<?php
if (!defined('ABSPATH')) { exit; }

/** Deliberate, deterministic HTML conversion. Does not execute scripts or shortcodes. */
final class Zire_AI_Markdown {
    private string $base;
    public function __construct(string $base) { $this->base = $base; }

    public function extract(string $html, string $selector = '', array $exclude_selectors = []): array {
        $this->block_links = 0;
        $this->tables = 0;
        $this->disclosures = 0;
        $this->contexts = 0;
        $this->decorative_headings = 0;
        $this->decorative_symbols = 0;
        if (!class_exists('DOMDocument')) { throw new RuntimeException('PHP DOM/XML extension is required. Ask your host to enable it.'); }
        $previous = libxml_use_internal_errors(true);
        $dom = new DOMDocument();
        $dom->loadHTML('<?xml encoding="UTF-8">' . $html, LIBXML_NONET | LIBXML_NOERROR | LIBXML_NOWARNING);
        libxml_clear_errors(); libxml_use_internal_errors($previous);
        $xp = new DOMXPath($dom);
        $seo_title = '';
        $title_node = $dom->getElementsByTagName('title')->item(0);
        if ($title_node) {
            $seo_title = trim((string) preg_replace('/\s+/u', ' ', $title_node->textContent));
        }
        $seo_description = '';
        $robots = [];
        foreach ($xp->query('//meta[@name]') as $meta) {
            $meta_name = strtolower($meta->getAttribute('name'));
            if (in_array($meta_name, ['robots','googlebot','bingbot'], true)) {
                $robots[] = strtolower($meta->getAttribute('content'));
            }
            if ($seo_description === '' && $meta_name === 'description') {
                $seo_description = trim((string) preg_replace('/\s+/u', ' ', $meta->getAttribute('content')));
            }
        }
        if ($seo_title === '') {
            $og_title = $xp->query('//meta[translate(@property,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")="og:title"]')->item(0);
            if ($og_title) { $seo_title = trim((string) preg_replace('/\s+/u', ' ', $og_title->getAttribute('content'))); }
        }
        if ($seo_description === '') {
            $og_description = $xp->query('//meta[translate(@property,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")="og:description"]')->item(0);
            if ($og_description) { $seo_description = trim((string) preg_replace('/\s+/u', ' ', $og_description->getAttribute('content'))); }
        }
        if (preg_match('/\b(noindex|none)\b/', implode(' ', $robots))) {
            throw new RuntimeException('Source HTML declares noindex. Publishing is blocked.');
        }
        $canonical = '';
        foreach ($xp->query('//link[@href]') as $link) {
            if (preg_match('/\bcanonical\b/i', $link->getAttribute('rel'))) { $canonical = $this->url($link->getAttribute('href')); break; }
        }
        $root = null;
        if ($selector !== '') {
            // A single tag, #id or .class only: no arbitrary XPath execution.
            if (!preg_match('/^(?:[a-z][a-z0-9-]*|[.#][a-zA-Z_][a-zA-Z0-9_-]*)$/', $selector)) {
                throw new RuntimeException('Use one selector: main, article, #content or .entry-content.');
            }
            $query = $selector[0] === '#' ? '//*[@id="' . substr($selector, 1) . '"]' : ($selector[0] === '.' ? '//*[contains(concat(" ",normalize-space(@class)," ")," ' . substr($selector, 1) . ' ")]' : '//' . $selector);
            $root = $xp->query($query)->item(0);
            if (!$root) { throw new RuntimeException('Content selector did not match. Nothing has been published.'); }
        } else {
            foreach (['//main','//*[@role="main"]','//article','//*[contains(concat(" ",normalize-space(@class)," ")," entry-content ")]'] as $query) {
                $root = $xp->query($query)->item(0);
                if ($root) { break; }
            }
            if (!$root) { throw new RuntimeException('No main content container found. Set a content selector; the plugin will not export the whole page blindly.'); }
        }
        if (count($exclude_selectors) > 30) { throw new RuntimeException('Use no more than 30 exclusion selectors.'); }
        // Headers inside articles may contain real title/byline content and are retained.
