Why does my append dataframe in a function not work?

Question

This is a portion of my code:

df3      = pd.DataFrame(columns=["colA","colB"])

def someFunction(filePath,file,df): 
    df2      = pd.DataFrame(columns=["colA","colB"])
    df1      = pd.read_csv(filePath,header=0)
    var      = ["foo1","bar1"]

    df2['colA'] = ["foo2","bar2"]
    df2['colB'] = var

    df = df.append(df2)
    df1 = pd.DataFrame()
    df2 = pd.DataFrame()
    return df

for root, dirs, fileList in os.walk(someDir):
    for file in fileList:
        if <someCondition>:
            someFunction(filePath,file,df3)

print df3

On running the code, I don't see df2 getting appended to df3. Instead, the value of df3 within the function is the same as df3 and empty dataframe outside the function.

How do I consistently append df2 to df3 and get df3 to grow for every file processed?

In your example the `file` arg isn't used at all and the df2 is the same every time it's created - is this pertinent to the problem? — Toby Petty, Apr 06 '19 at 04:18
What happens if you concanate your Dataframes? frames = [df1, df3], df3 = pd.concat(frames, axis = 1) — hemanta, Apr 06 '19 at 04:18
I use `file` in generating `df2['colA']`. I use `df2` as a local variable that creates a sub-dataframe for every `file` and then appends to `df3` — kurious, Apr 06 '19 at 12:12
@hemanta: concat adds columns to the dataframe. I need to add rows — kurious, Apr 06 '19 at 12:17
Tried this `frames = [df2,df] df = pd.concat(frames, axis = 0)` Same result — kurious, Apr 06 '19 at 12:22
You are not mutating the input but returning a new copy. You need to mutate or else call as `df3 = somefunction(..., df)`. Reassigning the result _inside a function_ won't work without a `global` keyword. — Andras Deak -- Слава Україні, Apr 06 '19 at 12:35
I got the code to work with `df3 = someFunction(filePath,file,df3)`. If you post your response as the answer, I will accept it. — kurious, Apr 07 '19 at 02:24

Why does my append dataframe in a function not work?

0 Answers0