From a01ed981ceba0b05d68038c6c7a9d535ca115a67 Mon Sep 17 00:00:00 2001
From: ConnorSkees <39542938+ConnorSkees@users.noreply.github.com>
Date: Sun, 24 May 2020 07:43:54 -0400
Subject: [PATCH] make parsing of toplevel tokens more robust

---
 src/lib.rs            |  7 +++++++
 src/stylesheet.rs     | 45 ++++++++++++++++++++++++++++---------------
 tests/error.rs        | 34 +++++++++++++++++++++++---------
 tests/get-function.rs |  2 +-
 tests/selectors.rs    | 11 +++++++++++
 5 files changed, 73 insertions(+), 26 deletions(-)

diff --git a/src/lib.rs b/src/lib.rs
index d34c7f5..dfc500a 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -400,5 +400,12 @@ pub(crate) fn eat_expr<I: Iterator<Item = Token>>(
             _ => values.push(toks.next().unwrap()),
         };
     }
+
+    // if `values` is not empty, there was an unexpected toplevel token
+    // that should be part of a selector
+    if let Some(v) = values.pop() {
+        return Err(("expected \"{\".", v.pos).into());
+    }
+
     Ok(None)
 }
diff --git a/src/stylesheet.rs b/src/stylesheet.rs
index 550a19b..c5df0a8 100644
--- a/src/stylesheet.rs
+++ b/src/stylesheet.rs
@@ -151,23 +151,40 @@ struct StyleSheetParser<'a> {
     path: &'a Path,
 }
 
-fn is_selector_char(c: char) -> bool {
-    c.is_alphanumeric()
-        || matches!(
-            c,
-            '_' | '-' | '[' | '#' | ':' | '*' | '%' | '.' | '>' | '\\' | '+'
-        )
-}
-
 impl<'a> StyleSheetParser<'a> {
     fn parse_toplevel(mut self) -> SassResult<(Vec<Spanned<Stmt>>, Scope)> {
         let mut rules: Vec<Spanned<Stmt>> = Vec::new();
+        devour_whitespace(self.lexer);
         while let Some(Token { kind, .. }) = self.lexer.peek() {
             match kind {
-                _ if is_selector_char(*kind) => {
+                'a'..='z'
+                | 'A'..='Z'
+                | '0'..='9'
+                | '['
+                | '\\'
+                | ']'
+                | '^'
+                | '_'
+                | '-'
+                | '#'
+                | ':'
+                | '*'
+                | '%'
+                | '.'
+                | '>'
+                | '+'
+                | '='
+                | ','
+                | '('
+                | ')'
+                | '<'
+                | '?'
+                | '~'
+                | '|'
+                | '\u{7f}'..=std::char::MAX => {
                     rules.extend(self.eat_rules(&Selector::new(), &mut Scope::new())?)
                 }
-                '\t' | '\n' | ' ' => {
+                '\t' | '\n' | ' ' | ';' => {
                     self.lexer.next();
                     continue;
                 }
@@ -331,20 +348,16 @@ impl<'a> StyleSheetParser<'a> {
                     )
                         .into())
                 }
-                c if c.is_control() => {
+                '\u{0}'..='\u{8}' | '\u{b}'..='\u{1f}' => {
                     return Err(("expected selector.", self.lexer.next().unwrap().pos).into());
                 }
-                ',' | '!' | '(' | ')' => {
-                    return Err(("expected \"{\".", self.lexer.next().unwrap().pos).into());
-                }
-                '{' => {
+                '{' | '!' => {
                     return Err(("expected \"}\".", self.lexer.next().unwrap().pos).into());
                 }
                 '`' | '\'' | '"' => {
                     return Err(("expected selector.", self.lexer.next().unwrap().pos).into());
                 }
                 '}' => return Err(("unmatched \"}\".", self.lexer.next().unwrap().pos).into()),
-                _ => todo!("unexpected toplevel token: {:?}", kind),
             };
         }
         Ok((rules, GLOBAL_SCOPE.with(|s| s.borrow().clone())))
diff --git a/tests/error.rs b/tests/error.rs
index d4ea7a6..04dec37 100644
--- a/tests/error.rs
+++ b/tests/error.rs
@@ -83,20 +83,15 @@ error!(
     "a {$a", "Error: expected \":\"."
 );
 error!(toplevel_comma, "a {},", "Error: expected \"{\".");
-error!(toplevel_exclamation, "! {}", "Error: expected \"{\".");
+error!(toplevel_exclamation_alone, "!", "Error: expected \"}\".");
+error!(toplevel_exclamation, "! {}", "Error: expected \"}\".");
 error!(toplevel_backtick, "` {}", "Error: expected selector.");
 error!(
     toplevel_open_curly_brace,
     "{ {color: red;}", "Error: expected \"}\"."
 );
-error!(
-    toplevel_open_paren,
-    "(", "Error: expected \"{\"."
-);
-error!(
-    toplevel_close_paren,
-    "(", "Error: expected \"{\"."
-);
+error!(toplevel_open_paren, "(", "Error: expected \"{\".");
+error!(toplevel_close_paren, "(", "Error: expected \"{\".");
 error!(
     backtick_in_value,
     "a {color:`red;}", "Error: Expected expression."
@@ -184,3 +179,24 @@ error!(
     nothing_after_gt,
     "a {color: 1 >", "Error: Expected expression."
 );
+error!(toplevel_eq_alone, "=", "Error: expected \"{\".");
+error!(toplevel_gt_alone, ">", "Error: expected \"{\".");
+error!(toplevel_lt_alone, "<", "Error: expected \"{\".");
+error!(toplevel_question_alone, "?", "Error: expected \"{\".");
+error!(toplevel_caret_alone, "^", "Error: expected \"{\".");
+test!(toplevel_gt_as_selector, "> {}", "");
+test!(toplevel_tilde_as_selector, "~ {}", "");
+error!(toplevel_lt_as_selector, "< {}", "Error: expected selector.");
+error!(
+    toplevel_question_as_selector,
+    "? {}", "Error: expected selector."
+);
+error!(
+    toplevel_caret_as_selector,
+    "^ {}", "Error: expected selector."
+);
+error!(toplevel_eq, "= {}", "Error: expected selector.");
+error!(value_after_style, "a {}a", "Error: expected \"{\".");
+test!(whitespace_after_style, "a {}\t\n ", "");
+test!(toplevel_semicolon, ";", "");
+test!(toplevel_semicolon_after_style, "a {};", "");
diff --git a/tests/get-function.rs b/tests/get-function.rs
index b752665..91da744 100644
--- a/tests/get-function.rs
+++ b/tests/get-function.rs
@@ -15,7 +15,7 @@ test!(
 test!(
     same_function_equal,
     "@function user-defined() {@return null}
-    a {b: get-function(user-defined) == get-function(user-defined)}s",
+    a {b: get-function(user-defined) == get-function(user-defined)}",
     "a {\n  b: true;\n}\n"
 );
 test!(
diff --git a/tests/selectors.rs b/tests/selectors.rs
index 70b70ed..606827c 100644
--- a/tests/selectors.rs
+++ b/tests/selectors.rs
@@ -530,3 +530,14 @@ test!(
     "+ {\n  color: &;\n}\n",
     "+ {\n  color: +;\n}\n"
 );
+error!(
+    #[ignore = "namespaces are not yet parsed correctly"]
+    empty_namespace,
+    "| {}", "Error: Expected identifier."
+);
+test!(
+    #[ignore = "namespaces are not yet parsed correctly"]
+    simple_namespace,
+    "|f {\n  color: &;\n}\n",
+    "|f {\n  color: |f;\n}\n"
+);