usher-manage-stack/app/Console/Commands/ImportHugoContent.php

<?php

namespace App\Console\Commands;

use App\Models\Article;
use App\Models\ArticleCategory;
use App\Models\ArticleTag;
use App\Models\Page;
use App\Models\User;
use Illuminate\Console\Command;
use Illuminate\Support\Facades\DB;
use Illuminate\Support\Facades\File;
use Illuminate\Support\Str;
use Symfony\Component\Yaml\Yaml;

class ImportHugoContent extends Command
{
    protected $signature = 'hugo:import
        {--path= : Path to Hugo content/zh directory}
        {--images= : Path to Hugo static/images directory}
        {--dry-run : Preview import without writing to database}';

    protected $description = 'Import Hugo site content (markdown files) into CMS articles and pages';

    private int $articlesCreated = 0;

    private int $pagesCreated = 0;

    private int $categoriesCreated = 0;

    private int $tagsCreated = 0;

    private int $imagescopied = 0;

    private int $skipped = 0;

    public function handle(): int
    {
        $contentPath = $this->option('path') ?: base_path('../usher-site/content/zh');
        $imagesPath = $this->option('images') ?: base_path('../usher-site/static/images');

        // Expand tilde
        $contentPath = str_replace('~', getenv('HOME'), $contentPath);
        $imagesPath = str_replace('~', getenv('HOME'), $imagesPath);

        if (! is_dir($contentPath)) {
            $this->error("Content directory not found: {$contentPath}");

            return self::FAILURE;
        }

        $isDryRun = $this->option('dry-run');
        if ($isDryRun) {
            $this->warn('DRY RUN — no changes will be written to database.');
        }

        $this->info("Importing Hugo content from: {$contentPath}");
        $this->newLine();

        // Get or create a system user for attribution
        $systemUser = User::where('email', 'admin@test.com')->first()
            ?? User::first();

        if (! $systemUser) {
            $this->error('No user found to attribute content to. Please create a user first.');

            return self::FAILURE;
        }

        $this->info("Attributing content to: {$systemUser->name} ({$systemUser->email})");
        $this->newLine();

        DB::beginTransaction();

        try {
            // Import articles by content type
            $contentTypes = [
                'blog' => Article::CONTENT_TYPE_BLOG,
                'notice' => Article::CONTENT_TYPE_NOTICE,
                'document' => Article::CONTENT_TYPE_DOCUMENT,
                'related_news' => Article::CONTENT_TYPE_RELATED_NEWS,
            ];

            foreach ($contentTypes as $dir => $contentType) {
                $dirPath = "{$contentPath}/{$dir}";
                if (is_dir($dirPath)) {
                    $this->importArticles($dirPath, $contentType, $systemUser, $isDryRun);
                } else {
                    $this->warn("Directory not found, skipping: {$dir}/");
                }
            }

            // Import static pages
            $pageDirectories = [
                'about' => '關於我們',
                'message' => '理事長的話',
                'structure' => '組織架構',
                'mission' => '協會任務',
                'contact' => '聯繫我們',
                'logo_represent' => 'Logo象徵',
            ];

            foreach ($pageDirectories as $dir => $fallbackTitle) {
                $dirPath = "{$contentPath}/{$dir}";
                if (is_dir($dirPath)) {
                    $this->importPage($dirPath, $dir, $fallbackTitle, $systemUser, $isDryRun);
                } else {
                    $this->warn("Page directory not found, skipping: {$dir}/");
                }
            }

