14-06-2026 2nd commit

This commit is contained in:
Ankit Malik
2026-06-15 13:52:24 +05:30
parent 166a2a1d19
commit e77941c23d
10 changed files with 2844 additions and 772 deletions
+195 -171
View File
@@ -13,238 +13,262 @@
# ///
from __future__ import annotations
import os
# import pyarrow
import sys
from datetime import date, timedelta, datetime
from datetime import date, datetime, timedelta
import polars as pl
import yaml
from dotenv import load_dotenv
from sqlalchemy import create_engine, text
from sqlalchemy.engine import Engine, URL
import clickhouse_connect
from log import log
from clickhouse_task.create_table import create_clickhouse_table
from clickhouse_task.delete_task import (
delete_existing_data,
truncate_table,
)
from clickhouse_task.load_table import load_to_clickhouse
from db_con.connection import (
build_sql_server_engine,
build_clickhouse_engine,
get_clickhouse_client,
)
from mids import (
MID_TABLE_COV,
MID_TABLE_COV1,
)
from clickhouse_task.create_table import *
from clickhouse_task.delete_task import *
from clickhouse_task.load_table import *
from db_con.connection import *
from mids import *
from masters.dimensions import *
from masters.bridge import *
from kpi.facts import *
# ==========================================================
# Helpers
# ==========================================================
def table_exists(
client,
table_name: str,
) -> bool:
return bool(
client.command(
f"EXISTS TABLE {table_name}"
)
)
def get_dataframe(
fn_name: str,
fetch_by: str,
sql_engine,
mids,
run_date,
):
fn = globals()[fn_name]
if fetch_by == "mids":
return fn(sql_engine, mids)
if fetch_by == "run_date":
return fn(sql_engine, run_date)
return fn(sql_engine)
# ==========================================================
# Main
# ==========================================================
def main():
log.info("=" * 80)
log.info("Hello from data-move Python data pipeline !")
log.info("Hello from data-move Python data pipeline!")
# ------------------------------------------------------
# Run Date
# ------------------------------------------------------
if len(sys.argv) > 1:
run_date = datetime.strptime(sys.argv[1], "%Y-%m-%d").date()
run_date = datetime.strptime(
sys.argv[1],
"%Y-%m-%d",
).date()
else:
run_date = date.today() - timedelta(days=1)
log.info(f"Data-pipeline running Date is -:{run_date}")
# connecting with both db servers sql-server
log.info(
"Pipeline Run Date: %s",
run_date,
)
# ------------------------------------------------------
# Connections
# ------------------------------------------------------
log.info(
"Connecting to databases..."
)
log.info("connecting with both db servers sql-serveras well as clickhouse DB")
sql_engine = build_sql_server_engine()
clickhouse_engine = build_clickhouse_engine()
client=get_clickhouse_client()
client = get_clickhouse_client()
log.info("Both databases connected successfully")
log.info(
"Database connections established"
)
mids = MID_TABLE_COV(sql_engine, run_date)
# ------------------------------------------------------
# Delete Keys
# ------------------------------------------------------
mids = MID_TABLE_COV(
sql_engine,
run_date,
)
emp_visit_df = MID_TABLE_COV1(
sql_engine,
run_date
run_date,
)
delete_existing_data(
client=client,
run_date=run_date,
mids=mids,
emp_visit_df=emp_visit_df,
)
# ------------------------------------------------------
# Config
# ------------------------------------------------------
with open(
"t.yml",
"r",
) as file:
mid_list = ",".join(map(str, mids))
conditions = {
"mids": f"MID IN ({mid_list})",
"j_plan": (
f"MONTH(VisitDate) = {run_date.month} "
f"AND YEAR(VisitDate) = {run_date.year}"
),
"mapping": (
f"CAST(Z.FromDate AS DATE) <= '{run_date}' "
f"AND CAST(Z.ToDate AS DATE) >= '{run_date}'"
),
"web": (
f"CAST(login_date AS DATE) = '{run_date}'"
),
"none": None,
}
# fetching polar df from sql-server
with open("tables.yml", "r") as file:
config = yaml.safe_load(file)
for table in config["tables"]:
table_name=table["name"]
table_type=table["type"]
operation=table["operation"]
