Example for multi-locale and multi-site Setup

[aarongerig]

How do you guys configure Dynamic Search to work with a multi-locale and multi-site setup? The page structure does look something like this:

Root
├── Site A
│   ├── de              # Search only German contents of Site A.
│   ├── fr              # Search only French contents of Site A.
│   ├── en              # Search only English contents of Site A.
│   └── ...
├── Site B
│   ├── en              # Search only English contents of Site B.
│   ├── zh              # Search only Chinese contents of Site B.
│   ├── da              # Search only Danish contents of Site B.
│   └── ...
└── ...

I'd be very much interested in how to filter contents by language and site, which is a requirement in that case (see comments in tree structure above).

My first intuition was to go for the web crawler approach, which would only crawl a certain site (via defined URL), but then I realized I have to create a context per site, which isn't ideal as well. Is there any best practice on how to solve such a use case?

Thanks for any tips!

[solverat]

There are several ways to go:

Context

Use multiple context blocks for each site. This allows you to completely configure your DS setup independently of each site. Depending on you're frontend request (/dynamic-search/context1/search?q=123 or /dynamic-search/context2/search?q=123 you're able to fetch site based results.

Field Flags

Add two fields to each of your definition (siteId and locale). So every document, you're going to index, is identifiable by those fields. In your output channel, you're then able to filter your result.

Example (ElasticSearch Provider):

$locale = $this->requestStack->getMainRequest()->getLocale();

$localeQuery = new TermQuery('locale', $locale);

$boolQuery = new BoolQuery();
$boolQuery->add($localeQuery, BoolQuery::FILTER);

// ... todo: add site query

[aarongerig]

Thank you for your answer.

Which data provider would you recommend for that? I remember that you once recommended to always use the Trinity data provider. But how can I "assign" assets or even objects to a certain site when using Trinity Data? Is there any example out there for that as well?

And last but not least, how can we apply filters on Lucene index based data? Is there an example too?

If I manage to create such a setup and make it work, I'm very happy to provide a PR with those code examples for documentation!