            // Copy images
            if (is_dir($imagesPath)) {
                $this->copyImages($imagesPath, $isDryRun);
            } else {
                $this->warn("Images directory not found: {$imagesPath}");
            }

            if ($isDryRun) {
                DB::rollBack();
                $this->newLine();
                $this->warn('DRY RUN complete — no changes were made.');
            } else {
                DB::commit();
            }
        } catch (\Exception $e) {
            DB::rollBack();
            $this->error("Import failed: {$e->getMessage()}");
            $this->error($e->getTraceAsString());

            return self::FAILURE;
        }

        $this->newLine();
        $this->info('=== Import Summary ===');
        $this->table(
            ['Item', 'Count'],
            [
                ['Articles created', $this->articlesCreated],
                ['Pages created', $this->pagesCreated],
                ['Categories created', $this->categoriesCreated],
                ['Tags created', $this->tagsCreated],
                ['Images copied', $this->imagescopied],
                ['Skipped (_index.md etc.)', $this->skipped],
            ]
        );

        return self::SUCCESS;
    }

    private function importArticles(string $dirPath, string $contentType, User $user, bool $isDryRun): void
    {
        $this->info("Importing {$contentType} articles from: {$dirPath}");

        $files = File::glob("{$dirPath}/*.md");

        foreach ($files as $filePath) {
            $filename = basename($filePath);

            // Skip _index.md files (Hugo section pages, not content)
            if ($filename === '_index.md') {
                $this->skipped++;

                continue;
            }

            $parsed = $this->parseMarkdownFile($filePath);
            if (! $parsed) {
                $this->warn("  Could not parse: {$filename}");
                $this->skipped++;

                continue;
            }

            $frontmatter = $parsed['frontmatter'];
            $body = $parsed['body'];

            // Skip drafts
            if (! empty($frontmatter['draft']) && $frontmatter['draft'] === true) {
                $this->line("  Skipping draft: {$filename}");
                $this->skipped++;

                continue;
            }

            $title = $frontmatter['title'] ?? pathinfo($filename, PATHINFO_FILENAME);

            // Prefer filename-based slug (Hugo filenames are usually descriptive ASCII)
            $filenameBase = pathinfo($filename, PATHINFO_FILENAME);
            $slug = Str::slug($filenameBase);
            if (empty($slug)) {
                $slug = Str::slug($title);
            }
            if (empty($slug)) {
                // For Chinese-only titles/filenames, use content_type + short hash
                $slug = $contentType.'-'.substr(md5($title), 0, 8);
            }

            // Ensure unique slug
            $originalSlug = $slug;
            $counter = 1;
            while (! $isDryRun && Article::withTrashed()->where('slug', $slug)->exists()) {
                $slug = "{$originalSlug}-{$counter}";
                $counter++;
            }

            $publishedAt = $this->parseDate($frontmatter['date'] ?? null);
            $description = $frontmatter['description'] ?? null;
            $summary = $frontmatter['summary'] ?? null;
            $author = $frontmatter['author'] ?? null;
            $image = $this->resolveImagePath($frontmatter['image'] ?? null);

            $this->line("  [{$contentType}] {$title} → {$slug}");

            if (! $isDryRun) {
                $article = Article::create([
                    'title' => $title,
                    'slug' => $slug,
                    'summary' => $summary,
                    'content' => $body,
                    'content_type' => $contentType,
                    'status' => Article::STATUS_PUBLISHED,
                    'access_level' => Article::ACCESS_LEVEL_PUBLIC,
                    'featured_image_path' => $image,
                    'author_name' => $author,
                    'meta_description' => $description ?: null,
                    'published_at' => $publishedAt ?? now(),
                    'created_by_user_id' => $user->id,
                    'view_count' => 0,
                ]);

                // Sync categories
                $categoryNames = $this->extractList($frontmatter, 'categories', 'category');
                foreach ($categoryNames as $catName) {
                    $category = $this->findOrCreateCategory($catName);
                    $article->categories()->attach($category->id);
                }

                // Sync tags
                $tagNames = $this->extractList($frontmatter, 'tags');
                foreach ($tagNames as $tagName) {
                    $tag = $this->findOrCreateTag($tagName);
                    $article->tags()->attach($tag->id);
                }
            }

