parser: fix unicode escape idents in cast position (#1127)

Author: sbdchd

crates/squawk_parser/src/grammar.rs

@@ -37,7 +37,7 @@ fn literal(p: &mut Parser<'_>) -> Option<CompletedMarker> {
     let m = p.start();
     if p.eat(UNICODE_ESC_STRING) {
         if p.eat(UESCAPE_KW) {
-            p.eat(STRING);
+            p.expect(STRING);
         }
     }
     // E021-03 string continuation syntax
@@ -1678,7 +1678,9 @@ fn path_segment(p: &mut Parser<'_>, kind: SyntaxKind) {
         // skip
     } else if p.at_ts(COL_LABEL_FIRST) {
         let m = p.start();
-        p.bump_any();
+        if !opt_ident(p) {
+            p.bump_any();
+        }
         let kind = if p.at(DOT) { NAME_REF } else { kind };
         m.complete(p, kind);
     } else {
@@ -2250,7 +2252,9 @@ fn index_expr(p: &mut Parser<'_>, lhs: CompletedMarker) -> CompletedMarker {
 fn name_ref_or_index(p: &mut Parser<'_>) {
     assert!(p.at(IDENT) || p.at_ts(TYPE_KEYWORDS) || p.at_ts(ALL_KEYWORDS) || p.at(INT_NUMBER));
     let m = p.start();
-    p.bump_any();
+    if !opt_ident(p) {
+        p.bump_any();
+    }
     m.complete(p, NAME_REF);
 }
 
@@ -14267,6 +14271,8 @@ fn create_extension(p: &mut Parser<'_>) -> CompletedMarker {
 }
 
 fn opt_ident(p: &mut Parser<'_>) -> bool {
+    // handle cases like:
+    // U&"!0069!006E!0074!0038" UESCAPE '!'
     if p.eat(IDENT) {
         if p.eat(UESCAPE_KW) {
             p.expect(STRING);

crates/squawk_parser/tests/data/ok/select.sql

@@ -509,6 +509,12 @@ ORDER BY sensor_id, day;
 select U&"d!0061t!+000061" UESCAPE '!';
 SELECT U&' \' UESCAPE '!';
 
+-- select with uescape cast
+select 2::U&"!0069!006E!0074!0038" UESCAPE '!' from t;
+
+-- select with uescape on qualified field
+select t.U&"!0061" UESCAPE '!' from t;
+
 -- select_from_user_table
 select * from user;
 

crates/squawk_parser/tests/snapshots/tests__select_ok.snap

@@ -6277,6 +6277,65 @@ SOURCE_FILE
             STRING "'!'"
   SEMICOLON ";"
   WHITESPACE "\n\n"
+  COMMENT "-- select with uescape cast"
+  WHITESPACE "\n"
+  SELECT
+    SELECT_CLAUSE
+      SELECT_KW "select"
+      WHITESPACE " "
+      TARGET_LIST
+        TARGET
+          CAST_EXPR
+            LITERAL
+              INT_NUMBER "2"
+            COLON_COLON
+              COLON ":"
+              COLON ":"
+            PATH_TYPE
+              PATH
+                PATH_SEGMENT
+                  NAME_REF
+                    IDENT "U&\"!0069!006E!0074!0038\""
+                    WHITESPACE " "
+                    UESCAPE_KW "UESCAPE"
+                    WHITESPACE " "
+                    STRING "'!'"
+    WHITESPACE " "
+    FROM_CLAUSE
+      FROM_KW "from"
+      WHITESPACE " "
+      FROM_ITEM
+        NAME_REF
+          IDENT "t"
+  SEMICOLON ";"
+  WHITESPACE "\n\n"
+  COMMENT "-- select with uescape on qualified field"
+  WHITESPACE "\n"
+  SELECT
+    SELECT_CLAUSE
+      SELECT_KW "select"
+      WHITESPACE " "
+      TARGET_LIST
+        TARGET
+          FIELD_EXPR
+            NAME_REF
+              IDENT "t"
+            DOT "."
+            NAME_REF
+              IDENT "U&\"!0061\""
+              WHITESPACE " "
+              UESCAPE_KW "UESCAPE"
+              WHITESPACE " "
+              STRING "'!'"
+    WHITESPACE " "
+    FROM_CLAUSE
+      FROM_KW "from"
+      WHITESPACE " "
+      FROM_ITEM
+        NAME_REF
+          IDENT "t"
+  SEMICOLON ";"
+  WHITESPACE "\n\n"
   COMMENT "-- select_from_user_table"
   WHITESPACE "\n"
   SELECT