Non-canonical URL being sent for product recaching

[quangdo-aligent]

For example, products can have a URL like: /occasions-and-events/anniversary/golden-anniversary/decorative-metallic-champagne-gold-placemat-chrtwpm06.html in addition to their canonical URLS (e.g. /decorative-metallic-champagne-gold-placemat-chrtwpm06.html in this case).

It seems like due to the way that the request_path product field is populated in the Product collection, it will always use the URL rewrite with the newest ID that matches the following criteria:

select * from url_rewrite where entity_id=46223 and store_id=1 and redirect_type=0 and entity_type='product' order by url_rewrite_id;
+----------------+-------------+-----------+----------------------------------------------------------------------------------------------------------------+--------------------------------------------+---------------+----------+-------------+------------------+-----------------------+
| url_rewrite_id | entity_type | entity_id | request_path                                                                                                   | target_path                                | redirect_type | store_id | description | is_autogenerated | metadata              |
+----------------+-------------+-----------+----------------------------------------------------------------------------------------------------------------+--------------------------------------------+---------------+----------+-------------+------------------+-----------------------+
|       72115604 | product     |     46223 | decorative-metallic-champagne-gold-placemat-chrtwpm06.html                                                     | catalog/product/view/id/46223              |             0 |        1 | NULL        |                1 | NULL                  |
|       72115607 | product     |     46223 | occasions-and-events/christmas/decorative-metallic-champagne-gold-placemat-chrtwpm06.html                      | catalog/product/view/id/46223/category/277 |             0 |        1 | NULL        |                1 | {"category_id":"277"} |
|       72115610 | product     |     46223 | themed-parties/gangster/decorative-metallic-champagne-gold-placemat-chrtwpm06.html                             | catalog/product/view/id/46223/category/255 |             0 |        1 | NULL        |                1 | {"category_id":"255"} |
|       72115613 | product     |     46223 | coloured-party-supplies/gold/decorative-metallic-champagne-gold-placemat-chrtwpm06.html                        | catalog/product/view/id/46223/category/318 |             0 |        1 | NULL        |                1 | {"category_id":"318"} |
|       72115616 | product     |     46223 | fringe-colours/fringe-gold/decorative-metallic-champagne-gold-placemat-chrtwpm06.html                          | catalog/product/view/id/46223/category/795 |             0 |        1 | NULL        |                1 | {"category_id":"795"} |
|       72115619 | product     |     46223 | occasions-and-events/anniversary/golden-anniversary/decorative-metallic-champagne-gold-placemat-chrtwpm06.html | catalog/product/view/id/46223/category/821 |             0 |        1 | NULL        |                1 | {"category_id":"821"} |
+----------------+-------------+-----------+----------------------------------------------------------------------------------------------------------------+--------------------------------------------+---------------+----------+-------------+------------------+-----------------------+

The relevant part of the code in \Magento\Catalog\Model\ResourceModel\Product\Collection::_addUrlRewrite:

The for loop starts from the lowest ID to the highest ID so occasions-and-events/anniversary/golden-anniversary/decorative-metallic-champagne-gold-placemat-chrtwpm06.html would be the final value set as the request_path.

Requirements:

• We need to be using the canonical URL for re-caching instead of relying on the request_path field of each product due to the way it is being generated.

I'm thinking we probably don't want to use request_path at all and rather rely on the product's URL key.

[jehanben]

The changes have been fixed and merged into the main branch.

PR: #13