Files
ASE/transform_file.py
2024-12-01 11:47:58 +01:00

148 lines
5.3 KiB
Python
Executable File

#!/usr/bin/env python3
import sys
import os
import re
import json
import psycopg2
import logging
from sqlalchemy import create_engine, text
from utils.time import timestamp_fmt as ts
from utils.config import set_config as setting
def extract_value(patterns, source, default='Not Defined'):
ip = {}
for pattern in patterns:
pattern = f'r"{pattern}:\s*(\d{1,3}(?:\.\d{1,3}){3})"'
matches = re.search(pattern, source, re.IGNORECASE)
if matches:
ip.update({pattern: matches.group(1)})
else:
ip.update({pattern: default})
return ip
def write_db(engine, records):
with engine.connect() as conn:
conn.execute(text("""
INSERT INTO dataraw (unit_name, unit_type, tool_name, tool_type, ip_centralina, ip_gateway, event_timestamp, battery_level, temperature, nodes_jsonb)
VALUES
""" + ",".join([
f"(:{i}_unit_name, :{i}_unit_type, :{i}_tool_name, :{i}_tool_type, :{i}_ip_centralina, :{i}_ip_gateway, :{i}_event_timestamp, :{i}_battery_level, :{i}_temperature, :{i}_nodes_jsonb)"
for i in range(len(records))
]) + """
ON CONFLICT ON CONSTRAINT dataraw_unique
DO UPDATE SET
unit_type = EXCLUDED.unit_type,
tool_type = EXCLUDED.tool_type,
ip_centralina = EXCLUDED.ip_centralina,
ip_gateway = EXCLUDED.ip_gateway,
battery_level = EXCLUDED.battery_level,
temperature = EXCLUDED.temperature,
nodes_jsonb = EXCLUDED.nodes_jsonb;
"""), {f"{i}_{key}": value for i, record in enumerate(records) for key, value in record.items()})
conn.commit()
def elab_csv(engine, cfg):
with engine.connect() as conn:
cur = conn.cursor()
cur.execute(f'select unit_name, unit_type, tool_name, tool_type, tool_data from {cfg.dbrectable } r ')
unit_name, unit_type, tool_name, tool_type, tool_data = cur.fetchone()
data_list = str(tool_data).strip("('{\"").strip("\"}\',)").split('","')
# Estrarre le informazioni degli ip dalla header
infos = extract_value(cfg.csv_infos, data_list[:9])
records = []
# Elabora le righe dei dati a partire dalla riga 8 in poi
for line in data_list:
if ";|;" in line:
# Rimuovi spazi bianchi o caratteri di nuova riga
input_data = line.strip().replace('\\n', '')
# Suddividi la stringa in sezioni usando ";|;" come separatore
parts = input_data.split(';|;')
# Verifica che ci siano almeno tre parti (timestamp, misure e nodi)
if len(parts) < 3:
print(f"Riga non valida: {input_data}")
continue
# Estrai la data/ora e le prime misurazioni
timestamp = parts[0]
measurements = parts[1]
# Estrai i valori di ciascun nodo e formatta i dati come JSON
nodes = parts[2:]
node_list = []
for i, node_data in enumerate(nodes, start=1):
node_dict = {"num": i}
# Dividi ogni nodo in valori separati da ";"
node_values = node_data.split(';')
for j, value in enumerate(node_values, start=0):
# Imposta i valori a -9999 se trovi "Dis."
node_dict['val' + str(j)] = -9999 if value == "Dis." else float(value)
node_list.append(node_dict)
# Prepara i dati per l'inserimento/aggiornamento
record = {
"unit_name": unit_name.upper(),
"unit_type": unit_type.upper(),
"tool_name": tool_name.upper(),
"tool_type": tool_type.upper(),
"ip_centralina": infos['IP'],
"ip_subnet": infos['Subnet'],
"ip_gateway": infos['Gateway'],
"Web_port": infos['Web port'],
"Ftp_port": infos['Ftp port'],
"event_timestamp": timestamp,
"battery_level": float(measurements.split(';')[0]),
"temperature": float(measurements.split(';')[1]),
"nodes_jsonb": json.dumps(node_list) # Converti la lista di dizionari in una stringa JSON
}
records.append(record)
# Se abbiamo raggiunto 500 record, esegui l'inserimento in batch
if len(records) >= 500:
print("raggiunti 500 record scrivo sul db")
write_db(engine, records)
records = []
write_db(engine, records)
def main():
# Load the configuration settings
cfg = setting.config()
try:
# Configura la connessione al database PostgreSQL
engine = create_engine(f'postgresql://{cfg.dbuser}:{cfg.dbpass}@{cfg.dbhost}:{cfg.dbport}/{cfg.dbschema}')
# Configure logging
logging.basicConfig(
format="%(asctime)s %(message)s",
filename=cfg.logfilename,
level=logging.INFO,
)
elab_csv(engine, cfg)
except KeyboardInterrupt:
logging.info(
"Info: {}.".format("Shutdown requested...exiting")
)
except Exception:
print(
"{} - PID {:>5} >> Error: {}.".format(
ts.timestamp("log"), os.getpid(), sys.exc_info()[1]
)
)
if __name__ == "__main__":
main()