esun3d/app/code/Esun/MultiLocaleTranslation/Model/Html/TranslatableTextExtractor.php

<?php

declare(strict_types=1);

namespace Esun\MultiLocaleTranslation\Model\Html;

class TranslatableTextExtractor
{
    private const IGNORED_TAGS = ['code', 'noscript', 'script', 'style', 'svg', 'template'];

    /**
     * @return array<int, array{offset:int,length:int,text:string}>
     */
    public function extract(string $html, int $maxFragmentBytes): array
    {
        $ranges = $this->findTextRanges($html);
        $fragments = [];

        foreach ($ranges as [$rangeOffset, $rangeLength]) {
            foreach ($this->findEncodedTextRanges($html, $rangeOffset, $rangeLength) as [$offset, $length]) {
                $value = substr($html, $offset, $length);
                if ($value === '' || preg_match('/\S/u', $value) !== 1) {
                    continue;
                }

                preg_match('/\A(\s*)/u', $value, $leadingMatch);
                preg_match('/(\s*)\z/u', $value, $trailingMatch);
                $leadingBytes = strlen($leadingMatch[1] ?? '');
                $trailingBytes = strlen($trailingMatch[1] ?? '');
                $textLength = strlen($value) - $leadingBytes - $trailingBytes;
                if ($textLength <= 0) {
                    continue;
                }

                $textOffset = $offset + $leadingBytes;
                $text = substr($html, $textOffset, $textLength);
                foreach ($this->splitText($text, $maxFragmentBytes) as [$partOffset, $part]) {
                    $fragments[] = [
                        'offset' => $textOffset + $partOffset,
                        'length' => strlen($part),
                        'text' => $part,
                    ];
                }
            }
        }

        return $fragments;
    }

    /**
     * @return array<int, array{0:int,1:int}>
     */
    private function findTextRanges(string $html): array
    {
        $ranges = [];
        $length = strlen($html);
        $cursor = 0;
        $textStart = 0;
        $ignoredTag = null;

        while ($cursor < $length) {
            $tagStart = strpos($html, '<', $cursor);
            if ($tagStart === false) {
                break;
            }

            $tagEnd = $this->findTagEnd($html, $tagStart);
            if ($tagEnd === null) {
                $cursor = $tagStart + 1;
                continue;
            }

            $tag = substr($html, $tagStart, $tagEnd - $tagStart);
            if ($ignoredTag === null && $tagStart > $textStart) {
                $ranges[] = [$textStart, $tagStart - $textStart];
            }

            if (preg_match('/^<\s*(\/?)\s*([a-z][a-z0-9:-]*)/i', $tag, $match) === 1) {
                $isClosing = $match[1] === '/';
                $tagName = strtolower($match[2]);
                if ($ignoredTag === null && !$isClosing && in_array($tagName, self::IGNORED_TAGS, true)
                    && !str_ends_with(rtrim($tag), '/>')
                ) {
                    $ignoredTag = $tagName;
                } elseif ($ignoredTag === $tagName && $isClosing) {
                    $ignoredTag = null;
                }
            }

            $cursor = $tagEnd;
            $textStart = $tagEnd;
        }

        if ($ignoredTag === null && $textStart < $length) {
            $ranges[] = [$textStart, $length - $textStart];
        }

        return $ranges;
    }

    private function findTagEnd(string $html, int $start): ?int
    {
        if (substr($html, $start, 4) === '<!--') {
            $end = strpos($html, '-->', $start + 4);
            return $end === false ? null : $end + 3;
        }
        if (substr($html, $start, 9) === '<![CDATA[') {
            $end = strpos($html, ']]>', $start + 9);
            return $end === false ? null : $end + 3;
        }
        if (preg_match('/\A<\s*(?:\/?\s*[a-z]|!doctype\b|\?)/i', substr($html, $start, 32)) !== 1) {
            return null;
        }

