handle syntax errors more gracefully

src/parser.rs

@@ -34,11 +34,21 @@ pub enum ParseError {
     UnexpectedEOF,
     /// UnexpectedWanted is used when specific tokens are expected, but the end of file is reached
     UnexpectedEOFWanted(Box<[SyntaxKind]>),
+    Expected(TextRange, String),
 }
 
 impl fmt::Display for ParseError {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         match self {
+            ParseError::Expected(range, description) => {
+                write!(
+                    f,
+                    "expected {} at {}..{}",
+                    description,
+                    usize::from(range.start()),
+                    usize::from(range.end())
+                )
+            }
             ParseError::Unexpected(range) => {
                 write!(f, "error node at {}..{}", usize::from(range.start()), usize::from(range.end()))
             }
@@ -291,7 +301,15 @@ where
         match self.peek() {
             Some(TOKEN_DYNAMIC_START) => self.parse_dynamic(),
             Some(TOKEN_STRING_START) => self.parse_string(),
-            _ => self.expect_ident(),
+            Some(TOKEN_IDENT) => self.expect_ident(),
+            _ => {
+                let start = self.start_error_node();
+                let end = self.finish_error_node();
+                self.errors.push(ParseError::Expected(
+                    TextRange::new(start, end),
+                    "attribute".to_string(),
+                ));
+            }
         }
     }
     fn parse_attr(&mut self) {
@@ -391,7 +409,9 @@ where
                     self.parse_attr();
                     self.expect(TOKEN_ASSIGN);
                     self.parse_expr();
-                    self.expect(TOKEN_SEMICOLON);
+                    if let Some(TOKEN_SEMICOLON) = self.peek() {
+                        self.bump();
+                    }
                     self.finish_node();
                 }
             }
@@ -696,6 +716,16 @@ where
                 self.finish_node();
                 checkpoint
             }
+            Some(TOKEN_SEMICOLON) => {
+                let checkpoint = self.checkpoint();
+                let start = self.start_error_node();
+                let end = self.finish_error_node();
+                self.errors.push(ParseError::Expected(
+                    TextRange::new(start, end),
+                    "expression".to_string(),
+                ));
+                checkpoint
+            }
             _ => self.parse_math(),
         }
     }

test_data/general/error.expect

@@ -3,9 +3,7 @@ error: unexpected TOKEN_STRING_START at 78..98, wanted any of [TOKEN_ASSIGN]
 error: unexpected TOKEN_CURLY_B_CLOSE at 162..166, wanted any of [TOKEN_ASSIGN]
 error: unexpected end of file, wanted any of [TOKEN_ASSIGN]
 error: unexpected end of file
-error: unexpected end of file, wanted any of [TOKEN_SEMICOLON]
 error: unexpected end of file
-error: unexpected end of file, wanted any of [TOKEN_SEMICOLON]
 error: unexpected end of file
 NODE_ROOT 0..166 {
   NODE_ATTR_SET 0..166 {

test_data/parser/invalid_syntax/1.expect

@@ -1,10 +1,12 @@
 error: unexpected TOKEN_SQUARE_B_CLOSE at 0..1, wanted any of [TOKEN_PAREN_OPEN, TOKEN_REC, TOKEN_CURLY_B_OPEN, TOKEN_SQUARE_B_OPEN, TOKEN_DYNAMIC_START, TOKEN_STRING_START, TOKEN_IDENT]
-error: unexpected end of file, wanted any of [TOKEN_IDENT]
+error: expected attribute at 2..2
 NODE_ROOT 0..2 {
   NODE_SELECT 0..2 {
     NODE_ERROR 0..1 {
       TOKEN_SQUARE_B_CLOSE("]") 0..1
     }
     TOKEN_DOT(".") 1..2
+    NODE_ERROR 2..2 {
+    }
   }
 }

