חילוץ פריטים מחשבוניות

def _calculate_extraction_score(items, confidence):
    # Base confidence weight (40%)
    score = confidence * 0.4

    # Completeness score per item (40%)
    for item in items:
        item_score = 0
        if item.get('description'): item_score += 0.2
        if item.get('quantity'): item_score += 0.15
        if item.get('unit_price'): item_score += 0.15
        if item.get('line_total'): item_score += 0.15
        if item.get('product_code') or item.get('sku'): item_score += 0.1
        if item.get('vat_rate'): item_score += 0.1
        ...

    # Bonus for number of items (20%)
    items_bonus = min(len(items) * 0.02, 0.2)

    return min(score + completeness + items_bonus, 1.0)

// Response Example
{
    "success": true,
    "data": {
        "items": [
            {
                "id": 1,
                "line_number": 1,
                "description": "שקי מלט 50 ק\"ג",
                "description_english": "Cement bags 50 kg",
                "original_language": "he",
                "product_code": "BLD-001",
                "sku": "BLD-001",
                "quantity": 20,
                "unit_price": 45.00,
                "line_total": 900.00,
                "unit_of_measure": "ק\"ג",
                "unit_of_measure_normalized": "kg",
                "extraction_method": "hybrid",
                "ai_provider": "google",
                "confidence_score": 0.92
            }
        ],
        "total_items": 6,
        "extraction_method": "hybrid",
        "confidence": 0.92
    }
}

// Request Body
{
    "use_ai": true,
    "ai_provider": "google",  // optional: "google", "openai", "anthropic"
    "force": true             // optional: re-extract even if exists
}

// Response
{
    "success": true,
    "data": {
        "items_extracted": 6,
        "extraction_method": "hybrid",
        "ai_provider": "google",
        "confidence": 0.92,
        "document_language": "he"
    }
}

# AI API Keys
GOOGLE_AI_API_KEY=AIzaSy...
OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...

# OCR Configuration
OCR_DEFAULT_LANGUAGE=heb+eng
OCR_CONFIDENCE_THRESHOLD=0.7

# Line Items Extraction
LINE_ITEMS_USE_AI=true
LINE_ITEMS_AI_PROVIDER=google  # google, openai, anthropic, auto
LINE_ITEMS_HYBRID_MODE=true

from app.services.ocr_service import ocr_service

# Extract line items from OCR text
invoice_text = """
חשבונית מס' 1087
מק"ט        פריט                 כמות   מחיר    סה"כ
BLD-001     שקי מלט 50 ק"ג        20   45.00    900.00
BLD-002     בלוקים 20x20x40      100    8.50    850.00
"""

# Hybrid extraction (AI + Regex)
result = ocr_service.extract_line_items(invoice_text)

print(f"Method: {result['extraction_method']}")
print(f"AI Provider: {result.get('ai_provider', 'N/A')}")
print(f"Items found: {len(result['items'])}")
print(f"Confidence: {result['confidence']:.0%}")

for item in result['items']:
    print(f"  - {item['sku']}: {item['description']} = {item['line_total']}")

// Get line items from document
async function getLineItems(documentId) {
    const response = await fetch(
        `/api/email-scanning/documents/${documentId}/line-items`,
        {
            credentials: 'include',
            headers: { 'Content-Type': 'application/json' }
        }
    );

    const result = await response.json();
    const data = result.data || result;

    console.log(`Extraction method: ${data.extraction_method}`);
    console.log(`Items: ${data.items.length}`);

    data.items.forEach(item => {
        console.log(`${item.sku}: ${item.description}`);
        if (item.description_english) {
            console.log(`  (EN: ${item.description_english})`);
        }
    });
}

# Output Example
Method: hybrid
AI Provider: google
Items found: 6
Confidence: 92%

  - BLD-001: שקי מלט 50 ק"ג = 900.0
    (EN: Cement bags 50 kg)
  - BLD-002: בלוקים 20x20x40 = 850.0
    (EN: Blocks 20x20x40)
  - PLM-100: צינור PVC 4 אינץ' = 325.0
    (EN: PVC pipe 4 inch)