app.py

from flask import Flask, request, jsonify, send_from_directory
import requests
import threading
import time
import uuid
import json
import os
from urllib.parse import urlparse
import socket

app = Flask(__name__, static_folder='.', static_url_path='')

# --- Configuration ---
DATA_DIR = "/data"  # Writable directory on Hugging Face Spaces
if not os.path.exists(DATA_DIR) and not os.environ.get('SPACE_ID'): # For local dev if /data isn't preset
    print(f"Warning: {DATA_DIR} not found. Using current directory for data.json (local dev mode).")
    DATA_DIR = "." # Fallback to current dir for local testing if /data does not exist
    
DATA_FILE = os.path.join(DATA_DIR, "data.json")
PING_INTERVAL_SECONDS = 60
HISTORY_DURATION_SECONDS = 60 * 60

monitored_urls_store = {}
lock = threading.Lock()

# --- Helper Functions ---
def ensure_data_dir_exists():
    """Ensures the data directory exists."""
    if DATA_DIR == ".": # No need to create current directory
        return
    if not os.path.exists(DATA_DIR):
        try:
            os.makedirs(DATA_DIR)
            print(f"Created data directory: {DATA_DIR}")
        except OSError as e:
            print(f"Error creating data directory {DATA_DIR}: {e}. Data persistence may fail.")

def save_data_to_json():
    with lock:
        ensure_data_dir_exists() # Ensure directory exists before attempting to write
        serializable_data = {}
        for url_id, data in monitored_urls_store.items():
            s_data = data.copy()
            s_data.pop("_thread", None)
            s_data.pop("_stop_event", None)
            serializable_data[url_id] = s_data
        try:
            with open(DATA_FILE, 'w') as f:
                json.dump(serializable_data, f, indent=2)
        except IOError as e:
            print(f"Error saving data to {DATA_FILE}: {e}") # This is where your error was logged

def load_data_from_json():
    global monitored_urls_store
    ensure_data_dir_exists() # Ensure directory exists before attempting to read
    if os.path.exists(DATA_FILE):
        try:
            with open(DATA_FILE, 'r') as f:
                loaded_json_data = json.load(f)
            
            temp_store = {}
            for url_id_key, data_item in loaded_json_data.items():
                data_item.setdefault('id', url_id_key)
                current_id = data_item['id']
                data_item.setdefault('status', 'pending')
                data_item.setdefault('ip', data_item.get('ip', 'N/A'))
                data_item.setdefault('responseTime', None)
                data_item.setdefault('lastChecked', None)
                data_item.setdefault('history', data_item.get('history', []))
                temp_store[current_id] = data_item
            
            with lock:
                monitored_urls_store = temp_store

        except json.JSONDecodeError:
            print(f"Warning: Could not decode {DATA_FILE}. Starting with an empty list.")
            with lock: monitored_urls_store = {}
        except Exception as e:
            print(f"Error loading data from {DATA_FILE}: {e}. Starting fresh.")
            with lock: monitored_urls_store = {}
    else:
        print(f"{DATA_FILE} not found. Starting with an empty list.")
        with lock: monitored_urls_store = {}
    
    url_ids_to_start_monitoring = []
    with lock:
        url_ids_to_start_monitoring = list(monitored_urls_store.keys())

    for url_id in url_ids_to_start_monitoring:
        start_url_monitoring_thread(url_id)

def get_host_ip_address(hostname_str):
    try:
        socket.inet_aton(hostname_str)
        return hostname_str 
    except OSError:
        try:
            ip_address = socket.gethostbyname(hostname_str)
            return ip_address
        except socket.gaierror:
            print(f"Could not resolve hostname: {hostname_str}")
            return 'N/A'
    except Exception as e:
        print(f"Error processing hostname/IP for {hostname_str}: {e}")
        return 'N/A'

def prune_url_history(url_data_entry):
    cutoff_time = time.time() - HISTORY_DURATION_SECONDS
    url_data_entry['history'] = [
        entry for entry in url_data_entry.get('history', []) if entry['timestamp'] >= cutoff_time
    ]