            $this->articlesCreated++;
        }
    }

    private function importPage(string $dirPath, string $slug, string $fallbackTitle, User $user, bool $isDryRun): void
    {
        $indexFile = "{$dirPath}/_index.md";

        if (! file_exists($indexFile)) {
            $this->warn("  No _index.md in {$slug}/, skipping.");
            $this->skipped++;

            return;
        }

        $parsed = $this->parseMarkdownFile($indexFile);
        if (! $parsed) {
            $this->warn("  Could not parse: {$slug}/_index.md");
            $this->skipped++;

            return;
        }

        $frontmatter = $parsed['frontmatter'];
        $body = $parsed['body'];

        $title = $frontmatter['title'] ?? $fallbackTitle;
        $description = $frontmatter['description'] ?? null;

        $this->info("  [page] {$title} → {$slug}");

        $parentPage = null;

        if (! $isDryRun) {
            // Check for existing page with same slug
            if (Page::withTrashed()->where('slug', $slug)->exists()) {
                $this->warn("    Page with slug '{$slug}' already exists, skipping.");
                $this->skipped++;

                return;
            }

            $parentPage = Page::create([
                'title' => $title,
                'slug' => $slug,
                'content' => $body,
                'status' => Page::STATUS_PUBLISHED,
                'meta_description' => $description ?: null,
                'published_at' => now(),
                'created_by_user_id' => $user->id,
                'sort_order' => 0,
            ]);
        }

        $this->pagesCreated++;

        // Import child pages (non-_index.md files in the same directory)
        $childFiles = File::glob("{$dirPath}/*.md");
        $sortOrder = 1;

        foreach ($childFiles as $childFile) {
            $childFilename = basename($childFile);
            if ($childFilename === '_index.md') {
                continue;
            }

            $childParsed = $this->parseMarkdownFile($childFile);
            if (! $childParsed) {
                $this->skipped++;

                continue;
            }

            $childFm = $childParsed['frontmatter'];
            $childBody = $childParsed['body'];
            $childTitle = $childFm['title'] ?? pathinfo($childFilename, PATHINFO_FILENAME);
            $childSlug = Str::slug(pathinfo($childFilename, PATHINFO_FILENAME));
            if (empty($childSlug)) {
                $childSlug = Str::slug($childTitle);
            }
            if (empty($childSlug)) {
                $childSlug = 'page-'.md5($childTitle);
            }

            $this->line("    [child] {$childTitle} → {$childSlug}");

            if (! $isDryRun) {
                $originalChildSlug = $childSlug;
                $counter = 1;
                while (Page::withTrashed()->where('slug', $childSlug)->exists()) {
                    $childSlug = "{$originalChildSlug}-{$counter}";
                    $counter++;
                }

                Page::create([
                    'title' => $childTitle,
                    'slug' => $childSlug,
                    'content' => $childBody,
                    'status' => Page::STATUS_PUBLISHED,
                    'meta_description' => $childFm['description'] ?? null,
                    'parent_id' => $parentPage->id,
                    'sort_order' => $sortOrder,
                    'published_at' => now(),
                    'created_by_user_id' => $user->id,
                ]);
            }

            $this->pagesCreated++;
            $sortOrder++;
        }
    }

    private function copyImages(string $imagesPath, bool $isDryRun): void
    {
        $this->info("Copying images from: {$imagesPath}");

        $destPath = storage_path('app/public/migrated-images');

        if (! $isDryRun && ! is_dir($destPath)) {
            File::makeDirectory($destPath, 0755, true);
        }

        $files = File::allFiles($imagesPath);

        foreach ($files as $file) {
            $relativePath = $file->getRelativePathname();
            $destFile = "{$destPath}/{$relativePath}";

            $this->line("  {$relativePath}");

            if (! $isDryRun) {
                $destDir = dirname($destFile);
                if (! is_dir($destDir)) {
                    File::makeDirectory($destDir, 0755, true);
                }
                File::copy($file->getPathname(), $destFile);
            }

