parser: warn about empty / invalid params & empty quoted idents

crates/squawk_parser/src/lexed_str.rs

@@ -1,6 +1,6 @@
 // based on https://github.com/rust-lang/rust-analyzer/blob/d8887c0758bbd2d5f752d5bd405d4491e90e7ed6/crates/parser/src/lexed_str.rs
 
-use std::ops;
+use std::{num::IntErrorKind, ops};
 
 use squawk_lexer::tokenize;
 
@@ -120,6 +120,18 @@ struct Converter<'a> {
     offset: usize,
 }
 
+fn is_empty_quoted_ident(token_text: &str) -> bool {
+    let inner = if let Some(stripped) = token_text
+        .strip_prefix(['u', 'U'])
+        .and_then(|s| s.strip_prefix('&'))
+    {
+        stripped
+    } else {
+        token_text
+    };
+    inner == "\"\""
+}
+
 impl<'a> Converter<'a> {
     fn new(text: &'a str) -> Self {
         Self {
@@ -213,7 +225,17 @@ impl<'a> Converter<'a> {
                 squawk_lexer::TokenKind::PositionalParam {
                     trailing_junk_start,
                 } => {
-                    if (*trailing_junk_start as usize) < token_text.len() {
+                    let digits = &token_text[1..*trailing_junk_start as usize];
+                    if digits.is_empty() {
+                        err = "missing parameter number";
+                        err_range = Some(0..1);
+                    } else if digits
+                        .parse::<i32>()
+                        .is_err_and(|err| matches!(err.kind(), IntErrorKind::PosOverflow))
+                    {
+                        err = "parameter number too large";
+                        err_range = Some(0..*trailing_junk_start);
+                    } else if (*trailing_junk_start as usize) < token_text.len() {
                         err = "trailing junk after positional parameter";
                         err_range = Some(*trailing_junk_start..token_text.len() as u32);
                     }
@@ -222,6 +244,8 @@ impl<'a> Converter<'a> {
                 squawk_lexer::TokenKind::QuotedIdent { terminated } => {
                     if !terminated {
                         err = "Missing trailing \" to terminate the quoted identifier"
+                    } else if is_empty_quoted_ident(token_text) {
+                        err = "empty delimited identifier";
                     }
                     SyntaxKind::IDENT
                 }

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

@@ -28,3 +28,9 @@ SELECT 1_000._5;
 SELECT 1_000.5_;
 SELECT 1_000.5e_1;
 SELECT $0_1;
+SELECT $2147483648;
+-- empty placeholder
+SELECT $;
+SELECT $0111111111111111111111111111111111111111111111111111;
+SELECT "";
+SELECT U&"";

crates/squawk_parser/tests/snapshots/tests__select_literal_err.snap

@@ -309,6 +309,58 @@ SOURCE_FILE
             POSITIONAL_PARAM "$0_1"
     SEMICOLON ";"
   WHITESPACE "\n"
+  SELECT
+    SELECT_CLAUSE
+      SELECT_KW "SELECT"
+      WHITESPACE " "
+      TARGET_LIST
+        TARGET
+          LITERAL
+            POSITIONAL_PARAM "$2147483648"
+    SEMICOLON ";"
+  WHITESPACE "\n"
+  COMMENT "-- empty placeholder"
+  WHITESPACE "\n"
+  SELECT
+    SELECT_CLAUSE
+      SELECT_KW "SELECT"
+      WHITESPACE " "
+      TARGET_LIST
+        TARGET
+          LITERAL
+            POSITIONAL_PARAM "$"
+    SEMICOLON ";"
+  WHITESPACE "\n"
+  SELECT
+    SELECT_CLAUSE
+      SELECT_KW "SELECT"
+      WHITESPACE " "
+      TARGET_LIST
+        TARGET
+          LITERAL
+            POSITIONAL_PARAM "$0111111111111111111111111111111111111111111111111111"
+    SEMICOLON ";"
+  WHITESPACE "\n"
+  SELECT
+    SELECT_CLAUSE
+      SELECT_KW "SELECT"
+      WHITESPACE " "
+      TARGET_LIST
+        TARGET
+          NAME_REF
+            IDENT "\"\""
+    SEMICOLON ";"
+  WHITESPACE "\n"
+  SELECT
+    SELECT_CLAUSE
+      SELECT_KW "SELECT"
+      WHITESPACE " "
+      TARGET_LIST
+        TARGET
+          NAME_REF
+            IDENT "U&\"\""
+    SEMICOLON ";"
+  WHITESPACE "\n"
 ---
 error[syntax-error]: trailing junk after numeric literal
   ╭▸ 
@@ -458,3 +510,23 @@ error[syntax-error]: trailing junk after positional parameter
    ╭▸ 
 30 │ SELECT $0_1;
    ╰╴         ━━
+error[syntax-error]: parameter number too large
+   ╭▸ 
+31 │ SELECT $2147483648;
+   ╰╴       ━━━━━━━━━━━
+error[syntax-error]: missing parameter number
+   ╭▸ 
+33 │ SELECT $;
+   ╰╴       ━
+error[syntax-error]: parameter number too large
+   ╭▸ 
+34 │ SELECT $0111111111111111111111111111111111111111111111111111;
+   ╰╴       ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+error[syntax-error]: empty delimited identifier
+   ╭▸ 
+35 │ SELECT "";
+   ╰╴       ━━
+error[syntax-error]: empty delimited identifier
+   ╭▸ 
+36 │ SELECT U&"";
+   ╰╴       ━━━━