Improve string literal parsing and add unit tests

config/syntax.toml

@@ -12,7 +12,7 @@ atom_patterns = [
 ]
 string_patterns = [
   # Double-quoted strings
-  '"((\\")|[^"])*"',
+  '"((\\.)|[^"])*"',
   # Single quoted strings
   "'((\\\\')|[^'])*'",
   # Backtick strings
@@ -57,7 +57,7 @@ atom_patterns = [
 ]
 string_patterns = [
   # Double-quoted strings
-  '"((\\")|[^"])*"',
+  '"((\\.)|[^"])*"',
   # Single quoted strings
   "'((\\\\')|[^'])*'",
 ]
@@ -80,7 +80,7 @@ atom_patterns = [
 ]
 string_patterns = [
   # Double-quoted strings
-  '"((\\")|[^"])*"',
+  '"((\\.)|[^"])*"',
 ]
 comment_patterns = [
   ';.*',
@@ -123,7 +123,7 @@ atom_patterns = [
 ]
 string_patterns = [
   # Double-quoted strings
-  '"((\\")|[^"])*"',
+  '"((\\.)|[^"])*"',
 ]
 comment_patterns = [
   # Mult-line comments
@@ -146,7 +146,7 @@ atom_patterns = [
 ]
 string_patterns = [
   # Double-quoted strings
-  '"((\\")|[^"])*"',
+  '"((\\.)|[^"])*"',
 ]
 comment_patterns = [
   ';.*',
@@ -176,7 +176,7 @@ atom_patterns = [
 ]
 string_patterns = [
   # Double-quoted strings
-  '"((\\")|[^"])*"',
+  '"((\\.)|[^"])*"',
 ]
 comment_patterns = [
   # Single line comments

src/parse.rs

@@ -470,6 +470,71 @@ mod tests {
         );
     }
 
+    #[test]
+    fn test_parse_string_escaped_doublequote() {
+        let arena = Arena::new();
+
+        assert_syntaxes(
+            // "\""
+            &parse(&arena, "\"\\\"\"", &lang()),
+            &[Syntax::new_atom(
+                &arena,
+                vec![SingleLineSpan {
+                    line: 0.into(),
+                    start_col: 0,
+                    end_col: 4,
+                }],
+                "\"\\\"\"",
+            )],
+        );
+    }
+
+    #[test]
+    fn test_parse_string_escaped_backlash() {
+        let arena = Arena::new();
+
+        assert_syntaxes(
+            // "\\"
+            &parse(&arena, "\"\\\\\"", &lang()),
+            &[Syntax::new_atom(
+                &arena,
+                vec![SingleLineSpan {
+                    line: 0.into(),
+                    start_col: 0,
+                    end_col: 4,
+                }],
+                "\"\\\\\"",
+            )],
+        );
+    }
+
+    #[test]
+    fn test_parse_string_escaped_backlash_and_second_string() {
+        let arena = Arena::new();
+
+        assert_syntaxes(
+            // "\\" "a"
+            &parse(&arena, "\"\\\\\" \"a\"", &lang()),
+            &[Syntax::new_atom(
+                &arena,
+                vec![SingleLineSpan {
+                    line: 0.into(),
+                    start_col: 0,
+                    end_col: 4,
+                }],
+                "\"\\\\\"",
+            ), Syntax::new_atom(
+                &arena,
+                vec![SingleLineSpan {
+                    line: 0.into(),
+                    start_col: 5,
+                    end_col: 8,
+                }],
+                "\"a\"",
+            )],
+        );
+    }
+
     #[test]
     fn test_parse_multiple() {
         let arena = Arena::new();