            $this->imagescopied++;
        }
    }

    /**
     * Parse a markdown file into frontmatter array and body string.
     */
    private function parseMarkdownFile(string $filePath): ?array
    {
        $content = file_get_contents($filePath);
        if ($content === false) {
            return null;
        }

        // Match YAML frontmatter between --- delimiters
        if (! preg_match('/\A---\s*\n(.*?)\n---\s*\n?(.*)\z/s', $content, $matches)) {
            // No frontmatter — treat entire content as body
            return [
                'frontmatter' => [],
                'body' => trim($content),
            ];
        }

        $yamlString = $matches[1];
        $body = trim($matches[2]);

        try {
            // Strip YAML comments (lines starting with #) before parsing
            $yamlLines = array_filter(
                explode("\n", $yamlString),
                fn ($line) => ! preg_match('/^\s*#/', $line)
            );
            $cleanYaml = implode("\n", $yamlLines);
            $frontmatter = Yaml::parse($cleanYaml) ?? [];
        } catch (\Exception $e) {
            $this->warn("  YAML parse error in {$filePath}: {$e->getMessage()}");
            $frontmatter = [];
        }

        return [
            'frontmatter' => $frontmatter,
            'body' => $body,
        ];
    }

    /**
     * Parse a date value from Hugo frontmatter.
     * YAML parser may convert ISO 8601 dates to Unix timestamps (int).
     *
     * @param  string|int|null  $dateValue
     */
    private function parseDate(mixed $dateValue): ?\Carbon\Carbon
    {
        if ($dateValue === null || $dateValue === '') {
            return null;
        }

        try {
            if (is_int($dateValue) || is_float($dateValue)) {
                return \Carbon\Carbon::createFromTimestamp((int) $dateValue);
            }

            return \Carbon\Carbon::parse((string) $dateValue);
        } catch (\Exception $e) {
            $this->warn("  Could not parse date: {$dateValue}");

            return null;
        }
    }

    /**
     * Extract a list of values from frontmatter.
     * Hugo uses both plural array (categories: ["a", "b"]) and singular string (category: "a").
     */
    private function extractList(array $frontmatter, string $pluralKey, ?string $singularKey = null): array
    {
        $items = [];

        if (! empty($frontmatter[$pluralKey])) {
            $value = $frontmatter[$pluralKey];
            $items = is_array($value) ? $value : [$value];
        } elseif ($singularKey && ! empty($frontmatter[$singularKey])) {
            $value = $frontmatter[$singularKey];
            $items = is_array($value) ? $value : [$value];
        }

        // Clean up strings
        return array_filter(array_map('trim', $items));
    }

    private function findOrCreateCategory(string $name): ArticleCategory
    {
        $slug = Str::slug($name) ?: 'category-'.mt_rand(100, 999);

        $category = ArticleCategory::where('slug', $slug)->first();

        if (! $category) {
            $category = ArticleCategory::create([
                'name' => $name,
                'slug' => $slug,
                'sort_order' => 0,
            ]);
            $this->categoriesCreated++;
            $this->line("    Created category: {$name}");
        }

        return $category;
    }

    private function findOrCreateTag(string $name): ArticleTag
    {
        $slug = Str::slug($name) ?: 'tag-'.mt_rand(100, 999);

        $tag = ArticleTag::where('slug', $slug)->first();

        if (! $tag) {
            $tag = ArticleTag::create([
                'name' => $name,
                'slug' => $slug,
            ]);
            $this->tagsCreated++;
            $this->line("    Created tag: {$name}");
        }

        return $tag;
    }

    /**
     * Convert Hugo image paths to migrated storage paths.
     * Hugo: "images/blog/Update.jpg" → "migrated-images/blog/Update.jpg"
     */
    private function resolveImagePath(?string $hugoPath): ?string
    {
        if (empty($hugoPath)) {
            return null;
        }

        // Remove leading "images/" prefix
        $relativePath = preg_replace('#^images/#', '', $hugoPath);

        return "migrated-images/{$relativePath}";
    }
}