Added migration code and pipelines to migrate Magento pages to Statamic

composer.json

@@ -32,6 +32,7 @@
     "require": {
         "php": "^8.4",
         "justbetter/statamic-glide-directive": "dev-feature/statamic-6",
+        "league/html-to-markdown": "^5.1",
         "rapidez/blade-directives": "^1.0",
         "rapidez/core": "^5.0",
         "rapidez/sitemap": "^5.0",

config/rapidez/statamic/migration.php

@@ -0,0 +1,32 @@
+<?php
+
+use Rapidez\Statamic\Pipelines\BardData\FixListItems;
+use Rapidez\Statamic\Pipelines\Html\HtmlThroughMarkdown;
+use Rapidez\Statamic\Pipelines\Html\TransformIframes;
+use Rapidez\Statamic\Pipelines\Html\TransformRapidezContent;
+use Rapidez\Statamic\Pipelines\Markdown\TransformImages;
+
+return [
+    /**
+     * These are the pipelines your migrations will run through in order to format Magento data
+     * into data Statamic can hold.
+     */
+    'pipelines' => [
+        'clean_html' => [
+            /**
+             * These pipelines are executed by the CleanHtml action **in this order** 
+             */
+            'html' => [
+                TransformRapidezContent::class,
+                TransformIframes::class,
+                HtmlThroughMarkdown::class,
+            ],
+            'markdown' => [
+                TransformImages::class,
+            ]
+        ],
+        'convert_field' => [
+            FixListItems::class
+        ]
+    ]
+];

src/Actions/CleanHtml.php

@@ -0,0 +1,23 @@
+<?php
+
+namespace Rapidez\Statamic\Actions;
+
+use Illuminate\Pipeline\Pipeline;
+
+class CleanHtml
+{
+    public function __construct(private Pipeline $pipeline)
+    {
+    }
+    /**
+     * Transform HTML to statamic compatible content
+     */
+    public function execute(string $html): string
+    {
+        return $this->pipeline
+            ->send($html)
+            ->via('handle')
+            ->through(config('rapidez.statamic.migration.pipelines.clean_html.html', []))
+            ->thenReturn();
+    }
+}

src/Actions/ConvertField.php

@@ -0,0 +1,73 @@
+<?php
+
+namespace Rapidez\Statamic\Actions;
+
+use Illuminate\Pipeline\Pipeline;
+use InvalidArgumentException;
+use Statamic\Entries\Entry;
+use Statamic\Fieldtypes\Bard;
+use Statamic\Fieldtypes\Bard\Augmentor;
+use Statamic\Fieldtypes\Replicator;
+
+class ConvertField
+{
+    public function __construct(
+        private readonly CleanHtml $cleanHtml, 
+        private readonly Pipeline $pipeline
+    ) {}
+
+    /**
+     * Transform HTML to statamic compatible content
+     */
+    public function execute(Entry $entry, string $html, ?string $fieldName = 'page_builder'): Entry
+    {
+        $fieldType = $entry->blueprint()->field($fieldName)?->fieldType();
+        if ($fieldType === null) {
+            throw new InvalidArgumentException(__('Entry with blueprint ":blueprint" does not contain field ":fieldname"', ['blueprint' => $entry->blueprint()->handle, 'fieldname' => $fieldName]));
+        }
+
+        if (!$fieldType instanceof Replicator || (!$fieldType instanceof Bard && !$fieldType->flattenedSetsConfig()->hasAny('content'))) {
+            throw new InvalidArgumentException(__('":fieldname" on Blueprint ":blueprint" is not a Bard field or is not a replicator containing a Bard field', ['blueprint' => $entry->blueprint()->handle, 'fieldname' => $fieldName]));
+        }
+
+        $isReplicator = false;
+        if (!($fieldType instanceof Bard) && $fieldType instanceof Replicator) {
+            $isReplicator = true;
+        }
+
+        $html = $this->cleanHtml->execute($html);
+        if (trim($html) === '') {
+            return $entry;
+        }
+
+        $augmentor = new Augmentor($fieldType instanceof Bard ? $fieldType : resolve(Bard::class));
+        $data = $augmentor->renderHtmlToProsemirror($html);
+        $content = $this->pipeline
+            ->send($data['content'])
+            ->via('handle')
+            ->through(config('rapidez.statamic.migration.pipelines.convert_field', []))
+            ->thenReturn();
+
+        $currentContent = $isReplicator ? data_get($entry->get($fieldName), '0.content.content') : $entry->get($fieldName);
+        if (
+            $currentContent
+            // We must check by rendering to html since the array structure changes, but the html output does not.
+            && $augmentor->renderProsemirrorToHtml(['content' => $currentContent]) == $augmentor->renderProsemirrorToHtml(['content' => $content])
+        ) {
+            return $entry;
+        }
+
+        if ($isReplicator) {
+            // TODO: Currently we only support a fieldset named "content", containing a field called "content" which is a Bard field.
+            // Ideally we should determine the needed path by nesting fields in the Replicator until we find a Bard field.
+            $content = [[
+                'type' => 'content',
+                'content' => ['content' => $content]
+            ]];
+        }
+
+        $entry->set($fieldName, $content);
+
+        return $entry;
+    }
+}

