From d531211030908bdc347db097d6e52b2e8d835386 Mon Sep 17 00:00:00 2001
From: Ryan Malia <ryanmalia@gmail.com>
Date: Thu, 12 Mar 2026 00:14:21 -0700
Subject: [PATCH] fix: allow hyphenated words in vec/hyde queries (#383)

The validateSemanticQuery regex rejected any hyphen followed by a word
character, blocking common compound words (real-time, multi-client,
kebab-case identifiers like better-sqlite3). Tighten the check to only
match negation syntax at token boundaries (start of string or after
whitespace).

See https://github.com/tobi/qmd/issues/383

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 src/store.ts                   |  5 +--
 test/structured-search.test.ts | 58 +++++++++++++++++++++++++++++++++-
 2 files changed, 60 insertions(+), 3 deletions(-)

diff --git a/src/store.ts b/src/store.ts
index aa5fae4f..e8e5aa1f 100644
--- a/src/store.ts
+++ b/src/store.ts
@@ -2597,8 +2597,9 @@ function buildFTS5Query(query: string): string | null {
  * Returns error message if invalid, null if valid.
  */
 export function validateSemanticQuery(query: string): string | null {
-  // Check for negation syntax
-  if (/-\w/.test(query) || /-"/.test(query)) {
+  // Check for negation syntax — only at token boundaries (start of string or after whitespace).
+  // Hyphenated words like "real-time" or "write-ahead" must not trigger this.
+  if (/(^|\s)-[\w"]/.test(query)) {
     return 'Negation (-term) is not supported in vec/hyde queries. Use lex for exclusions.';
   }
   return null;
diff --git a/test/structured-search.test.ts b/test/structured-search.test.ts
index 5c4e97f3..764d745e 100644
--- a/test/structured-search.test.ts
+++ b/test/structured-search.test.ts
@@ -361,17 +361,73 @@ describe("lex query syntax", () => {
       expect(validateSemanticQuery("what is the CAP theorem")).toBeNull();
     });
 
-    test("rejects negation syntax", () => {
+    test("rejects negation at start of query", () => {
+      expect(validateSemanticQuery("-redis connection pooling")).toContain("Negation");
+    });
+
+    test("rejects negation after space", () => {
       expect(validateSemanticQuery("performance -sports")).toContain("Negation");
+    });
+
+    test("rejects negated quoted phrase", () => {
       expect(validateSemanticQuery('-"exact phrase"')).toContain("Negation");
     });
 
+    test("rejects multiple negations", () => {
+      expect(validateSemanticQuery("error handling -java -python")).toContain("Negation");
+    });
+
+    test("rejects negation after leading whitespace", () => {
+      expect(validateSemanticQuery("  -term at start")).toContain("Negation");
+    });
+
+    test("rejects negation after tab", () => {
+      expect(validateSemanticQuery("foo\t-bar")).toContain("Negation");
+    });
+
+    test("accepts hyphenated compound words", () => {
+      expect(validateSemanticQuery("long-lived server shared across clients")).toBeNull();
+      expect(validateSemanticQuery("real-time voice processing pipeline")).toBeNull();
+      expect(validateSemanticQuery("how does the rate-limiter handle burst traffic")).toBeNull();
+      expect(validateSemanticQuery("self-hosted deployment options")).toBeNull();
+      expect(validateSemanticQuery("multi-client session architecture")).toBeNull();
+      expect(validateSemanticQuery("cross-platform compatibility")).toBeNull();
+      expect(validateSemanticQuery("non-blocking I/O model")).toBeNull();
+      expect(validateSemanticQuery("in-memory caching strategy")).toBeNull();
+      expect(validateSemanticQuery("write-ahead log for crash recovery")).toBeNull();
+      expect(validateSemanticQuery("copy-on-write semantics")).toBeNull();
+    });
+
+    test("accepts multiple hyphens in a phrase", () => {
+      expect(validateSemanticQuery("state-of-the-art embedding models")).toBeNull();
+      expect(validateSemanticQuery("end-to-end testing")).toBeNull();
+      expect(validateSemanticQuery("man-in-the-middle attack prevention")).toBeNull();
+    });
+
+    test("accepts multiple hyphenated words in one query", () => {
+      expect(validateSemanticQuery("built-in vs add-on features")).toBeNull();
+    });
+
+    test("accepts short hyphenated terms", () => {
+      expect(validateSemanticQuery("A-B testing for ML models")).toBeNull();
+      expect(validateSemanticQuery("e-commerce platform")).toBeNull();
+    });
+
+    test("accepts bare hyphen without word character", () => {
+      expect(validateSemanticQuery("-")).toBeNull();
+    });
 
     test("accepts hyde-style hypothetical answers", () => {
       expect(validateSemanticQuery(
         "The CAP theorem states that a distributed system cannot simultaneously provide consistency, availability, and partition tolerance."
       )).toBeNull();
     });
+
+    test("accepts hyde with hyphenated words", () => {
+      expect(validateSemanticQuery(
+        "HTTP transport runs a single long-lived daemon shared across all clients, avoiding per-session model re-loading."
+      )).toBeNull();
+    });
   });
 
   describe("validateLexQuery", () => {