r) else ""
 
-    # Identifikatoren aller Konten des Nutzers -> Konto-ID (für Transfer-Erkennung).
+        d = parse_date(cell("date"), template.date_format)
+        if template.amount_mode == AmountMode.debit_credit:
+            debit = parse_amount(cell("debit"), template.decimal_sep)
+            credit = parse_amount(cell("credit"), template.decimal_sep)
+            amount = (credit or 0) - (abs(debit) if debit else 0)
+        elif template.amount_mode == AmountMode.abs_expense:
+            a = parse_amount(cell("amount"), template.decimal_sep)
+            amount = -abs(a) if a is not None else None
+        else:
+            amount = parse_amount(cell("amount"), template.decimal_sep)
+
+        out.append({
+            "date": d.isoformat() if d else None,
+            "amount_cents": amount,
+            "payee": cell("payee") or None,
+            "purpose": cell("purpose") or None,
+            "counterparty_iban": normalize_iban(cell("counterparty_iban")) or None,
+        })
+    return out
+
+
+def detect_rows(db: Session, user: User, account: Account, candidates: list[dict]) -> list[dict]:
+    """Reichert Kandidaten an: Typ, Auto-Kategorie, interner Transfer, Duplikat.
+
+    Wird von CSV- UND Beleg-Import genutzt (gleiche Prüf-/Buchungslogik)."""
     idents: dict[str, int] = {}
     for value_norm, acc_id in (
         db.query(AccountIdentifier.value_norm, AccountIdentifier.account_id)
@@ -177,39 +206,25 @@ def build_preview(db: Session, user: User, account: Account, template: ImportTem
         .order_by(CategorizationRule.priority.desc(), CategorizationRule.hits.desc())
         .all()
     )
-
     existing = {
         h[0]
         for h in db.query(Transaction.import_hash).filter(
             Transaction.owner_id == user.id, Transaction.import_hash.isnot(None)
         )
     }
-    seen_in_file: set[str] = set()
+    seen: set[str] = set()
 
     out: list[dict] = []
-    for i, r in enumerate(rows):
-        def cell(field: str) -> str:
-            c = col(field)
-            return r[c].strip() if c is not None and c < len(r) else ""
-
-        d = parse_date(cell("date"), template.date_format)
-
-        if template.amount_mode == AmountMode.debit_credit:
-            debit = parse_amount(cell("debit"), template.decimal_sep)
-            credit = parse_amount(cell("credit"), template.decimal_sep)
-            amount = (credit or 0) - (abs(debit) if debit else 0)
-        elif template.amount_mode == AmountMode.abs_expense:
-            a = parse_amount(cell("amount"), template.decimal_sep)
-            amount = -abs(a) if a is not None else None
-        else:
-            amount = parse_amount(cell("amount"), template.decimal_sep)
-
-        payee = cell("payee")
-        purpose = cell("purpose")
-        iban = normalize_iban(cell("counterparty_iban"))
+    for i, cand in enumerate(candidates):
+        d_iso = cand.get("date")
+        amount = cand.get("amount_cents")
+        payee = cand.get("payee")
+        purpose = cand.get("purpose")
+        iban = cand.get("counterparty_iban")
+        d = date.fromisoformat(d_iso) if d_iso else None
 
         valid = d is not None and amount is not None and amount != 0
-        ttype = "transfer" if False else ("expense" if (amount or 0) < 0 else "income")
+        ttype = "expense" if (amount or 0) < 0 else "income"
         category_id: int | None = None
         to_account_id: int | None = None
 
@@ -218,11 +233,11 @@ def build_preview(db: Session, user: User, account: Account, template: ImportTem
             ttype = "transfer"
             to_account_id = other
         else:
-            category_id = _apply_rules(rules, payee, purpose)
+            category_id = _apply_rules(rules, payee or "", purpose or "")
 
         h = import_hash(account.id, d, amount, payee)
-        is_dup = h in existing or h in seen_in_file
-        seen_in_file.add(h)
+        is_dup = h in existing or h in seen
+        seen.add(h)
 
         out.append({
             "index": i,
@@ -242,6 +257,10 @@ def build_preview(db: Session, user: User, account: Account, template: ImportTem
     return out
 
 
+def build_preview(db: Session, user: User, account: Account, template: ImportTemplate, data: bytes) -> list[dict]:
+    return detect_rows(db, user, account, parse_file_to_candidates(data, template))
+
+
 # --- Verbuchen -------------------------------------------------------------