fix: pragma parsing, add more tests

paradigmxyz · May 24, 2024 · 6c7623f · 6c7623f
1 parent 8f13a6f
commit 6c7623f
Show file tree

Hide file tree

Showing 12 changed files with 145 additions and 47 deletions.
diff --git a/crates/ast/src/token.rs b/crates/ast/src/token.rs
@@ -327,9 +327,11 @@ impl TokenKind {
         match self {
             Eq | Lt | Le | EqEq | Ne | Ge | Gt | AndAnd | OrOr | Not | Tilde | Walrus
             | PlusPlus | MinusMinus | StarStar | BinOp(_) | BinOpEq(_) | At | Dot | Comma
-            | Semi | Colon | Arrow | FatArrow | Question => true,
+            | Colon | Arrow | FatArrow | Question => true,
 
-            OpenDelim(..) | CloseDelim(..) | Literal(..) | Comment(..) | Ident(..) | Eof => false,
+            OpenDelim(..) | CloseDelim(..) | Literal(..) | Comment(..) | Ident(..) | Semi | Eof => {
+                false
+            }
         }
     }
 

diff --git a/crates/parse/src/lexer/cursor/mod.rs b/crates/parse/src/lexer/cursor/mod.rs
@@ -265,20 +265,20 @@ impl<'a> Cursor<'a> {
             // by field/method access (`12.foo()`)
             '.' if !is_id_start(self.second()) => {
                 self.bump();
-                let empty_fraction = !self.eat_decimal_digits();
+                self.eat_decimal_digits();
                 let empty_exponent = match self.first() {
                     'e' | 'E' => {
                         self.bump();
                         !self.eat_exponent()
                     }
                     _ => false,
                 };
-                RawLiteralKind::Rational { base, empty_fraction, empty_exponent }
+                RawLiteralKind::Rational { base, empty_exponent }
             }
             'e' | 'E' => {
                 self.bump();
                 let empty_exponent = !self.eat_exponent();
-                RawLiteralKind::Rational { base, empty_fraction: false, empty_exponent }
+                RawLiteralKind::Rational { base, empty_exponent }
             }
             _ => RawLiteralKind::Int { base, empty_int: false },
         }

diff --git a/crates/parse/src/lexer/cursor/token.rs b/crates/parse/src/lexer/cursor/token.rs
@@ -122,8 +122,8 @@ pub enum RawTokenKind {
 pub enum RawLiteralKind {
     /// `123`, `0x123`; empty_int: `0x`
     Int { base: Base, empty_int: bool },
-    /// `123.321`, `1.2e3`; empty_fraction: `1.`, `1.e2`; empty_exponent: `2e`, `2.3e`
-    Rational { base: Base, empty_fraction: bool, empty_exponent: bool },
+    /// `123.321`, `1.2e3`; empty_exponent: `2e`, `2.3e`
+    Rational { base: Base, empty_exponent: bool },
     /// `"abc"`, `"abc`; `unicode"abc"`, `unicode"abc`
     Str { terminated: bool, unicode: bool },
     /// `hex"abc"`, `hex"abc`

diff --git a/crates/parse/src/lexer/mod.rs b/crates/parse/src/lexer/mod.rs
@@ -345,12 +345,7 @@ impl<'sess, 'src> Lexer<'sess, 'src> {
                     (TokenLitKind::Integer, self.symbol_from_to(start, end))
                 }
             }
-            RawLiteralKind::Rational { base, empty_fraction, empty_exponent } => {
-                if empty_fraction {
-                    let span = self.new_span(start, self.pos);
-                    self.dcx().err("expected at least one digit in fraction").span(span).emit();
-                }
-
+            RawLiteralKind::Rational { base, empty_exponent } => {
                 if empty_exponent {
                     let span = self.new_span(start, self.pos);
                     self.dcx().err("expected at least one digit in exponent").span(span).emit();

diff --git a/crates/parse/src/parser/item.rs b/crates/parse/src/parser/item.rs
@@ -1,8 +1,9 @@
 use super::{ExpectedToken, SeqSep};
 use crate::{PResult, Parser};
 use itertools::Itertools;
+use std::num::IntErrorKind;
 use sulk_ast::{ast::*, token::*};
-use sulk_interface::{error_code, kw, sym, Ident};
+use sulk_interface::{error_code, kw, sym, Ident, Span};
 
