### Load data to time dimension
This task populates the `dim_time` table. \
Dimension table is truncated before inserting data.
We run data validation that expects to find a total of \
24 rows for 366 days (2016 was a leap year).
"""
# Add task to Load data into facts table
fact_weather_task = LoadFactOperator(task_id='load_weather_fact_table',
dag=dag,
redshift_conn_id=db_conn_name,
dest_table="fact_weather",
sql_query=SqlQueries.fact_weather_insert,
provide_context=True)
fact_weather_task.doc_md = """\
### Load weather facts
This task populates the `fact_weather` table with \
numerical data.
"""
# Add task to check that we have data in the facts table
run_quality_checks = DataQualityOperator(
task_id='run_data_quality_checks',
dag=dag,
redshift_conn_id=db_conn_name,
table="fact_weather",
sql_query=SqlQueries.row_count,
equals=366 * 24, # we expect to find hourly data for one year
)
run_quality_checks.doc_md = """\
This task populates the `dim_payment_types` table. \
Each task execution appends any rows that are not \
previously found in the table.
"""
# Add data to facts table
fact_trips_task = LoadFactOperator(
task_id='load_trips_fact_table',
dag=dag,
redshift_conn_id=db_conn_name,
dest_table="fact_trips",
source_table=
'staging_trips_{{ macros.ds_format(yesterday_ds, "%Y-%m-%d", "%Y_%m") }}', # noqa
sql_query=SqlQueries.fact_trips_insert,
provide_context=True)
fact_trips_task.doc_md = """\
### Load taxi trips facts
This task populates the `fact_trips` table with \
numerical data. The task writes the sum of \
rows in original table and staging table to a \
xcom called `expected_count`. This information \
is used in the next task when validating inserted data.
"""
# Validate data insert
run_quality_checks = DataQualityOperator(
task_id='run_data_quality_checks',
dag=dag,
redshift_conn_id=db_conn_name,
table="fact_trips",
sql_query=SqlQueries.row_count,
