sample_data/generate_samples.py

"""
Generate sample document images for demonstration and testing.
"""

from PIL import Image, ImageDraw, ImageFont
import os

def create_sample_receipt(path: str):
    os.makedirs(os.path.dirname(path), exist_ok=True)
    img = Image.new("RGB", (600, 750), color="#FFFFFF")
    draw = ImageDraw.Draw(img)

    # Header
    draw.rectangle([(20, 20), (580, 730)], outline="#D0D5DD", width=2)
    draw.text((220, 40), "TECH COFFEE ROASTERS", fill="#101828")
    draw.text((235, 65), "123 Innovation Way, Suite 400", fill="#475467")
    draw.text((265, 85), "Tel: (555) 019-2834", fill="#475467")

    draw.line([(40, 115), (560, 115)], fill="#E4E7EC", width=1)
    
    # Metadata
    draw.text((40, 130), "Receipt #: TCR-2026-8942", fill="#101828")
    draw.text((40, 150), "Date: 2026-08-26 14:22", fill="#475467")
    draw.text((40, 170), "Server: Alex M.", fill="#475467")

    draw.line([(40, 200), (560, 200)], fill="#101828", width=2)
    
    # Table Header
    draw.text((40, 215), "ITEM", fill="#101828")
    draw.text((320, 215), "QTY", fill="#101828")
    draw.text((420, 215), "PRICE", fill="#101828")
    draw.text((500, 215), "TOTAL", fill="#101828")
    
    draw.line([(40, 240), (560, 240)], fill="#E4E7EC", width=1)

    # Items
    items = [
        ("Single Origin Pour Over", "2", "$4.50", "$9.00"),
        ("Oat Milk Vanilla Latte", "1", "$5.75", "$5.75"),
        ("Almond Croissant", "2", "$4.25", "$8.50"),
        ("Avocado Sourdough Toast", "1", "$12.00", "$12.00"),
        ("Matcha Green Tea", "1", "$5.00", "$5.00"),
    ]
    y = 260
    for name, qty, price, total in items:
        draw.text((40, y), name, fill="#344054")
        draw.text((330, y), qty, fill="#344054")
        draw.text((420, y), price, fill="#344054")
        draw.text((500, y), total, fill="#344054")
        y += 35

    draw.line([(40, y + 10), (560, y + 10)], fill="#E4E7EC", width=1)
    y += 30

    # Totals
    draw.text((380, y), "Subtotal:", fill="#475467")
    draw.text((500, y), "$40.25", fill="#101828")
    y += 25
    draw.text((380, y), "Tax (8.25%):", fill="#475467")
    draw.text((500, y), "$3.32", fill="#101828")
    y += 25
    draw.text((380, y), "Total Due:", fill="#101828")
    draw.text((500, y), "$43.57", fill="#101828")

    draw.line([(40, y + 40), (560, y + 40)], fill="#101828", width=2)
    y += 60
    draw.text((200, y), "Thank you for your visit!", fill="#475467")
    draw.text((180, y + 25), "Scan QR code to join our rewards program", fill="#98A2B3")

    img.save(path)
    print(f"Generated sample receipt at: {path}")

def create_sample_table_doc(path: str):
    os.makedirs(os.path.dirname(path), exist_ok=True)
    img = Image.new("RGB", (700, 500), color="#FFFFFF")
    draw = ImageDraw.Draw(img)

    draw.rectangle([(20, 20), (680, 480)], outline="#E4E7EC", width=2)
    draw.text((40, 40), "QUARTERLY PERFORMANCE REPORT 2026", fill="#101828")
    draw.text((40, 65), "Department: AI Research & Systems", fill="#475467")

    # Draw table
    table_top = 110
    draw.rectangle([(40, table_top), (660, table_top + 40)], fill="#F8F9FA", outline="#D0D5DD")
    draw.text((50, table_top + 12), "Model Name", fill="#101828")
    draw.text((220, table_top + 12), "Throughput (t/s)", fill="#101828")
    draw.text((380, table_top + 12), "Accuracy (%)", fill="#101828")
    draw.text((520, table_top + 12), "Latency (ms)", fill="#101828")

    rows = [
        ("DeepSeek-OCR", "142.5", "96.8%", "18.2"),
        ("Qwen3-VL", "118.0", "97.4%", "24.1"),
        ("PP-StructureV3", "88.4", "94.9%", "31.0"),
        ("NuExtract3", "130.2", "96.1%", "21.5"),
        ("PP-OCRv5", "210.6", "93.8%", "11.4"),
        ("olmOCR", "95.7", "98.2%", "29.8"),
    ]

    curr_y = table_top + 40
    for name, tp, acc, lat in rows:
        draw.rectangle([(40, curr_y), (660, curr_y + 35)], outline="#E4E7EC")
        draw.text((50, curr_y + 10), name, fill="#344054")
        draw.text((220, curr_y + 10), tp, fill="#344054")
        draw.text((380, curr_y + 10), acc, fill="#344054")
        draw.text((520, curr_y + 10), lat, fill="#344054")
        curr_y += 35

    draw.text((40, curr_y + 30), "Summary: All models achieved target latency within SLA thresholds.", fill="#475467")
    img.save(path)
    print(f"Generated sample table doc at: {path}")

if __name__ == "__main__":
    create_sample_receipt("d:/OCR/sample_data/sample_receipt.png")
    create_sample_table_doc("d:/OCR/sample_data/sample_table.png")