<?php
namespace A1LocalMdi;

use DOMDocument;
use DOMElement;
use DOMXPath;
use A1LocalMdi\Vendor\League\HTMLToMarkdown\HtmlConverter;
use A1LocalMdi\Vendor\League\HTMLToMarkdown\Converter\TableConverter;

final class Document {
    /** Convert the anonymous theme output, including blocks, shortcodes and server-rendered builders. */
    public static function convert(string $html, \WP_Post $post) {
        if (!class_exists(DOMDocument::class)) { return new \WP_Error('a1local_mdi_dom', 'The PHP DOM extension is required.'); }
        if (strlen($html) > 2097152) { return new \WP_Error('a1local_mdi_size', 'This page exceeds the 2 MB conversion limit.'); }
        $dom = new DOMDocument('1.0', 'UTF-8');
        $previous = libxml_use_internal_errors(true);
        // Never load external entities. The encoding declaration keeps non-ASCII content intact.
        $loaded = $dom->loadHTML('<?xml encoding="UTF-8">' . $html, LIBXML_NONET | LIBXML_NOERROR | LIBXML_NOWARNING);
        libxml_clear_errors();
        libxml_use_internal_errors($previous);
        if (!$loaded) { return new \WP_Error('a1local_mdi_html', 'The page could not be read as HTML.'); }
        $xpath = new DOMXPath($dom);
        foreach ($xpath->query('//meta[@name]') as $meta) {
            if (in_array(strtolower($meta->getAttribute('name')), ['robots', 'googlebot'], true) && preg_match('/\b(?:noindex|none)\b/i', $meta->getAttribute('content'))) {
                return new \WP_Error('a1local_mdi_noindex', 'This page asks crawlers not to index it. Markdown is excluded.');
            }
        }

        // An explicit exclusion covers its whole subtree, including schema.
        foreach (iterator_to_array($xpath->query('//*[@data-a1local-mdi-exclude or @data-mfa-exclude]')) as $node) {
            if ($node->parentNode) { $node->parentNode->removeChild($node); }
        }
        $structured = [];
        foreach ($xpath->query('//script[@type="application/ld+json"]') as $script) {
            $data = json_decode($script->textContent, true);
            if (json_last_error() === JSON_ERROR_NONE && is_array($data)) { $structured[] = $data; }
        }
        $remove = '//script|//style|//noscript|//template|//svg|//nav|//aside|//form|//button|//input|//select|//textarea|//iframe|//comment()|//*[@hidden]|//*[@aria-hidden="true"]|//*[@data-a1local-mdi-exclude or @data-mfa-exclude]';
        $nodes = iterator_to_array($xpath->query($remove));
        foreach ($nodes as $node) { if ($node->parentNode) { $node->parentNode->removeChild($node); } }
        foreach (iterator_to_array($xpath->query('//*[@style]')) as $node) {
            if (preg_match('/(?:^|;)\s*(?:display\s*:\s*none|visibility\s*:\s*hidden)\s*(?:!important)?\s*(?:;|$)/i', $node->getAttribute('style'))) {
                $node->parentNode->removeChild($node);
            }
        }

        $root = null;
        $selection = '';
        $selectors = [
            'main' => '//main',
            'main landmark' => '//*[@role="main"]',
            'article' => '//article',
            'post content' => '//*[contains(concat(" ",normalize-space(@class)," ")," entry-content ") or contains(concat(" ",normalize-space(@class)," ")," wp-block-post-content ")]',
            'page body' => '//body',
        ];
        foreach ($selectors as $label => $expression) {
            $matches = $xpath->query($expression);
            if ($matches->length) { $root = $matches->item(0); $selection = $label; break; }
        }
        if (!$root) { return new \WP_Error('a1local_mdi_content', 'No readable content container was found.'); }
        // Site chrome outside the selected main content is not part of the document.
        if ($selection === 'page body') {
            foreach (iterator_to_array($xpath->query('.//header|.//footer', $root)) as $node) { $node->parentNode->removeChild($node); }
        }
        $plain = trim(preg_replace('/\s+/u', ' ', $root->textContent));
        if (strlen($plain) < 20) { return new \WP_Error('a1local_mdi_empty', 'The selected page content is empty or too short. Check the page template.'); }
        $canonical = get_permalink($post);
        foreach ($xpath->query('.//*[@href or @src]', $root) as $element) {
            foreach (['href', 'src'] as $attr) {
                if (!$element->hasAttribute($attr)) { continue; }
                $value = trim($element->getAttribute($attr));
                if ($value === '') { continue; }
                $absolute = \WP_Http::make_absolute_url($value, $canonical);
                $element->setAttribute($attr, esc_url_raw($absolute, ['http', 'https', 'mailto', 'tel']));
            }
        }
        $converter = new HtmlConverter(['strip_tags' => true, 'hard_break' => true, 'header_style' => 'atx', 'remove_nodes' => 'script style iframe']);
        $converter->getEnvironment()->addConverter(new TableConverter());
        $body = trim($converter->convert($dom->saveHTML($root)));
        if ($body === '') { return new \WP_Error('a1local_mdi_empty', 'Conversion produced no readable content.'); }
        $metadata = [
            'title' => html_entity_decode(get_the_title($post), ENT_QUOTES | ENT_HTML5, 'UTF-8'),
            'url' => $canonical,
            'updated' => get_post_modified_time('c', true, $post),
            'language' => get_bloginfo('language'),
        ];
        $markdown = "---\n";
        foreach ($metadata as $key => $value) { $markdown .= $key . ': ' . wp_json_encode($value, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE) . "\n"; }
        $markdown .= "---\n\n" . $body . "\n";
        if ($structured) {
            // Encode backticks to stop JSON-LD strings from terminating a Markdown code fence.
            $json = str_replace('`', '\\u0060', wp_json_encode(count($structured) === 1 ? $structured[0] : $structured, JSON_PRETTY_PRINT | JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE));
            $markdown .= "\n## Structured data\n\n```json\n" . $json . "\n```\n";
        }
        return ['markdown' => $markdown, 'source_bytes' => strlen($html), 'markdown_bytes' => strlen($markdown), 'container' => $selection, 'structured_count' => count($structured)];
    }
}