parser: initial pl/pgsql support

crates/squawk_ide/src/binder.rs

@@ -6,7 +6,6 @@ use rustc_hash::FxHashMap;
 use smallvec::SmallVec;
 use squawk_syntax::{SyntaxNodePtr, ast, ast::AstNode};
 
-use crate::literals::literal_string_value;
 use crate::name::{AsName, schema_and_func_name};
 use crate::scope::Scope;
 use crate::symbols::{Name, Schema, Symbol, SymbolKind};
@@ -1775,7 +1774,7 @@ fn bind_prepare_transaction(b: &mut Binder, prepare: ast::PrepareTransaction) {
     let Some(literal) = prepare.literal() else {
         return;
     };
-    let Some(transaction_id) = literal_string_value(&literal) else {
+    let Some(transaction_id) = literal.string_value() else {
         return;
     };
 
@@ -1787,7 +1786,7 @@ fn bind_prepared_transaction_ref(b: &mut Binder, literal: Option<ast::Literal>)
     let Some(literal) = literal else {
         return;
     };
-    let Some(transaction_id) = literal_string_value(&literal) else {
+    let Some(transaction_id) = literal.string_value() else {
         return;
     };
 
@@ -2063,7 +2062,7 @@ fn bind_select_set_config(b: &mut Binder, select: &ast::Select, position: TextSi
     let Some(ast::Expr::Literal(setting_name_literal)) = args.next().and_then(|a| a.expr()) else {
         return;
     };
-    let Some(setting_name) = literal_string_value(&setting_name_literal) else {
+    let Some(setting_name) = setting_name_literal.string_value() else {
         return;
     };
     if !setting_name.eq_ignore_ascii_case("search_path") {
@@ -2073,7 +2072,7 @@ fn bind_select_set_config(b: &mut Binder, select: &ast::Select, position: TextSi
     let Some(ast::Expr::Literal(new_value_literal)) = args.next().and_then(|a| a.expr()) else {
         return;
     };
-    let Some(new_value) = literal_string_value(&new_value_literal) else {
+    let Some(new_value) = new_value_literal.string_value() else {
         return;
     };
 

crates/squawk_ide/src/hover.rs

@@ -6,7 +6,6 @@ use crate::file::InFile;
 use crate::infer::{infer_type_from_expr, infer_type_from_literal};
 use crate::literals::binary_digits_to_hex;
 use crate::literals::hex_digits_to_binary;
-use crate::literals::literal_string_value;
 use crate::location::{Location, LocationKind};
 use crate::name;
 use crate::offsets::token_from_offset;
@@ -275,7 +274,7 @@ fn hover_literal(literal: &ast::Literal) -> Option<Hover> {
         return None;
     }
 
-    let value = literal_string_value(literal)?;
+    let value = literal.string_value()?;
     let ty = infer_type_from_literal(literal)?.to_string();
 
     let comment = match kind {

crates/squawk_ide/src/literals.rs

@@ -1,10 +1,3 @@
-use squawk_syntax::{
-    SyntaxKind,
-    ast::{self, AstNode},
-    quote::{strip_dollar_quotes, strip_prefixed_quotes, strip_quotes, strip_unicode_esc_prefix},
-    unescape::{decode_esc_string, decode_plain_string, decode_unicode_esc_string, uescape_char},
-};
-
 #[derive(Clone, Copy, PartialEq, Eq)]
 pub(crate) enum IntegerRadix {
     Binary,
@@ -86,88 +79,3 @@ pub(crate) fn hex_digits_to_binary(digits: &str) -> Option<String> {
 
     Some(out)
 }
-
-#[derive(Clone, Copy, PartialEq, Eq)]
-enum StringDecoding {
-    BitOrByte,
-    EscString,
-    UnicodeEscString,
-}
-
-pub(crate) fn literal_string_value(literal: &ast::Literal) -> Option<String> {
-    let escape_char = unicode_escape_char(literal);
-    let mut out = String::with_capacity(literal.syntax().text().len().into());
-    let mut decoding: Option<StringDecoding> = None;
-
-    for element in literal.syntax().children_with_tokens() {
-        let Some(token) = element.into_token() else {
-            continue;
-        };
-        match token.kind() {
-            SyntaxKind::ESC_STRING => {
-                let inner = strip_prefixed_quotes(token.text(), ['e', 'E'])?;
-                decode_esc_string(inner, &mut out);
-                decoding = Some(StringDecoding::EscString);
-            }
-            SyntaxKind::UNICODE_ESC_STRING => {
-                let inner = strip_unicode_esc_prefix(token.text())?;
-                decode_unicode_esc_string(inner, escape_char, &mut out);
-                decoding = Some(StringDecoding::UnicodeEscString);
-            }
-            SyntaxKind::BIT_STRING => {
-                let inner = strip_prefixed_quotes(token.text(), ['b', 'B'])?;
-                out.push_str(inner);
-                decoding = Some(StringDecoding::BitOrByte);
-            }
-            SyntaxKind::BYTE_STRING => {
-                let inner = strip_prefixed_quotes(token.text(), ['x', 'X'])?;
-                out.push_str(inner);
-                decoding = Some(StringDecoding::BitOrByte);
-            }
-            SyntaxKind::DOLLAR_QUOTED_STRING => {
-                let inner = strip_dollar_quotes(token.text())?;
-                out.push_str(inner);
-                return Some(out);
-            }
-            SyntaxKind::NATIONAL_STRING => {
-                let inner = strip_prefixed_quotes(token.text(), ['n', 'N'])?;
-                decode_plain_string(inner, &mut out);
-            }
-            SyntaxKind::STRING => {
-                let inner = strip_quotes(token.text())?;
-                match decoding {
-                    Some(StringDecoding::EscString) => decode_esc_string(inner, &mut out),
-                    Some(StringDecoding::UnicodeEscString) => {
-                        decode_unicode_esc_string(inner, escape_char, &mut out)
-                    }
-                    Some(StringDecoding::BitOrByte) => out.push_str(inner),
-                    None => decode_plain_string(inner, &mut out),
-                }
-            }
-            SyntaxKind::UESCAPE_KW => break,
-            _ => (),
-        }
-    }
-
-    Some(out)
-}
-
-fn unicode_escape_char(literal: &ast::Literal) -> char {
-    let mut seen_uescape = false;
-    for element in literal.syntax().children_with_tokens() {
-        let Some(token) = element.into_token() else {
-            continue;
-        };
-        match token.kind() {
-            SyntaxKind::UESCAPE_KW => seen_uescape = true,
-            SyntaxKind::STRING if seen_uescape => {
-                if let Some(ch) = uescape_char(token.text()) {
-                    return ch;
-                }
-                return '\\';
-            }
-            _ => (),
-        }
-    }
-    '\\'
-}

