Idneo code¶

In [1]:

import os

os.chdir("..")

import os os.chdir("..")

In [2]:

%pip install -r tools/vehicle/requirements.txt

%pip install -r tools/vehicle/requirements.txt

Requirement already satisfied: async-timeout==4.0.3 in /opt/conda/lib/python3.10/site-packages (from -r tools/vehicle/requirements.txt (line 1)) (4.0.3)
Requirement already satisfied: redis==5.0.0 in /opt/conda/lib/python3.10/site-packages (from -r tools/vehicle/requirements.txt (line 2)) (5.0.0)
WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv
Note: you may need to restart the kernel to use updated packages.

In [81]:

import pandas as pd
from tools.vehicle.src.csv_parser import proccess_csv
from tools.vehicle.src.data_handler import order_by_ts_relative
import matplotlib.pyplot as plt
import argparse
import time
import redis
from lightgbm import LGBMClassifier
import pandas as pd
import ast
import redis
from influxdb_client import InfluxDBClient, BucketsApi, Point
from influxdb_client.client.organizations_api import OrganizationsApi
from influxdb_client.client.write_api import SYNCHRONOUS

import pandas as pd from tools.vehicle.src.csv_parser import proccess_csv from tools.vehicle.src.data_handler import order_by_ts_relative import matplotlib.pyplot as plt import argparse import time import redis from lightgbm import LGBMClassifier import pandas as pd import ast import redis from influxdb_client import InfluxDBClient, BucketsApi, Point from influxdb_client.client.organizations_api import OrganizationsApi from influxdb_client.client.write_api import SYNCHRONOUS

In [4]:

data_sets = [
    "DS1_stopped_with_ignition_on_22Feb24_115812.csv",
    "DS1_stopped_with_ignition_on_25Jan24_124019.csv",
    "DS1_stopped_with_ignition_on_25Jan24_151531.csv",
    "DS1_stopped_with_ignition_on_25Mar24_153740.CSV",
    "DS2_national_road_90km_h_max_25Jan24_153019.csv",
    "DS2_national_road_90km_h_max_25Mar24_133516.CSV",
    "DS3_highway_120km_h_max_22Feb24_121145.csv",
    "DS3_highway_120km_h_max_25Mar24_154857.csv",
]
file = "tools/vehicle/datasets/ateca_R4_2.0l_TDI/" + data_sets[0]
# Pre-process csv dataset
data_ordered_by_sensor = proccess_csv(file)

# Timestamp sorting
data_ordered_by_time = order_by_ts_relative(data_ordered_by_sensor)

data_sets = [ "DS1_stopped_with_ignition_on_22Feb24_115812.csv", "DS1_stopped_with_ignition_on_25Jan24_124019.csv", "DS1_stopped_with_ignition_on_25Jan24_151531.csv", "DS1_stopped_with_ignition_on_25Mar24_153740.CSV", "DS2_national_road_90km_h_max_25Jan24_153019.csv", "DS2_national_road_90km_h_max_25Mar24_133516.CSV", "DS3_highway_120km_h_max_22Feb24_121145.csv", "DS3_highway_120km_h_max_25Mar24_154857.csv", ] file = "tools/vehicle/datasets/ateca_R4_2.0l_TDI/" + data_sets[0] # Pre-process csv dataset data_ordered_by_sensor = proccess_csv(file) # Timestamp sorting data_ordered_by_time = order_by_ts_relative(data_ordered_by_sensor)

['Vehicle speed', 'Time since engine start', 'Normed load value', 'Accelerator pedal position', 'Engine torque', 'Oil fill level', 'Engine oil temperature', 'Fuel level', 'Fuel consumption', 'Brake pressure', 'Engaged gear: raw signal-Bits 0-7', 'Efficiency of the SCR catalytic converter']
Dataset has begun after  0 seconds...

In [5]:

data_ordered_by_sensor[0].keys()

data_ordered_by_sensor[0].keys()

Out[5]:

dict_keys(['sensor', 'unit', 'values'])

In [6]:

data_ordered_by_sensor[0]["sensor"], data_ordered_by_sensor[0][
    "unit"
], data_ordered_by_sensor[0]["values"][:5]

data_ordered_by_sensor[0]["sensor"], data_ordered_by_sensor[0][ "unit" ], data_ordered_by_sensor[0]["values"][:5]

Out[6]:

('Vehicle_speed',
 ' km/h',
 [{'ts_relative': 1.74, 'value': 0.0},
  {'ts_relative': 3.19, 'value': 0.0},
  {'ts_relative': 4.69, 'value': 0.0},
  {'ts_relative': 6.17, 'value': 0.0},
  {'ts_relative': 7.61, 'value': 0.0}])