        $length = strlen($html);
        $quote = null;
        for ($index = $start + 1; $index < $length; $index++) {
            $character = $html[$index];
            if ($quote !== null) {
                if ($character === $quote) {
                    $quote = null;
                }
                continue;
            }
            if ($character === '"' || $character === "'") {
                $quote = $character;
            } elseif ($character === '>') {
                return $index + 1;
            }
        }

        return null;
    }

    /**
     * Split a normal HTML text range again when Page Builder stores HTML tags as &lt;...&gt;.
     *
     * @return array<int, array{0:int,1:int}>
     */
    private function findEncodedTextRanges(string $html, int $offset, int $length): array
    {
        $ranges = [];
        $limit = $offset + $length;
        $cursor = $offset;
        $textStart = $offset;
        $ignoredTag = null;
        $foundTag = false;

        while ($cursor < $limit) {
            $tagStart = stripos($html, '&lt;', $cursor);
            if ($tagStart === false || $tagStart >= $limit) {
                break;
            }

            $tagEnd = $this->findEncodedTagEnd($html, $tagStart, $limit);
            if ($tagEnd === null) {
                $cursor = $tagStart + 4;
                continue;
            }

            $encodedTag = substr($html, $tagStart, $tagEnd - $tagStart);
            $tag = html_entity_decode($encodedTag, ENT_QUOTES | ENT_HTML5, 'UTF-8');
            if (preg_match('/^<\s*(?:\/?\s*[a-z][a-z0-9:-]*|!--|!doctype\b|\?)/i', $tag) !== 1) {
                $cursor = $tagStart + 4;
                continue;
            }

            $foundTag = true;
            if ($ignoredTag === null && $tagStart > $textStart) {
                $ranges[] = [$textStart, $tagStart - $textStart];
            }

            if (preg_match('/^<\s*(\/?)\s*([a-z][a-z0-9:-]*)/i', $tag, $match) === 1) {
                $isClosing = $match[1] === '/';
                $tagName = strtolower($match[2]);
                if ($ignoredTag === null && !$isClosing && in_array($tagName, self::IGNORED_TAGS, true)
                    && !str_ends_with(rtrim($tag), '/>')
                ) {
                    $ignoredTag = $tagName;
                } elseif ($ignoredTag === $tagName && $isClosing) {
                    $ignoredTag = null;
                }
            }

            $cursor = $tagEnd;
            $textStart = $tagEnd;
        }

        if (!$foundTag) {
            return [[$offset, $length]];
        }
        if ($ignoredTag === null && $textStart < $limit) {
            $ranges[] = [$textStart, $limit - $textStart];
        }

        return $ranges;
    }

    private function findEncodedTagEnd(string $html, int $start, int $limit): ?int
    {
        if (strncasecmp(substr($html, $start, 7), '&lt;!--', 7) === 0) {
            $end = stripos($html, '--&gt;', $start + 7);
            return $end === false || $end >= $limit ? null : $end + 6;
        }

        $quote = null;
        for ($index = $start + 4; $index < $limit; $index++) {
            $character = $html[$index];
            if ($quote !== null) {
                if ($character === $quote) {
                    $quote = null;
                }
                continue;
            }
            if ($character === '"' || $character === "'") {
                $quote = $character;
            } elseif (strncasecmp(substr($html, $index, 4), '&gt;', 4) === 0) {
                return $index + 4;
            }
        }

        return null;
    }

    /**
     * @return array<int, array{0:int,1:string}>
     */
    private function splitText(string $text, int $maxBytes): array
    {
        $parts = [];
        $offset = 0;
        $length = strlen($text);

        while ($offset < $length) {
            $remaining = substr($text, $offset);
            if (strlen($remaining) <= $maxBytes) {
                $parts[] = [$offset, $remaining];
                break;
            }

            $candidate = mb_strcut($remaining, 0, $maxBytes, 'UTF-8');
            $splitAt = strlen($candidate);
            if (preg_match_all('/\s+/u', $candidate, $matches, PREG_OFFSET_CAPTURE) > 0) {
                $last = end($matches[0]);
                $whitespaceEnd = $last[1] + strlen($last[0]);
                if ($whitespaceEnd >= (int)($maxBytes / 2)) {
                    $splitAt = $whitespaceEnd;
                }
            }

            $parts[] = [$offset, substr($text, $offset, $splitAt)];
            $offset += $splitAt;
        }

        return $parts;
    }
}