Duplicated titles and descriptions in search index for chunks without parent book #159

Background

In #154, we addressed the issue of missing pages in the search index. The root cause was that some index entries lacked a title property (sdesc). We resolved this by applying the same solution used for the manual content: using the description (ldesc) as the title.

To avoid duplicating text in both the title and description fields, we now pull the description from the parent <book>. For example:

In this example, the title "Type" was taken from the page description, while the new description ("Language Reference") comes from the parent <book>. You can see the implementation here:

Issue

Some entries, like extension main pages (e.g. book.strings, book.zip) and top-level pages (e.g. copyright, getting-started, security), don’t have a parent <book>. In these cases, the description is being reused as the title, resulting in duplicate content:

Proposed fix

While some entries lack a parent <book>, every entry has at least one parent <set>. The root entry itself is a set called "PHP Manual".

The proposed solution is to fall back to the first <set> in the hierarchy when no <book> is found:

I have a working implementation and will submit a PR soon.