parser: fix non-unicode ident parsing + cte error recovery

crates/squawk_fmt/tests/after/create_schema.snap

@@ -4,6 +4,8 @@ input_file: crates/squawk_fmt/tests/before/create_schema.sql
 ---
 create schema app;
 
+create schema تست;
+
 create schema if not exists
   extraordinarily_long_application_reporting_and_data_warehouse_schema
   authorization extraordinarily_long_application_reporting_and_data_warehouse_owner_role;

crates/squawk_fmt/tests/before/create_schema.sql

@@ -1,5 +1,7 @@
 create schema app;
 
+create schema تست;
+
 create schema if not exists extraordinarily_long_application_reporting_and_data_warehouse_schema authorization extraordinarily_long_application_reporting_and_data_warehouse_owner_role;
 
 create schema authorization current_user;

crates/squawk_parser/src/grammar.rs

@@ -3400,6 +3400,10 @@ fn with_query_clause(p: &mut Parser<'_>) -> Option<CompletedMarker> {
     let m = p.start();
     p.expect(WITH_KW);
     p.eat(RECURSIVE_KW);
+    if p.at_ts(WITH_FOLLOW) {
+        p.error("expected common table expression");
+        return Some(m.complete(p, WITH_CLAUSE));
+    }
     while !p.at(EOF) {
         with_query(p);
         if p.at(COMMA) && p.nth_at_ts(1, WITH_FOLLOW) {
@@ -3409,8 +3413,11 @@ fn with_query_clause(p: &mut Parser<'_>) -> Option<CompletedMarker> {
         if !p.eat(COMMA) {
             if p.at_ts(WITH_FOLLOW) || (p.at(L_PAREN) && p.nth_at_ts(1, PAREN_SELECT_FIRST)) {
                 break;
-            } else {
+            } else if p.at_ts(NAME_REF_FIRST) {
                 p.error("missing comma");
+            } else {
+                p.err_and_bump("missing comma");
+                break;
             }
         }
     }

crates/squawk_parser/tests/data/err/select_cte.sql

@@ -45,3 +45,6 @@ values (2);
 
 -- WITHOUT OVERLAPS is only valid w/ PRIMARY KEY or UNIQUE constraint
 with t (a without overlaps) as (select 1) select * from t;
+
+-- missing common table expression
+with select 1;

crates/squawk_parser/tests/snapshots/tests__select_cte_err.snap

@@ -457,6 +457,21 @@ SOURCE_FILE
             PATH_SEGMENT_REF
               IDENT "t"
     SEMICOLON ";"
+  WHITESPACE "\n\n"
+  COMMENT "-- missing common table expression"
+  WHITESPACE "\n"
+  SELECT
+    WITH_CLAUSE
+      WITH_KW "with"
+    WHITESPACE " "
+    SELECT_CLAUSE
+      SELECT_KW "select"
+      WHITESPACE " "
+      TARGET_LIST
+        TARGET
+          LITERAL
+            INT_NUMBER "1"
+    SEMICOLON ";"
   WHITESPACE "\n"
 ---
 error[syntax-error]: unexpected comma
@@ -491,3 +506,7 @@ error[syntax-error]: WITHOUT OVERLAPS is only allowed in PRIMARY KEY and UNIQUE
    ╭▸ 
 47 │ with t (a without overlaps) as (select 1) select * from t;
    ╰╴          ━
+error[syntax-error]: expected common table expression
+   ╭▸ 
+50 │ with select 1;
+   ╰╴    ━

crates/squawk_syntax/src/ast/node_ext.rs

@@ -1000,9 +1000,21 @@ impl ast::PathSegmentRef {
 }
 
 pub fn is_quoted_name_node(node: &SyntaxNode) -> bool {
-    let text = node.text();
-    let first = text.char_at(0.into());
-    let second = text.char_at(1.into());
+    let mut first = None;
+    let mut second = None;
+    node.text().for_each_chunk(|chunk| {
+        if second.is_some() {
+            return;
+        }
+        for ch in chunk.chars() {
+            if first.is_none() {
+                first = Some(ch);
+            } else {
+                second = Some(ch);
+                break;
+            }
+        }
+    });
     matches!(
         (first, second),
         (Some('u' | 'U'), Some('"')) | (Some('"'), Some(_))

crates/squawk_syntax/src/quote.rs

@@ -64,12 +64,12 @@ pub fn unquote_ident(node: &SyntaxNode) -> Option<String> {
 
     // see: https://www.postgresql.org/docs/18/sql-syntax-lexical.html#SQL-SYNTAX-IDENTIFIERS
     match chars.next() {
-        Some(c) if c.is_lowercase() || c == '_' => {}
+        Some(c) if c.is_ascii_lowercase() || !c.is_ascii() || c == '_' => {}
         _ => return None,
     }
 
     for c in chars {
-        if c.is_lowercase() || c.is_ascii_digit() || c == '_' || c == '$' {
+        if c.is_ascii_lowercase() || !c.is_ascii() || c.is_ascii_digit() || c == '_' || c == '$' {
             continue;
         }
         return None;
@@ -90,12 +90,12 @@ pub fn needs_quoting(text: &str) -> bool {
     let mut chars = text.chars();
 
     match chars.next() {
-        Some(c) if c.is_lowercase() || c == '_' => {}
+        Some(c) if c.is_ascii_lowercase() || !c.is_ascii() || c == '_' => {}
         _ => return true,
     }
 
     for c in chars {
-        if c.is_lowercase() || c.is_ascii_digit() || c == '_' || c == '$' {
+        if c.is_ascii_lowercase() || !c.is_ascii() || c.is_ascii_digit() || c == '_' || c == '$' {
             continue;
         }
         return true;
@@ -205,6 +205,7 @@ mod tests {
         assert_snapshot!(quote_ident("col_name"), @"col_name");
         assert_snapshot!(quote_ident("users"), @"users");
         assert_snapshot!(quote_ident("t2$"), @"t2$");
+        assert_snapshot!(quote_ident("تست"), @"تست");
     }
 
     #[test]