In [7]:

data_ordered_by_time[:5]

data_ordered_by_time[:5]

Out[7]:

[{0.14: {'sensor': 'Efficiency_of_the_SCR_catalytic_converter', 'value': 0.0}},
 {0.36: {'sensor': 'Time_since_engine_start', 'value': 0.0}},
 {0.46: {'sensor': 'Normed_load_value', 'value': 23.1}},
 {0.57: {'sensor': 'Accelerator_pedal_position', 'value': 14.5}},
 {0.69: {'sensor': 'Engine_torque', 'value': 48.4}}]

In [32]:

df = pd.read_csv(file)

df = pd.read_csv(file)

In [33]:

df.head()

df.head()

Out[33]:

	Unnamed: 0	STAMP	Vehicle speed	STAMP.1	Time since engine start	STAMP.2	Normed load value	STAMP.3	Accelerator pedal position	STAMP.4	...	Fuel level	STAMP.8	Fuel consumption	STAMP.9	Brake pressure	STAMP.10	Engaged gear: raw signal-Bits 0-7	STAMP.11	Efficiency of the SCR catalytic converter	Unnamed: 25
0	NaN	NaN	km/h	NaN	s	NaN	%	NaN	%	NaN	...	l	NaN	l/h	NaN	MPa	NaN		NaN		NaN
1	NaN	0.24	0	0.36	0	0.46	23.1	0.57	14.5	0.69	...	24.00	1.23	1.15	1.35	3.63	1.49	0	0.14	0.00000	NaN
2	NaN	1.74	0	1.86	1	1.95	32.5	2.07	14.5	2.19	...	24.00	2.69	1.19	2.81	3.63	2.95	0	1.64	0.00000	NaN
3	NaN	3.19	0	3.28	3	3.39	41.2	3.50	14.5	3.62	...	24.00	4.16	1.15	4.30	3.54	4.43	0	3.09	0.00000	NaN
4	NaN	4.69	0	4.80	4	4.91	34.9	5.02	14.5	5.15	...	24.00	5.68	1.07	5.80	3.51	5.93	0	4.58	0.00363	NaN

5 rows × 26 columns

In [34]:

df.drop(columns=["Unnamed: 0", "Unnamed: 25"], inplace=True)
df.drop(index=0, inplace=True)

df.drop(columns=["Unnamed: 0", "Unnamed: 25"], inplace=True) df.drop(index=0, inplace=True)

In [35]:

timestamp_columns = [col for col in df.columns if col.startswith("STAMP")]

timestamp_columns = [col for col in df.columns if col.startswith("STAMP")]

In [36]:

df[timestamp_columns].diff().hist()
plt.suptitle("Distribution of time stmap differences between samples from each sensor");

df[timestamp_columns].diff().hist() plt.suptitle("Distribution of time stmap differences between samples from each sensor");

Check if the timestamp values are not overlapping¶

Chce if the data format is correct and the data could be used in "tabular chunks" for prediction in the model, i.e. if the maximum timestamp in each row is lower than the minimum timestamp in the following row

In [37]:

temp = df[timestamp_columns].min(axis=1)[1:].reset_index(drop=True) - df[
    timestamp_columns
].max(axis=1)[:-1].reset_index(drop=True)

temp = df[timestamp_columns].min(axis=1)[1:].reset_index(drop=True) - df[ timestamp_columns ].max(axis=1)[:-1].reset_index(drop=True)

In [38]:

(temp > 0).sum()

(temp > 0).sum()

Out[38]:

428

In [39]:

(temp <= 0).sum()

(temp <= 0).sum()

Out[39]:

0

Model¶

Add some class to train the model¶

In [40]:

df_train = df.drop(columns=timestamp_columns)
# keep only the first timestamp column
df_train["timestamp"] = df["STAMP"]
df_train["class"] = 0
df_train.loc[:100, ["class"]] = 1
df_train.loc[:100, ["Normed load value"]] = 100

df_train = df.drop(columns=timestamp_columns) # keep only the first timestamp column df_train["timestamp"] = df["STAMP"] df_train["class"] = 0 df_train.loc[:100, ["class"]] = 1 df_train.loc[:100, ["Normed load value"]] = 100

In [ ]:

df_train[df_train.drop(columns=["class"]).columns] = df_train[
    df_train.drop(columns=["class"]).columns
].astype(float)

df_train[df_train.drop(columns=["class"]).columns] = df_train[ df_train.drop(columns=["class"]).columns ].astype(float)

