Files
api_utils_converse_v2/background/finstitutions/trading/tick_in_stateful.py
T
2025-02-28 13:35:45 +05:30

609 lines
22 KiB
Python

"""
AUTHOR:
Khushal P Soonderji
DATE:
Wednesday, 1st Jan., 2025.
OBJECTIVE:
To get live updates from Zerodha and push them to Kafka.
REFERENCES:
01. YouTube Webinar: https://www.youtube.com/watch?v=9vzd289Eedk
02. Official Example (GitHub): https://github.com/zerodha/pykiteconnect/blob/master/examples/threaded_ticker.py
DOWNLOADS:
N/A
"""
# *****************************************************************************************************************
# ***** ****
# *** IMPORT ***
# ***** ****
# *****************************************************************************************************************
# To make sibling directories accessible for imports:
import sys
sys.path.append(".")
sys.path.append("..")
# System-level activities:
import io
import os
# My utils:
from utils_v2.string import json
from utils_v2.string import regex
from utils_v2.system import files
from utils_v2.date_time import date_time
from utils_v2.database.async_mongo_v2 import AsyncMongo
from utils_v2.cache.async_redis_cache_v2 import AsyncRedisCache
from utils_v2.queue.kafka.controllers.kafka import ProducerKafka, ConsumerKafka
from utils_v2.serialization.json_serializer import JSONSerializer
# To make HTTP calls:
import httpx
import socket
# To work with date and time:
import datetime
import time
# To work with tabulated data:
import pandas as pd
# Controllers:
from controllers_v2.finstitutions.trading.all_trading import AllTradingController
# Models:
from models.core.auth_token import CoreAuthTokenModel
from models.finstitutions.trading.symbols import TradingSymbol
from models.finstitutions.trading.ticks import TradingTick
# To work with Zerodha's Kite platform:
from kiteconnect import KiteConnect, KiteTicker
# To work with datatypes:
from typing import List
# For debugging:
from icecream import IceCreamDebugger
# For asynchronous operations:
import asyncio
# *****************************************************************************************************************
# ***** ****
# *** MACROS / ONE-TIME INIT ***
# ***** ****
# *****************************************************************************************************************
# To make API calls:
http_client = httpx.Client(
limits = httpx.Limits(
max_connections = 100, # ............ Maximum number of connections allowed in the pool.
max_keepalive_connections = 50, # ... Maximum number of connections that can be kept alive.
),
timeout = httpx.Timeout(
pool = 120.0, # .... Time to wait for a free connection from the pool.
connect = 2.5, # ... Time to wait for establishing a connection to the server.
write = 10.0, # .... Time to wait for sending data.
read = 9.9 # ....... Time to wait for receiving data.
)
)
# For debugging:
printer = IceCreamDebugger(prefix = "Tick-In (Sful) | ", includeContext = True)
no_context_printer = IceCreamDebugger(prefix = "Tick-In (Sful) | ", includeContext = False)
# *****************************************************************************************************************
# ***** ****
# *** VARIABLES ***
# ***** ****
# *****************************************************************************************************************
# To identify this process:
SERVER_HOSTNAME = str(socket.gethostname())
TOKEN_KEY = None
AUTH_TOKEN = None
# Pertaining to the behaviour of this script:
SCRIPT_DATA = {}
# For kafka:
kafka_producer: ProducerKafka | None = None
kafka_consumer: ConsumerKafka | None = None
TOTAL_TICK_COUNT = 0
TICKS_SINCE_FLUSH = 0
# For Zerodha-Kite:
ZERODHA_INSTRUMENT_TOKENS = []
ZERODHA_INSTRUMENT_LOOKUP = {}
# *****************************************************************************************************************
# ***** ****
# *** FUNCTIONS ***
# ***** ****
# *****************************************************************************************************************
def ticks_to_kafka(ticks: List[TradingTick]) -> int:
"""
This function simply takes the ticks received from the stockbroker's servers and sends them to the internal Kafka
queue for various end consumption use cases.
