df비교

import pandas as pd
import numpy as np

# 예시 데이터 (NaN 포함)
df1 = pd.DataFrame({'A': [1, 2, np.nan], 'B': ['a', 'b', 'c']}, index=[0, 1, 2])
df2 = pd.DataFrame({'A': [1, 2, 4], 'B': ['a', 'x', np.nan]}, index=[1, 2, 3])

# 1. merge 실행 (outer join)
merged = pd.merge(
    df1.reset_index(drop=True), 
    df2.reset_index(drop=True), 
    how='outer', 
    indicator='_source',
    on=list(df1.columns),
    suffixes=('', '_y')
)

# 2. 다른 행 필터링
diff_rows = merged[merged['_source'] != 'both'].copy()
diff_rows['_source'] = diff_rows['_source'].replace({
    'left_only': 'df1', 
    'right_only': 'df2'
})

# 3. NaN 안전 처리 + 하이라이트
for idx, row in diff_rows.iterrows():
    source = row['_source']
    other_df = df2 if source == 'df1' else df1
    pos = idx
    
    try:
        other_row = other_df.iloc[pos]
        for col in df1.columns:
            val = row[col]
            other_val = other_row[col]
            
            # NaN 비교 안전 처리 (pd.isna() 사용)
            if (pd.isna(val) and not pd.isna(other_val)) or \
               (not pd.isna(val) and pd.isna(other_val)) or \
               (not pd.isna(val) and not pd.isna(other_val) and val != other_val):
                diff_rows.at[idx, col] = f"{val} ({source})" if not pd.isna(val) else f"NaN ({source})"
    except IndexError:
        pass

# 4. 최종 결과
diff_rows = diff_rows[df1.columns.tolist() + ['_source']]
print(diff_rows)