Some bundle products not being indexed when sharing same child

[sambolek]

When the same child product is associated to multiple bundle products, only one of the bundle products is saved in ES. This causes bundle products to not appear in categories / search if they share the same child product.

Preconditions

Not relevant

Magento Version : Magento 2.2.1

ElasticSuite Version : 2.4.x-dev @ 55becb9

Environment : Either

Third party modules : Not relevant

Steps to reproduce

1. Create category "Test"
2. Create simple products "Cell phone", "Cell phone 2" and "Cable" that are in stock, enabled etc
3. Create one bundle product "Cell pack 1" containing "Cell phone" and "Cable" and assign it to category "Test"
4. Create another bundle product "Cell pack 2" containing "Cell phone 2" and "Cable" and assign it to category "Test"
5. Open category "Test"

Expected result

1. Both "Cell pack 1" and "Cell pack 2" products should be visible in category

Actual result

1. Only one "Cell pack" product is visible in category

The actual cause for this is in https://github.com/Smile-SA/elasticsuite/blob/2.4.x/src/module-elasticsuite-catalog/Model/ResourceModel/Product/Indexer/Fulltext/Datasource/AttributeData.php#L208 - the "group by" statement groups query by child products, meaning if one child product is assigned to multiple bundle products, all except the first connection are lost (group concat affects only "configurable_attributes" field.

Believe the query result should be grouped by parent_product_id as well?

Query before:

SELECT parent.entity_id AS parent_id, child.entity_id AS child_id, GROUP_CONCAT(DISTINCT super_table.attribute_id SEPARATOR ',') AS configurable_attributes FROM catalog_product_bundle_selection AS main
INNER JOIN catalog_product_entity AS parent ON parent.entity_id = main.parent_product_id
INNER JOIN catalog_product_entity AS child ON child.entity_id = main.product_id
LEFT JOIN catalog_product_super_attribute AS super_table ON super_table.product_id = main.parent_product_id
INNER JOIN catalog_product_website AS websites ON websites.product_id = main.product_id AND websites.website_id = '1'
WHERE (parent.entity_id in (2399, 1534)) GROUP BY main.product_id

Result: http://prntscr.com/hh0ewj

---

Query after:

SELECT parent.entity_id AS parent_id, child.entity_id AS child_id, GROUP_CONCAT(DISTINCT super_table.attribute_id SEPARATOR ',') AS configurable_attributes FROM catalog_product_bundle_selection AS main
INNER JOIN catalog_product_entity AS parent ON parent.entity_id = main.parent_product_id
INNER JOIN catalog_product_entity AS child ON child.entity_id = main.product_id
LEFT JOIN catalog_product_super_attribute AS super_table ON super_table.product_id = main.parent_product_id
INNER JOIN catalog_product_website AS websites ON websites.product_id = main.product_id AND websites.website_id = '1'
WHERE (parent.entity_id in (2399, 1534)) GROUP BY main.parent_product_id, main.product_id

Result after: http://prntscr.com/hh0fcs

---

I'll be submitting a pull request with this fix implemented in a few minutes. However, I'm not sure if this affects configurable products in some way (got no products with "configurable_attributes" filled to test with).

[sambolek]

Note that both 2399, 1534 products are bundle products in my example above.

[afoucret]

Your PR have been merged in branch 2.4.x and will be part of the next release.

Thanks for your contrib.