:param ticks: The modeled ticks that must be pushed to the Kafka queue.
:return: The count of the ticks that were successfully pushed to the queue.
"""
# Declare the required global variables:
global TOTAL_TICK_COUNT
global TICKS_SINCE_FLUSH
# Start with basic variables:
total_count = len(ticks)
success_count = 0
failure_count = 0
# Push out the ticks to the queue:
for tick in ticks:
# Flush the existing messages (if needed):
TOTAL_TICK_COUNT += 1
TICKS_SINCE_FLUSH += 1
if TICKS_SINCE_FLUSH > 50_000:
kafka_producer.flush(timeout = 0.0)
TICKS_SINCE_FLUSH = 0
# Push this one tick to the queue:
success = kafka_producer.produce(value = tick.summary)
kafka_producer.client.poll(0)
if success: success_count += 1
else: failure_count += 1
# Debugging print:
ticks_str = f"| TICKS: {total_count:6,} | PRDC'D: {success_count:6,} | TOT: {TOTAL_TICK_COUNT:10,} |"
no_context_printer(ticks_str)
# Done here:
return success_count
# ---------------------------------------------------------------------------------------------------------------------
def on_zerodha_kite_connect(ws, response) -> None:
"""
To do something when we get connected to Zerodha's Kite API successfully.
:param ws: The websocket object that is connected to Zerodha.
:param response: Some input from Zerodha's library.
:return: None
"""
printer("Zerodha Kite WS connected.")
ws.subscribe(ZERODHA_INSTRUMENT_TOKENS)
ws.set_mode(ws.MODE_FULL, ZERODHA_INSTRUMENT_TOKENS)
printer("Zerodha Kite instruments subscribed.", len(ZERODHA_INSTRUMENT_TOKENS))
# ---------------------------------------------------------------------------------------------------------------------
def on_zerodha_kite_ticks(ws, ticks) -> None:
"""
The function that gets called when Zerodha's tick updates come in.
:param ws: The websocket object that is connected to Zerodha.
:param ticks: The actual tick data received from Zerodha's Kite platform.
:return: None
"""
# Model the raw input ticks:
ticks = TradingTick.from_zerodha_kite(
ticks = ticks,
instrument_lookup = ZERODHA_INSTRUMENT_LOOKUP,
received_ts = date_time.get_current_utc_date_time(as_string = False)
)
# Send the ticks to Kafka:
success_count = ticks_to_kafka(ticks)
# ---------------------------------------------------------------------------------------------------------------------
def setup_zerodha_kite_feed(
auth_token: CoreAuthTokenModel,
total_instruments: List[dict]
) -> bool:
"""
Sets up the websocket for Zerodha's Kite platform.
:param auth_token: The auth-token model to use to se the feed up.
:param total_instruments: The list of dicts that describe the instruments we need to subscribe to.
:return: True if successful, else False.
"""
# Ensure that we filter out duplicate records:
total_instruments_df = pd.DataFrame(total_instruments)
# print(total_instruments_df.to_string())
total_instruments_df.drop_duplicates(subset = "broker_token", keep = "first", inplace = True)
total_instruments = total_instruments_df.to_dict(orient = "records")
