feat: dodaj dodatne troškove s OCR sinkronizacijom

Dodana je tablica DODATNI TROŠKOVI za putni nalog (backend model + API +
frontend modal) s izračunom ukupnog iznosa za isplatu.

Upload računa sada odmah upisuje redak u dodatne troškove, a asinkroni OCR
(Celery) naknadno pokušava popuniti broj računa i iznos iz slike/PDF-a.

Uvedena je invalidacija PDF cache-a nakon promjena računa/dodatnih troškova
kako bi generirani PDF uvijek prikazivao najnovije podatke.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

backend/modules/fleet/tasks.py

@@ -8,7 +8,14 @@ import logging
 from io import BytesIO
 from pathlib import Path
 import base64
+import re
+from decimal import Decimal, InvalidOperation
 from PIL import Image, UnidentifiedImageError
+from django.db import transaction
+from pypdf import PdfReader
+from pypdf.errors import PdfReadError
+import pytesseract
+from pytesseract import TesseractNotFoundError
 from reportlab.lib.pagesizes import A4
 from reportlab.lib.utils import ImageReader
 from reportlab.pdfgen import canvas
@@ -223,6 +230,148 @@ def build_work_order_invoices_pdf_task(work_order_id):
     }
 
 
+def _parse_decimal(value):
+    if value in (None, ''):
+        return Decimal('0.00')
+    try:
+        normalized = str(value).strip().replace('€', '').replace(' ', '').replace(',', '.')
+        return Decimal(normalized)
+    except (InvalidOperation, ValueError, TypeError):
+        return Decimal('0.00')
+
+
+def _extract_invoice_text(invoice):
+    if not invoice.image:
+        return ''
+    suffix = Path(invoice.image.name or '').suffix.lower()
+    if suffix == '.pdf':
+        invoice.image.open('rb')
+        try:
+            reader = PdfReader(invoice.image)
+            return '\n'.join((page.extract_text() or '') for page in reader.pages)
+        finally:
+            invoice.image.close()
+
+    invoice.image.open('rb')
+    try:
+        with Image.open(invoice.image) as source:
+            image = source.convert('RGB')
+            return pytesseract.image_to_string(image, lang='hrv+eng')
+    finally:
+        invoice.image.close()
+
+
+def _extract_invoice_number(text, fallback):
+    if not text:
+        return fallback
+    patterns = [
+        r'(?i)(?:broj\s*računa|račun\s*broj|invoice\s*no\.?)\s*[:#]?\s*([A-Z0-9\/\-\._]+)',
+        r'(?i)(?:broj)\s*[:#]?\s*([A-Z0-9\/\-\._]{4,})',
+    ]
+    for pattern in patterns:
+        match = re.search(pattern, text)
+        if match:
+            candidate = (match.group(1) or '').strip()
+            if candidate:
+                return candidate
+    return fallback
+
+
+def _extract_total_amount(text):
+    if not text:
+        return Decimal('0.00')
+
+    def _line_candidates(line):
+        number_pattern = r'(?<!\d)(\d{1,3}(?:[.\s]\d{3})*(?:,\d{2})|\d+(?:[.,]\d{2}))(?!\d)'
+        return [_parse_decimal(match) for match in re.findall(number_pattern, line)]
+
+    preferred_lines = []
+    for line in text.splitlines():
+        lower = line.lower()
+        if any(token in lower for token in ('ukupno', 'za platiti', 'iznos', 'total')):
+            preferred_lines.append(line)
+
+    for line in preferred_lines:
+        candidates = [value for value in _line_candidates(line) if value > Decimal('0.00')]
+        if candidates:
+            return max(candidates)
+
+    all_candidates = []
+    for line in text.splitlines():
+        all_candidates.extend([value for value in _line_candidates(line) if value > Decimal('0.00')])
+    if all_candidates:
+        return max(all_candidates)
+    return Decimal('0.00')
+
+
+@shared_task
+def process_work_order_invoice_ocr(invoice_id):
+    from .models import WorkOrderInvoice, WorkOrderAdditionalCostsTable, GeneratedWorkOrderPdf
+
+    invoice = (
+        WorkOrderInvoice.objects
+        .select_related('work_order')
+        .filter(pk=invoice_id, is_active=True)
+        .first()
+    )
+    if invoice is None or invoice.work_order_id is None:
+        return {"status": "skipped", "reason": "invoice-not-found"}
+
+    try:
+        text = _extract_invoice_text(invoice)
+    except (UnidentifiedImageError, OSError, TesseractNotFoundError, ValueError, RuntimeError, PdfReadError):
+        text = ''
+
+    attachment_name = Path(invoice.image.name).name if invoice.image and getattr(invoice.image, 'name', '') else ''
+    row = {
+        'naziv': (invoice.naziv_racuna or '').strip() or attachment_name or f'Račun {invoice.pk}',
+        'broj_racuna': _extract_invoice_number(text, fallback=str(invoice.pk)),
+        'ukupan_iznos': f"{_extract_total_amount(text):.2f}",
+        'prilog': attachment_name,
+        'source_invoice_id': str(invoice.pk),
+    }
+
+    with transaction.atomic():
+        table, _ = WorkOrderAdditionalCostsTable.objects.select_for_update().get_or_create(
+            work_order=invoice.work_order,
+            defaults={'data': {'rows': []}},
+        )
+        existing_rows = table.data.get('rows', []) if isinstance(table.data, dict) else []
+        existing_row = next(
+            (
+                item for item in existing_rows
+                if isinstance(item, dict) and str(item.get('source_invoice_id', '')) == str(invoice.pk)
+            ),
+            None,
+        )
+        normalized_rows = [
+            item for item in existing_rows
+            if isinstance(item, dict) and str(item.get('source_invoice_id', '')) != str(invoice.pk)
+        ]
+        if existing_row and row['ukupan_iznos'] == '0.00':
+            existing_amount = str(existing_row.get('ukupan_iznos', '') or '').strip()
+            if existing_amount:
+                row['ukupan_iznos'] = existing_amount
+        if existing_row and not row['broj_racuna']:
+            row['broj_racuna'] = str(existing_row.get('broj_racuna', '') or '').strip()
+        normalized_rows.append(row)
+        total = sum((_parse_decimal(item.get('ukupan_iznos')) for item in normalized_rows if isinstance(item, dict)), Decimal('0.00'))
+        table.data = {'rows': normalized_rows}
+        table.total_for_payout = total.quantize(Decimal('0.01'))
+        table.save(update_fields=['data', 'total_for_payout', 'updated_at'])
+        GeneratedWorkOrderPdf.objects.filter(
+            is_active=True,
+            work_order=invoice.work_order,
+            pdf_type__in=['work_order', 'invoices'],
+        ).update(
+            is_active=False,
+            status='failed',
+            error_message='PDF cache invalidiran zbog OCR ažuriranja dodatnih troškova.',
+        )
+
+    return {"status": "ok", "invoice_id": str(invoice.pk)}
+
+
 @shared_task
 def cleanup_expired_generated_pdfs_task():
     now = timezone.now()