In [ ]:

# Remove special characters from column names
df_train.columns = df_train.columns.str.replace("[^A-Za-z0-9]+", "_", regex=True)

# Remove special characters from column names df_train.columns = df_train.columns.str.replace("[^A-Za-z0-9]+", "_", regex=True)

In [51]:

# add some vehicle id
df_train["vehicle_id"] = "abc123"

# add some vehicle id df_train["vehicle_id"] = "abc123"

Train the model¶

In [52]:

model = LGBMClassifier()
model.fit(
    df_train.drop(columns=["class", "timestamp", "vehicle_id"]), df_train["class"]
)

model = LGBMClassifier() model.fit( df_train.drop(columns=["class", "timestamp", "vehicle_id"]), df_train["class"] )

[LightGBM] [Info] Number of positive: 100, number of negative: 329
[LightGBM] [Info] Auto-choosing col-wise multi-threading, the overhead of testing was 0.000439 seconds.
You can set `force_col_wise=true` to remove the overhead.
[LightGBM] [Info] Total Bins 516
[LightGBM] [Info] Number of data points in the train set: 429, number of used features: 9
[LightGBM] [Info] [binary:BoostFromScore]: pavg=0.233100 -> initscore=-1.190888
[LightGBM] [Info] Start training from score -1.190888
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf
[LightGBM] [Warning] No further splits with positive gain, best gain: -inf

Out[52]:

LGBMClassifier()

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

In [117]:

df_pred = df_train.iloc[95:105][["timestamp", "vehicle_id"]].reset_index(drop=True)
df_pred["prediction"] = model.predict(
    df_train.iloc[95:105]
    .drop(columns=["class", "timestamp", "vehicle_id"])
    .reset_index(drop=True)
)

df_pred["timestamp"] = pd.to_datetime(
    time.time() - 3600 + df_pred["timestamp"], unit="s"
)

df_pred = df_train.iloc[95:105][["timestamp", "vehicle_id"]].reset_index(drop=True) df_pred["prediction"] = model.predict( df_train.iloc[95:105] .drop(columns=["class", "timestamp", "vehicle_id"]) .reset_index(drop=True) ) df_pred["timestamp"] = pd.to_datetime( time.time() - 3600 + df_pred["timestamp"], unit="s" )

In [118]:

df_pred

df_pred

Out[118]:

	timestamp	vehicle_id	prediction
0	2024-07-30 11:56:40.044199467	abc123	1
1	2024-07-30 11:56:41.464199305	abc123	1
2	2024-07-30 11:56:42.954199314	abc123	1
3	2024-07-30 11:56:44.474199295	abc123	1
4	2024-07-30 11:56:45.914199352	abc123	1
5	2024-07-30 11:56:47.334199429	abc123	0
6	2024-07-30 11:56:48.894199371	abc123	0
7	2024-07-30 11:56:50.384199381	abc123	0
8	2024-07-30 11:56:51.864199400	abc123	0
9	2024-07-30 11:56:53.384199381	abc123	0

In [120]:

# # https://github.com/influxdata/helm-charts/blob/master/charts/influxdb2/values.yaml
# INFLUXDB_IP = "10.152.183.219"
# INFLUXDB_PORT = 80
# INFLUXDB_USER = "admin"
# INFLUXDB_PASS = "admin_pass"

# influxdbClient = InfluxDBClient(url=f"http://{INFLUXDB_IP}:{str(INFLUXDB_PORT)}", username=INFLUXDB_USER, password=INFLUXDB_PASS, org="influxdata")
# # influxdbClient = InfluxDBClient(url=f"http://{INFLUXDB_IP}:{str(INFLUXDB_PORT)}", token="admin_token", org="influxdata")

# write_api = influxdbClient.write_api(write_options=SYNCHRONOUS)

# Function to convert a DataFrame row to an InfluxDB Point
def df_row_to_point(row):
    return (
        Point("prediction")
        .tag("vehicle_id", row["vehicle_id"])
        .field("value", row["prediction"])
        .time(row["timestamp"])
    )