src/Commands/MigrateCmsPages.php

@@ -0,0 +1,134 @@
+<?php
+
+namespace Rapidez\Statamic\Commands;
+
+use Statamic\Facades\Site;
+use Illuminate\Console\Command;
+use Illuminate\Support\Collection;
+use Illuminate\Support\Facades\DB;
+use Rapidez\Core\Facades\Rapidez;
+use Rapidez\Core\Models\Page;
+use Rapidez\Statamic\Actions\ConvertField;
+use Statamic\Eloquent\Entries\EntryQueryBuilder;
+use Statamic\Facades\Entry;
+
+class MigrateCmsPages extends Command
+{
+    /**
+     * The name and signature of the console command.
+     *
+     * @var string
+     */
+    protected $signature = 'statamic-content-migration:migrate-cms-pages {--identifiers=} {--identifier-type=whitelist : Type of identifier, whitelist/blacklist}';
+
+    /**
+     * The console command description.
+     *
+     * @var string
+     */
+    protected $description = 'Migrate magento CMS pages into the "pages" collection';
+
+    private ConvertField $converter;
+
+    /** @var Collection<int,\Statamic\Sites\Site> $sitePerStoreId */
+    private Collection $sitePerStoreId;
+
+    /**
+     * Execute the console command.
+     */
+    public function handle(ConvertField $converter): void
+    {
+        $this->converter = $converter;
+
+        $this->sitePerStoreId = Site::all()
+            ->filter(fn(\Statamic\Sites\Site $site) => $site->attribute('magento_store_id'))
+            ->mapWithKeys(fn(\Statamic\Sites\Site $site, $key): array => [$site->attribute('magento_store_id') => $site]);
+
+        foreach ($this->sitePerStoreId as $storeId => $site) {
+            Rapidez::setStore($storeId);
+
+            $this->createCmsPages($site);
+        }
+
+        $this->output->writeln(__('Don\'t forget to disable the pages in Magento or remove the cms page fallback route! ":line"', ['line' => 'Rapidez::removeFallbackRoute(CmsPageController::class);']));
+    }
+
+    public function createCmsPages(\Statamic\Sites\Site $site): void
+    {
+        /** @var Page $cmsPageModel */
+        $cmsPageModel = config('rapidez.models.page');
+        $pagesQuery = $cmsPageModel::query();
+
+        if ($this->option('identifiers')) {
+            $identifiers = is_array($this->option('identifiers')) ? $this->option('identifiers') : explode(',', $this->option('identifiers'));
+            $pagesQuery->when($this->option('identifier-type') === 'blacklist', fn($q) => $q->whereNotIn('identifier', $identifiers));
+            $pagesQuery->when($this->option('identifier-type') !== 'blacklist', fn($q) => $q->whereIn('identifier', $identifiers));
+        }
+
+        $this->output->progressStart($pagesQuery->count());
+
+        foreach ($pagesQuery->lazy() as $page) {
+            $this->output->progressAdvance();
+
+            /** @var EntryQueryBuilder $query */
+            $query = Entry::query();
+            if ($query
+                ->where('collection', 'pages')
+                ->where('slug', $page->identifier)
+                ->where('site', $site->handle)
+                ->exists()
+            ) {
+                continue;
+            }
+
+            $this->createCmsPage($page, $site);
+        }
+
+        $this->output->progressFinish();
+    }
+
+    private function createCmsPage($page, \Statamic\Sites\Site $site): void
+    {
+        if (trim((string) $page->content) === '') {
+            return;
+        }
+
+        $pageStores = DB::table('cms_page_store')->where('page_id', $page->page_id)->pluck('store_id');
+
+        /** @var EntryQueryBuilder $query */
+        $query = Entry::query();
+        $query
+            ->where('collection', 'pages')
+            ->where('slug', $page->identifier);
+        if ($page->store_id !== 0) {
+            $query->whereIn('site', $this->sitePerStoreId->filter(fn(\Statamic\Sites\Site $aSite): bool => in_array($aSite->attribute('magento_store_id'), $pageStores->toArray()))->map(fn(\Statamic\Sites\Site $aSite) => $aSite->handle()));
+        }
+
+        /** @var ?\Statamic\Entries\Entry $originEntry */
+        $originEntry = $query->first();
+
+        if ($originEntry !== null) {
+            /** @var \Statamic\Entries\Entry $entry */
+            $entry = $originEntry->makeLocalization($site->handle);
+        } else {
+            /** @var \Statamic\Entries\Entry $entry */
+            $entry = Entry::make();
+            $entry
+                ->collection('pages')
+                ->locale($site)
+                ->published(1)
+                ->slug($page->identifier);
+        }
+
+        $entry->title = Rapidez::content($page->title); // @phpstan-ignore property.notFound
+
+        $this->converter->execute($entry, $page->content, 'page_builder');
+
+        $entry->seo = [ // @phpstan-ignore property.notFound
+            'title' => Rapidez::content($page->meta_title),
+            'description' => Rapidez::content($page->meta_description),
+        ];
+
+        $entry->save();
+    }
+}

