bot_detector.py

# bot_detector.py
import re
import numpy as np
from datetime import datetime
from sklearn.cluster import DBSCAN

# Feature Extraction Functions
def is_generic_name(username):
    return bool(re.match(r'^[a-z]+\d{3,}$', username)) or 'user' in username.lower()

def is_new_account(created_date):
    return (datetime.now() - created_date).days < 7 if created_date else False

def detect_generic_text(comments):
    generic_phrases = ['great post!', 'awesome', 'nice', 'cool', 'thanks for sharing']
    return [c for c in comments if any(p in c.lower() for p in generic_phrases)]

def analyze_timing(timestamps):
    deltas = np.diff(sorted(timestamps))
    return np.mean(deltas), np.std(deltas)

def cluster_engagers(engagements):
    # Simple clustering based on engagement patterns
    features = [[e['followers'], e['posts_count'], e['engagement_freq']] 
               for e in engagements]
    return DBSCAN(eps=0.5, min_samples=3).fit_predict(features)

# Main Detection Function
def analyze_post(post_url):
    # Data collection would happen here
    engagements = fetch_engagements(post_url)  
    
    # Analysis pipeline
    results = {
        'suspicious_profiles': [],
        'duplicate_comments': [],
        'time_analysis': {},
        'content_analysis': {}
    }
    
    if engagements:
        # Profile analysis
        results['suspicious_profiles'] = [e for e in engagements 
                                         if is_generic_name(e['username']) or 
                                         is_new_account(e['created_at'])]
        
        # Timing analysis
        timestamps = [e['timestamp'] for e in engagements if e['timestamp']]
        if timestamps:
            results['time_analysis']['mean_interval'], results['time_analysis']['std_dev'] = analyze_timing(timestamps)
        
        # Content analysis
        comments = [e['comment'] for e in engagements if e['comment']]
        results['content_analysis']['generic_comments'] = detect_generic_text(comments)
        results['content_analysis']['duplicate_comments'] = find_duplicates(comments)
        
        # Cluster analysis
        results['cluster_analysis'] = cluster_engagers(engagements).tolist()
    
    return results