v0.3.10

Fixed

• WASM merge URL lookup now handles normalized URL formats — multi-key Map with normalized variants prevents result stub fallback; misses logged as [scolta:merge] WASM URL lookup missed.
• Title deduplication threshold lowered to 0.6 Jaccard — reduces duplicate titles slipping through, with secondary condition for short-title pairs sharing ≥3 words.
• AI Overview headings now render as HTML — #, ##, and ### markdown headings in AI summaries were falling through to <p> tags and displaying as raw # text. formatSummary() now maps them to <h3>/<h4>/<h5> elements.
• AI summary now describes post-expansion results — summarizeResults() was firing in parallel with the expansion merge, so the AI described the Phase 1 literal-keyword ranking while the displayed results showed the semantically-reordered Phase 2 ranking. Summarization is now deferred until after mergeExpandedSearchResults() completes. A searchVersion staleness check prevents summarizing results from a superseded search.
• Relative URLs from pagefind index are absolutized before use — both the summarize API call and result card <a> href attributes now prepend window.location.origin when the stored URL starts with /, so links work correctly when ContentItem stores relative paths.
• stripHtml() now decodes HTML entities — the previous regex-only implementation left entities like ’ intact, causing escapeHtml() to double-encode them and display literal entity strings in titles and excerpts. stripHtml() now uses DOM parsing to both strip tags and decode entities.
• AI summary text invisible on themes with element-level p { color } rules — themes that apply p { color: ... } or li { color: ... } at element specificity (0,0,1) overrode inherited color from .scolta-ai-summary-text. Added explicit color: var(--scolta-text) on .scolta-ai-summary-text p and .scolta-ai-summary-text li to win the specificity race.
• get_the_title() double-encoding of HTML entities — WordPress's wptexturize() converts straight quotes to HTML entities (‘ etc.), and when PagefindHtmlBuilder called htmlspecialchars() the & was re-encoded to &, rendering ‘ as literal text. get_the_title() output is now decoded with html_entity_decode(..., ENT_QUOTES | ENT_HTML5, 'UTF-8') before being passed to ContentItem.

Changed

• scolta_content_item filter now fires in both indexer pipelines — previously the filter only fired in the PHP indexer pipeline (via Scolta_Content_Gatherer). It now also fires in the binary indexer pipeline (via Scolta_Content_Source). Returning null from the filter excludes the post from indexing in both pipelines. Returning a modified ContentItem replaces the item to be indexed. This is the recommended hook for demo sites to exclude About, Contact, and other non-content pages.