From aed787f44b2a4fa2bd5bc925dc718d58d96f1519 Mon Sep 17 00:00:00 2001
From: ak2k <19240940+ak2k@users.noreply.github.com>
Date: Sun, 5 Jul 2026 00:21:18 -0400
Subject: [PATCH] test(ck-engine): guard empty-corpus lexical rebuild

Add a regression test covering the transition from an empty corpus (which
builds an empty tantivy index) to a populated one: the corpus fingerprint
changes, so the next lexical search must rebuild and surface the new files
rather than trusting the empty index.

The test uses a plain #[tokio::test] with a TempDir-scoped .ck, matching the
other lexical-search tests: the unique temp path keys its own index (in-tree,
or a path-hashed subdir when CK_INDEX_DIR relocates it), so it stays isolated
without serial_test or removing the CK_INDEX_DIR env var.
---
 ck-engine/src/lib.rs | 39 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)

diff --git a/ck-engine/src/lib.rs b/ck-engine/src/lib.rs
index 1e0309d..3d6beef 100644
--- a/ck-engine/src/lib.rs
+++ b/ck-engine/src/lib.rs
@@ -2747,4 +2747,43 @@ mod tests {
         // Top score is normalized to 1.0, exactly as before this patch.
         assert!((results[0].score - 1.0).abs() < 1e-6);
     }
+
+    #[tokio::test]
+    async fn test_lexical_search_rebuilds_empty_stub() {
+        // A lexical search over a directory with no indexable files builds an
+        // empty tantivy index. Once real files appear the corpus fingerprint
+        // changes, so a later search must rebuild and surface them rather than
+        // trusting the empty index. The unique TempDir path keys its own index
+        // (in-tree .ck, or a path-hashed subdir under CK_INDEX_DIR when set), so
+        // the test needs no env manipulation or serialization to stay isolated.
+        let temp_dir = TempDir::new().unwrap();
+        fs::create_dir_all(temp_dir.path().join(".ck")).unwrap();
+
+        let options = SearchOptions {
+            mode: SearchMode::Lexical,
+            query: "zebra".to_string(),
+            path: temp_dir.path().to_path_buf(),
+            recursive: true,
+            ..Default::default()
+        };
+
+        // First search builds an empty index (no files yet).
+        let empty = lexical_search(&options).await.unwrap();
+        assert!(empty.is_empty());
+
+        // A real file appears; the next search must find it.
+        fs::write(
+            temp_dir.path().join("mod.py"),
+            "def gamma():\n    zebra = 3\n    return zebra\n",
+        )
+        .unwrap();
+
+        let results = lexical_search(&options).await.unwrap();
+        assert!(
+            results
+                .iter()
+                .any(|r| r.file.file_name().unwrap() == "mod.py"),
+            "empty tantivy stub was trusted; expected a rebuild to find the new file"
+        );
+    }
 }