fix(cypher): refuse a WITH wider than a binding can carry

execute_return_star_after_with sized its column arrays at CYP_MAX_VARS
while the parser allowed up to CBM_SZ_32 items. A WITH naming 17 to 32
aliases parsed cleanly, then answered 16 columns and dropped the rest
with no error. The caller could not tell that from a query that genuinely
had 16 columns.

Widening those arrays to 32 is the wrong fix and would make the result
worse. A binding holds exactly CYP_MAX_VARS variables, and
with_add_vbinding_var drops any alias past the 16th on the way in. The
projection would then report 20 columns of which 4 are always blank —
silent wrong data in place of a silent short answer. The 16 was not an
arbitrary undersize; it matched what the binding can carry.

So bound the WITH where the RETURN is already bounded, in
parse_return_or_with, and refuse the query instead of answering it
short. The clamp in execute_return_star_after_with stays: it can no
longer fire, and it is what keeps that function correct if the bound
ever moves.

The test asserts both halves — a 20-alias WITH is refused, and a
16-alias WITH still succeeds with 16 columns — so the guard rejects only
what the binding genuinely cannot hold.

Signed-off-by: Joshua Richter <jrichter5781@gmail.com>

Author: CaptainMittens

src/cypher/cypher.c

@@ -1802,8 +1802,16 @@ static int parse_return_or_with(parser_t *p, cbm_return_clause_t **out, bool is_
     /* Projection is materialized per row into fixed-width stack arrays sized at
      * CBM_SZ_32 columns (execute_return_simple and its siblings). Bound the
      * parsed item count to that width so an over-wide RETURN is rejected here
-     * instead of writing past those arrays downstream. */
-    if (r->count > CBM_SZ_32) {
+     * instead of writing past those arrays downstream.
+     *
+     * WITH is bounded tighter, by CYP_MAX_VARS. Every item a WITH projects
+     * becomes one variable of the binding that carries the rest of the query,
+     * and binding_t holds exactly CYP_MAX_VARS variables. A wider WITH used to
+     * parse, then lose every alias past the 16th in with_add_vbinding_var and
+     * answer with silently blank columns. Refuse it here, the same way an
+     * over-wide RETURN is refused, so the caller sees an error instead of a
+     * short or empty result. */
+    if (r->count > (is_with ? CYP_MAX_VARS : CBM_SZ_32)) {
         free_return_clause(r);
         return CBM_NOT_FOUND;
     }
@@ -4286,6 +4294,8 @@ static void execute_return_star_after_with(cbm_query_t *q, binding_t *bindings,
     cbm_return_clause_t *wc = q->with_clause;
     char name_bufs[CYP_MAX_VARS][CBM_SZ_128];
     const char *cols[CYP_MAX_VARS];
+    /* parse_return_or_with refuses a WITH wider than CYP_MAX_VARS, so this
+     * clamp cannot fire. It stays as the bound this function relies on. */
     int col_n = wc->count < CYP_MAX_VARS ? wc->count : CYP_MAX_VARS;
     for (int i = 0; i < col_n; i++) {
         cols[i] = resolve_item_alias(&wc->items[i], name_bufs[i], sizeof(name_bufs[i]));

tests/test_cypher.c

@@ -3026,6 +3026,45 @@ TEST(cypher_return_star_after_with_names_aliases) {
     PASS();
 }
 
+TEST(cypher_wide_with_refused_not_truncated) {
+    /* Every item a WITH projects becomes one variable of the binding that
+     * carries the rest of the query, and a binding holds CYP_MAX_VARS (16) of
+     * them. A 20-alias WITH used to parse, drop aliases 17 to 20 inside
+     * with_add_vbinding_var, and answer RETURN * with 16 columns and no error —
+     * a short result the caller could not tell from a complete one. It has to
+     * be refused at parse time instead. */
+    char query[1024];
+    int off = snprintf(query, sizeof(query), "MATCH (f:Function) WITH ");
+    for (int i = 0; i < 20; i++) { /* 20 > CYP_MAX_VARS (16) */
+        off += snprintf(query + off, sizeof(query) - (size_t)off, "%sf.name AS c%d", i ? ", " : "",
+                        i);
+    }
+    snprintf(query + off, sizeof(query) - (size_t)off, " RETURN *");
+
+    cbm_store_t *s = setup_cypher_store();
+    cbm_cypher_result_t r = {0};
+    int rc = cbm_cypher_execute(s, query, "test", 0, &r);
+    ASSERT_TRUE(rc != 0); /* refused, not silently narrowed to 16 columns */
+    cbm_cypher_result_free(&r);
+
+    /* The width just under the bound still works, so the guard rejects only
+     * what the binding genuinely cannot carry. */
+    char ok_query[1024];
+    off = snprintf(ok_query, sizeof(ok_query), "MATCH (f:Function) WITH ");
+    for (int i = 0; i < 16; i++) {
+        off += snprintf(ok_query + off, sizeof(ok_query) - (size_t)off, "%sf.name AS c%d",
+                        i ? ", " : "", i);
+    }
+    snprintf(ok_query + off, sizeof(ok_query) - (size_t)off, " RETURN *");
+    cbm_cypher_result_t r16 = {0};
+    ASSERT_EQ(cbm_cypher_execute(s, ok_query, "test", 0, &r16), 0);
+    ASSERT_EQ(r16.col_count, 16);
+    cbm_cypher_result_free(&r16);
+
+    cbm_store_close(s);
+    PASS();
+}
+
 TEST(cypher_parse_neq) {
     cbm_query_t *q = NULL;
     char *err = NULL;
@@ -4336,6 +4375,7 @@ SUITE(cypher) {
     RUN_TEST(cypher_exec_return_star);
     RUN_TEST(cypher_return_star_dedups_repeated_pattern_var);
     RUN_TEST(cypher_return_star_after_with_names_aliases);
+    RUN_TEST(cypher_wide_with_refused_not_truncated);
     RUN_TEST(cypher_parse_neq);
     RUN_TEST(cypher_parse_in);
     RUN_TEST(cypher_parse_is_null);