MD Umbau

2026-05-13 11:57:52 +02:00
parent 6abf6f9c3d
commit f4511b9213
76 changed files with 4494 additions and 1940 deletions
--- a/kb-markdown-importer/includes/Markdown/MarkdownRenderer.php
+++ b/kb-markdown-importer/includes/Markdown/MarkdownRenderer.php
@@ -0,0 +1,190 @@
+<?php
+declare(strict_types=1);
+
+namespace KbMarkdownImporter\Markdown;
+
+use KbMarkdownImporter\Frontend\UrlBuilder;
+
+final class MarkdownRenderer
+{
+    public function render(string $markdown, array $context = []): string
+    {
+        $lines = preg_split('/\R/', str_replace(["\r\n", "\r"], "\n", $markdown)) ?: [];
+        $html = [];
+        $paragraph = [];
+        $listType = '';
+        $codeFence = false;
+        $code = [];
+
+        $flushParagraph = function () use (&$html, &$paragraph, $context): void {
+            if (! $paragraph) {
+                return;
+            }
+
+            $html[] = '<p>' . $this->renderInline(trim(implode(' ', $paragraph)), $context) . '</p>';
+            $paragraph = [];
+        };
+
+        $closeList = function () use (&$html, &$listType): void {
+            if ('' === $listType) {
+                return;
+            }
+
+            $html[] = '</' . $listType . '>';
+            $listType = '';
+        };
+
+        foreach ($lines as $line) {
+            if (preg_match('/^\s*```/', $line)) {
+                if ($codeFence) {
+                    $html[] = '<pre><code>' . esc_html(implode("\n", $code)) . '</code></pre>';
+                    $code = [];
+                    $codeFence = false;
+                } else {
+                    $flushParagraph();
+                    $closeList();
+                    $codeFence = true;
+                }
+                continue;
+            }
+
+            if ($codeFence) {
+                $code[] = $line;
+                continue;
+            }
+
+            if ('' === trim($line)) {
+                $flushParagraph();
+                $closeList();
+                continue;
+            }
+
+            if (preg_match('/^(#{1,6})\s+(.+)$/', $line, $matches)) {
+                $flushParagraph();
+                $closeList();
+                $level = strlen($matches[1]);
+                $text = trim($matches[2]);
+                $id = sanitize_title(wp_strip_all_tags($text));
+                $html[] = sprintf('<h%d id="%s">%s</h%d>', $level, esc_attr($id), $this->renderInline($text, $context), $level);
+                continue;
+            }
+
+            if (preg_match('/^\s*[-*]\s+(.+)$/', $line, $matches)) {
+                $flushParagraph();
+                if ('ul' !== $listType) {
+                    $closeList();
+                    $html[] = '<ul>';
+                    $listType = 'ul';
+                }
+                $html[] = '<li>' . $this->renderInline(trim($matches[1]), $context) . '</li>';
+                continue;
+            }
+
+            if (preg_match('/^\s*\d+\.\s+(.+)$/', $line, $matches)) {
+                $flushParagraph();
+                if ('ol' !== $listType) {
+                    $closeList();
+                    $html[] = '<ol>';
+                    $listType = 'ol';
+                }
+                $html[] = '<li>' . $this->renderInline(trim($matches[1]), $context) . '</li>';
+                continue;
+            }
+
+            if (preg_match('/^>\s?(.*)$/', $line, $matches)) {
+                $flushParagraph();
+                $closeList();
+                $html[] = '<blockquote><p>' . $this->renderInline(trim($matches[1]), $context) . '</p></blockquote>';
+                continue;
+            }
+
+            $paragraph[] = trim($line);
+        }
+
+        if ($codeFence) {
+            $html[] = '<pre><code>' . esc_html(implode("\n", $code)) . '</code></pre>';
+        }
+
+        $flushParagraph();
+        $closeList();
+
+        return wp_kses_post(implode("\n", $html));
+    }
+
+    private function renderInline(string $text, array $context): string
+    {
+        $escaped = esc_html($text);
+
+        $escaped = preg_replace_callback('/!\[([^\]]*)\]\(([^)]+)\)/', function (array $matches) use ($context): string {
+            $alt = html_entity_decode($matches[1], ENT_QUOTES);
+            $src = html_entity_decode($matches[2], ENT_QUOTES);
+            $url = $this->resolveImageUrl($src, (array) ($context['images'] ?? []));
+            $image = sprintf('<img src="%s" alt="%s">', esc_url($url ?: $src), esc_attr($alt));
+
+            if ($url && ! empty($context['lightbox'])) {
+                return sprintf('<a href="%s" class="kb-lightbox">%s</a>', esc_url($url), $image);
+            }
+
+            return $image;
+        }, $escaped) ?? $escaped;
+
+        $escaped = preg_replace_callback('/(?<!!)\[([^\]]+)\]\(([^)]+)\)/', function (array $matches) use ($context): string {
+            $label = $this->renderInline($matches[1], $context);
+            $href = html_entity_decode($matches[2], ENT_QUOTES);
+            $url = $this->rewriteLink($href, $context) ?: $href;
+
+            return sprintf('<a href="%s">%s</a>', esc_url($url), $label);
+        }, $escaped) ?? $escaped;
+
+        $escaped = preg_replace('/`([^`]+)`/', '<code>$1</code>', $escaped) ?? $escaped;
+        $escaped = preg_replace('/\*\*([^*]+)\*\*/', '<strong>$1</strong>', $escaped) ?? $escaped;
+        $escaped = preg_replace('/\*([^*]+)\*/', '<em>$1</em>', $escaped) ?? $escaped;
+
+        return $escaped;
+    }
+
+    private function rewriteLink(string $href, array $context): string
+    {
+        if ('' === $href || str_starts_with($href, '#') || preg_match('#^(?:https?:|mailto:|tel:)#i', $href)) {
+            return '';
+        }
+
+        $parts = wp_parse_url($href);
+        if (! is_array($parts)) {
+            return '';
+        }
+
+        $path = (string) ($parts['path'] ?? '');
+        if (! preg_match('/\.md$/i', $path)) {
+            return '';
+        }
+
+        $page = preg_replace('/\.md$/i', '', basename($path)) ?: basename($path);
+        $slug = in_array(strtolower($page), ['doku', 'index'], true) ? '' : sanitize_title($page);
+        $fragment = isset($parts['fragment']) ? '#' . sanitize_title((string) $parts['fragment']) : '';
+
+        return UrlBuilder::page((string) $context['product_slug'], (string) $context['version_slug'], $slug) . $fragment;
+    }
+
+    private function resolveImageUrl(string $src, array $images): string
+    {
+        if ('' === $src || preg_match('#^(?:https?:|data:)#i', $src)) {
+            return '';
+        }
+
+        $candidates = array_unique([
+            $src,
+            ltrim($src, '/'),
+            basename($src),
+            preg_replace('#^images/#', '', $src) ?: $src,
+        ]);
+
+        foreach ($candidates as $candidate) {
+            if (isset($images[$candidate])) {
+                return (string) $images[$candidate];
+            }
+        }
+
+        return '';
+    }
+}