search

proudcity / wp-proudcity

The ProudCity WordPress platform
https://proudcity.com
Other
21 stars 9 forks source link

Data pollution issue with Elastic #2605

Open curtismchale opened 2 weeks ago

curtismchale commented 2 weeks ago

Builds

Source

Describe the bug

The events page has some events that were not created by a known user and have the date 1969. When we look into the Elastic index the post_type is not labelled correctly which means there is no date information and it seems to default to Unix epoch time.

QA

Notes:

Links:

*

kevindherman commented 2 weeks ago

Petaluma is having a similar issue with old page content rather than locations. Here's the support ticket: https://proudcity.zendesk.com/agent/tickets/7090

curtismchale commented 2 weeks ago

This has ended up being a content pollution issue with Elastic search seeing content as the incorrect post_type thus it shows as an event. @aschmoe suggested the following places as ideas for where we need to modify teaser code as follows.

Probably would need halting (if you’re in save_post or doing an autosave) in a couple places:
1. Prevent the WP-Query from happening at all? https://github.com/proudcity/wp-proud-core/blob/master/modules/proud-teasers/proud-teasers.php#L176
2. Prevent the content from being printed https://github.com/proudcity/wp-proud-core/blob/master/modules/proud-teasers/proud-teasers.php#L979
3. Maybe prevent filters from bring printed https://github.com/proudcity/wp-proud-core/blob/master/modules/proud-teasers/proud-teasers.php#L1016

I wonder why Elastic is involved at all in showing content in the Event List widget as it has no option to show content from other sites thus shouldn't have elastic involved at all.

Activating wp-proud-search-elastic makes the duplicate content come back on beta.