Script to parse ChatGPT conversations.json file to analyze how many messages were sent by ChatGPT and users, and how many conversations were created per year and month.

parse_chatgpt_conversations.py

# Script to parse ChatGPT conversations.json file to analyze how many messages were sent by ChatGPT and users, and how many conversations were created per year and month.
# conversations.json is a JSON file containing a list of conversations. You can get this by starting a data export in ChatGPT settings.

import json
from collections import defaultdict
from datetime import datetime

# Load the JSON file
file_path = "conversations.json"
with open(file_path, "r") as file:
    conversations = json.load(file)

# Initialize counters
chatgpt_message_count = 0
user_message_count = 0

# Initialize conversation counters
conversations_per_year = defaultdict(int)
conversations_per_month = defaultdict(int)

# Initialize message counters per year and month
chatgpt_messages_per_year = defaultdict(int)
user_messages_per_year = defaultdict(int)
chatgpt_messages_per_month = defaultdict(int)
user_messages_per_month = defaultdict(int)

# Iterate through each conversation in the JSON list
for conversation in conversations:
    conversation_timestamp = conversation.get("create_time", 0)
    if conversation_timestamp:
        conversation_date = datetime.fromtimestamp(conversation_timestamp)
        year = conversation_date.year
        month = conversation_date.strftime("%Y-%m")

        conversations_per_year[year] += 1
        conversations_per_month[month] += 1

    messages_mapping = conversation.get("mapping", {})

    # Iterate through each message in the mapping
    for key, value in messages_mapping.items():
        message = value.get("message", {})
        if message:  # Check if the message exists
            author = message.get("author", {}).get("role", "unknown")
            message_timestamp = message.get("create_time", 0)
            if message_timestamp:
                message_date = datetime.fromtimestamp(message_timestamp)
                message_year = message_date.year
                message_month = message_date.strftime("%Y-%m")

                if author == "tool" or author == "assistant":
                    chatgpt_message_count += 1
                    chatgpt_messages_per_year[message_year] += 1
                    chatgpt_messages_per_month[message_month] += 1
                if author == "user":
                    user_message_count += 1
                    user_messages_per_year[message_year] += 1
                    user_messages_per_month[message_month] += 1

print(f"Total ChatGPT messages: {chatgpt_message_count}")
print(f"Total user messages: {user_message_count}")

conversations_per_year = dict(sorted(conversations_per_year.items()))
print("Conversations per year:")
for year, count in conversations_per_year.items():
    print(f"{year}: {count}")

conversations_per_month = dict(sorted(conversations_per_month.items()))
print("Conversations per month:")
for month, count in conversations_per_month.items():
    print(f"{month}: {count}")

chatgpt_messages_per_year = dict(sorted(chatgpt_messages_per_year.items()))
print("ChatGPT messages per year:")
for year, count in chatgpt_messages_per_year.items():
    print(f"{year}: {count}")

user_messages_per_year = dict(sorted(user_messages_per_year.items()))
print("User messages per year:")
for year, count in user_messages_per_year.items():
    print(f"{year}: {count}")

chatgpt_messages_per_month = dict(sorted(chatgpt_messages_per_month.items()))
print("ChatGPT messages per month:")
for month, count in chatgpt_messages_per_month.items():
    print(f"{month}: {count}")

user_messages_per_month = dict(sorted(user_messages_per_month.items()))
print("User messages per month:")
for month, count in user_messages_per_month.items():
    print(f"{month}: {count}")