test_data/parser/invalid_syntax/4.expect

@@ -0,0 +1,47 @@
+error: expected expression at 12..12
+NODE_ROOT 0..40 {
+  NODE_BIN_OP 0..40 {
+    NODE_ATTR_SET 0..28 {
+      TOKEN_CURLY_B_OPEN("{") 0..1
+      TOKEN_WHITESPACE("\n    ") 1..6
+      NODE_KEY_VALUE 6..13 {
+        NODE_KEY 6..9 {
+          NODE_IDENT 6..9 {
+            TOKEN_IDENT("foo") 6..9
+          }
+        }
+        TOKEN_WHITESPACE(" ") 9..10
+        TOKEN_ASSIGN("=") 10..11
+        TOKEN_WHITESPACE(" ") 11..12
+        NODE_ERROR 12..12 {
+        }
+        TOKEN_SEMICOLON(";") 12..13
+      }
+      TOKEN_WHITESPACE("\n    ") 13..18
+      NODE_KEY_VALUE 18..26 {
+        NODE_KEY 18..21 {
+          NODE_IDENT 18..21 {
+            TOKEN_IDENT("bar") 18..21
+          }
+        }
+        TOKEN_WHITESPACE(" ") 21..22
+        TOKEN_ASSIGN("=") 22..23
+        TOKEN_WHITESPACE(" ") 23..24
+        NODE_LITERAL 24..25 {
+          TOKEN_INTEGER("1") 24..25
+        }
+        TOKEN_SEMICOLON(";") 25..26
+      }
+      TOKEN_WHITESPACE("\n") 26..27
+      TOKEN_CURLY_B_CLOSE("}") 27..28
+    }
+    TOKEN_WHITESPACE(" ") 28..29
+    TOKEN_UPDATE("//") 29..31
+    TOKEN_WHITESPACE(" ") 31..32
+    NODE_ATTR_SET 32..40 {
+      TOKEN_CURLY_B_OPEN("{") 32..33
+      TOKEN_WHITESPACE("\n    \n") 33..39
+      TOKEN_CURLY_B_CLOSE("}") 39..40
+    }
+  }
+}

test_data/parser/invalid_syntax/4.nix

@@ -0,0 +1,6 @@
+{
+    foo = ;
+    bar = 1;
+} // {
+
+}

test_data/parser/invalid_syntax/5.expect

@@ -0,0 +1,64 @@
+error: expected attribute at 37..37
+NODE_ROOT 0..57 {
+  NODE_ATTR_SET 0..57 {
+    TOKEN_REC("rec") 0..3
+    TOKEN_WHITESPACE(" ") 3..4
+    TOKEN_CURLY_B_OPEN("{") 4..5
+    TOKEN_WHITESPACE("\n    ") 5..10
+    NODE_KEY_VALUE 10..22 {
+      NODE_KEY 10..17 {
+        NODE_IDENT 10..13 {
+          TOKEN_IDENT("bar") 10..13
+        }
+        TOKEN_DOT(".") 13..14
+        NODE_IDENT 14..17 {
+          TOKEN_IDENT("xyz") 14..17
+        }
+      }
+      TOKEN_WHITESPACE(" ") 17..18
+      TOKEN_ASSIGN("=") 18..19
+      TOKEN_WHITESPACE(" ") 19..20
+      NODE_LITERAL 20..21 {
+        TOKEN_INTEGER("1") 20..21
+      }
+      TOKEN_SEMICOLON(";") 21..22
+    }
+    TOKEN_WHITESPACE("\n    ") 22..27
+    NODE_KEY_VALUE 27..38 {
+      NODE_KEY 27..30 {
+        NODE_IDENT 27..30 {
+          TOKEN_IDENT("foo") 27..30
+        }
+      }
+      TOKEN_WHITESPACE(" ") 30..31
+      TOKEN_ASSIGN("=") 31..32
+      TOKEN_WHITESPACE(" ") 32..33
+      NODE_SELECT 33..37 {
+        NODE_IDENT 33..36 {
+          TOKEN_IDENT("bar") 33..36
+        }
+        TOKEN_DOT(".") 36..37
+        NODE_ERROR 37..37 {
+        }
+      }
+      TOKEN_SEMICOLON(";") 37..38
+    }
+    TOKEN_WHITESPACE("\n    ") 38..43
+    NODE_KEY_VALUE 43..55 {
+      NODE_KEY 43..46 {
+        NODE_IDENT 43..46 {
+          TOKEN_IDENT("abc") 43..46
+        }
+      }
+      TOKEN_WHITESPACE(" ") 46..47
+      TOKEN_ASSIGN("=") 47..48
+      TOKEN_WHITESPACE(" ") 48..49
+      NODE_LITERAL 49..54 {
+        TOKEN_INTEGER("12345") 49..54
+      }
+      TOKEN_SEMICOLON(";") 54..55
+    }
+    TOKEN_WHITESPACE("\n") 55..56
+    TOKEN_CURLY_B_CLOSE("}") 56..57
+  }
+}

