From 2876a11c329f92d83c334b95c4cb6704357dc21b Mon Sep 17 00:00:00 2001
From: Tilak Madichetti <tilak.madichetti@gmail.com>
Date: Thu, 12 Dec 2024 22:29:54 +0530
Subject: [PATCH] feat: underscores and literals validation (#165)

* feat: remove trailing underscores

* literals

* fix

* fix: remove print statement

* fix: suggestions

* demo expr recursed

* Update crates/sema/src/ast_passes.rs

---------

Co-authored-by: DaniPopes <57450786+DaniPopes@users.noreply.github.com>
---
 crates/sema/src/ast_passes.rs                | 45 ++++++++++++---
 tests/ui/resolve/literals_underscores.sol    | 14 +++++
 tests/ui/resolve/literals_underscores.stderr | 58 ++++++++++++++++++++
 3 files changed, 110 insertions(+), 7 deletions(-)
 create mode 100644 tests/ui/resolve/literals_underscores.sol
 create mode 100644 tests/ui/resolve/literals_underscores.stderr

diff --git a/crates/sema/src/ast_passes.rs b/crates/sema/src/ast_passes.rs
index cc007f47..58e77ac8 100644
--- a/crates/sema/src/ast_passes.rs
+++ b/crates/sema/src/ast_passes.rs
@@ -60,6 +60,38 @@ impl<'sess> AstValidator<'sess, '_> {
                 .emit();
         }
     }
+
+    fn check_underscores_in_number_literals(&self, lit: &ast::Lit) {
+        let ast::LitKind::Number(_) = lit.kind else {
+            return;
+        };
+        let literal_span = lit.span;
+        let literal_str = lit.symbol.as_str();
+
+        let error_help_msg = {
+            if literal_str.ends_with('_') {
+                Some("remove trailing underscores")
+            } else if literal_str.contains("__") {
+                Some("only 1 consecutive underscore `_` is allowed between digits")
+            } else if literal_str.contains("._") || literal_str.contains("_.") {
+                Some("remove underscores in front of the fraction part")
+            } else if literal_str.contains("_e") {
+                Some("remove underscores at the end of the mantissa")
+            } else if literal_str.contains("e_") {
+                Some("remove underscores in front of the exponent")
+            } else {
+                None
+            }
+        };
+
+        if let Some(error_help_msg) = error_help_msg {
+            self.dcx()
+                .err("invalid use of underscores in number literal")
+                .span(literal_span)
+                .help(error_help_msg)
+                .emit();
+        }
+    }
 }
 
 impl<'ast> Visit<'ast> for AstValidator<'_, 'ast> {
@@ -329,12 +361,11 @@ impl<'ast> Visit<'ast> for AstValidator<'_, 'ast> {
         self.walk_using_directive(using)
     }
 
-    // Intentionally override unused default implementations to reduce bloat.
-    fn visit_expr(&mut self, _expr: &'ast ast::Expr<'ast>) -> ControlFlow<Self::BreakValue> {
-        ControlFlow::Continue(())
-    }
-
-    fn visit_ty(&mut self, _ty: &'ast ast::Type<'ast>) -> ControlFlow<Self::BreakValue> {
-        ControlFlow::Continue(())
+    fn visit_expr(&mut self, expr: &'ast ast::Expr<'ast>) -> ControlFlow<Self::BreakValue> {
+        let ast::Expr { kind, .. } = expr;
+        if let ast::ExprKind::Lit(lit, _) = kind {
+            self.check_underscores_in_number_literals(lit);
+        }
+        self.walk_expr(expr)
     }
 }
diff --git a/tests/ui/resolve/literals_underscores.sol b/tests/ui/resolve/literals_underscores.sol
new file mode 100644
index 00000000..99270d26
--- /dev/null
+++ b/tests/ui/resolve/literals_underscores.sol
@@ -0,0 +1,14 @@
+contract LT {
+    uint256 a = 1000_; //~ERROR: invalid use of underscores in number literal
+    uint256 b = 100__0; //~ERROR: invalid use of underscores in number literal
+    uint256 c = 1_.4e10; //~ERROR: invalid use of underscores in number literal
+    uint256 d = 3.4_e10; //~ERROR: invalid use of underscores in number literal
+    uint256 e = 3.4e_10; //~ERROR: invalid use of underscores in number literal
+
+    // exception
+    uint256 f = 3._4e10; // Does not show up
+
+    uint256 g = 1_.4e10 + 3.4e_10; //~ERROR: invalid use of underscores in number literal
+    //~^ERROR: invalid use of underscores in number literal
+
+}
diff --git a/tests/ui/resolve/literals_underscores.stderr b/tests/ui/resolve/literals_underscores.stderr
new file mode 100644
index 00000000..1a13d92c
--- /dev/null
+++ b/tests/ui/resolve/literals_underscores.stderr
@@ -0,0 +1,58 @@
+error: invalid use of underscores in number literal
+  --> ROOT/tests/ui/resolve/literals_underscores.sol:LL:CC
+   |
+LL |     uint256 a = 1000_;
+   |                 ^^^^^
+   |
+   = help: remove trailing underscores
+
+error: invalid use of underscores in number literal
+  --> ROOT/tests/ui/resolve/literals_underscores.sol:LL:CC
+   |
+LL |     uint256 b = 100__0;
+   |                 ^^^^^^
+   |
+   = help: only 1 consecutive underscore `_` is allowed between digits
+
+error: invalid use of underscores in number literal
+  --> ROOT/tests/ui/resolve/literals_underscores.sol:LL:CC
+   |
+LL |     uint256 c = 1_.4e10;
+   |                 ^^^^^^^
+   |
+   = help: remove underscores in front of the fraction part
+
+error: invalid use of underscores in number literal
+  --> ROOT/tests/ui/resolve/literals_underscores.sol:LL:CC
+   |
+LL |     uint256 d = 3.4_e10;
+   |                 ^^^^^^^
+   |
+   = help: remove underscores at the end of the mantissa
+
+error: invalid use of underscores in number literal
+  --> ROOT/tests/ui/resolve/literals_underscores.sol:LL:CC
+   |
+LL |     uint256 e = 3.4e_10;
+   |                 ^^^^^^^
+   |
+   = help: remove underscores in front of the exponent
+
+error: invalid use of underscores in number literal
+  --> ROOT/tests/ui/resolve/literals_underscores.sol:LL:CC
+   |
+LL |     uint256 g = 1_.4e10 + 3.4e_10;
+   |                 ^^^^^^^
+   |
+   = help: remove underscores in front of the fraction part
+
+error: invalid use of underscores in number literal
+  --> ROOT/tests/ui/resolve/literals_underscores.sol:LL:CC
+   |
+LL |     uint256 g = 1_.4e10 + 3.4e_10;
+   |                           ^^^^^^^
+   |
+   = help: remove underscores in front of the exponent
+
+error: aborting due to 7 previous errors
+