Add tests

tests/test_detailed.py

@@ -0,0 +1,34 @@
+from src.search.vector_search import (
+    search_vector_with_expansion,
+    search_vector_with_reranking,
+)
+
+if __name__ == "__main__":
+    query = "wireless headphones"
+    print(f"=== Testing: {query} ===\n")
+
+    # Stage 3: Query Expansion
+    print("STAGE 3: Query Expansion Results")
+    results_3 = search_vector_with_expansion(query)
+    for i, r in enumerate(results_3[:5], 1):
+        print(f"{i}. [{r['score']:.4f}] {r['product_name']}")
+
+    print("\n" + "=" * 80 + "\n")
+
+    # Stage 4: Reranking
+    print("STAGE 4: Reranking Results")
+    results_4 = search_vector_with_reranking(query)
+    for i, r in enumerate(results_4[:5], 1):
+        print(f"{i}. [{r['score']:.4f}] {r['product_name']}")
+
+    print("\n" + "=" * 80 + "\n")
+    print("Analysis:")
+    print(
+        f"Stage 3 Top-1: {results_3[0]['score']:.4f} - {results_3[0]['product_name']}"
+    )
+    print(
+        f"Stage 4 Top-1: {results_4[0]['score']:.4f} - {results_4[0]['product_name']}"
+    )
+
+    if results_3[0]["product_name"] != results_4[0]["product_name"]:
+        print("\n⚠️  Reranker changed the top result!")

tests/test_stages.py

@@ -0,0 +1,34 @@
+from src.app import search_stage_2, search_stage_3, search_stage_4
+
+
+if __name__ == "__main__":
+    # Test with a few different queries
+    test_queries = [
+        "outdoor activities",
+        "science experiments for kids",
+        "wireless headphones",
+    ]
+
+    for query in test_queries:
+        print(f"\n=== Testing: {query} ===")
+
+        _, metrics_2 = search_stage_2(query)
+        _, metrics_3 = search_stage_3(query)
+        _, metrics_4 = search_stage_4(query)
+
+        print(
+            f'Stage 2 (Vector):          Top-1: {metrics_2["top1_score"]:.4f}, Top-5: {metrics_2["top5_avg"]:.4f}, Latency: {metrics_2["latency_ms"]}ms'
+        )
+        print(
+            f'Stage 3 (+ Expansion):     Top-1: {metrics_3["top1_score"]:.4f}, Top-5: {metrics_3["top5_avg"]:.4f}, Latency: {metrics_3["latency_ms"]}ms'
+        )
+        print(
+            f'Stage 4 (+ Reranking):     Top-1: {metrics_4["top1_score"]:.4f}, Top-5: {metrics_4["top5_avg"]:.4f}, Latency: {metrics_4["latency_ms"]}ms'
+        )
+
+        if metrics_4["top1_score"] < metrics_3["top1_score"]:
+            diff = metrics_3["top1_score"] - metrics_4["top1_score"]
+            print(f"⚠️  WARNING: Reranking decreased top-1 score by {diff:.4f}")
+        else:
+            diff = metrics_4["top1_score"] - metrics_3["top1_score"]
+            print(f"✓ Reranking improved top-1 score by {diff:.4f}")