My goal is to pass a string of a file path from one @task into Emailoperators , so I can apply logic with the dataset that I will read from the file path to build up my operators that will send emails. My code looks like this:

from airflow import DAG
from airflow.decorators import dag, task
from airflow.utils.dates import days_ago
from datetime import datetime

import polars as pl

@dag(
    dag_id="data_to_taskgroup_dag",
    start_date=days_ago(2),
    schedule_interval=None,
)
def data_to_taskgroup_dag():
    @task
    def produce_file_path():
        file_path = "path/to/your/dataframe.csv"  
        return file_path

    @task
    def process_dataframe(file_path):
        df = pl.read_csv(file_path)

        for _, row in df.iter_rows():
            email_subject = f"Email for {row['column_name']}"
            email_body = f"Email body: {row['another_column']}"

            EmailOperator(
                task_id=f"send_email_{row['column_name']}",
                to='[email protected]',
                subject=email_subject,
                html_content=email_body
            ).execute(context=kwargs)

    file_path = produce_file_path()
    process_dataframe(file_path)

the problem of this approach is that I am nesting emailoperators into process_dataframe task, and so I cannot really see the emailops in the UI, the nesting is mainly done to be able to get the xcom from upstream task, if i dont decorate process_dataframe and keep as plain function this line will fail df = pl.read_csv(file_path) because now file_path is not a string but a plainxcomarg. What is the approach/practice for this kind of situation ?