condition=table["condition"]
c = conditions.get(condition)
log.info("=" * 80)
log.info("TABLE=%s | TYPE=%s | OPERATION=%s",
table_name,
table_type,
operation)
# ------------------------------------------------------
# Process Tables
# ------------------------------------------------------
fn=f"fetch_{table_name}"
for table in config["tables"]:
table_name = table["name"]
operation = table["operation"]
fetch_by = table["fetch_by"]
if operation == "DELETE+INSERT" :
if fetch_by == "mids":
df = globals()[fn](sql_engine, mids)
check_query=f"EXISTS TABLE {table_name}"
exists = client.command(check_query)
if exists == 0 :
create_clickhouse_table(df, table_name, clickhouse_engine)
log.info("=" * 80)
log.info(
"Processing Table: %s",
table_name,
)
else:
truncate_table(client , table_name )
log.info(f"Truncate a ClickHouse table - {table_name}")
try:
load_to_clickhouse(client=client,table_name=table_name,df=df)
# ------------------------------------------
# Fetch Data
# ------------------------------------------
elif fetch_by == "run_date":
df = globals()[fn](sql_engine, run_date)
check_query=f"EXISTS TABLE {table_name}"
exists = client.command(check_query)
if exists == 0 :
create_clickhouse_table(df, table_name, clickhouse_engine)
fn_name = f"fetch_{table_name}"
else:
truncate_table(client , table_name )
log.info(f"Truncate a ClickHouse table - {table_name}")
df = get_dataframe(
fn_name=fn_name,
fetch_by=fetch_by,
sql_engine=sql_engine,
mids=mids,
run_date=run_date,
)
if df.is_empty():
log.warning(
"%s returned no rows",
table_name,
)
continue
log.info(
"Fetched %s rows",
len(df),
)
# ------------------------------------------
# Create Table If Missing
# ------------------------------------------
exists = table_exists(
client,
table_name,
)
if not exists:
log.info(
"Creating table %s",
table_name,
)
create_clickhouse_table(
df=df,
table_name=table_name,
clickhouse_engine=clickhouse_engine,
)
# ------------------------------------------
# Existing Table Logic
# ------------------------------------------
else:
if operation == "DELETE+INSERT":
truncate_table(
client,
table_name,
)
load_to_clickhouse(client=client,table_name=table_name,df=df)
else:
df = globals()[fn](sql_engine)
check_query=f"EXISTS TABLE {table_name}"
exists = client.command(check_query)
if exists == 0 :
create_clickhouse_table(df, table_name, clickhouse_engine)
else:
truncate_table(client , table_name )
log.info(f"Truncate a ClickHouse table - {table_name}")
delete_existing_data(
client=client,
table_name=table_name,
run_date=run_date,
mids=mids,
emp_visit_df=emp_visit_df,
)
load_to_clickhouse(client=client,table_name=table_name,df=df)
else:
if fetch_by == "mids":
df = globals()[fn](sql_engine, mids)
check_query=f"EXISTS TABLE {table_name}"
exists = client.command(check_query)
if exists == 0 :
create_clickhouse_table(df, table_name, clickhouse_engine)
# ------------------------------------------
# Load Data
# ------------------------------------------
else:
load_to_clickhouse(
client=client,
table_name=table_name,
df=df,
)
delete_rows(client, table_name, c )
log.info(
"%s loaded successfully (%s rows)",
table_name,
len(df),
)
load_to_clickhouse(client=client,table_name=table_name,df=df)
elif fetch_by == "run_date":
df = globals()[fn](sql_engine, run_date)
check_query=f"EXISTS TABLE {table_name}"
exists = client.command(check_query)
if exists == 0 :
create_clickhouse_table(df, table_name, clickhouse_engine)
else:
delete_rows(client, table_name, c)
load_to_clickhouse(client=client,table_name=table_name,df=df)
else:
df = globals()[fn](sql_engine)
check_query=f"EXISTS TABLE {table_name}"
exists = client.command(check_query)
if exists == 0 :
create_clickhouse_table(df, table_name, clickhouse_engine)
else:
truncate_table(client , table_name )
log.info(f"Truncate a ClickHouse table - {table_name}")
load_to_clickhouse(client=client,table_name=table_name,df=df)
except Exception:
log.exception(
"Failed processing table %s",
table_name,
)
raise
log.info("=" * 80)
log.info("Pipeline Completed Successfully")
log.info("=" * 80)
if __name__ == "__main__":