src/Pipelines/BardData/FixListItems.php

@@ -0,0 +1,57 @@
+<?php
+
+namespace Rapidez\Statamic\Pipelines\BardData;
+
+use Closure;
+
+class FixListItems
+{
+    private const array BLOCK_NODE_TYPES = [
+        'paragraph', 'heading', 'bulletList', 'orderedList',
+        'codeBlock', 'blockquote', 'horizontalRule',
+    ];
+
+    public function handle(array $data, Closure $next): array
+    {
+        return $next($this->fixListItems($data));
+    }
+
+    /**
+     * Recursively walk any value (array, nested arrays) and fix listItem nodes
+     * whose content consists entirely of inline nodes (e.g. bare "text" nodes)
+     * by wrapping them in a paragraph node, as Statamic's Bard editor requires.
+     */
+    public function fixListItems(mixed $data): mixed
+    {
+        if (!is_array($data)) {
+            return $data;
+        }
+
+        if (($data['type'] ?? null) === 'listItem' && isset($data['content'])) {
+            $data['content'] = $this->wrapInlineContentInParagraph($data['content']);
+        }
+
+        foreach ($data as $key => $value) {
+            $data[$key] = $this->fixListItems($value);
+        }
+
+        return $data;
+    }
+
+    /**
+     * @param array<int, mixed> $content
+     * @return array<int, mixed>
+     */
+    private function wrapInlineContentInParagraph(array $content): array
+    {
+        $allInline = collect($content)->every(
+            fn (mixed $node): bool => !in_array($node['type'] ?? null, self::BLOCK_NODE_TYPES, true)
+        );
+
+        if (!$allInline) {
+            return $content;
+        }
+
+        return [['type' => 'paragraph', 'content' => $content]];
+    }
+}

src/Pipelines/Html/HtmlThroughMarkdown.php

@@ -0,0 +1,32 @@
+<?php
+
+namespace Rapidez\Statamic\Pipelines\Html;
+
+use Closure;
+use Illuminate\Pipeline\Pipeline;
+use Illuminate\Support\Str;
+use League\HTMLToMarkdown\Converter\TableConverter;
+use League\HTMLToMarkdown\HtmlConverter;
+
+class HtmlThroughMarkdown
+{
+    private readonly ?HtmlConverter $converter;
+
+    public function __construct(private Pipeline $pipeline)
+    {
+        $this->converter = new HtmlConverter(['strip_tags' => true]);
+        $this->converter->getEnvironment()->addConverter(new TableConverter());
+    }
+
+    public function handle(string $html, Closure $next): string
+    {
+        $markdown = $this->converter->convert($html);
+        $markdown = $this->pipeline
+            ->send($markdown)
+            ->via('handle')
+            ->through(config('rapidez.statamic.migration.pipelines.clean_html.markdown', []))
+            ->thenReturn();
+
+        return $next(Str::markdown($markdown));
+    }
+}

src/Pipelines/Html/TransformIframes.php

@@ -0,0 +1,27 @@
+<?php
+
+namespace Rapidez\Statamic\Pipelines\Html;
+
+use Closure;
+
+class TransformIframes
+{
+    public function handle(string $html, Closure $next): string
+    {
+        return $next(preg_replace_callback('#<iframe[^>]*src="(https?://[^"]+)"[^>]*>(?:</iframe>)?#', $this->transformIframe(...), $html));
+    }
+
+    /**
+     * @param array<int, string> $matches
+     */
+    private function transformIframe(array $matches): string
+    {
+        $value = (string) $matches[1];
+
+        if (str_contains($value, 'youtube.com/embed/')) {
+            $value = str_replace('youtube.com/embed/', 'youtube.com/watch?v=', $value);
+        }
+
+        return '<a href="'.$value.'">'.$value.'</a>';
+    }
+}