        $remove = [];
        foreach ($xp->query('.//script|.//style|.//nav|.//footer|.//form|.//iframe|.//svg|.//noscript|.//template|.//button|.//*[@hidden]|.//*[@aria-hidden="true"]|.//*[@data-aeohelper-exclude]|.//*[@data-zire-ai-exclude]', $root) as $node) { $remove[] = $node; }
        foreach ($exclude_selectors as $exclude_selector) {
            $exclude_selector = trim((string) $exclude_selector);
            if (!preg_match('/^[.#][a-zA-Z_][a-zA-Z0-9_-]*$/', $exclude_selector)) {
                throw new RuntimeException('Exclusion selectors must be one simple .class or #id per line.');
            }
            $query = $exclude_selector[0] === '#'
                ? './/*[@id="' . substr($exclude_selector, 1) . '"]'
                : './/*[contains(concat(" ",normalize-space(@class)," ")," ' . substr($exclude_selector, 1) . ' ")]';
            foreach ($xp->query($query, $root) as $node) { $remove[] = $node; }
        }
        foreach ($xp->query('.//*[@style]', $root) as $node) {
            if (preg_match('/(?:display\s*:\s*none|visibility\s*:\s*hidden)/i', $node->getAttribute('style'))) { $remove[] = $node; }
        }
        foreach ($remove as $node) { if ($node->parentNode) { $node->parentNode->removeChild($node); } }
        $duplicate_h1_removed = $this->deduplicate_h1($root);
        $headings = [];
        foreach ($xp->query('.//h1|.//h2|.//h3|.//h4|.//h5|.//h6', $root) as $heading) {
            $heading_text = trim((string) preg_replace('/\s+/u', ' ', $heading->textContent));
            if ($this->heading_key($heading_text) !== '') {
                $headings[] = strtoupper($heading->nodeName) . ': ' . $heading_text;
            }
        }
        $markdown = trim(preg_replace('/\n{3,}/', "\n\n", $this->node($root)));
        if (strlen($markdown) < 80) { throw new RuntimeException('Extracted content is too short. Check the selector or whether content requires JavaScript.'); }
        return ['markdown' => $markdown, 'headings' => $headings, 'canonical' => $canonical, 'seo_title' => $seo_title, 'seo_description' => $seo_description, 'bytes' => strlen($markdown), 'words' => count(preg_split('/\s+/u', $markdown, -1, PREG_SPLIT_NO_EMPTY)), 'selector' => $selector ?: $root->nodeName, 'exclude_selectors' => array_values($exclude_selectors), 'diagnostics' => ['duplicate_h1_removed'=>$duplicate_h1_removed,'decorative_headings_removed'=>$this->decorative_headings,'decorative_symbols_removed'=>$this->decorative_symbols,'block_links_rewritten'=>$this->block_links,'tables'=>$this->tables,'disclosures'=>$this->disclosures,'contexts'=>$this->contexts]];
    }
    private int $block_links = 0;
    private int $tables = 0;
    private int $disclosures = 0;
    private int $contexts = 0;
    private int $decorative_headings = 0;
    private int $decorative_symbols = 0;
    private function children(DOMNode $node): string {
        $out = ''; foreach ($node->childNodes as $child) { $out .= $this->node($child); } return $out;
    }
    private function node(DOMNode $node): string {
        if ($node instanceof DOMText) {
            return $this->escape_text($node->textContent);
        }
        if (!($node instanceof DOMElement)) { return ''; }
        $tag = strtolower($node->tagName);
        $plain_text = trim((string) preg_replace('/\s+/u', ' ', $node->textContent));
        if ($plain_text === '#' && !in_array($tag, ['pre','code'], true)) {
            $this->decorative_symbols++;
            return '';
        }
        $context_prefix = $this->context_prefix($node);
        $result = '';
        if ($tag === 'pre') {
            $code = trim($node->textContent, "\r\n");
            preg_match_all('/`+/', $code, $runs);
            $length = 3; foreach ($runs[0] as $run) { $length = max($length, strlen($run) + 1); }
            $fence = str_repeat('`', $length);
            $result = "\n\n" . $fence . "\n" . $code . "\n" . $fence . "\n\n";
            return $this->with_context($context_prefix, $result);
        }
        if ($node->hasAttribute('data-aeohelper-table') || $node->hasAttribute('data-zire-ai-table')) {
            $this->tables++;
            return $this->with_context($context_prefix, $this->structured_table($node));
        }