# Declare the required global variables:
global ZERODHA_INSTRUMENT_TOKENS
global ZERODHA_INSTRUMENT_LOOKUP
# Filter the input instruments. This has to be tuned to match what you get from the API.
valid_broker_symbols = []
valid_broker_tokens = []
for i in total_instruments:
symbol = i["symbol"]
broker_token = i["broker_token"]
if broker_token is not None and i["source"] == "zerodha":
valid_broker_symbols.append(symbol)
valid_broker_tokens.append(str(broker_token))
# Debugging print:
printer(len(total_instruments), len(valid_broker_symbols), len(valid_broker_tokens))
# Create an instance of Zerodha's Kite connection:
kite = KiteConnect(api_key = auth_token.auth["apiKey"])
kite.set_access_token(auth_token.token["accessToken"])
# Get the entire list of instruments:
instruments = []
instruments += kite.instruments(exchange = "NSE")
instruments += kite.instruments(exchange = "NFO")
instruments += kite.instruments(exchange = "BSE")
instruments += kite.instruments(exchange = "BFO")
instruments += kite.instruments(exchange = "MCX")
instruments += kite.instruments(exchange = "CDS")
instruments += kite.instruments(exchange = "BCD")
# Convert the loaded instruments to their modelled form:
instruments = [TradingSymbol.from_zerodha_kite(i) for i in instruments]
# Create the lookup:
# invalid_broken_token_count = 0
# print("INVALID BROKEN TOKENS:\n")
all_temp_instr = {}
pop_count = 0
for i in instruments:
if str(i.brokerToken) in valid_broker_tokens:
ZERODHA_INSTRUMENT_TOKENS.append(i.brokerToken)
ZERODHA_INSTRUMENT_LOOKUP[i.brokerToken] = i.model_dump()
all_temp_instr[i.brokerToken] = i
pop_count += 1
print(f"Populated: {pop_count}")
invalid_broken_token_count = 0
not_found_broker_tokens = []
for broker_token, broker_symbol in zip(valid_broker_tokens, valid_broker_symbols):
broker_token = int(broker_token)
if ZERODHA_INSTRUMENT_LOOKUP.get(broker_token) is None:
not_found_broker_tokens.append(str(broker_token))
try: print(f"FAILED: {broker_token: ^15} | {all_temp_instr[broker_token].symbol: ^30} | {all_temp_instr[broker_token].exchange}")
except: print(f"FAILED: {broker_token: ^15} | {broker_symbol: ^30} | ")
invalid_broken_token_count += 1
print("\nTOTAL:", invalid_broken_token_count)
not_found_df = total_instruments_df[total_instruments_df["broker_token"].isin(not_found_broker_tokens)]
not_found_df['expiry_date'] = pd.to_datetime(not_found_df['expiry_date'])
not_found_df['expiry_date'] = not_found_df['expiry_date'].dt.strftime('%Y-%m-%d')
print(not_found_df.to_string())
while True: pass
printer("Zerodha instruments loaded.")
# If there are no symbols or too many symbols, we return with failure:
if not (1 <= len(ZERODHA_INSTRUMENT_TOKENS) <= 3_000):
printer("Zerodha must have min. 1 and max. 3,000 symbols in one WS.", len(ZERODHA_INSTRUMENT_TOKENS))
return False
# Create the websocket to Zerodha:
kite_ws = KiteTicker(
api_key = auth_token.auth["apiKey"],
access_token = auth_token.token["accessToken"]
)
# Assign the callbacks:
kite_ws.on_connect = on_zerodha_kite_connect
kite_ws.on_ticks = on_zerodha_kite_ticks
# Run the websocket in a background thread and release the main thread:
kite_ws.connect(threaded = True)
# Done here:
return True
# ---------------------------------------------------------------------------------------------------------------------
async def get_auth_token(
script_cred: dict,
token_key: str,
debug: bool
) -> CoreAuthTokenModel | None:
"""
We need the full auth-token model to run the tick-ingestion script. This is a separate special coroutine that will
use the existing async code to fetch the full document from the database.
:param script_cred: The credentials of the script (needed to connect to the database).
:param token_key: The token 'key' whose full auth-token model needs to be fetched.
:param debug: Whether, or not, you would like to print debugging details.
:return: The full auth-token of the account if found, or None if not found.