def execute_url_check(url_id_to_check):
    url_config_snapshot = None
    with lock:
        if url_id_to_check not in monitored_urls_store: return
        
        current_url_data = monitored_urls_store[url_id_to_check]
        if current_url_data.get('_stop_event') and current_url_data['_stop_event'].is_set(): return
        
        print(f"Checking {current_url_data['url']} (ID: {url_id_to_check})...")
        current_url_data['status'] = 'checking'
        url_config_snapshot = current_url_data.copy()

    if not url_config_snapshot: return

    check_start_time = time.perf_counter()
    final_check_status = 'error'
    http_response_time_ms = None
    http_headers = {'User-Agent': 'URLPinger/1.0 (HuggingFace Space Bot)'} 
    
    try:
        try:
            head_response = requests.head(url_config_snapshot['url'], timeout=10, allow_redirects=True, headers=http_headers)
            if 200 <= head_response.status_code < 400:
                 final_check_status = 'ok'
            else:
                print(f"HEAD for {url_config_snapshot['url']} returned {head_response.status_code}. Trying GET.")
        except requests.exceptions.Timeout:
            print(f"HEAD timeout for {url_config_snapshot['url']}. Trying GET...")
        except requests.RequestException as e_head:
            print(f"HEAD failed for {url_config_snapshot['url']}: {e_head}. Trying GET...")

        if final_check_status != 'ok':
            try:
                get_response = requests.get(url_config_snapshot['url'], timeout=15, allow_redirects=True, headers=http_headers)
                if get_response.ok:
                    final_check_status = 'ok'
                else:
                    print(f"GET for {url_config_snapshot['url']} status: {get_response.status_code}")
                    final_check_status = 'error' 
            except requests.exceptions.Timeout:
                print(f"GET timeout for {url_config_snapshot['url']}")
                final_check_status = 'error'
            except requests.RequestException as e_get:
                print(f"GET failed for {url_config_snapshot['url']}: {e_get}")
                final_check_status = 'error'
        
        if final_check_status == 'ok':
            http_response_time_ms = (time.perf_counter() - check_start_time) * 1000

    except Exception as e: 
        print(f"Outer check exception for {url_config_snapshot['url']}: {e}")
        final_check_status = 'error'

    with lock:
        if url_id_to_check not in monitored_urls_store: return
        
        live_url_data = monitored_urls_store[url_id_to_check] 
        live_url_data['status'] = final_check_status
        live_url_data['responseTime'] = round(http_response_time_ms) if http_response_time_ms is not None else None
        live_url_data['lastChecked'] = time.strftime("%Y-%m-%dT%H:%M:%SZ", time.gmtime())
        
        current_history_list = live_url_data.get('history', [])
        current_history_list.append({'timestamp': time.time(), 'status': final_check_status})
        live_url_data['history'] = current_history_list
        prune_url_history(live_url_data)
        
        save_data_to_json() # This will now try to save to /data/data.json
        print(f"Finished check for {live_url_data['url']}: {final_check_status}, {http_response_time_ms} ms")

def pinger_thread_function(url_id_param, stop_event_param):
    while not stop_event_param.is_set():
        execute_url_check(url_id_param)
        for _ in range(PING_INTERVAL_SECONDS): 
            if stop_event_param.is_set(): break
            time.sleep(1)
    print(f"PingerThread for {url_id_param} stopped.")

def start_url_monitoring_thread(target_url_id):
    with lock:
        if target_url_id not in monitored_urls_store:
            print(f"Cannot start monitoring: URL ID {target_url_id} not found.")
            return

        url_data_entry = monitored_urls_store[target_url_id]

        if "_thread" in url_data_entry and url_data_entry["_thread"].is_alive():
            print(f"Monitor for URL ID {target_url_id} already running. Attempting to restart.")
            if "_stop_event" in url_data_entry and url_data_entry["_stop_event"]:
                url_data_entry["_stop_event"].set()
            url_data_entry["_thread"].join(timeout=3)

        new_stop_event = threading.Event()
        new_thread = threading.Thread(target=pinger_thread_function, args=(target_url_id, new_stop_event), daemon=True)
        
