>>> [!note] Migrated issue <!-- Drupal.org comment --> <!-- Migrated from issue #2930145. --> Reported by: [DeFr](https://www.drupal.org/user/59710) >>> <h3>Problem/Motivation</h3> <p>Currently, the module uses an InboundPathProcessor to listen on every incoming request, and if the request path matches one of the active facet source path, alter the path to pretend that it's in fact occuring on the facet source path. This has a lot of drawbacks, some of then fixed, some of them with pending patches, some of them not reported in the issue queue yet. Listing a few of them as examples:</p> <ul> <li><span class="drupalorg-gitlab-issue-link drupalorg-gitlab-link-wrapper"><a href="https://git.drupalcode.org/project/facets_pretty_paths/-/work_items/2862038" class="drupalorg-gitlab-link">https://git.drupalcode.org/project/facets_pretty_paths/-/work_items/2862038</a></span> because redirect thinks that all the requests are on the same path. The fix that went there disables redirect completly for those path, which is more than what's really needed and will prevent some normalization from happening when it'd be kinda nice to still keep it: trailing slash removal, alias normalization, ...</li> <li><span class="drupalorg-gitlab-issue-link drupalorg-gitlab-link-wrapper"><a href="https://git.drupalcode.org/project/facets_pretty_paths/-/work_items/2897756" class="drupalorg-gitlab-link">https://git.drupalcode.org/project/facets_pretty_paths/-/work_items/2897756</a></span> because Views tries to regenerate a path and has no data to do it correctly</li> <li><span class="drupalorg-gitlab-issue-link drupalorg-gitlab-link-wrapper"><a href="https://git.drupalcode.org/project/facets_pretty_paths/-/work_items/2779971" class="drupalorg-gitlab-link">https://git.drupalcode.org/project/facets_pretty_paths/-/work_items/2779971</a></span> because the parameters are lost</li> <li><span class="drupalorg-gitlab-issue-link drupalorg-gitlab-link-wrapper"><a href="https://git.drupalcode.org/project/facets_pretty_paths/-/work_items/2820741" class="drupalorg-gitlab-link">https://git.drupalcode.org/project/facets_pretty_paths/-/work_items/2820741</a></span> because language prefix doesn't work by default due to using "base:"</li> <li><span class="drupalorg-gitlab-issue-link drupalorg-gitlab-link-wrapper"><a href="https://git.drupalcode.org/project/facets_pretty_paths/-/work_items/2897757" class="drupalorg-gitlab-link">https://git.drupalcode.org/project/facets_pretty_paths/-/work_items/2897757</a></span> because the alias normalization loses the filters</li> <li>Not reported in the issue queue yet: language switcher links are broken (leads to the unfiltered search), links for search with facets in a menu are translated into unfiltered search links, links generated for facets that have an alias doesn't correctly use the alias, and probably quite a few more</li> </ul> <p>Those issues can probably all be fixed individually, but I took it as a symptom of a bigger problem with the module architecture, and thus would like to propose a revamp.</p> <h3>Proposed resolution</h3> <p>Instead of having an inbound path alter service, add a route subscriber, that adds an additional parameters to all the facet source routes ; make that parameter receive the whole / parts ( a bit tricky due to <span class="drupalorg-gitlab-issue-link project-issue-status-info project-issue-status-13"><a href="https://www.drupal.org/project/drupal/issues/2741939" title="Status: Needs work">#2741939: Cannot use a / in route parameter</a></span>, workaround included in the patch ) ; use that parameter in the url processor plugin instead of trying to parse the request path.</p> <p>Pros:</p> <ul> <li>All of core and contrib correctly deals with route parameters, so all of the above bugs get fixed magically.</li> <li>Performance wise, the facets source are only instantiated when the router is rebuilt, instead of every incoming request</li> <li>A lot less path processing, Symfony handles the route by itself, and hands us a properly formatted filter query string</li> </ul> <p>Cons: Until the core bug is fixed, this needs a kinda ugly workaround in the RouteSubscriber, which will imply a cap on the number of active facets.</p>