crates/squawk_lexer/src/lib.rs

@@ -106,6 +106,11 @@ impl Cursor<'_> {
                 TokenKind::Literal { kind: literal_kind }
             }
             '.' => match self.first() {
+                // https://github.com/postgres/postgres/blob/db0c96cc18aec417101e37e59fcc53d4bf647915/src/backend/parser/scan.l#L353
+                '.' => {
+                    self.bump();
+                    TokenKind::DotDot
+                }
                 '0'..='9' => {
                     let literal_kind = self.number('.');
                     TokenKind::Literal { kind: literal_kind }
@@ -302,7 +307,9 @@ impl Cursor<'_> {
         };
 
         match self.first() {
-            '.' => {
+            // `1..10` lexes as 1, .., 10 rather than `1.` followed by `.10`
+            // https://github.com/postgres/postgres/blob/db0c96cc18aec417101e37e59fcc53d4bf647915/src/backend/parser/scan.l#L410
+            '.' if self.second() != '.' => {
                 self.bump();
                 self.eat_fractional()
             }
@@ -794,6 +801,63 @@ $foo$hello$world$bar$
         "#)
     }
 
+    #[test]
+    fn dot_dot_is_its_own_token() {
+        assert_debug_snapshot!(lex("1..10 1 .. 10 1.5..2 1...2 .. a[1..2]"), @r#"
+        [
+            "1" @ Literal { kind: Int { base: Decimal, empty_int: false, trailing_junk_start: 1 } },
+            ".." @ DotDot,
+            "10" @ Literal { kind: Int { base: Decimal, empty_int: false, trailing_junk_start: 2 } },
+            " " @ Whitespace,
+            "1" @ Literal { kind: Int { base: Decimal, empty_int: false, trailing_junk_start: 1 } },
+            " " @ Whitespace,
+            ".." @ DotDot,
+            " " @ Whitespace,
+            "10" @ Literal { kind: Int { base: Decimal, empty_int: false, trailing_junk_start: 2 } },
+            " " @ Whitespace,
+            "1.5" @ Literal { kind: Numeric { empty_exponent_start: None, trailing_junk_start: 3 } },
+            ".." @ DotDot,
+            "2" @ Literal { kind: Int { base: Decimal, empty_int: false, trailing_junk_start: 1 } },
+            " " @ Whitespace,
+            "1" @ Literal { kind: Int { base: Decimal, empty_int: false, trailing_junk_start: 1 } },
+            ".." @ DotDot,
+            ".2" @ Literal { kind: Numeric { empty_exponent_start: None, trailing_junk_start: 2 } },
+            " " @ Whitespace,
+            ".." @ DotDot,
+            " " @ Whitespace,
+            "a" @ Ident,
+            "[" @ OpenBracket,
+            "1" @ Literal { kind: Int { base: Decimal, empty_int: false, trailing_junk_start: 1 } },
+            ".." @ DotDot,
+            "2" @ Literal { kind: Int { base: Decimal, empty_int: false, trailing_junk_start: 1 } },
+            "]" @ CloseBracket,
+        ]
+        "#);
+    }
+
+    #[test]
+    fn dot_dot_does_not_disturb_neighboring_numerics() {
+        assert_debug_snapshot!(lex("1. .10 0 . .5 a.b 1.5"), @r#"
+        [
+            "1." @ Literal { kind: Numeric { empty_exponent_start: None, trailing_junk_start: 2 } },
+            " " @ Whitespace,
+            ".10" @ Literal { kind: Numeric { empty_exponent_start: None, trailing_junk_start: 3 } },
+            " " @ Whitespace,
+            "0" @ Literal { kind: Int { base: Decimal, empty_int: false, trailing_junk_start: 1 } },
+            " " @ Whitespace,
+            "." @ Dot,
+            " " @ Whitespace,
+            ".5" @ Literal { kind: Numeric { empty_exponent_start: None, trailing_junk_start: 2 } },
+            " " @ Whitespace,
+            "a" @ Ident,
+            "." @ Dot,
+            "b" @ Ident,
+            " " @ Whitespace,
+            "1.5" @ Literal { kind: Numeric { empty_exponent_start: None, trailing_junk_start: 3 } },
+        ]
+        "#);
+    }
+
     #[test]
     fn numeric_exponent_underscore_after_sign() {
         assert_debug_snapshot!(lex("1e+_2 1e-_2 1.0e+_2 .1e+_2"), @r#"

crates/squawk_lexer/src/token.rs

@@ -35,6 +35,8 @@ pub enum TokenKind {
     Colon,
     /// `.`
     Dot,
+    /// `..`
+    DotDot,
     /// `=`
     Eq,
     /// `>`