wms-app/scripts/adjust_docx_whitespace.php

<?php

declare(strict_types=1);

if ($argc !== 2) {
    fwrite(STDERR, "Usage: php adjust_docx_whitespace.php <document.docx>\n");
    exit(2);
}

$path = $argv[1];
$wordNs = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main';
$zip = new ZipArchive();
if ($zip->open($path) !== true) {
    throw new RuntimeException("Cannot open DOCX: {$path}");
}

$documentXml = $zip->getFromName('word/document.xml');
if ($documentXml === false) {
    throw new RuntimeException('word/document.xml is missing');
}

$doc = new DOMDocument('1.0', 'UTF-8');
$doc->preserveWhiteSpace = false;
$doc->formatOutput = false;
if (!$doc->loadXML($documentXml)) {
    throw new RuntimeException('word/document.xml is invalid');
}

$xpath = new DOMXPath($doc);
$xpath->registerNamespace('w', $wordNs);

function attr(DOMElement $element, string $name, string $value, string $ns): void
{
    $element->setAttributeNS($ns, 'w:' . $name, $value);
}

function child(DOMDocument $doc, DOMElement $parent, string $localName, string $ns): DOMElement
{
    foreach ($parent->childNodes as $node) {
        if ($node instanceof DOMElement && $node->namespaceURI === $ns && $node->localName === $localName) {
            return $node;
        }
    }
    $element = $doc->createElementNS($ns, 'w:' . $localName);
    $parent->appendChild($element);
    return $element;
}

foreach ($xpath->query('//w:body/w:p | //w:body/w:tbl//w:p') as $paragraph) {
    if (!$paragraph instanceof DOMElement) {
        continue;
    }

    $text = '';
    foreach ($xpath->query('.//w:t', $paragraph) as $textNode) {
        $text .= $textNode->textContent;
    }
    $text = trim(preg_replace('/\s+/u', ' ', $text) ?? '');

    $pPrNode = $xpath->query('./w:pPr', $paragraph)->item(0);
    if (!$pPrNode instanceof DOMElement) {
        $pPrNode = $doc->createElementNS($wordNs, 'w:pPr');
        $paragraph->insertBefore($pPrNode, $paragraph->firstChild);
    }
    $spacing = child($doc, $pPrNode, 'spacing', $wordNs);

    $before = '0';
    $after = '30';
    $line = '320';

    if ($text === '') {
        $after = '20';
        $line = '120';
    } elseif (preg_match('/^[1-7]\.\s/u', $text)) {
        $before = '120';
        $after = '45';
        $line = '330';
    } elseif (preg_match('/^[1-7]\.\d\s/u', $text)) {
        $before = '70';
        $after = '30';
        $line = '320';
    } elseif (str_starts_with($text, '•')) {
        $after = '15';
        $line = '310';
    } elseif (str_starts_with($text, 'วันที่ 5 มกราคม')) {
        $before = '40';
        $after = '100';
        $line = '330';
    } elseif (str_starts_with($text, 'เรื่อง') || str_starts_with($text, 'เรียน')) {
        $after = '55';
        $line = '330';
    } elseif (str_starts_with($text, 'ลงชื่อ')) {
        $before = '160';
        $after = '10';
        $line = '310';
    } elseif (str_starts_with($text, '(') || str_starts_with($text, 'ตำแหน่ง') || str_starts_with($text, 'วันที่ _') || str_starts_with($text, 'ผู้มีอำนาจ')) {
        $after = '10';
        $line = '310';
    } elseif (str_starts_with($text, 'บริษัท บี.อาร์.เอ็น.') || str_starts_with($text, 'เอกสารฉบับนี้') || str_starts_with($text, 'ผลส่งมอบประกอบ') || str_starts_with($text, 'การเปลี่ยนแปลง') || str_starts_with($text, 'จึงจัดทำ')) {
        $after = '55';
        $line = '330';
    }

    attr($spacing, 'before', $before, $wordNs);
    attr($spacing, 'after', $after, $wordNs);
    attr($spacing, 'line', $line, $wordNs);
    attr($spacing, 'lineRule', 'auto', $wordNs);
}

foreach ($xpath->query('//w:sectPr/w:pgMar') as $margin) {
    if (!$margin instanceof DOMElement) {
        continue;
    }
    attr($margin, 'top', '850', $wordNs);
    attr($margin, 'bottom', '850', $wordNs);
    attr($margin, 'left', '1080', $wordNs);
    attr($margin, 'right', '1080', $wordNs);
    attr($margin, 'header', '400', $wordNs);
    attr($margin, 'footer', '400', $wordNs);
}

foreach ($xpath->query('//w:tbl/w:tblPr') as $tableProperties) {
    if (!$tableProperties instanceof DOMElement) {
        continue;
    }
    $margins = child($doc, $tableProperties, 'tblCellMar', $wordNs);
    foreach (['top' => '45', 'left' => '70', 'bottom' => '45', 'right' => '70'] as $side => $width) {
        $edge = child($doc, $margins, $side, $wordNs);
        attr($edge, 'w', $width, $wordNs);
        attr($edge, 'type', 'dxa', $wordNs);
    }
}

$updatedXml = $doc->saveXML();
if ($updatedXml === false) {
    throw new RuntimeException('Cannot serialize adjusted document XML');
}
if (!$zip->addFromString('word/document.xml', $updatedXml)) {
    throw new RuntimeException('Cannot update word/document.xml');
}
$zip->close();

echo $path . PHP_EOL;