From cd7bb3b1e0c35d22a6e18db69f3702a3362c1b77 Mon Sep 17 00:00:00 2001
From: Sven Geboers <geboers.sven@gmail.com>
Date: Sun, 22 Mar 2026 22:54:54 +0100
Subject: [PATCH] fix(explorer): filter to uniform-dim windows before PCA to
 prevent np.vstack shape mismatch

---
 explorer.py | 44 +++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 43 insertions(+), 1 deletion(-)

diff --git a/explorer.py b/explorer.py
index 9e709a3..38255dd 100644
--- a/explorer.py
+++ b/explorer.py
@@ -72,6 +72,46 @@ def get_available_windows(db_path: str) -> List[str]:
         con.close()
 
 
+@st.cache_data(show_spinner=False)
+def get_uniform_dim_windows(db_path: str) -> List[str]:
+    """Return only windows whose vector dimension equals the most common dimension.
+
+    np.vstack requires all vectors to have the same shape. Early or small windows
+    have lower SVD rank (dim < 50). This helper filters to only windows at the
+    dominant (max-count) dimension so compute_2d_axes never sees mixed shapes.
+    """
+    con = duckdb.connect(database=db_path, read_only=True)
+    try:
+        rows = con.execute(
+            """
+            WITH window_dims AS (
+                SELECT DISTINCT ON (window_id)
+                    window_id,
+                    json_array_length(vector) AS dim
+                FROM svd_vectors
+                WHERE entity_type = 'mp'
+                ORDER BY window_id
+            ),
+            dim_counts AS (
+                SELECT dim, COUNT(*) AS cnt FROM window_dims GROUP BY dim
+            ),
+            dominant AS (
+                SELECT dim FROM dim_counts ORDER BY cnt DESC, dim DESC LIMIT 1
+            )
+            SELECT wd.window_id
+            FROM window_dims wd
+            JOIN dominant d ON wd.dim = d.dim
+            ORDER BY wd.window_id
+            """
+        ).fetchall()
+        return [r[0] for r in rows]
+    except Exception:
+        logger.exception("Failed to query uniform-dim windows")
+        return []
+    finally:
+        con.close()
+
+
 @st.cache_data(show_spinner="2D posities berekenen (kan even duren)…")
 def load_positions(
     db_path: str, window_size: str = "quarterly"
@@ -84,7 +124,9 @@ def load_positions(
     """
     from analysis.political_axis import compute_2d_axes
 
-    available = get_available_windows(db_path)
+    # Only use windows where all vectors share the same dimension (dim=50).
+    # Mixed-dim windows cause np.vstack to fail in compute_2d_axes.
+    available = get_uniform_dim_windows(db_path)
     if window_size == "annual":
         # Keep only Q4 windows (one representative window per year)
         available = [w for w in available if w.endswith("-Q4")]