        url_data_entry["_thread"] = new_thread
        url_data_entry["_stop_event"] = new_stop_event
        
        new_thread.start()
        print(f"Started/Restarted monitoring for URL ID {target_url_id}: {url_data_entry['url']}")

def stop_url_monitoring_thread(target_url_id):
    with lock:
        if target_url_id in monitored_urls_store:
            url_data_entry = monitored_urls_store[target_url_id]
            if "_thread" in url_data_entry and url_data_entry["_thread"].is_alive():
                print(f"Signaling stop for monitor thread of URL ID {target_url_id}")
                if "_stop_event" in url_data_entry and url_data_entry["_stop_event"]:
                    url_data_entry["_stop_event"].set()
            url_data_entry.pop("_thread", None)
            url_data_entry.pop("_stop_event", None)

# --- API Endpoints ---
@app.route('/')
def serve_index():
    return send_from_directory(app.static_folder, 'index.html')

@app.route('/api/urls', methods=['GET'])
def get_all_urls():
    with lock:
        response_list = []
        for data_item in monitored_urls_store.values():
            display_item = data_item.copy()
            display_item.pop("_thread", None)
            display_item.pop("_stop_event", None)
            response_list.append(display_item)
    return jsonify(response_list)

@app.route('/api/urls', methods=['POST'])
def add_new_url():
    request_data = request.get_json()
    if not request_data or 'url' not in request_data:
        return jsonify({"error": "URL is required"}), 400
    
    input_url = request_data['url'].strip()

    if not input_url.startswith('http://') and not input_url.startswith('https://'):
        input_url = 'https://' + input_url
    
    try:
        parsed_input_url = urlparse(input_url)
        if not parsed_input_url.scheme or not parsed_input_url.netloc:
            raise ValueError("Invalid URL structure")
        url_hostname = parsed_input_url.hostname
    except ValueError:
        return jsonify({"error": "Invalid URL format"}), 400

    with lock:
        normalized_new_url = input_url.rstrip('/').lower()
        for existing_url_id in list(monitored_urls_store.keys()):
            existing_url_data = monitored_urls_store.get(existing_url_id)
            if existing_url_data and existing_url_data['url'].rstrip('/').lower() == normalized_new_url:
                return jsonify({"error": "URL already monitored"}), 409

        new_url_id = str(uuid.uuid4())
        resolved_ip = get_host_ip_address(url_hostname) if url_hostname else 'N/A'

        url_entry_to_add = {
            "id": new_url_id, "url": input_url, "status": 'pending', 
            "ip": resolved_ip, "responseTime": None, "lastChecked": None, "history": []
        }
        
        response_payload = url_entry_to_add.copy()
        monitored_urls_store[new_url_id] = url_entry_to_add
        save_data_to_json() 
    
    start_url_monitoring_thread(new_url_id)
    return jsonify(response_payload), 201

@app.route('/api/urls/<string:target_url_id>', methods=['DELETE'])
def delete_existing_url(target_url_id):
    with lock:
        if target_url_id in monitored_urls_store:
            stop_url_monitoring_thread(target_url_id) 
            removed_url_entry = monitored_urls_store.pop(target_url_id)
            save_data_to_json()
            
            response_data = removed_url_entry.copy()
            response_data.pop("_thread", None)
            response_data.pop("_stop_event", None)
            print(f"Deleted URL ID {target_url_id}")
            return jsonify({"message": "URL removed", "url": response_data}), 200
        else:
            return jsonify({"error": "URL not found"}), 404

# --- Main Execution / Gunicorn Entry Point ---
if os.environ.get('WERKZEUG_RUN_MAIN') != 'true':
    ensure_data_dir_exists() # Ensure data dir exists before loading
    load_data_from_json() 

if __name__ == '__main__':
    if os.environ.get('WERKZEUG_RUN_MAIN') == 'true':
        ensure_data_dir_exists() # Ensure data dir exists before loading in reloader
        load_data_from_json()
    # The DATA_DIR fallback for local dev will be used here if /data doesn't exist
    app.run(debug=True, host='0.0.0.0', port=7860)