import pandas as pd
import getString
import getNLP
import os

def one_hot_encode_objects(df,nlp,columns):
    object_cols = df.select_dtypes(include='object').columns
    for col in object_cols:
        if col not in nlp and col not in columns and "label" not in col:
            if df[col].apply(lambda x: isinstance(x, (list, tuple, dict, set)) or hasattr(x, '__array__')).any():
                print(f"Skipping column '{col}' due to unhashable values.")
                continue
            dummies = pd.get_dummies(df[col], prefix=col).astype(int)
            df = pd.concat([df, dummies], axis=1)
            df = df.drop(columns=[col])
    print(df.columns)
    return df

def fixEmpty(df):
    df.replace(['undefined', 'null', 'NaN', 'None'], pd.NA, inplace=True)
    for col in df.columns:
        if df[col].dtype == 'object':
            df[col] = df[col].fillna('Unknown')
        else:
            df[col] = df[col].fillna(df[col].mean())
    
    return df

def preprocessing(query):
    os.makedirs("./processed",exist_ok=True)
    df=pd.read_csv("final/"+query+".csv")
    # print(df.head())
    df=fixEmpty(df)
    preDict,col,nlp=getString.getCodes(query)
    if len(col)>0:
        for new_col, expr in preDict.items():
            df[new_col] = eval(expr)
        df.drop(columns=col, inplace=True)
    if len(nlp)>0:
        df=getNLP.wordEmbed(df,nlp)
        # print(df.columns)
    df=one_hot_encode_objects(df,nlp,col)
    # df = df.astype('float32')
    df.to_csv("./processed/"+query+".csv", index=False)
    # print(df.head())
    # print(df.info())
    
# preprocessing("twitter sentiment analysis")