ReachDigital_CategoryAttributeDeduplication

Removes duplicate store-specific category attribute values that are identical to global
(store_id=0) values.

Background

Q: Do you know of any known issue or bug in Magento 2.4.x where category attribute values are
written to store-scope (without this intentionally being done by an admin user)?

A: Yes, this is a well-known and long-standing issue in Magento 2.

The main culprit: When an admin user edits a category while viewing a specific store scope (not
"All Store Views"), Magento tends to save all attribute values to that store scope, even if:

• The values weren't actually changed
• The "Use Default Value" checkbox is checked

This happens because the category save controller/processor doesn't properly filter out unchanged
values or respect the use_default flags in certain scenarios. Over time, this leads to massive
accumulation of store-scoped rows that are identical to the global values.

Common triggers:

1. Simply opening a category in a store view and clicking Save
2. Mass actions on categories
3. Programmatic saves that don't explicitly set store_id = 0
4. Import processes that don't handle scope correctly

Related GitHub issues:

• https://github.com/magento/magento2/issues/13743

This module provides automated cleanup of these duplicate values.

Features

• Deduplicates store-specific category attribute values that match global values
• Configurable triggers: after category save (observer) and/or via cron job
• Console command for manual execution
• Dry-run mode to preview changes without deleting data
• Verbose logging option for detailed audit trail
• Configurable attribute exclusion list
• Separate log file: var/log/category_attribute_deduplication.log

Configuration

Navigate to Stores > Configuration > Reach Digital > Category Attribute Deduplication

Note: url_path is always excluded from deduplication regardless of configuration.

Console Command

The console command always runs in dry-run mode by default (ignores config setting).

# Dry-run for all categories (default)
bin/magento reach:category:deduplicate-attributes

# Actually delete duplicates
bin/magento reach:category:deduplicate-attributes --apply

# Process specific category (dry-run)
bin/magento reach:category:deduplicate-attributes --category-id=123

# Process specific category and apply changes
bin/magento reach:category:deduplicate-attributes --category-id=123 --apply

Logging

All actions are logged to var/log/category_attribute_deduplication.log.

Non-verbose mode: Logs entity IDs and attribute codes with counts.

[DRY-RUN] name: 5 duplicates in entity_ids: 10, 15, 20, 25, 30

Verbose mode: Logs detailed information about each duplicate.

[DRY-RUN] Duplicate: entity_id=10, store_id=1, attribute=name, value_id=123, ...

Installation

bin/magento module:enable ReachDigital_CategoryAttributeDeduplication
bin/magento setup:upgrade
bin/magento cache:flush

Prompt

Implement a Magento extension under app/code that can deduplicate these store-specific values for
categories.

Some requirements:
- Needs to have a logger that writes everything it does to its own separate log file. Maybe a
  verbose mode (behind config) that logs literally everything, and a shorter mode where only
  entity_ids along with attribute codes are logged?
- Should exclude url_path attribute. If you know of any other attributes that should be excluded,
  let me know.
- Needs to have a dry-run mode where it only logs what it /would/ do
- Needs to be triggered for a single category after saving (use the right event though, do we need
  the commit after event?)
- Needs to be triggered for all categories from a cronjob and console command
- Add configs to enable or disable either cronjob and observer trigger (i.e. separate configs).
  Leave both disabled by default.

Please include this prompt in the module README and keep it up to date as I provide more
instructions

---

Can you remove the exclusion of url_key? I think it should be okay to deduplicate this - if it's
the same at global and store level, then it must still remain consistent with the url_rewrite
table. If it's not, then this was already an issue, so I don't think deduplicating makes it worse
(or better).

---

Can you change the console command name so that is prefixed with reach-digital?

---

Adjust line-lengths in the README to break at 100 chars

---

Can you move the adminhtml configs to their own section and put them under the Reach Digital tab?
Don't forget to update default value paths and the constants in the code etc.

---

Can you change the console command prefix to `reach`? This is the one we most commonly use

---

Make the console command completely ignore the dry-run config, and make it use dry-run by default.
Invert the existing --dry-run option so can apply deduplication after checking with the default
dry-run mode