Parser errors on malformed C# code different in playground
nibab opened this issue · 2 comments
While the following malformed code
using System; class Program { static void Main() { Console.WriteLine(\"Hello, World); } }
works fine in the playground (ie the error occurs much lower in the tree) https://tree-sitter.github.io/tree-sitter/playground. see:
compilation_unit [0, 0] - [1, 0]
using_directive [0, 0] - [0, 13]
identifier [0, 6] - [0, 12]
class_declaration [0, 14] - [0, 89]
name: identifier [0, 20] - [0, 27]
body: declaration_list [0, 28] - [0, 89]
method_declaration [0, 30] - [0, 87]
modifier [0, 30] - [0, 36]
type: void_keyword [0, 37] - [0, 41]
name: identifier [0, 42] - [0, 46]
parameters: parameter_list [0, 46] - [0, 48]
body: block [0, 49] - [0, 87]
expression_statement [0, 51] - [0, 85]
invocation_expression [0, 51] - [0, 84]
function: member_access_expression [0, 51] - [0, 68]
expression: identifier [0, 51] - [0, 58]
name: identifier [0, 59] - [0, 68]
arguments: argument_list [0, 68] - [0, 84]
ERROR [0, 69] - [0, 71]
escape_sequence [0, 69] - [0, 71]
argument [0, 71] - [0, 76]
identifier [0, 71] - [0, 76]
argument [0, 78] - [0, 83]
identifier [0, 78] - [0, 83]
this is what the grammar file from tree-sitter-c-sharp-0.20.0 outputs (ie error parsing the method definition):
compilation_unit: [0] - [88]
using_directive: [0] - [13]
using: [0] - [5]
identifier: [6] - [12]
;: [12] - [13]
ERROR: [14] - [88]
class: [14] - [19]
identifier: [20] - [27]
{: [28] - [29]
modifier: [30] - [36]
static: [30] - [36]
predefined_type: [37] - [41]
identifier: [42] - [46]
parameter_list: [46] - [48]
(: [46] - [47]
): [47] - [48]
{: [49] - [50]
member_access_expression: [51] - [68]
identifier: [51] - [58]
.: [58] - [59]
identifier: [59] - [68]
(: [68] - [69]
": [69] - [70]
string_literal_fragment: [70] - [88]
im curious why that is. it is obviously difficult to know what grammar the playground is using and it's possible that this is some old grammar, although I would argue that being able to parse the method declaration is significantly better.
for reference, searching the method declaration with roslyn through:
public class MethodWalker : CSharpSyntaxWalker
{
public List<MethodData> Methods { get; } = new List<MethodData>();
public override void VisitMethodDeclaration(MethodDeclarationSyntax node)
{
string methodName = node.Identifier.ValueText;
string methodBody = node.Body?.ToString();
Methods.Add(new MethodData(methodName, methodBody));
base.VisitMethodDeclaration(node);
}
}will also find the method declaration.
I tried parsing this input with the latest version of the grammar in the master branch, and I got the following. The tree seems to match the one returned by the playground.
❯ tree-sitter parse a.cs
(compilation_unit [0, 0] - [0, 89]
(using_directive [0, 0] - [0, 13]
name: (identifier [0, 6] - [0, 12]))
(class_declaration [0, 14] - [0, 89]
name: (identifier [0, 20] - [0, 27])
body: (declaration_list [0, 28] - [0, 89]
(method_declaration [0, 30] - [0, 87]
(modifier [0, 30] - [0, 36])
type: (predefined_type [0, 37] - [0, 41])
name: (identifier [0, 42] - [0, 46])
parameters: (parameter_list [0, 46] - [0, 48])
body: (block [0, 49] - [0, 87]
(expression_statement [0, 51] - [0, 85]
(invocation_expression [0, 51] - [0, 84]
function: (member_access_expression [0, 51] - [0, 68]
expression: (identifier [0, 51] - [0, 58])
name: (identifier [0, 59] - [0, 68]))
arguments: (argument_list [0, 68] - [0, 84]
(ERROR [0, 69] - [0, 71]
(escape_sequence [0, 69] - [0, 71]))
(argument [0, 71] - [0, 76]
(identifier [0, 71] - [0, 76]))
(argument [0, 78] - [0, 83]
(identifier [0, 78] - [0, 83]))))))))))
a.cs 0 ms (ERROR [0, 69] - [0, 71])
We should probably publish a new version. I can do the GitHub and npm side but as for the Rust crate... 🤷🏻