Home assistant python data downloader

Below is a simple script that pulls sensor data from a home assistant installation and writes it to a csv file.

The first time the script is ran it will pull all the historical data up until that point, subsequent runs will only grab new data not present in the data.csv (or whatever you wish to name it) file. This script is useful in conjunction with a excel spreadsheet that can enable further processing of the data, all that's needed is to run the script to grab new data and then refresh the spreadsheet.

Feel free to leave a comment at the bottom of this page with any issues or suggestions you might have.

Download here, or simply run the python below.

from requests import get
from datetime import *
import os 

uri = "http://HOME_ASSISTANT_IP:PORT"
long_lived_access_token = "YOUR_ACCESS_TOKEN" #Can be created in user profile
sensor_id = "sensor.YOUR_SENSOR_ID" #Can be found in the details of an entity
filename = "data.csv"

def get_data(sensor):
    url = uri + "/api/states/" + sensor
    headers = {
        "Authorization": "Bearer " + long_lived_access_token,
        "content-type": "application/json",
    }
    response = get(url, headers=headers)
    if '{"message":"Entity not found."}' == response.text:
        print("Sensor not found")
        return
    return response.text

def get_attrib(response,attrib):
    index = response.index(attrib)+len(attrib)+3
    return response[index:index+8].split('"')[0]

def get_val(sensor):
    response = get_data(sensor)
    return get_attrib(response,"state")+get_attrib(response,"unit_of_measurement")
    
def get_history(sensor,timestamp):
    print("Starting from:",timestamp,"\n")
    url = uri + "/api/history/period/" + timestamp
    params = {
        "filter_entity_id": sensor,
        "minimal_response": True,
        "no_attributes": True,
        "end_time": datetime.now()
    }
    headers = {
        "Authorization": "Bearer " + long_lived_access_token,
        "content-type": "application/json",
    }
    response = get(url, headers=headers, params=params)
    values = response.text.replace("},{","\n")
    values = values.replace('"state":',"")
    values = values.replace('"last_changed":',"")
    values = values.replace('"',"")
    try:
        return values[values.index("\n")+1:-3]
    except:
        return ""

dir_path = os.path.dirname(os.path.realpath(__file__))
last_line = ""
start_date_str = "2000-01-01T00:00:00+00:00"

try:
    last_line = open(dir_path+"/"+filename).readlines()[-1]
except:
    print("Creating file")
    open(dir_path+"/"+filename,'x').write("Value,Date")
print("Writing in dir",dir_path)

if last_line != "":
    last_date = last_line.split(',')[1]
    if last_date[0] == '"':
        last_date = last_date[1:]
    start_date = datetime.strptime(last_date.split(".")[0], "%Y-%m-%dT%H:%M:%S")
    start_date += timedelta(seconds=1)
    start_date_str = str(start_date)+"+00:00"
values = get_history(sensor_id,start_date_str)
print(values)
no_values = values.count(",")
if no_values > 0:
    open(dir_path+"/"+filename,'a').write("\n"+values)
print(no_values,"new values added")