michaelschiemer/tests/debug/test-bit-order-in-codewords.php

<?php

declare(strict_types=1);

require_once __DIR__ . '/../../vendor/autoload.php';

use App\Framework\QrCode\QrCodeGenerator;
use App\Framework\QrCode\ValueObjects\ErrorCorrectionLevel;
use App\Framework\QrCode\ValueObjects\QrCodeConfig;
use App\Framework\QrCode\ValueObjects\QrCodeVersion;
use App\Framework\QrCode\ValueObjects\EncodingMode;

echo "=== Testing Bit Order in Codewords ===\n\n";

$testData = 'HELLO WORLD';
$config = new QrCodeConfig(
    version: QrCodeVersion::fromNumber(1),
    errorCorrectionLevel: ErrorCorrectionLevel::M,
    encodingMode: EncodingMode::BYTE
);

// Expected encoding:
// Mode: 0100 (4 bits)
// Count: 00001011 (8 bits = 11)
// So first codeword should be: 01000000 = 64

echo "Expected encoding:\n";
echo "  Mode (4 bits): 0100\n";
echo "  Count (8 bits): 00001011\n";
echo "  First codeword (8 bits): 01000000 = 64\n\n";

// But wait - if we read bits MSB-first, the first codeword should contain:
// Bits 0-7: 01000000 = 64
// But mode indicator is only 4 bits, so we need to look at the first 4 bits only

// Let's check what we actually get
$matrix = QrCodeGenerator::generate($testData, $config);
$size = $matrix->getSize();

// Extract format info
$formatCols = [0, 1, 2, 3, 4, 5, 7, 8, $size - 1, $size - 2, $size - 3, $size - 4, $size - 5, $size - 6, $size - 7];
$formatH = '';
foreach ($formatCols as $col) {
    $formatH .= $matrix->getModuleAt(8, $col)->isDark() ? '1' : '0';
}

$xorMask = "101010000010010";
$unmasked = '';
for ($i = 0; $i < 15; $i++) {
    $unmasked .= (int)$formatH[$i] ^ (int)$xorMask[$i];
}

$maskBits = substr($unmasked, 2, 3);
$maskPattern = bindec($maskBits);

use App\Framework\QrCode\Masking\MaskPattern as MaskPatternEnum;

$mask = match($maskPattern) {
    0 => MaskPatternEnum::PATTERN_0,
    1 => MaskPatternEnum::PATTERN_1,
    2 => MaskPatternEnum::PATTERN_2,
    3 => MaskPatternEnum::PATTERN_3,
    4 => MaskPatternEnum::PATTERN_4,
    5 => MaskPatternEnum::PATTERN_5,
    6 => MaskPatternEnum::PATTERN_6,
    7 => MaskPatternEnum::PATTERN_7,
};

// Read first 16 bits (mode + count)
$dataBits = [];
$bitCount = 0;

for ($col = $size - 1; $col >= 1 && $bitCount < 16; $col -= 2) {
    if ($col === 6) {
        $col--;
    }
    
    $upward = ((int) (($size - 1 - $col) / 2) % 2) === 0;
    
    for ($i = 0; $i < $size && $bitCount < 16; $i++) {
        $row = $upward ? ($size - 1 - $i) : $i;
        
        for ($c = 0; $c < 2 && $bitCount < 16; $c++) {
            $currentCol = $col - $c;
            
            // Skip function patterns
            if (($row <= 8 && $currentCol <= 8) ||
                ($row <= 7 && $currentCol >= $size - 8) ||
                ($row >= $size - 8 && $currentCol <= 7) ||
                $row === 6 || $currentCol === 6 ||
                ($row === 8 && (($currentCol >= 0 && $currentCol <= 5) || $currentCol === 7 || $currentCol === 8 || $currentCol >= $size - 8)) ||
                ($currentCol === 8 && (($row >= 0 && $row <= 5) || $row === 7 || $row === 8 || $row >= $size - 7)) ||
                ($row === 13 && $currentCol === 8)) {
                continue;
            }
            
            $maskedBit = $matrix->getModuleAt($row, $currentCol)->isDark() ? 1 : 0;
            $unmaskedBit = $mask->shouldInvert($row, $currentCol) ? (1 - $maskedBit) : $maskedBit;
            $dataBits[] = $unmaskedBit;
            $bitCount++;
        }
    }
}

$bitString = implode('', $dataBits);
echo "First 16 bits read: {$bitString}\n";

// First 4 bits = mode
$modeBits = substr($bitString, 0, 4);
$mode = bindec($modeBits);
echo "  Mode (bits 0-3): {$modeBits} = {$mode} (expected: 0100 = 4)\n";

// Next 8 bits = count
$countBits = substr($bitString, 4, 8);
$count = bindec($countBits);
echo "  Count (bits 4-11): {$countBits} = {$count} (expected: 00001011 = 11)\n\n";

if ($mode === 4 && $count === 11) {
    echo "✅ Mode and count are CORRECT!\n";
    echo "\nThe problem is that we're reading whole codewords (8 bits) instead of\n";
    echo "reading the mode indicator (4 bits) separately!\n";
} else {
    echo "❌ Mode or count is WRONG!\n";
}

// Now let's see what the first codeword actually is
$firstCodewordBits = substr($bitString, 0, 8);
$firstCodeword = bindec($firstCodewordBits);
echo "\nFirst codeword (bits 0-7): {$firstCodewordBits} = {$firstCodeword}\n";
echo "This is WRONG because it includes mode (4 bits) + part of count (4 bits)\n";