 impl<'a> Parser<'a> {
     /// Parses a source unit.
@@ -500,8 +501,8 @@ impl<'a> Parser<'a> {
             // 0.1 .2
             let lit = self.token.lit().unwrap();
             let (mj, mn) = lit.symbol.as_str().split_once('.').unwrap();
-            major = SemverVersionNumber::Number(self.parse_u32(mj)?);
-            minor = Some(SemverVersionNumber::Number(self.parse_u32(mn)?));
+            major = SemverVersionNumber::Number(self.parse_u32(mj, self.token.span));
+            minor = Some(SemverVersionNumber::Number(self.parse_u32(mn, self.token.span)));
             self.bump();
 
             patch =
@@ -514,8 +515,8 @@ impl<'a> Parser<'a> {
                     // *. 1.2
                     let lit = self.token.lit().unwrap();
                     let (mn, p) = lit.symbol.as_str().split_once('.').unwrap();
-                    minor = Some(SemverVersionNumber::Number(self.parse_u32(mn)?));
-                    patch = Some(SemverVersionNumber::Number(self.parse_u32(p)?));
+                    minor = Some(SemverVersionNumber::Number(self.parse_u32(mn, self.token.span)));
+                    patch = Some(SemverVersionNumber::Number(self.parse_u32(p, self.token.span)));
                     self.bump();
                 } else {
                     // *.1 .2
@@ -548,13 +549,22 @@ impl<'a> Parser<'a> {
             self.expected_tokens.push(ExpectedToken::VersionNumber);
             return self.unexpected();
         };
-        let value = self.parse_u32(symbol.as_str()).map_err(|e| e.span(span))?;
+        let value = self.parse_u32(symbol.as_str(), span);
         self.bump();
         Ok(SemverVersionNumber::Number(value))
     }
 
-    fn parse_u32(&mut self, s: &str) -> PResult<'a, u32> {
-        s.parse::<u32>().map_err(|e| self.dcx().err(e.to_string()))
+    fn parse_u32(&mut self, s: &str, span: Span) -> u32 {
+        match s.parse::<u32>() {
+            Ok(n) => n,
+            Err(e) => match e.kind() {
+                IntErrorKind::Empty => 0,
+                _ => {
+                    self.dcx().err(e.to_string()).span(span).emit();
+                    u32::MAX
+                }
+            },
+        }
     }
 
     /// Parses an import directive.

diff --git a/crates/sema/src/ast_validation.rs b/crates/sema/src/ast_validation.rs
@@ -40,9 +40,7 @@ impl<'ast, 'sess> Visit<'ast> for AstValidator<'sess> {
                 if name.name != sym::solidity {
                     let msg = "only `solidity` is supported as a version pragma";
                     self.dcx().err(msg).span(name.span).emit();
-                    // return;
                 }
-                // TODO: Check or ignore version?
             }
             ast::PragmaTokens::Custom(name, value) => {
                 let name = name.as_str();

