Datasets

A dataset is a catalog ready to be imported: its products and collections (products.yaml, collections.yaml), their images (media/) and a dataset.yaml describing it. A shop is created from a dataset, then its fixtures are generated and loaded as usual: several shops can share one dataset, and a dataset is built (AI calls, downloads) once.

cellar/
├── dataset.yaml       name, description, suggested theme…
├── products.yaml      image: media/<code>.webp
├── collections.yaml
└── media/             the images, in WebP

In a dataset, the image of a product or a collection is a path relative to the dataset directory: the fixtures generation copies it instead of downloading it. A URL still works, and is downloaded.

Ready-made datasets

sylius-starter/datasets ships ready-made datasets, images included, so a demo shop needs no AI call nor download:

Dataset	Catalog	Suggested theme
automotive	Eighteen car parts, tyres, tools and workshop equipment	Torque
bloom	Eighteen beauty products and tools, on six shelves	Bloom
cellar	Seventeen wines from fictional estates, in seven regions	Cellar
culture	Eighteen books, games, instruments and collectibles	Culture
depot	Eighteen trade supplies and equipment, in six aisles	Depot
groove	Sixteen fictional vinyl records, in six genres	Groove
home	Eighteen pieces of furniture, lighting, linen and tableware	Hearth
outdoor	Eighteen pieces of hiking, climbing and camping gear	Summit
pets	Eighteen foods, treats, toys, beds and care products for pets	Paws
retro-gaming	Sixteen retro consoles, games and accessories, on six shelves	Retro Gaming
roastery	Sixteen specialty coffees and brewing tools, on six shelves	Roastery
supermarket	Nine everyday groceries, one per supermarket aisle	Drive
teahouse	Sixteen fine teas and pieces of teaware, on six shelves	Teahouse
tech	Sixteen devices from fictional brands, in six departments	Tech
video-games	Sixteen fictional video games, in six genres	Arena

To add one, drop its directory in src/Datasets/datasets/ (a downloaded dataset has that layout). Every product needs a local image, a collection may have one (shown on the homepage category tiles): the package tests check them. Another package can ship its own datasets by registering its directory from an autoloaded file:

use SyliusStarter\Core\Dataset\DatasetDirectories;

DatasetDirectories::register('my-datasets', __DIR__ . '/../datasets');

Build a dataset

castor sylius:import:dataset:build --name="Tea House" --description="Loose leaf teas and tea ware"
castor sylius:import:dataset:build --source=existing --name="Tea House" --url=https://www.example-tea.com

The ai source generates the catalog with AI (--limit, 20 products at most); other sources are registered by the project, such as existing (the fetch of an existing site) in the Sales Kit. The dataset is written to .castor/import/datasets/<slug>/ (the slug of the name, or --dataset), then the image of every product and collection is downloaded or generated into its media/. Building it again replaces its catalog.

castor sylius:import:dataset:list            # the datasets, with their number of products and images
castor sylius:import:dataset:images tea-house # download or generate the images still missing
castor sylius:import:dataset:delete tea-house # built datasets only, the shops created from it are kept

A source is a callable that writes the catalog of the slug it receives, in an import context whose var dir is the datasets one (the shop import functions then write to the dataset):

use SyliusStarter\Import\Dataset\DatasetBuilders;

DatasetBuilders::register('existing', static function (string $slug, string $name, string $description, ?string $url, int $limit): void {
    fetch_import_data($url, $name, $description, $slug);
});

Download a dataset

castor sylius:import:dataset:export cellar                   # .castor/import/exports/cellar.zip
castor sylius:import:dataset:export tea-house --output=tea.zip

The zip holds a <slug>/ directory: unzipped in src/Datasets/datasets/, it becomes a ready-made dataset.

Create a shop from a dataset

castor sylius:import:shop:create cellar                                    # shop "cellar"
castor sylius:import:shop:create cellar --name="Cave du Port" --description="…"  # shop "cave-du-port"
castor sylius:import:fixtures:generate --project=cave-du-port
castor sylius:import:fixtures:load --project=cave-du-port

The dataset is copied into .castor/import/var/<shop>/, and project.yaml records the dataset it comes from (dataset: cellar). The shop then lives on its own: deleting or rebuilding the dataset leaves it unchanged.