parser.py

# parser.py
import ast

def get_category(node):
    """Determine the category of an AST node."""
    if isinstance(node, (ast.Import, ast.ImportFrom)):
        return 'import'
    elif isinstance(node, (ast.Assign, ast.AnnAssign, ast.AugAssign)):
        return 'assignment'
    elif isinstance(node, ast.FunctionDef):
        return 'function'
    elif isinstance(node, ast.AsyncFunctionDef):
        return 'async_function'
    elif isinstance(node, ast.ClassDef):
        return 'class'
    elif isinstance(node, ast.Expr):
        return 'expression'
    else:
        return 'other'

def parse_node(node, lines, prev_end, level=0):
    """Recursively parse an AST node and its children, assigning hierarchy levels."""
    parts = []
    start_line = getattr(node, 'lineno', prev_end + 1)
    end_line = getattr(node, 'end_lineno', start_line)

    # Handle spacers before the node
    if start_line > prev_end + 1:
        spacer_lines = lines[prev_end:start_line - 1]
        parts.append({
            'category': 'spacer',
            'source': ''.join(spacer_lines),
            'location': (prev_end + 1, start_line - 1),
            'level': level
        })

    # Capture the node's source
    stmt_lines = lines[start_line - 1:end_line]
    parts.append({
        'category': get_category(node),
        'source': ''.join(stmt_lines),
        'location': (start_line, end_line),
        'level': level
    })

    # Process nested nodes (e.g., class or function bodies)
    if hasattr(node, 'body'):
        nested_prev_end = end_line - 1
        for child in node.body:
            child_parts = parse_node(child, lines, nested_prev_end, level + 1)
            parts.extend(child_parts)
            nested_prev_end = child_parts[-1]['location'][1]

    return parts

def parse_python_code(code):
    """Parse Python code string and return parts with hierarchy."""
    lines = code.splitlines(keepends=True)
    try:
        tree = ast.parse(code)
    except SyntaxError:
        return [{'category': 'error', 'source': 'Invalid Python code', 'location': (1, 1), 'level': 0}]

    parts = []
    prev_end = 0

    for stmt in tree.body:
        stmt_parts = parse_node(stmt, lines, prev_end)
        parts.extend(stmt_parts)
        prev_end = stmt_parts[-1]['location'][1]

    # Capture trailing spacers
    if prev_end < len(lines):
        remaining_lines = lines[prev_end:]
        parts.append({
            'category': 'spacer',
            'source': ''.join(remaining_lines),
            'location': (prev_end + 1, len(lines) + 1),
            'level': 0
        })

    return parts