Parse chat export from GitHub Copilot

extract.py

import json
from typing import Dict, Any, List, Optional

def get_invocation_message(item: Dict[str, Any]) -> Optional[str]:
    """Safely extracts invocation message whether it's a string or dict."""
    invocation_msg_data = item.get("invocationMessage")
    if not invocation_msg_data:
        return None
    if isinstance(invocation_msg_data, str):
        return invocation_msg_data
    return invocation_msg_data.get("value")

def process_chat_data(filepath: str) -> None:
    """Loads and processes chat response data from a JSON file."""
    try:
        with open(filepath, "r", encoding="utf-8") as f:
            data = json.load(f)
    except (FileNotFoundError, json.JSONDecodeError) as e:
        print(f"Error loading '{filepath}': {e}")
        return

    # Safely navigate the JSON tree
    requests = data.get("requests", [])
    if not requests:
        return
        
    response = requests[0].get("response", [])
    
    for item in response:
        kind = item.get("kind")
        
        if not kind:
            handle_raw_text(item)
            continue

        match kind:
            case "thinking":
                handle_thinking(item)
            case "toolInvocationSerialized":
                handle_tool_invocation(item, response)
            case "mcpServersStarting" | "undoStop" | "progressTaskSerialized" | "codeblockUri" | "textEditGroup" | "inlineReference":
                continue
            case _:
                print(item)

def handle_raw_text(item: Dict[str, Any]) -> None:
    """Handles items that do not have a specific 'kind' categorization."""
    value = item.get("value")
    if value and value.strip() not in ("", "```"):
        print(value)

def handle_thinking(item: Dict[str, Any]) -> None:
    """Formats and prints the thinking block."""
    value = item.get("value")
    if value and value.strip():
        print(f"<thinking>\n{value}\n</thinking>")

def handle_tool_invocation(item: Dict[str, Any], full_response: List[Dict[str, Any]]) -> None:
    """Routes different tool invocations to their specific handlers."""
    tool_data = item.get("toolSpecificData")
    
    if not tool_data:
        invocation_msg = get_invocation_message(item)
        if invocation_msg:
            print(f"<toolCall>\n{invocation_msg}\n</toolCall>")
        return

    tool_kind = tool_data.get("kind")
    match tool_kind:
        case "todoList":
            handle_todo_list(tool_data.get("todoList"))
        case "subagent":
            handle_subagent(item, full_response)
        case "input":
            if "subAgentInvocationId" not in item:
                invocation_msg = get_invocation_message(item)
                if invocation_msg:
                    print(f"<toolCall>\n{invocation_msg}\n</toolCall>")
        case "terminal":
            if "subAgentInvocationId" not in item:
                cmd = tool_data.get("commandLine", {}).get("original")
                if cmd:
                    print(f"<terminal>\n{cmd}\n</terminal>")
        case _:
            print(item)

def handle_todo_list(todo_list: Any) -> None:
    """Formats and prints the todo list."""
    if not isinstance(todo_list, list):
        print(json.dumps(todo_list, indent=2, ensure_ascii=False))
        return
        
    status_icon = {
        "not-started": "[ ]",
        "in-progress": "[~]",
        "completed": "[x]",
    }
    
    print("<todoList>")
    for todo in todo_list:
        title = todo.get("title", "").strip()
        status = todo.get("status", "unknown")
        icon = status_icon.get(status, "•")
        print(f"{icon} {title}")
    print("</todoList>")

def handle_subagent(item: Dict[str, Any], full_response: List[Dict[str, Any]]) -> None:
    """Processes subagent operations and groups related tool calls."""
    tool_call_id = item.get("toolCallId", "")
    tool_data = item.get("toolSpecificData", {})
    
    print("<subagent>")
    prompt = tool_data.get("prompt", "")
    if prompt:
        print(f"<prompt>\n{prompt}\n</prompt>")

    print("<items>")
    for related in full_response:
        # Only process items linked to this specific subagent
        if related.get("subAgentInvocationId") != tool_call_id:
            continue
            
        invocation_msg = get_invocation_message(related)
        if invocation_msg:
            print(f"<toolCall>{invocation_msg}</toolCall>")
            
        related_tool_data = related.get("toolSpecificData", {})
        if related_tool_data.get("kind") == "terminal":
            cmd = related_tool_data.get("commandLine", {}).get("original")
            if cmd:
                print(f"<terminal>\n{cmd}\n</terminal>")
    print("</items>")

    result = tool_data.get("result", "")
    if result:
        print(f"<result>\n{result}\n</result>")
    print("</subagent>")

process_chat_data("./chat.json")

copilot has new format for chatsession jsonl

try out this extension https://marketplace.visualstudio.com/manage/publishers/imperium-dev