        if ($tag === 'table') {
            $this->tables++;
            return $this->with_context($context_prefix, $this->table($node));
        }
        if ($tag === 'details') {
            $this->disclosures++;
            return $this->with_context($context_prefix, $this->details($node));
        }
        if ($tag === 'ul' || $tag === 'ol') {
            $out = "\n\n"; $i = $node->hasAttribute('start') ? (int) $node->getAttribute('start') : 1;
            foreach ($node->childNodes as $li) {
                if (!($li instanceof DOMElement) || $li->tagName !== 'li') { continue; }
                $list_prefix = $tag === 'ol' ? ($i++) . '. ' : '- ';
                $text = trim($this->children($li));
                $out .= $list_prefix . str_replace("\n", "\n" . str_repeat(' ', strlen($list_prefix)), $text) . "\n";
            }
            return $this->with_context($context_prefix, $out . "\n");
        }
        $text = $this->children($node);
        if (preg_match('/^h([1-6])$/', $tag, $m)) {
            $heading = trim((string) preg_replace('/\s+/u', ' ', $text));
            if ($heading === '' || !preg_match('/[\p{L}\p{N}]/u', $heading)) {
                $this->decorative_headings++;
                return '';
            }
            $result = "\n\n" . str_repeat('#', (int) $m[1]) . ' ' . $heading . "\n\n";
        } elseif ($tag === 'a') {
            $result = $this->anchor($node, $text);
        } elseif ($tag === 'img') {
            $alt = trim($node->getAttribute('alt'));
            $url = $this->url($node->getAttribute('src'));
            $result = $alt && $url ? '![' . str_replace([']','['], '', $alt) . '](' . str_replace([' ',')','('], ['%20','%29','%28'], $url) . ')' : '';
        } elseif ($tag === 'strong' || $tag === 'b') {
            $text = trim($text);
            $result = $text === '' ? '' : ($this->has_ancestor_tag($node, ['strong','b']) ? $text : '**' . $text . '**');
        } elseif ($tag === 'em' || $tag === 'i') {
            $text = trim($text);
            $result = $text === '' ? '' : ($this->has_ancestor_tag($node, ['em','i']) ? $text : '*' . $text . '*');
        } elseif ($tag === 'code') {
            $result = '`` ' . str_replace('``', '', $node->textContent) . ' ``';
        } elseif ($tag === 'br') {
            $result = "\n";
        } elseif ($tag === 'hr') {
            $result = "\n\n---\n\n";
        } elseif ($tag === 'blockquote') {
            $result = "\n\n> " . str_replace("\n", "\n> ", trim($text)) . "\n\n";
        } elseif ($tag === 'summary') {
            $result = trim($this->summary_text($node));
        } elseif (in_array($tag, ['p','div','section','article','main','header','aside','figure','figcaption','dt','dd'], true)) {
            $result = "\n\n" . trim($text) . "\n\n";
        } else {
            $result = $text;
        }
        return $this->with_context($context_prefix, $result);
    }

    private function escape_text(string $text): string {
        $text = (string) preg_replace('/\s+/u', ' ', $text);
        return (string) preg_replace('/([\\\\`*\[\]<>])/', '\\\\$1', $text);
    }

    private function context_prefix(DOMElement $node): string {
        if (!$node->hasAttribute('data-aeohelper-context') && !$node->hasAttribute('data-zire-ai-context')) { return ''; }
        $attribute = $node->hasAttribute('data-aeohelper-context') ? 'data-aeohelper-context' : 'data-zire-ai-context';
        $context = trim((string) preg_replace('/\s+/u', ' ', $node->getAttribute($attribute)));
        if ($context === '') { return ''; }
        return "\n\n**" . trim($this->escape_text($context)) . "**\n\n";
    }

