chunk_python_code.py

import ast
from langchain.schema import Document 

def chunk_python_code_with_metadata(python_code, file_path):
    """
    Entry point method to process the Python file.
    It invokes the iterate_ast function.
    """
    documents = []
   #print(f"Processing file: {file_path}")
  
    _iterate_ast(python_code, documents, file_path)
    # Determine usage based on the file_path
    if file_path.startswith("kadi_apy/lib/"):
        usage = "kadi-apy python library"
    elif file_path.startswith("kadi_apy/cli/"):
        usage = "kadi-apy python cli library"
    else:
        usage = "undefined"

    # Add metadata for usage to all documents
    for doc in documents:
        doc.metadata["source"] = file_path
        doc.metadata["usage"] = usage  # Add the determined usage metadata
        #print(doc)
    return documents


def _iterate_ast(python_code, documents, file_path):
    """
    Parses the AST of the given Python file and delegates
    handling to specific methods based on node types.
    """
    tree = ast.parse(python_code, filename=file_path)

    first_level_nodes = list(ast.iter_child_nodes(tree))

    # Check if there are no first-level nodes
    if not first_level_nodes:
        documents.extend(
            _chunk_nodeless_python_code(documents, python_code, file_path))
        return

    all_imports = all(isinstance(node, (ast.Import, ast.ImportFrom)) for node in first_level_nodes)
    if all_imports:
        documents.extend(
            _chunk_import_only_python_code(documents, python_code, file_path))
    
    # Iterate over first-level nodes
    for first_level_node in ast.iter_child_nodes(tree):
        
        if isinstance(first_level_node, ast.ClassDef):
            documents.extend(
                _handle_first_level_class(first_level_node, documents, python_code))
        elif isinstance(first_level_node, ast.FunctionDef):
            documents.extend(
                _chunk_first_level_func_node(first_level_node, documents, python_code))
        elif isinstance(first_level_node, ast.Assign):
            documents.extend(
                _chunk_first_level_assign_node(first_level_node, documents, python_code))
        else: 
            documents.extend(
                _handle_notdefined_case(python_code))
             

def _chunk_import_only_python_code(python_code, file_path):
    """
    Handles cases where the first-level nodes are only imports.
    """
    documents = []
    if file_path.endswith("__init__.py"):
        type = "__init__-file"
    else:
        type = "undefined"

    # Create metadata without imports
    metadata = {"type": type}

    # Create and store a Document with the full source code
    doc = Document(
        page_content=python_code,
        metadata=metadata
    )
    documents.append(doc)
    return documents



def _handle_notdefined_case(python_code):
    documents = []
    documents.extend(
        chunk_python_code_by_character)  
    return documents
    
from langchain.text_splitter import RecursiveCharacterTextSplitter


def _chunk_python_code_by_character(python_code):
    documents = []
    text_splitter = RecursiveCharacterTextSplitter(
        chunk_size=512,         
        chunk_overlap=128,      
        separators=[]
    )
    
    chunks = text_splitter.split_text(python_code)
    
    for chunk in chunks:
        doc = Document(
            page_content=chunk
        )
        documents.append(doc)
    
    return documents

    

def _chunk_nodeless_python_code(python_code, file_path):
    """
    Handles cases where no top-level nodes are found in the AST.
    """
    documents = []
    if file_path.endswith("__init__.py"):
        type = "__init__-file"
    else:
        type = "undefined"

    # Create metadata without imports
    metadata = {"type": type}

    # Create and store a Document with the full source code
    doc = Document(
        page_content=python_code,
        metadata=metadata
    )
    documents.append(doc)

    return documents

def _chunk_first_level_assign_node(ast_node, python_code):
    
    """
    Handles assignment statements at the first level of the AST.
    """
    documents = []
    assign_start_line = ast_node.lineno
    assign_end_line = ast_node.end_lineno
    assign_source = '\n'.join(python_code.splitlines()[assign_start_line-1:assign_end_line])

    # Create metadata without imports
    metadata = {"type": "Assign"}

    # Create and store Document for the assignment
    doc = Document(
        page_content=assign_source,
        metadata=metadata
    )
    documents.append(doc)

    return documents

def _handle_first_level_class(ast_node , python_code):
    """
    Handles classes at the first level of the AST.
    """
    documents = []
    class_start_line = ast_node.lineno
    class_body_lines = [child.lineno for child in ast_node.body if isinstance(child, ast.FunctionDef)]
    class_end_line = min(class_body_lines, default=ast_node.end_lineno) - 1
    class_source = '\n'.join(python_code.splitlines()[class_start_line-1:class_end_line])

    metadata = {
        "type": "class",
        "class": ast_node.name,
        "visibility": "public"
    }

    # Create and store Document for the class
    doc = Document(
        page_content=class_source,
        metadata=metadata
    )
    documents.append(doc)

    # Handle methods within the class
    for second_level_node in ast.iter_child_nodes(ast_node):
        if isinstance(second_level_node, ast.FunctionDef):
            method_start_line = (
                second_level_node.decorator_list[0].lineno
                if second_level_node.decorator_list else second_level_node.lineno
            )
            method_end_line = second_level_node.end_lineno
            method_source = '\n'.join(python_code.splitlines()[method_start_line-1:method_end_line])

            visibility = "internal" if second_level_node.name.startswith("_") else "public"

            doc = Document(
                page_content=method_source,
                metadata={
                    "type": "method",
                    "method": second_level_node.name,
                    "visibility": visibility,
                    "class": ast_node.name
                }
            )
            documents.append(doc)

    return documents

def _chunk_first_level_func_node(ast_node, python_code):
    """
    Handles functions at the first level of the AST.
    """
    documents = []
    function_start_line = (
        ast_node.decorator_list[0].lineno
        if ast_node.decorator_list else ast_node.lineno
    )
    function_end_line = ast_node.end_lineno
    function_source = '\n'.join(python_code.splitlines()[function_start_line-1:function_end_line])

    visibility = "internal" if ast_node.name.startswith("_") else "public"

    is_command = any(
        decorator.id == "apy_command"
        for decorator in ast_node.decorator_list
        if hasattr(decorator, "id")
    )

    metadata = {
        "type": "command" if is_command else "function",
        "visibility": visibility
    }
    if is_command:
        metadata["command"] = ast_node.name
    else:
        metadata["method"] = ast_node.name

    doc = Document(
        page_content=function_source,
        metadata=metadata
    )
    documents.append(doc)

    return documents