PDPS-2029: Index ACF custom fields into a new text-2 resource field

[ykulynych]

ACF field data (e.g. Branch pages' location, hours, contact details) was never read by the sync pipeline, which only sent post_content as text-1. Adds AcfAdapter to map every ACF field type into plain text, sent alongside post_content as a new text-2 PLAIN field on each indexed resource, so this content becomes searchable.

[StrawHat-Dery]

Overall, the core ACF indexing flow looks good and the existing WordPress content remains unaffected. I verified that standard post/page content continues to index into text-1, while supported ACF content is added to text-2. Updating a non-empty ACF value also updates the existing text-2 field as expected.

Automated checks also passed successfully:

PHP lint: PASS
PHPUnit: PASS — 120 tests / 684 assertions

During runtime testing, however, I confirmed one issue that I believe should be addressed before merge:

Should Fix/High Value — stale text-2 after ACF content is cleared

When an ACF value is removed completely, the previously indexed text-2 value remains in PAR and continues to be searchable. Re-adding a non-empty value updates text-2 again, so the issue appears specific to the empty/cleared state not removing the existing upstream field.

I also confirmed a second issue with ACF Page Link fields. A real Page Link pointing to another published WordPress page was not included in text-2, while a normal ACF Text field on the same resource continued to sync successfully. The current Page Link handling appears to expect a post ID/object even though the formatted ACF value is a URL/permalink. I’ve left an inline comment with more detail.

One additional static-review item is the Flexible Content handling, which appears likely to miss real ACF layout structures due to the layout-name/key lookup. I wasn’t able to runtime-test that path because it requires functionality outside the free ACF setup used for this review, so I would treat that as a high-value follow-up rather than a runtime-confirmed blocker.

Aside from those items, the core ACF-to-text-2 implementation behaved as expected in the scenarios tested.

[StrawHat-Dery]

Follow-up review for 34aed39.

The peer-review fixes look good, and the previously reported runtime issues have now been re-tested successfully.

Runtime validation

• Stale text-2 — Fixed. Clearing all ACF values from PAR Widget Test removed the previously indexed text-2 content.
• Page Link indexing — Fixed. Re-adding Branch Name and a real Page Link value resulted in both values appearing correctly under text-2.
• Previously non-ACF resource — Passed. Adding ACF values to Sample Page, which previously had no text-2, successfully created and populated text-2.

text-1 remained unaffected, and supported ACF values continue to be flattened into a single text-2 field as intended.

Remaining follow-ups

• Flexible Content was not runtime-tested because this environment does not have ACF Pro. The updated layout-name lookup looks correct from static review, and no concrete regression was identified.
• Alternate return-format handling for single taxonomy values, image/file/user/select variants, and Page Link archive URLs remains unchanged and can be tracked separately.

Merge readiness

No blocking or high-value issues remain from this review.

Ready to merge from my side.