    private function with_context(string $context, string $content): string {
        if (trim($content) === '') { return ''; }
        if ($context !== '') { $this->contexts++; }
        return $context . $content;
    }

    private function has_ancestor_tag(DOMElement $node, array $tags): bool {
        $parent = $node->parentNode;
        while ($parent instanceof DOMElement) {
            if (in_array(strtolower($parent->tagName), $tags, true)) { return true; }
            $parent = $parent->parentNode;
        }
        return false;
    }
    private function anchor(DOMElement $anchor, string $text): string {
        $url = $this->url($anchor->getAttribute('href'));
        if (!$url || trim($text) === '') { return $text; }
        $target = str_replace([' ',')','('], ['%20','%29','%28'], $url);
        foreach (['h1','h2','h3','h4','h5','h6','p','div','section','article','ul','ol','table','details'] as $block_tag) {
            if ($anchor->getElementsByTagName($block_tag)->length < 1) { continue; }
            $this->block_links++;
            $label = '';
            foreach (['h1','h2','h3','h4','h5','h6'] as $heading_tag) {
                $heading = $anchor->getElementsByTagName($heading_tag)->item(0);
                if ($heading) { $label = trim(preg_replace('/\s+/u', ' ', $heading->textContent)); break; }
            }
            if ($label === '') { $label = 'View linked page'; }
            $label = trim(str_replace(['[',']'], '', $label));
            return "\n\n" . trim($text) . "\n\n[" . $label . " →](" . $target . ")\n\n";
        }
        return '[' . trim($text) . '](' . $target . ')';
    }

    private function deduplicate_h1(DOMNode $root): int {
        $headings = [];
        if ($root instanceof DOMElement && strtolower($root->tagName) === 'h1') { $headings[] = $root; }
        if ($root instanceof DOMElement) {
            foreach ($root->getElementsByTagName('h1') as $heading) { $headings[] = $heading; }
        }
        $keep = []; $remove = [];
        for ($i = count($headings) - 1; $i >= 0; $i--) {
            $key = $this->heading_key($headings[$i]->textContent);
            if ($key === '' || !isset($keep[$key])) { $keep[$key] = true; continue; }
            $remove[] = $headings[$i];
        }
        foreach ($remove as $heading) { if ($heading->parentNode) { $heading->parentNode->removeChild($heading); } }
        return count($remove);
    }

