From 6e5898e7663e5810afb0c8e16ec6b975dd81adf8 Mon Sep 17 00:00:00 2001
From: Cameron Cordes <cameronc.dev@gmail.com>
Date: Sun, 14 Jun 2026 02:20:06 -0400
Subject: [PATCH] Unified search: rank within filtered set instead of
 pre-thresholding CLIP
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

When structured filters are present they're the constraint and CLIP only ranks
within the candidate set, so drop the global similarity threshold for that
case. Previously the 0.2 whole-library threshold ran BEFORE intersecting with
the filters, discarding filter-matching photos that scored just under it (e.g.
a 2022 beach photo at 0.18) — producing after_struct_filter=0 even when matches
existed. Plain semantic (no filters) keeps the user's threshold.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---
 src/unified_search.rs | 28 +++++++++++++++++++++-------
 1 file changed, 21 insertions(+), 7 deletions(-)

diff --git a/src/unified_search.rs b/src/unified_search.rs
index bb6344c..9eea405 100644
--- a/src/unified_search.rs
+++ b/src/unified_search.rs
@@ -364,13 +364,27 @@ pub async fn unified_search<TagD: TagDao>(
     // ── 4. Rank ──
     match semantic {
         Some(ref sem) => {
-            // Semantic term present: CLIP-rank, then keep only hits that pass
-            // the structured filters (by content_hash).
-            let scored =
-                match score_photos(&state, &exif_dao, sem, &library_ids, threshold, None).await {
-                    Ok(s) => s,
-                    Err(e) => return score_error_response(e),
-                };
+            // When structured filters are present they ARE the constraint —
+            // CLIP only ranks within the candidate set. So drop the global
+            // similarity threshold (it's tuned for whole-library search and
+            // would pre-discard filter-matching photos that scored just under
+            // it — e.g. a 2022 beach photo at 0.18 — before the intersection
+            // ever runs). With no filters, keep the user's threshold for the
+            // plain semantic case.
+            let clip_threshold = if has_struct { -1.0 } else { threshold };
+            let scored = match score_photos(
+                &state,
+                &exif_dao,
+                sem,
+                &library_ids,
+                clip_threshold,
+                None,
+            )
+            .await
+            {
+                Ok(s) => s,
+                Err(e) => return score_error_response(e),
+            };
             let considered = scored.considered;
             let clip_hits = scored.hits.len();
             let hits: Vec<(f32, String)> = if has_struct {