test_data/parser/invalid_syntax/5.nix

@@ -0,0 +1,5 @@
+rec {
+    bar.xyz = 1;
+    foo = bar.;
+    abc = 12345;
+}

test_data/parser/invalid_syntax/6.expect

@@ -0,0 +1,64 @@
+error: expected attribute at 37..37
+NODE_ROOT 0..57 {
+  NODE_ATTR_SET 0..57 {
+    TOKEN_REC("rec") 0..3
+    TOKEN_WHITESPACE(" ") 3..4
+    TOKEN_CURLY_B_OPEN("{") 4..5
+    TOKEN_WHITESPACE("\n    ") 5..10
+    NODE_KEY_VALUE 10..22 {
+      NODE_KEY 10..17 {
+        NODE_IDENT 10..13 {
+          TOKEN_IDENT("bar") 10..13
+        }
+        TOKEN_DOT(".") 13..14
+        NODE_IDENT 14..17 {
+          TOKEN_IDENT("xyz") 14..17
+        }
+      }
+      TOKEN_WHITESPACE(" ") 17..18
+      TOKEN_ASSIGN("=") 18..19
+      TOKEN_WHITESPACE(" ") 19..20
+      NODE_LITERAL 20..21 {
+        TOKEN_INTEGER("1") 20..21
+      }
+      TOKEN_SEMICOLON(";") 21..22
+    }
+    TOKEN_WHITESPACE("\n    ") 22..27
+    NODE_KEY_VALUE 27..38 {
+      NODE_KEY 27..30 {
+        NODE_IDENT 27..30 {
+          TOKEN_IDENT("foo") 27..30
+        }
+      }
+      TOKEN_WHITESPACE(" ") 30..31
+      TOKEN_ASSIGN("=") 31..32
+      TOKEN_WHITESPACE(" ") 32..33
+      NODE_SELECT 33..37 {
+        NODE_IDENT 33..36 {
+          TOKEN_IDENT("bar") 33..36
+        }
+        TOKEN_DOT(".") 36..37
+        NODE_ERROR 37..37 {
+        }
+      }
+      TOKEN_SEMICOLON(";") 37..38
+    }
+    TOKEN_WHITESPACE("\n    ") 38..43
+    NODE_KEY_VALUE 43..55 {
+      NODE_KEY 43..46 {
+        NODE_IDENT 43..46 {
+          TOKEN_IDENT("abc") 43..46
+        }
+      }
+      TOKEN_WHITESPACE(" ") 46..47
+      TOKEN_ASSIGN("=") 47..48
+      TOKEN_WHITESPACE(" ") 48..49
+      NODE_LITERAL 49..54 {
+        TOKEN_INTEGER("12345") 49..54
+      }
+      TOKEN_SEMICOLON(";") 54..55
+    }
+    TOKEN_WHITESPACE("\n") 55..56
+    TOKEN_CURLY_B_CLOSE("}") 56..57
+  }
+}

test_data/parser/invalid_syntax/6.nix

@@ -0,0 +1,5 @@
+rec {
+    bar.xyz = 1;
+    foo = bar.;
+    abc = 12345;
+}