about summary refs log tree commit diff
path: root/src
diff options
context:
space:
mode:
Diffstat (limited to 'src')
-rw-r--r--src/search/native.rs32
1 files changed, 20 insertions, 12 deletions
diff --git a/src/search/native.rs b/src/search/native.rs
index a7b2fb5..665477a 100644
--- a/src/search/native.rs
+++ b/src/search/native.rs
@@ -267,7 +267,7 @@ mod tests {
     #[test]
     fn boolean_prefixes_parse_to_exact_index_terms() {
         let mut qp = build_query_parser();
-        let terms = |query: &str| -> std::collections::BTreeSet<String> {
+        let mut terms = |query: &str| -> std::collections::BTreeSet<String> {
             qp.parse_query_checked::<&str>(query, None, None)
                 .unwrap_or_else(|e| panic!("{query} must parse: {e:?}"))
                 .unique_terms()
@@ -288,21 +288,29 @@ mod tests {
         assert!(terms("chr:1").contains("XC1"));
         assert!(terms("peakchr:1").contains("XPC1"));
         // search.py:204-209, 222-223: group values lowercase (xapian
-        // does not stem boolean fields starting with a capital).
+        // does not stem boolean fields starting with a capital):
+        // B6 -> b6, concatenated behind XG. (The r1-test G4 smoke's
+        // XGbxd was the same rule for value BXD, not a stem.)
         let b6 = terms("group:B6");
-        assert!(b6.contains("XGbxd"), "group:B6 must lowercase to XGbxd, got {b6:?}");
-        assert!(!b6.contains("XGBXD"), "group must not emit the raw-case term");
+        assert!(
+            b6.contains("XGb6"),
+            "group:B6 must lowercase to XGb6, got {b6:?}"
+        );
+        assert!(!b6.contains("XGB6"), "group must not emit the raw-case term");
     }
 
     /// The probabilistic-prefix registrations (C3.3): description, rif,
     /// wiki and the uppercase aliases must parse as weighted field
     /// terms, not free text (search.py:229-233). With STEM_ALL_Z the
-    /// field term is the prefix + Z-stemmed value (e.g. XDZinsulin),
-    /// so assert on the prefix, the invariant the registration owns.
+    /// field term is Z+<prefix>+stem — ZXDinsulin, ZXRF…, ZXWK… — the
+    /// exact shape the production indexer's TermGenerator emits for
+    /// these fields (confirmed against the dev index: term
+    /// ZXWKdiabet exists with 24 gene documents, ZXRF-prefixed terms
+    /// abound; d.v. the delve probes in LANE-REPORT).
     #[test]
     fn probabilistic_prefixes_and_aliases_parse_as_field_terms() {
         let mut qp = build_query_parser();
-        let prefixed = |query: &str, prefix: &str| {
+        let mut prefixed = |query: &str, prefix: &str| {
             let set = qp
                 .parse_query_checked::<&str>(query, None, None)
                 .unwrap_or_else(|e| panic!("{query} must parse: {e:?}"))
@@ -311,11 +319,11 @@ mod tests {
                 .collect::<std::collections::BTreeSet<_>>();
             set.iter().any(|t| t.starts_with(prefix))
         };
-        assert!(prefixed("description:insulin", "XD"));
-        assert!(prefixed("rif:insulin", "XRF"));
-        assert!(prefixed("RIF:insulin", "XRF"));
-        assert!(prefixed("wiki:diabetes", "XWK"));
-        assert!(prefixed("WIKI:diabetes", "XWK"));
+        assert!(prefixed("description:insulin", "ZXD"));
+        assert!(prefixed("rif:insulin", "ZXRF"));
+        assert!(prefixed("RIF:insulin", "ZXRF"));
+        assert!(prefixed("wiki:diabetes", "ZXWK"));
+        assert!(prefixed("WIKI:diabetes", "ZXWK"));
     }
 
     /// The index-open guards, without touching a real index.