translator/parse.py

from tree_sitter import Parser
# tree-sitter-c-sharp pip package provides the C# language.
from tree_sitter_languages import get_language  # helper wrapper (optional)
import re



def parse_csharp(code: str) -> dict:
    """
    Parses a C# script and returns a simple AST-like dict:
    {
        "classes": [
            {
                "name": "Foo",
                "methods": [
                    {"name": "Update", "body": "..."}
                ]
            }
        ]
    }
    """
    result = {"classes": []}

    try:
        # try tree-sitter approach
        from tree_sitter import Parser
        from tree_sitter_c_sharp import CSHARP_LANGUAGE
        parser = Parser()
        parser.set_language(CSHARP_LANGUAGE)
        tree = parser.parse(bytes(code, "utf8"))
        root = tree.root_node

        def traverse(node):
            classes = []
            for child in node.children:
                if child.type == "class_declaration":
                    class_name = None
                    methods = []
                    # get class name
                    for grand in child.children:
                        if grand.type == "identifier":
                            class_name = code[grand.start_byte:grand.end_byte]
                        if grand.type == "method_declaration":
                            method_name = None
                            body = ""
                            for g in grand.children:
                                if g.type == "identifier":
                                    method_name = code[g.start_byte:g.end_byte]
                                if g.type == "block":
                                    body = code[g.start_byte:g.end_byte]
                            methods.append({"name": method_name, "body": body})
                    classes.append({"name": class_name, "methods": methods})
                # recurse into namespaces or other nodes
                classes.extend(traverse(child))
            return classes

        result["classes"] = traverse(root)
        return result

    except Exception:
        # Fallback using regex
        classes = []
        class_matches = re.finditer(r'class\s+(\w+)\s*:\s*MonoBehaviour\s*{', code)
        for m in class_matches:
            cls = m.group(1)
            methods = []
            for mm in re.finditer(r'(void|public)\s+(Start|Update|FixedUpdate)\s*\([^\)]*\)\s*{', code):
                method_name = mm.group(2)
                start = mm.end()
                depth = 1
                i = start
                while i < len(code) and depth > 0:
                    if code[i] == '{':
                        depth += 1
                    elif code[i] == '}':
                        depth -= 1
                    i += 1
                body = code[start:i-1].strip()
                methods.append({"name": method_name, "body": body})
            classes.append({"name": cls, "methods": methods})
        result["classes"] = classes
        return result