filter.py

import glob
import os
import json

# Get the current working directory
current_directory = os.getcwd()

# Define the pattern for the JSON files we are interested in
file_pattern = os.path.join(current_directory, '2023-*.json')

# Function to extract message texts from a JSON file
def extract_messages(json_file_path):
    with open(json_file_path, 'r') as file:
        json_data = json.load(file)
        return [message['text'] for message in json_data if 'text' in message]

# Find all matching JSON files in the current directory
json_files = glob.glob(file_pattern)

# Dictionary to hold file names and their messages
messages_per_file = {}

# Loop through the files and extract messages
for json_file in json_files:
    messages = extract_messages(json_file)
    if messages:  # Only add to the dictionary if messages were found
        messages_per_file[os.path.basename(json_file)] = messages

# Define the name of the output text file
output_file_path = os.path.join(current_directory, 'extracted_messages.txt')

# Check if we found any messages and write them to the text file
if messages_per_file:
    with open(output_file_path, 'w') as output_file:
        for file_name, messages in messages_per_file.items():
            output_file.write(f'Messages from {file_name}:\n')
            for message in messages:
                output_file.write(message + '\n')
            output_file.write('\n')  # Add an extra newline for separation between files
    print(f"Messages have been written to {output_file_path}")
else:
    print("No messages found in the files.")