    private function heading_key(string $text): string {
        $text = function_exists('mb_strtolower') ? mb_strtolower($text, 'UTF-8') : strtolower($text);
        return trim((string) preg_replace('/[^\p{L}\p{N}]+/u', '', $text));
    }
    private function details(DOMElement $details): string {
        $summary = null;
        foreach ($details->childNodes as $child) {
            if ($child instanceof DOMElement && strtolower($child->tagName) === 'summary') { $summary = $child; break; }
        }
        if (!$summary) { return "\n\n" . trim($this->children($details)) . "\n\n"; }
        $question = trim($this->summary_text($summary));
        $answer = '';
        foreach ($details->childNodes as $child) {
            if ($child === $summary) { continue; }
            $answer .= $this->node($child);
        }
        $answer = trim($answer);
        if ($question === '') { return $answer === '' ? '' : "\n\n" . $answer . "\n\n"; }
        return "\n\n### " . $question . ($answer === '' ? '' : "\n\n" . $answer) . "\n\n";
    }
    private function summary_text(DOMNode $summary): string {
        $out = '';
        foreach ($summary->childNodes as $child) {
            if ($child instanceof DOMText) { $out .= $this->node($child); continue; }
            if (!($child instanceof DOMElement)) { continue; }
            // Themes commonly add a span containing only a +/minus/chevron as the visual toggle.
            $plain = trim(preg_replace('/\s+/u', ' ', $child->textContent));
            if ($plain !== '' && preg_match('/^[+\x{2212}\x{2013}\x{2014}\x{203A}\x{2039}\x{2303}\x{2304}\x{25B2}\x{25BC}\x{276F}\x{2191}\x{2193}]$/u', $plain)) { continue; }
            $out .= $this->node($child);
        }
        return trim($out);
    }
    private function table(DOMElement $table): string {
        $rows = []; $width = 0;
        foreach ($table->getElementsByTagName('tr') as $row) {
            $cells = [];
            foreach ($row->childNodes as $cell) {
                if ($cell instanceof DOMElement && in_array($cell->tagName, ['th','td'], true)) {
                    $cells[] = str_replace(['|',"\n"], ['\\|',' '], trim($this->children($cell)));
                }
            }
            if ($cells) { $rows[] = $cells; $width = max($width, count($cells)); }
        }
        if (!$rows) { return ''; }
        $out = "\n\n";
        foreach ($rows as $i => $cells) {
            $out .= '| ' . implode(' | ', array_pad($cells, $width, '')) . " |\n";
            if ($i === 0) { $out .= '| ' . implode(' | ', array_fill(0, $width, '---')) . " |\n"; }
        }
        return $out . "\n\n";
    }
	    /**
     * Convert an explicitly marked visual grid into a Markdown table.
     *
     * The first direct child is treated as the header row.
     * Each subsequent direct child is treated as one table row.
     * Direct child elements inside each row are treated as cells.
     * A row containing only text becomes a labelled group row.
     *
     * Usage:
     * <div data-aeohelper-table>
     *     <div>
     *         <div>Feature</div>
     *         <div>Plan one</div>
     *         <div>Plan two</div>
     *     </div>
     *     <div>
     *         <div>Projects</div>
     *         <div>1</div>
     *         <div>3</div>
     *     </div>
     * </div>
     */
    private function structured_table(DOMElement $table): string {
        $rows  = [];
        $width = 0;

        foreach ($table->childNodes as $row) {
            if (!($row instanceof DOMElement)) {
                continue;
            }

            $cells = [];

            foreach ($row->childNodes as $cell) {
                if (!($cell instanceof DOMElement)) {
                    continue;
                }

                $value = trim(
                    preg_replace(
                        '/\s+/u',
                        ' ',
                        $this->children($cell)
                    )
                );

                $value = str_replace('|', '\\|', $value);

                if (preg_match('/^[\x{2713}\x{2714}]$/u', $value)) {
                    $value = 'Included';
                } elseif (preg_match('/^[\x{2013}\x{2014}-]$/u', $value)) {
                    $value = 'Not included';
                }

                $cells[] = $value;
            }

            /*
             * A direct child containing text but no child elements
             * is treated as a group label.
             */
            if (!$cells) {
                $label = trim(
                    preg_replace(
                        '/\s+/u',
                        ' ',
                        $this->children($row)
                    )
                );

                if ($label === '') {
                    continue;
                }

                $cells[] = '**'
                    . str_replace('|', '\\|', $label)
                    . '**';
            }

            $rows[] = $cells;
            $width  = max($width, count($cells));
        }

        /*
         * If the marked element is not shaped like a table,
         * preserve its ordinary content rather than losing it.
         */
        if (!$rows || $width < 2) {
            return $this->children($table);
        }

        $out = "\n\n";

        foreach ($rows as $index => $cells) {
            $out .= '| '
                . implode(
                    ' | ',
                    array_pad($cells, $width, '')
                )
                . " |\n";

            if ($index === 0) {
                $out .= '| '
                    . implode(
                        ' | ',
                        array_fill(0, $width, '---')
                    )
                    . " |\n";
            }
        }

        return $out . "\n\n";
    }

    private function url(string $url): string {
        $url = trim($url);
        if (!$url || preg_match('/^(javascript|data|vbscript):/i', $url)) { return ''; }
        if (preg_match('/^https?:\/\//i', $url) || preg_match('/^(mailto|tel):/i', $url)) { return $url; }
        if (preg_match('/^[a-z][a-z0-9+.-]*:/i', $url)) { return ''; }
        return WP_Http::make_absolute_url($url, $this->base);
    }
}