# Convert DataFrame to InfluxDB Points and write
points = [df_row_to_point(row) for index, row in df_pred.iterrows()]

# Write points to InfluxDB
write_api.write(bucket="default", record=points)

# # https://github.com/influxdata/helm-charts/blob/master/charts/influxdb2/values.yaml # INFLUXDB_IP = "10.152.183.219" # INFLUXDB_PORT = 80 # INFLUXDB_USER = "admin" # INFLUXDB_PASS = "admin_pass" # influxdbClient = InfluxDBClient(url=f"http://{INFLUXDB_IP}:{str(INFLUXDB_PORT)}", username=INFLUXDB_USER, password=INFLUXDB_PASS, org="influxdata") # # influxdbClient = InfluxDBClient(url=f"http://{INFLUXDB_IP}:{str(INFLUXDB_PORT)}", token="admin_token", org="influxdata") # write_api = influxdbClient.write_api(write_options=SYNCHRONOUS) # Function to convert a DataFrame row to an InfluxDB Point def df_row_to_point(row): return ( Point("prediction") .tag("vehicle_id", row["vehicle_id"]) .field("value", row["prediction"]) .time(row["timestamp"]) ) # Convert DataFrame to InfluxDB Points and write points = [df_row_to_point(row) for index, row in df_pred.iterrows()] # Write points to InfluxDB write_api.write(bucket="default", record=points)

Query/collect data from redis and send it to influxdb¶

This is a case when the data is not forwarded directly to inference service, i.e. legacy solution made due to previous agreements in the project

Subcsribe to channel and define actions¶

In [65]:

REDIS_PASSWORD = "redis"
REDIS_IP = "10.152.183.169"
REDIS_PORT = 6379
INFLUXDB_IP = "10.152.183.216"
INFLUXDB_PORT = 8086
INFLUXDB_USER = "admin"
INFLUXDB_PASS = "admin"

redisClient = redis.Redis(host=REDIS_IP, password=REDIS_PASSWORD, port=REDIS_PORT)
influxdbClient = InfluxDBClient(
    url=f"http://{INFLUXDB_PORT}:{str(INFLUXDB_PORT)}",
    username=INFLUXDB_USER,
    password=INFLUXDB_PASS,
)


def handle_message(message):
    print(f"New data arrived:\n{message['data']}")
    # Convert the message data (bytes) to string
    json_data = message["data"].decode("utf-8")
    # Convert JSON string to DataFrame
    df = pd.read_json(json_data, orient="split")
    print(df)
    df_pred = model.predict(df.drop(columns=["class", "timestamp", "vehicle_id"]))
    print(f"Predictions:\n{df_pred}")


# Subscribe to the "idneo_v2x" channel
pubsub = redisClient.pubsub()
pubsub.subscribe(**{"idneo_v2x": handle_message})

# Wait for messages
pubsub.run_in_thread(sleep_time=1)

REDIS_PASSWORD = "redis" REDIS_IP = "10.152.183.169" REDIS_PORT = 6379 INFLUXDB_IP = "10.152.183.216" INFLUXDB_PORT = 8086 INFLUXDB_USER = "admin" INFLUXDB_PASS = "admin" redisClient = redis.Redis(host=REDIS_IP, password=REDIS_PASSWORD, port=REDIS_PORT) influxdbClient = InfluxDBClient( url=f"http://{INFLUXDB_PORT}:{str(INFLUXDB_PORT)}", username=INFLUXDB_USER, password=INFLUXDB_PASS, ) def handle_message(message): print(f"New data arrived:\n{message['data']}") # Convert the message data (bytes) to string json_data = message["data"].decode("utf-8") # Convert JSON string to DataFrame df = pd.read_json(json_data, orient="split") print(df) df_pred = model.predict(df.drop(columns=["class", "timestamp", "vehicle_id"])) print(f"Predictions:\n{df_pred}") # Subscribe to the "idneo_v2x" channel pubsub = redisClient.pubsub() pubsub.subscribe(**{"idneo_v2x": handle_message}) # Wait for messages pubsub.run_in_thread(sleep_time=1)

Out[65]:

<PubSubWorkerThread(Thread-6, started 128202013926976)>

Publish the data into the channel and make predictions¶

In [ ]:

df_test_json = df_train.iloc[95:105].reset_index(drop=True).to_json(orient="split")


time.sleep(2)  # Wait for a few seconds before publishing
redisClient.publish("idneo_v2x", df_test_json)

df_test_json = df_train.iloc[95:105].reset_index(drop=True).to_json(orient="split") time.sleep(2) # Wait for a few seconds before publishing redisClient.publish("idneo_v2x", df_test_json)

Send predictions to InfluxDB¶

• https://www.influxdata.com/blog/writing-data-to-influxdb-with-python/
• https://gist.github.com/noahcrowley/941e87422cd6fc43b0e9e8f0d0877836/#file-write_test-py

In [ ]: