261 lines
10 KiB
Python
261 lines
10 KiB
Python
|
|
import logging
|
|
import pymysql
|
|
from datetime import datetime
|
|
from urllib.parse import unquote, parse_qs
|
|
from django.utils import timezone
|
|
import html
|
|
from hotels.models import Room, Hotel
|
|
from .models import UserActivityLog, ExternalDBSettings
|
|
from touchh.utils.log import CustomLogger
|
|
from concurrent.futures import ThreadPoolExecutor, TimeoutError
|
|
from decouple import config
|
|
|
|
class DatabaseConnector:
|
|
def __init__(self, db_settings_id):
|
|
self.db_settings_id = db_settings_id
|
|
self.logger = CustomLogger(name="DatabaseConnector", log_level="DEBUG").get_logger()
|
|
self.connection = None
|
|
self.db_settings = self.get_db_settings()
|
|
|
|
def get_db_settings(self):
|
|
try:
|
|
settings = ExternalDBSettings.objects.get(id=self.db_settings_id)
|
|
self.logger.info(f"Retrieved DB settings: {settings}")
|
|
return {
|
|
"host": settings.host,
|
|
"port": settings.port,
|
|
"user": settings.user,
|
|
"password": settings.password,
|
|
"database": settings.database,
|
|
"table_name": settings.table_name
|
|
|
|
}
|
|
except ExternalDBSettings.DoesNotExist:
|
|
self.logger.error(f"Settings with ID {self.db_settings_id} not found.")
|
|
raise ValueError("Invalid db_settings_id")
|
|
except Exception as e:
|
|
self.logger.error(f"Error retrieving settings: {e}")
|
|
raise
|
|
|
|
def connect(self):
|
|
try:
|
|
self.logger.info(f"Connecting to DB with settings: {self.db_settings}")
|
|
self.connection = pymysql.connect(
|
|
host=self.db_settings["host"],
|
|
port=self.db_settings["port"],
|
|
user=self.db_settings["user"],
|
|
password=self.db_settings["password"],
|
|
database=self.db_settings["database"],
|
|
charset="utf8mb4",
|
|
cursorclass=pymysql.cursors.DictCursor,
|
|
)
|
|
self.logger.info("Database connection established successfully.")
|
|
except pymysql.err.OperationalError as e:
|
|
self.logger.error(f"Operational error during DB connection: {e}")
|
|
raise ConnectionError(f"Operational error: {e}")
|
|
except Exception as e:
|
|
self.logger.error(f"Unexpected database connection error: {e}")
|
|
raise ConnectionError(e)
|
|
|
|
def execute_query(self, query):
|
|
try:
|
|
with self.connection.cursor() as cursor:
|
|
cursor.execute(query)
|
|
return cursor.fetchall()
|
|
except Exception as e:
|
|
self.logger.error(f"Query execution error: {e}")
|
|
return []
|
|
|
|
def close(self):
|
|
if self.connection:
|
|
self.connection.close()
|
|
self.logger.info("Database connection closed.")
|
|
|
|
|
|
class DataProcessor:
|
|
def __init__(self, logger):
|
|
self.logger = logger
|
|
|
|
def decode_html_entities(self, text):
|
|
if text and isinstance(text, str):
|
|
return html.unescape(unquote(text))
|
|
return text
|
|
|
|
def parse_datetime(self, dt_str):
|
|
try:
|
|
if isinstance(dt_str, datetime):
|
|
return timezone.make_aware(dt_str) if timezone.is_naive(dt_str) else dt_str
|
|
if dt_str:
|
|
return timezone.make_aware(datetime.strptime(dt_str, "%Y-%m-%d %H:%M:%S"))
|
|
except Exception as e:
|
|
self.logger.error(f"Datetime parsing error: {e}")
|
|
return None
|
|
|
|
def url_parameters_parser(self, url_parameters):
|
|
try:
|
|
if not url_parameters:
|
|
return {}
|
|
decoded_params = unquote(url_parameters)
|
|
parsed_params = parse_qs(decoded_params)
|
|
return {key: value[0] for key, value in parsed_params.items()}
|
|
except Exception as e:
|
|
self.logger.error(f"URL parameters parsing error: {e}")
|
|
return {}
|
|
|
|
|
|
class HotelRoomManager:
|
|
def __init__(self, logger):
|
|
self.logger = logger
|
|
|
|
def get_or_create_hotel(self, hotel_id, page_title):
|
|
if not hotel_id:
|
|
self.logger.warning("Hotel creation skipped: missing hotel_id.")
|
|
return None
|
|
|
|
hotel, created = Hotel.objects.get_or_create(
|
|
hotel_id=hotel_id,
|
|
defaults={
|
|
"name": html.unescape(page_title) or f"Отель {hotel_id}",
|
|
"description": "Автоматически созданный отель",
|
|
}
|
|
)
|
|
if created:
|
|
self.logger.info(f"Hotel '{hotel.name}' created with hotel_id: {hotel_id}")
|
|
else:
|
|
self.logger.info(f"Hotel '{hotel.name}' already exists with hotel_id: {hotel_id}")
|
|
return hotel
|
|
|
|
def get_or_create_room(self, hotel, room_number):
|
|
if not hotel:
|
|
self.logger.warning("Room creation skipped: missing hotel.")
|
|
return None
|
|
|
|
if not room_number:
|
|
self.logger.warning(f"Room creation skipped: missing room_number for hotel {hotel.name}.")
|
|
return None
|
|
|
|
external_id = f"{hotel.hotel_id}_{room_number}".lower()
|
|
|
|
room, created = Room.objects.get_or_create(
|
|
hotel=hotel,
|
|
number=room_number,
|
|
defaults={
|
|
"external_id": external_id,
|
|
"description": "Автоматически созданная комната",
|
|
}
|
|
)
|
|
|
|
if created:
|
|
self.logger.info(f"Room '{room.number}' (external_id: {external_id}) created in hotel '{hotel.name}'")
|
|
else:
|
|
self.logger.info(f"Room '{room.number}' already exists in hotel '{hotel.name}'")
|
|
|
|
return room
|
|
|
|
|
|
class DataSyncManager:
|
|
def __init__(self, db_settings_id):
|
|
self.logger = CustomLogger(name="DataSyncManager", log_level="DEBUG").get_logger()
|
|
self.db_connector = DatabaseConnector(db_settings_id)
|
|
self.db_settings = self.db_connector.db_settings # Сохраняем настройки базы данных
|
|
self.data_processor = DataProcessor(self.logger)
|
|
self.hotel_manager = HotelRoomManager(self.logger)
|
|
|
|
def get_last_saved_record(self):
|
|
record = UserActivityLog.objects.order_by("-id").first()
|
|
return record.id if record else 0
|
|
|
|
def fetch_new_data(self, last_id):
|
|
query = f"""
|
|
SELECT * FROM `{self.db_settings.get('table_name')}`
|
|
WHERE id > {last_id}
|
|
AND url_parameters IS NOT NULL
|
|
AND url_parameters LIKE '%utm_medium%'
|
|
AND page_url IS NOT NULL
|
|
ORDER BY id ASC
|
|
LIMIT 1000;
|
|
"""
|
|
|
|
self.logger.info(f"Fetching new data with query: {query}")
|
|
return self.db_connector.execute_query(query)
|
|
|
|
def process_and_save_data(self, rows):
|
|
for row in rows:
|
|
try:
|
|
url_params = self.data_processor.decode_html_entities(row.get("url_parameters", ""))
|
|
params = self.data_processor.url_parameters_parser(url_params)
|
|
|
|
hotel_id = params.get("utm_content")
|
|
room_number = params.get("utm_term")
|
|
page_title = row.get("page_title")
|
|
external_id = row.get("id")
|
|
hits = row.get("hits") or 0
|
|
|
|
hotel = self.hotel_manager.get_or_create_hotel(hotel_id, page_title)
|
|
room = self.hotel_manager.get_or_create_room(hotel, room_number)
|
|
page_url = row.get("page_url")
|
|
|
|
if hits != 0 and page_title is not None:
|
|
UserActivityLog.objects.update_or_create(
|
|
external_id=external_id,
|
|
defaults={
|
|
"user_id": row.get("user_id") or 0,
|
|
"timestamp": row.get("timestamp"),
|
|
"date_time": row.get("date_time"),
|
|
"ip": row.get("ip") or "0.0.0.0",
|
|
"created": self.data_processor.parse_datetime(row.get("created")) or timezone.now(),
|
|
"url_parameters": url_params,
|
|
"page_id": room.id if room else None,
|
|
"page_title": html.unescape(page_title),
|
|
"hits": hits,
|
|
"page_url": html.unescape(page_url),
|
|
}
|
|
)
|
|
else:
|
|
self.logger.warning("Invalid data for UserActivityLog.")
|
|
self.logger.info(f"Record ID {external_id} processed successfully.")
|
|
except Exception as e:
|
|
self.logger.error(f"Error processing record ID {row.get('id')}: {e}")
|
|
|
|
def sync(self):
|
|
self.db_connector.connect()
|
|
try:
|
|
last_id = self.get_last_saved_record()
|
|
rows = self.fetch_new_data(last_id)
|
|
self.process_and_save_data(rows)
|
|
self.logger.info("Sync completed.")
|
|
finally:
|
|
self.db_connector.close()
|
|
|
|
|
|
def scheduled_sync():
|
|
logger = CustomLogger(name="DatabaseSyncScheduler", log_level="DEBUG").get_logger()
|
|
logger.info("Starting scheduled sync.")
|
|
|
|
active_db_settings = ExternalDBSettings.objects.filter(is_active=True)
|
|
if not active_db_settings.exists():
|
|
logger.warning("No active database connections found.")
|
|
return
|
|
|
|
logger.info(f"Found {len(active_db_settings)} active database connections.")
|
|
|
|
def sync_task(db_settings):
|
|
try:
|
|
logger.info(f"Syncing connection: {db_settings.name} (ID={db_settings.id})")
|
|
sync_manager = DataSyncManager(db_settings.id)
|
|
sync_manager.sync()
|
|
logger.info(f"Sync completed for connection: {db_settings}")
|
|
except Exception as e:
|
|
logger.error(f"Error syncing connection {db_settings}: {e}")
|
|
|
|
with ThreadPoolExecutor(max_workers=5) as executor:
|
|
futures = [executor.submit(sync_task, db_settings) for db_settings in active_db_settings]
|
|
for future in futures:
|
|
try:
|
|
future.result(timeout=300)
|
|
except TimeoutError:
|
|
logger.error("Sync task timed out.")
|
|
|
|
logger.info("Scheduled sync completed.")
|