diff --git a/tests/ui/parser/pragma_unknown.sol b/tests/ui/parser/pragma_unknown.sol
@@ -0,0 +1,11 @@
+
+pragma foo bar;
+//~^ ERROR unknown pragma
+pragma amogus;
+//~^ ERROR unknown pragma
+pragma amogus sus;
+//~^ ERROR unknown pragma
+pragma amogus 69;
+//~^ ERROR unknown pragma
+pragma amogus 69 diwqbn9ru3b2q945 390ru31290r 0qjr09wadm;
+//~^ ERROR unknown pragma
diff --git a/tests/ui/parser/pragma_unknown.stderr b/tests/ui/parser/pragma_unknown.stderr
@@ -0,0 +1,37 @@
+error: unknown pragma
+  --> $DIR/pragma_unknown.sol:2:1
+   |
+LL | pragma foo bar;
+   | ^^^^^^^^^^^^^^^
+   |
+
+error: unknown pragma
+  --> $DIR/pragma_unknown.sol:4:1
+   |
+LL | pragma amogus;
+   | ^^^^^^^^^^^^^^
+   |
+
+error: unknown pragma
+  --> $DIR/pragma_unknown.sol:6:1
+   |
+LL | pragma amogus sus;
+   | ^^^^^^^^^^^^^^^^^^
+   |
+
+error: unknown pragma
+  --> $DIR/pragma_unknown.sol:8:1
+   |
+LL | pragma amogus 69;
+   | ^^^^^^^^^^^^^^^^^
+   |
+
+error: unknown pragma
+  --> $DIR/pragma_unknown.sol:10:1
+   |
+LL | pragma amogus 69 diwqbn9ru3b2q945 390ru31290r 0qjr09wadm;
+   | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+   |
+
+error: aborting due to 5 previous errors
+
diff --git a/tests/ui/parser/pragma_valid.sol b/tests/ui/parser/pragma_valid.sol
@@ -0,0 +1,21 @@
+pragma abicoder v1;
+pragma abicoder v2;
+pragma abicoder "v1";
+pragma abicoder "v2";
+
+// These aren't accepted by solc.
+pragma "abicoder" v1;
+pragma "abicoder" v2;
+pragma "abicoder" "v1";
+pragma "abicoder" "v2";
+
+pragma experimental ABIEncoderV2;
+pragma experimental "ABIEncoderV2";
+pragma experimental SMTChecker;
+pragma experimental "SMTChecker";
+
+// These aren't accepted by solc.
+pragma "experimental" ABIEncoderV2;
+pragma "experimental" "ABIEncoderV2";
+pragma "experimental" SMTChecker;
+pragma "experimental" "SMTChecker";
diff --git a/tests/ui/parser/pragmas.sol → tests/ui/parser/pragma_versions.sol b/tests/ui/parser/pragmas.sol → tests/ui/parser/pragma_versions.sol
@@ -1,4 +1,3 @@
-// Commented out are lexed as empty literals "0."
 pragma solidity *.*.*;
 pragma solidity *.*.0 ;
 pragma solidity *.*.0;
@@ -7,7 +6,7 @@ pragma solidity *.0 .*;
 pragma solidity *.0 .0 ;
 pragma solidity *.0 .0;
 pragma solidity *.0 ;
-// pragma solidity *.0.*;
+pragma solidity *.0.*;
 pragma solidity *.0.0 ;
 pragma solidity *.0.0;
 pragma solidity *.0;
@@ -20,15 +19,15 @@ pragma solidity 0 .0 .*;
 pragma solidity 0 .0 .0 ;
 pragma solidity 0 .0 .0;
 pragma solidity 0 .0 ;
-// pragma solidity 0 .0.*;
+pragma solidity 0 .0.*;
 pragma solidity 0 .0.0 ;
 pragma solidity 0 .0.0;
 pragma solidity 0 .0;
 pragma solidity 0 ;
-// pragma solidity 0.*.*;
-// pragma solidity 0.*.0 ;
-// pragma solidity 0.*.0;
-// pragma solidity 0.*;
+pragma solidity 0.*.*;
+pragma solidity 0.*.0 ;
+pragma solidity 0.*.0;
+pragma solidity 0.*;
 pragma solidity 0.0 .*;
 pragma solidity 0.0 .0 ;
 pragma solidity 0.0 .0;
@@ -39,13 +38,17 @@ pragma solidity 0.0.0;
 pragma solidity 0.0;
 pragma solidity 0;
 
-// pragma foo bar;
-// ~^ ERROR unknown pragma
-pragma abicoder v2;
-// pragma solidity ^4294967295;
-// ~^ ERROR too large
-pragma solidity 0 - 1 0 - 2;
-//~^ ERROR ranges can only be combined using the || operator
-pragma abicoder "v2";
 pragma solidity ^0.5.16 =0.8.22 || >=0.8.21 <=2 ~1 0.6.2;
 pragma solidity 0.4 - 1 || 0.3 - 0.5.16;
+
+pragma solidity ^4294967295;
+pragma solidity ^4294967296;
+//~^ ERROR too large
+pragma solidity ^0.4294967296;
+//~^ ERROR too large
+
+pragma solidity 88_;
+//~^ ERROR invalid digit
+
+pragma solidity 0 - 1 0 - 2;
+//~^ ERROR ranges can only be combined using the || operator
diff --git a/tests/ui/parser/pragma_versions.stderr b/tests/ui/parser/pragma_versions.stderr
@@ -0,0 +1,30 @@
+error: number too large to fit in target type
+  --> $DIR/pragma_versions.sol:45:18
+   |
+LL | pragma solidity ^4294967296;
+   |                  ^^^^^^^^^^
+   |
+
+error: number too large to fit in target type
+  --> $DIR/pragma_versions.sol:47:18
+   |
+LL | pragma solidity ^0.4294967296;
+   |                  ^^^^^^^^^^^^
+   |
+
+error: invalid digit found in string
+  --> $DIR/pragma_versions.sol:50:17
+   |
+LL | pragma solidity 88_;
+   |                 ^^^
+   |
+
+error: ranges can only be combined using the || operator
+  --> $DIR/pragma_versions.sol:53:1
+   |
+LL | pragma solidity 0 - 1 0 - 2;
+   | ^^^^^^^^^^^^^^^^^^^^^
+   |
+
+error: aborting due to 4 previous errors
+
diff --git a/tests/ui/parser/pragmas.stderr b/tests/ui/parser/pragmas.stderr