fix(analytics): replace the control-character class with a code-point scan (no-control-regex)

The ASCII-only check used a regex character class of control characters. It
tripped eslint's no-control-regex, which the repo's quality gate enforces —
and the rule has a point: such a class is unreadable and is exactly the kind
of expression a reviewer nods at without checking. Same behaviour, written as
an explicit scan naming the permitted set.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

evidence/bi-p017/analyst-sql-live-bypass.json

@@ -2,7 +2,7 @@
   "prompt": "BI-P017",
   "claim": "Every bypass attempt, executed directly against a live ClickHouse as the tenant's restricted user WITH THE STATEMENT GUARD REMOVED, fails to return a foreign-tenant row, an ungranted-table row, or server metadata.",
   "clickhouse_version": "24.8.14.39",
-  "recorded_at": "2026-09-07T00:15:15.942Z",
+  "recorded_at": "2026-09-07T00:19:07.198Z",
   "base_table_rows_both_tenants": 100,
   "tenant_view_rows_alpha": 50,
   "attempts_total": 64,
@@ -607,7 +607,7 @@
       "leaked_foreign_tenant_row": false,
       "leaked_ungranted_table_row": false,
       "leaked_server_metadata": false,
-      "guard_verdict_for_reference": "refused: [AUTHORIZATION_DENIED] 0pt.analytics.sql: statement contains a non-ASCII character (U+0430) outside a string literal",
+      "guard_verdict_for_reference": "refused: [AUTHORIZATION_DENIED] 0pt.analytics.sql: statement contains a disallowed character (U+0430) outside a string literal",
       "engine_response_head": "Code: 62. DB::Exception: Syntax error: failed at position 17 ('�'): аlpha_sql.sales . Unrecognized token: '�'. (SYNTAX_ERROR) (version 24.8.14.39 (official buil"
     },
     {
@@ -619,7 +619,7 @@
       "leaked_foreign_tenant_row": false,
       "leaked_ungranted_table_row": false,
       "leaked_server_metadata": false,
-      "guard_verdict_for_reference": "refused: [AUTHORIZATION_DENIED] 0pt.analytics.sql: statement contains a non-ASCII character (U+200b) outside a string literal",
+      "guard_verdict_for_reference": "refused: [AUTHORIZATION_DENIED] 0pt.analytics.sql: statement contains a disallowed character (U+200B) outside a string literal",
       "engine_response_head": "Code: 62. DB::Exception: Syntax error: failed at position 24 ('.'): .tables . Expected one of: FINAL, SAMPLE, table, table function, subquery or list of joined"
     },
     {
@@ -631,7 +631,7 @@
       "leaked_foreign_tenant_row": false,
       "leaked_ungranted_table_row": false,
       "leaked_server_metadata": false,
-      "guard_verdict_for_reference": "refused: [AUTHORIZATION_DENIED] 0pt.analytics.sql: statement contains a non-ASCII character (U+0000) outside a string literal",
+      "guard_verdict_for_reference": "refused: [AUTHORIZATION_DENIED] 0pt.analytics.sql: statement contains a disallowed character (U+0000) outside a string literal",
       "engine_response_head": "Code: 62. DB::Exception: Syntax error: failed at position 40 ('\u0000'): \u0000 UNION ALL SELECT name FROM system.tables . Unrecognized token: '\u0000'. (SYNTAX_ERROR) (versio"
     },
     {
@@ -656,7 +656,7 @@
       "leaked_ungranted_table_row": false,
       "leaked_server_metadata": false,
       "guard_verdict_for_reference": "refused: [AUTHORIZATION_DENIED] 0pt.analytics.sql: \"SYSTEM\" is not permitted in a read-only analyst session",
-      "engine_response_head": "opt_analytics events f6d0a7de-8f53-4fe5-ad8f-7cff5fdd2005 MergeTree 0 ['/var/lib/clickhouse/store/f6d/f6d0a7de-8f53-4fe5-ad8f-7cff5fdd2005/'] /var/lib/clickhous"
+      "engine_response_head": "opt_analytics events 12706469-5605-436f-94ff-fc3c112aec6f MergeTree 0 ['/var/lib/clickhouse/store/127/12706469-5605-436f-94ff-fc3c112aec6f/'] /var/lib/clickhous"
     },
     {
       "id": "describe-subquery",

src/analytics/sql/statement-guard.ts

@@ -377,13 +377,21 @@ export function guardAnalystStatement(sql: string, opts: StatementGuardOptions):
   // Homoglyphs, zero-width joiners and RTL overrides stop being expressible
   // against the keyword scan. Values are unaffected: they live in literals,
   // which have already been removed at this point.
-  const nonAscii = /[^\x09\x0a\x0d\x20-\x7e]/.exec(stripped);
-  if (nonAscii) {
-    deny(
-      `statement contains a non-ASCII character (U+${nonAscii[0].codePointAt(0)?.toString(16).padStart(4, '0')}) outside a string literal`,
-    );
+  // Written as a code-point scan rather than a character class: a regex of
+  // control characters is unreadable, trips `no-control-regex`, and is the
+  // kind of expression a reviewer nods at without checking. Permitted: tab,
+  // LF, CR and printable ASCII. Everything else, NUL included, is refused.
+  for (let i = 0; i < stripped.length; i += 1) {
+    const code = stripped.charCodeAt(i);
+    const printable = code >= 0x20 && code <= 0x7e;
+    const allowedWhitespace = code === 0x09 || code === 0x0a || code === 0x0d;
+    if (!printable && !allowedWhitespace) {
+      deny(
+        `statement contains a disallowed character (U+${code.toString(16).toUpperCase().padStart(4, '0')}) ` +
+          `outside a string literal`,
+      );
+    }
   }
-  if (stripped.includes('\0')) deny('statement contains a NUL byte');
 
   // ── exactly one statement ──────────────────────────────────────────────
   const withoutTrailingSemi = stripped.replace(/;\s*$/, '');