"""
# ┳┳┓
# ┃┃┃┏┓┏┓┏┓┏┓
# ┛ ┗┗┛┛┗┗┫┗┛
# ┛
data_mongo = AsyncMongo(
connection_string = script_cred["mongoDb"]["data"]["connectionString"],
database_name = script_cred["mongoDb"]["data"]["dbName"],
max_connections = script_cred["mongoDb"]["data"]["poolSize"],
debug = debug
)
await data_mongo.connect()
# ┳┓ ┓• ┏┓ ┓
# ┣┫┏┓┏┫┓┏ ━━ ┃ ┏┓┏┣┓┏┓
# ┛┗┗ ┗┻┗┛ ┗┛┗┻┗┛┗┗
general_cache = AsyncRedisCache(
connection_string = script_cred["redisCache"]["general"]["connectionString"],
serializer = JSONSerializer(),
debug = debug,
debug_prefix = "General Cache | "
)
await general_cache.connect()
# ┏┓ ┓┓ ┓┏┏┓
# ┃ ┏┓┏┓╋┏┓┏┓┃┃┏┓┏┓┏ ┃┃┏┛
# ┗┛┗┛┛┗┗┛ ┗┛┗┗┗ ┛ ┛ ┗┛┗━
controller = AllTradingController(
cache = general_cache,
http_client = httpx.AsyncClient(timeout = 10.0),
alert_url = SCRIPT_DATA["alerts"]["url"],
debug = debug
)
# Fetch and return the auth-token model for the given key:
return await controller.get_token_from_key(
mongo_data_conn = data_mongo,
token_key = token_key
)
# ---------------------------------------------------------------------------------------------------------------------
def init(
script_id: str,
token_key: str,
debug: bool
) -> bool:
"""
To initialize all credentials, instances, and connectivity for this whole script.
:param script_id: The id to use to load cred and data from the internal service.
:param token_key: The key to identify the auth-token that must be used for connecting to the data feed.
:param debug: Whether, or not, you would like to print the debug messages.
:return: True if initialized successfully, else False.
"""
# Declare the required global variables:
global TOKEN_KEY
global AUTH_TOKEN
global SCRIPT_DATA
global kafka_producer
global kafka_consumer
# Basic stuff:
TOKEN_KEY = token_key
if not debug: printer.disable()
# ┏┓ ┓ ┓ ┳┓
# ┃ ┏┓┏┓┏┫ ┏┓┏┓┏┫ ┃┃┏┓╋┏┓
# ┗┛┛ ┗ ┗┻ ┗┻┛┗┗┻ ┻┛┗┻┗┗┻
# Get the script credentials:
response = http_client.get(
url = r"https://nexcom.ditscentre.in/internal/cred/get",
headers = {"X-Script-Id": script_id}
)
if response.status_code not in [200]:
print("FATAL: SCRIPT CREDENTIALS LOADING FAILED!")
return False
script_cred = response.json().get("data")
# Get the script data:
response = http_client.get(
url = r"https://nexcom.ditscentre.in/internal/data/get",
headers = {"X-Script-Id": script_id}
)
if response.status_code not in [200]:
print("FATAL: SCRIPT DATA LOADING FAILED!")
return False
SCRIPT_DATA = response.json().get("data")
# Done with this step:
printer("Cred and Data loaded.")
# ┓┏┓ ┏┓ ┏┓┓•
# ┃┫ ┏┓╋┃┏┏┓ ┃ ┃┓┏┓┏┓╋┏
# ┛┗┛┗┻┛┛┗┗┻ ┗┛┗┗┗ ┛┗┗┛
# Create the producer that will produce tick-by-tick data that it receives from the stockbroker's server:
producer_creds = script_cred["kafka"]["producer"]
kafka_producer = ProducerKafka(
config = ProducerKafka.create_config(
bootstrap_servers = producer_creds["config"]["bootstrapServers"],
security_protocol = producer_creds["config"].get("securityProtocol", "PLAINTEXT"),
ca_file = producer_creds["config"].get("caFile"),
cert_file = producer_creds["config"].get("certFile"),
key_file = producer_creds["config"].get("keyFile"),
client_id = f"{SERVER_HOSTNAME}_{TOKEN_KEY}",
acks = producer_creds["config"].get("acks", 1),
retries = producer_creds["config"].get("retries", 1),
linger_ms = producer_creds["config"].get("lingerMs", 0),
misc_json = {
"queue.buffering.max.messages": 2_00_000
}
),
topic = producer_creds["topic"],
serializer = JSONSerializer(),
debug = debug
)
if not kafka_producer.connect():
print("FATAL: KAFKA PRODUCER NOT CREATED!")
return False
printer("Kafka producer ready.")
