Good morning, I know one principle of

hamilton

is that you cannot redefine a node with the same name. However today, I noticed that this is possible in the contrived example below. It was accidentally accomplished via

load_from/inject_

. Is this intended behavior or should an error have been raise?

@subdag(
    scraper_a,
    inputs={"data_location": source("a_data_location")}
)
def a_raw(
    a_scraped_data: pd.DataFrame,
) -> pd.DataFrame:
    return a_scraped_data

@extract_fields({
    "a_processed": pd.DataFrame,
    "b_processed": pd.DataFrame,
    "c_processed": pd.DataFrame
})
@load_from.excel(path=..., inject_="a_raw")
def do_some_work(
    a_raw: pd.DataFrame,
    b_raw: pd.DataFrame,
    c_raw: pd.DataFrame,
) -> Dict[str, pd.DataFrame]: 
    # some work
    return {
        "a_processed": pd.DataFrame(...),
        "b_processed": pd.DataFrame(...),
        "c_processed": pd.DataFrame(...),
    }

Yep this seems like an edge case.

So yes, edge-case, but I don’t think this is something that should break. Specifically, i think that one of these (either

load_from

or

inject

is not creating a node called

a_raw

). Can you print out the visualization of this? Will help debug.

%%cell_to_module scraper_a -d
import pandas as pd
def a_scraped_data(a_data_location: str) -> pd.DataFrame:
    return pd.DataFrame(dict(a=list("xyx")))

a_data_location = "a_sample_data_path.csv"
pd.DataFrame(dict(a=list("abc"))).to_csv(a_data_location)

%%cell_to_module loaders -d

import pandas as pd
from typing import Dict
from hamilton.function_modifiers import subdag, source, extract_fields, load_from

import scraper_a

@subdag(
    scraper_a,
    inputs={"data_location": source("a_data_location")}
)
def a_raw(
    a_scraped_data: pd.DataFrame,
) -> pd.DataFrame:
    print(f"FROM SUBDAG: {a_scraped_data}")
    return a_scraped_data

@extract_fields({
    "a_processed": pd.DataFrame,
    # "b_processed": pd.DataFrame,
    # "c_processed": pd.DataFrame
})
@load_from.csv(path=source("a_data_location"), inject_="a_raw")
def do_some_work(
    a_raw: pd.DataFrame,
    # b_raw: pd.DataFrame,
    # c_raw: pd.DataFrame,
) -> Dict[str, pd.DataFrame]: 
    # some work
    return {
        "a_processed": pd.DataFrame(dict(a=[1,2,3])),
        # "b_processed": pd.DataFrame(dict(b=[3,4,5])),
        # "c_processed": pd.DataFrame(dict(c=[5,6,7])),
    }

import loaders
from hamilton import driver

dr = driver.Builder().with_modules(loaders).build()
dr.execute(["a_processed"], inputs={"a_data_location": a_data_location})

>>> {'a_processed':    a
 0  1
 1  2
 2  3}

So yeah, exactly