Facebook Posts and Page Likes Pull

This python code pulls the user posts and page likes from facebook raw data extracted from the Facebook Graph API. The code then saves the data to a CSV file.

"""
@author: Robert Vidigal, PhD (CSMaP-NYU)
"""
import pandas as pd
import glob
from tqdm import tqdm 

# Reading JSON files and converting to CSV
def readFiles(lang):
    outdir='/Users/rb5286/Downloads/facebook/'
    files =glob.glob(f'/Users/rb5286/Downloads/facebook/2022/{lang}/*json*')
    columns = ['message', 'shares', 'like.summary.total_count', 'love.summary.total_count', 'wow.summary.total_count',
           'haha.summary.total_count','sad.summary.total_count','angry.summary.total_count']
    likes_df = pd.DataFrame()
    posts_df = pd.DataFrame()
    
    for file in tqdm(files[:]):
        try:
            user_id=file.split('/')[-1].split('__')[0]
            df = pd.read_json(file,lines=True)
            likes = pd.read_json(df['likes'].tolist()[0],lines=True)
            if likes.empty or 'id' not in likes.columns: 
                print(f'{user_id} LIKES are empty')
                continue
            likes = likes[['name', 'id', 'is_media']]
            likes['userid']=user_id
            
            posts = pd.read_json(df['posts'].tolist()[0],lines=True)
            if posts.empty or 'like.summary.total_count' not in posts.columns: 
                print(f'{user_id} POSTS are empty')
                continue
            if 'message' not in posts.columns or 'shares' not in posts.columns or posts.empty:
                posts['message'] = None
                posts['shares'] = None
                posts = posts[columns]
                posts['userid']=user_id
            else: 
                posts = posts[columns]
                posts['userid']=user_id
            
        except Exception as e: 
            print(f'{file} has error {e}')
            continue
        
        likes_df = pd.concat([likes_df, likes], ignore_index=True)
        posts_df = pd.concat([posts_df, posts], ignore_index=True)
        

    likes_df.to_csv(f'{outdir}CSIP_Wave_2_facebook_{lang}_likes.csv',index=False)
    posts_df.to_csv(f'{outdir}CSIP_Wave_2_facebook_{lang}_posts.csv',index=False)
    print(len(likes_df), len(posts_df))

Running the defined function above.

for lan in ['en', 'sp']:
    readFiles(lan)