# Create the consumer that will listen to changes in watchlist:
consumer_creds = script_cred["kafka"]["consumer"]
kafka_consumer = ConsumerKafka(
config = ConsumerKafka.create_config(
bootstrap_servers = consumer_creds["config"]["bootstrapServers"],
group_id = f"{SERVER_HOSTNAME}_{TOKEN_KEY}",
auto_offset_reset = consumer_creds["config"].get("autoOffsetReset", "latest"),
security_protocol = consumer_creds["config"].get("securityProtocol", "PLAINTEXT"),
ca_file = consumer_creds["config"].get("caFile"),
cert_file = consumer_creds["config"].get("certFile"),
key_file = consumer_creds["config"].get("keyFile"),
client_id = f"{SERVER_HOSTNAME}_{TOKEN_KEY}"
),
topic = consumer_creds["topic"],
serializer = JSONSerializer(),
debug = debug
)
if not kafka_consumer.connect():
print("FATAL: KAFKA CONSUMER NOT CREATED!")
return False
printer("Kafka consumer ready.")
# ┏┓ ┓ ┏┳┓ ┓
# ┣┫┓┏╋┣┓━━ ┃ ┏┓┃┏┏┓┏┓
# ┛┗┗┻┗┛┗ ┻ ┗┛┛┗┗ ┛┗
AUTH_TOKEN = asyncio.run(get_auth_token(
script_cred = script_cred,
token_key = token_key,
debug = debug
))
if not AUTH_TOKEN:
print("FATAL: AUTH-TOKEN FETCHING FAILED!")
return False
printer("Auth-token fetched.")
# ┳┓ ┏┓ ┓
# ┃┃┏┓╋┏┓━━┣ ┏┓┏┓┏┫
# ┻┛┗┻┗┗┻ ┻ ┗ ┗ ┗┻
# First we load the tokens of interest that we need to monitor:
response = http_client.post(
url = r"https://api.thecaoffice.com/markets/watchlist/distincts",
json = {"tokenKey": token_key}
)
instruments_of_interest = response.json()["data"]["rs0"]
# Set the feed up by the stockbroker:
live_feed = False
if AUTH_TOKEN.client == "zerodhaKite":
live_feed = setup_zerodha_kite_feed(
auth_token = AUTH_TOKEN,
total_instruments = instruments_of_interest
)
else: print(f"FATAL: INVALID TRADING CLIENT ({AUTH_TOKEN.client})!")
# If our live feed setup failed:
if not live_feed:
print("FATAL: LIVE FEED SETUP FAILED!")
return False
printer("Live feed ready.")
# ┳┓
# ┃┃┏┓┏┓┏┓
# ┻┛┗┛┛┗┗
# If everything went well, we return with success:
printer("Initialization done.")
return True
# ---------------------------------------------------------------------------------------------------------------------
def main():
while True:
pass
# *****************************************************************************************************************
# ***** ****
# *** MAIN PROGRAM ***
# ***** ****
# *****************************************************************************************************************
if __name__ == "__main__":
# To get args. from the terminal:
import argparse
# Get the config. from the command-line:
parser = argparse.ArgumentParser(
description = (
"Create one stateful process that gets tick-by-tick updates from stock brokers "
"and produces them on the common Kafka broker. This process will be dedicated to one account."
)
)
parser.add_argument(
"-s", "--script-id",
dest = "script_id",
type = str,
help = "The id of this script (will affect the loaded config)."
)
parser.add_argument(
"-t", "--token-key", "--token-id",
dest = "token_key",
type = str,
help = "The 'key' to use to retrieve the auth-token for accessing the broker account."
)
parser.add_argument(
"-d", "--debug",
dest = "debug",
action = "store_true",
help = "Whether, or not, you want to see debugging messages in the terminal.",
default = False
)
args = parser.parse_args()
# Run the main script:
if init(
script_id = args.script_id,
token_key = args.token_key,
debug = args.debug
): main()