#!/usr/bin/env python3 from __future__ import annotations import argparse from pathlib import Path from PIL import Image, ImageDraw, ImageFont def resolve_font_pair(regular_override: Path | None, bold_override: Path | None) -> tuple[Path, Path]: if (regular_override is None) != (bold_override is None): raise SystemExit("--regular-font and --bold-font must be supplied together") if regular_override and bold_override: if not regular_override.is_file() or not bold_override.is_file(): raise SystemExit("one or both explicit font paths do not exist") return regular_override, bold_override families = ( ("DejaVuSans.ttf", "DejaVuSans-Bold.ttf"), ("LiberationSans-Regular.ttf", "LiberationSans-Bold.ttf"), ) roots = (Path("/usr/share/fonts"), Path("/usr/local/share/fonts")) for regular_name, bold_name in families: for root in roots: regular_matches = sorted(root.rglob(regular_name)) if root.is_dir() else [] bold_matches = sorted(root.rglob(bold_name)) if root.is_dir() else [] if regular_matches and bold_matches: return regular_matches[0], bold_matches[0] raise SystemExit("no DejaVu Sans or Liberation Sans font pair found; pass --regular-font and --bold-font") def main() -> None: parser = argparse.ArgumentParser(description="Generate a deterministic OCR validation page") parser.add_argument("--output", type=Path, required=True) parser.add_argument("--pdf-output", type=Path, help="also write the page as a 150-DPI PDF") parser.add_argument("--regular-font", type=Path, help="regular TrueType/OpenType font override") parser.add_argument("--bold-font", type=Path, help="bold TrueType/OpenType font override") args = parser.parse_args() regular_path, bold_path = resolve_font_pair(args.regular_font, args.bold_font) image = Image.new("RGB", (1600, 2000), "white") draw = ImageDraw.Draw(image) title = ImageFont.truetype(str(bold_path), 68) heading = ImageFont.truetype(str(bold_path), 38) body = ImageFont.truetype(str(regular_path), 34) small = ImageFont.truetype(str(regular_path), 28) draw.text((120, 100), "Unlimited OCR RDNA4 Smoke Test", font=title, fill="black") draw.line((120, 205, 1480, 205), fill="black", width=4) draw.text((120, 265), "Invoice No. OCR-2026-0731", font=body, fill="black") draw.text((120, 325), "Date: 31 July 2026", font=body, fill="black") draw.text((120, 385), "GPU target: AMD gfx1201", font=body, fill="black") draw.text((120, 505), "Items", font=heading, fill="black") left, top, right, bottom = 120, 580, 1480, 1050 columns = (left, 930, 1120, right) rows = (top, 690, 810, 930, bottom) for x in columns: draw.line((x, top, x, bottom), fill="black", width=3) for y in rows: draw.line((left, y, right, y), fill="black", width=3) draw.text((145, 610), "Item", font=heading, fill="black") draw.text((960, 610), "Qty", font=heading, fill="black") draw.text((1150, 610), "Price", font=heading, fill="black") draw.text((145, 725), "Document scan", font=body, fill="black") draw.text((980, 725), "2", font=body, fill="black") draw.text((1150, 725), "€19.50", font=body, fill="black") draw.text((145, 845), "Table extraction", font=body, fill="black") draw.text((980, 845), "1", font=body, fill="black") draw.text((1150, 845), "€9.50", font=body, fill="black") draw.text((145, 965), "Total", font=heading, fill="black") draw.text((1150, 965), "€48.50", font=heading, fill="black") draw.text((120, 1190), "Verification notes", font=heading, fill="black") notes = [ "• AMD Radeon RX 9070 XT", "• ROCm architecture: gfx1201", "• Expected checksum: RDNA4-OCR-PASS", "• Formula sample: E = mc²", ] for index, note in enumerate(notes): draw.text((150, 1270 + index * 70), note, font=body, fill="black") draw.line((120, 1780, 1480, 1780), fill="black", width=2) draw.text((120, 1820), "Synthetic validation page; no private data.", font=small, fill="black") args.output.parent.mkdir(parents=True, exist_ok=True) image.save(args.output, format="PNG", optimize=True) print(args.output) if args.pdf_output: args.pdf_output.parent.mkdir(parents=True, exist_ok=True) image.save(args.pdf_output, format="PDF", resolution=150.0) print(args.pdf_output) if __name__ == "__main__": main()