daily-ai-summary/daily_summary.py

import os
import subprocess
import requests
import smtplib
from email.message import EmailMessage
import time
import schedule

# --- CONFIGURATION (Loaded from environment variables) ---
OLLAMA_API = os.getenv("OLLAMA_API_URL", "http://localhost:11434/api/generate")
MODEL = os.getenv("OLLAMA_MODEL", "llama3.2:1b")
AI_SYSTEM_PROMPT = os.getenv("AI_SYSTEM_PROMPT", "You are a Linux Sysadmin. Summarize these logs briefly.")

MY_EMAIL = os.getenv("REPORT_TO_EMAIL")
FROM_EMAIL = os.getenv("REPORT_FROM_EMAIL")
SMTP_SERVER = os.getenv("SMTP_SERVER", "127.0.0.1")
SMTP_PORT = int(os.getenv("SMTP_PORT", 25))
SMTP_USER = os.getenv("SMTP_USER", "")
SMTP_PASS = os.getenv("SMTP_PASS", "")

JOURNAL_UNIT = os.getenv("JOURNAL_UNIT", "mailserver")
MAX_ERRORS = int(os.getenv("MAX_ERRORS_TO_ANALYZE", 50))

# Parse comma-separated keywords into a list
keyword_string = os.getenv("ERROR_KEYWORDS", "error,failed,warning,fatal,timeout")
KEYWORDS = [k.strip().lower() for k in keyword_string.split(',')]

# --- EXECUTION TOGGLES ---
DEBUG = os.getenv("DEBUG", "false").lower() in ("true", "1", "yes")
RUN_NOW = os.getenv("RUN_NOW", "false").lower() in ("true", "1", "yes")
REPORT_TIME = os.getenv("REPORT_TIME", "08:00")

def debug_print(message):
    """Prints only if DEBUG=true in .env"""
    if DEBUG:
        print(f"[DEBUG] {message}")

def get_yesterdays_logs():
    print(f"Fetching logs for '{JOURNAL_UNIT}' from yesterday...")
    debug_print(f"Running command: journalctl -u {JOURNAL_UNIT} --since yesterday --until today")
    
    result = subprocess.run(
        ['journalctl', '-u', JOURNAL_UNIT, '--since', 'yesterday', '--until', 'today'],
        capture_output=True, text=True
    )
    lines = result.stdout.splitlines()
    debug_print(f"Fetched {len(lines)} total lines from journalctl.")
    return lines

def process_logs(logs):
    total_lines = len(logs)
    errors = []
    
    debug_print(f"Filtering logs using keywords: {KEYWORDS}")
    for line in logs:
        line_lower = line.lower()
        if any(keyword in line_lower for keyword in KEYWORDS):
            errors.append(line)
            
    debug_print(f"Found {len(errors)} matching lines before truncation.")
    
    # Truncate to maximum allowed errors to save CPU time
    errors = errors[-MAX_ERRORS:] 
    debug_print(f"Truncated to {len(errors)} lines for AI processing (MAX_ERRORS={MAX_ERRORS}).")
    return total_lines, errors

def get_ai_summary(errors):
    if not errors:
        debug_print("Skipping AI API call because error list is empty.")
        return "No errors or warnings were found in the logs yesterday. Everything looks healthy!"
        
    print(f"Asking AI to summarize {len(errors)} errors... (This will take a few minutes)")
    
    error_text = "\n".join(errors)
    
    # Combine the system prompt from .env with the logs
    prompt = f"{AI_SYSTEM_PROMPT}\n\nLOGS:\n{error_text}"
    
    payload = {
        "model": MODEL,
        "prompt": prompt,
        "stream": False
    }
    
    debug_print(f"Sending prompt to Ollama at {OLLAMA_API} using model '{MODEL}'")
    debug_print(f"System Prompt used: {AI_SYSTEM_PROMPT}")
    debug_print(f"Total payload length: {len(prompt)} characters.")
    
    try:
        response = requests.post(OLLAMA_API, json=payload)
        debug_print(f"AI response status code: {response.status_code}")
        return response.json().get('response', 'Error: AI returned empty response.')
    except Exception as e:
        debug_print(f"AI Connection Exception: {str(e)}")
        return f"Failed to connect to local AI: {str(e)}"

def send_email_report(total_traffic, ai_summary):
    if not MY_EMAIL or not FROM_EMAIL:
        print("Skipping email: REPORT_TO_EMAIL or REPORT_FROM_EMAIL not set.")
        return

    print(f"Sending email report to {MY_EMAIL}...")
    debug_print(f"SMTP Server: {SMTP_SERVER}:{SMTP_PORT}")
    debug_print(f"SMTP Auth User: {'<configured>' if SMTP_USER else '<none>'}")
    
    msg = EmailMessage()
    msg['Subject'] = f"Daily AI Summary: {JOURNAL_UNIT}"
    msg['From'] = FROM_EMAIL
    msg['To'] = MY_EMAIL
    
    body = (
        f"DAILY SERVER REPORT: {JOURNAL_UNIT}\n"
        f"----------------------------------------\n"
        f"Total Log Lines Processed: {total_traffic}\n\n"
        f"AI Analysis of Errors/Warnings:\n"
        f"{ai_summary}\n"
    )
    
    msg.set_content(body)
    
    try:
        server = smtplib.SMTP(SMTP_SERVER, SMTP_PORT)
        if SMTP_USER:
            debug_print("Attempting STARTTLS and Login...")
            server.starttls()
            server.login(SMTP_USER, SMTP_PASS)
            
        debug_print("Transmitting email payload...")
        server.send_message(msg)
        server.quit()
        print("Email sent successfully!")
    except Exception as e:
        print(f"Failed to send email: {e}")
        debug_print(f"Email Exception: {str(e)}")

def job():
    print(f"\n--- Starting log analysis for {JOURNAL_UNIT} ---")
    logs = get_yesterdays_logs()
    total_traffic, errors = process_logs(logs)
    summary = get_ai_summary(errors)
    send_email_report(total_traffic, summary)
    print("--- Analysis complete. ---")

if __name__ == "__main__":
    print("Log Analyzer Container starting up...")
    
    if DEBUG:
        print("[DEBUG] Debug mode is ENABLED.")
    
    # 1. Register the permanent daily schedule first
    schedule.every().day.at(REPORT_TIME).do(job)
    print(f"Job permanently scheduled to run daily at {REPORT_TIME}.")
    
    # 2. Handle the immediate run if requested
    if RUN_NOW:
        print("RUN_NOW is enabled. Executing immediate startup run...")
        job()
        print("--- Immediate run finished. Returning to standard schedule. ---")
    else:
        print("RUN_NOW is disabled. Skipping immediate run.")
        
    # 3. Enter the permanent background loop
    print(f"Entering background scheduler loop. Waiting for {REPORT_TIME}...")
    while True:
        schedule.run_pending()
        time.sleep(60)