假设空单元格是NaN，使用fillna，mask，str.replace：

df['test'] = df['PM No_'].fillna(df['No_']
                                 .mask(df['Condition'].eq(9),
                                       df['No_'].str.replace('-0G$', '',
                                                             regex=True))
                                )

输出量：

PM No_  Condition           No_        test
0  GVSU10K20H          1    GVSU10K20H  GVSU10K20H
1    1Y002VMH          5   1Y002VMH-0H    1Y002VMH
2         NaN          9  2NX00ESRU-0G   2NX00ESRU
3      20WNGE          2        20WNGE      20WNGE
4         NaN          1     C33761-0S   C33761-0S

如果你有空字符串：

df['test'] = (df['PM No_']
              .mask(df['PM No_'].eq(''),
                    df['No_'].mask(df['Condition'].eq(9),
                                   df['No_'].str.replace('-0G$', '',
                                                         regex=True))
                    )
             )

输出量：

PM No_  Condition           No_        test
0  GVSU10K20H          1    GVSU10K20H  GVSU10K20H
1    1Y002VMH          5   1Y002VMH-0H    1Y002VMH
2                      9  2NX00ESRU-0G   2NX00ESRU
3      20WNGE          2        20WNGE      20WNGE
4                      1     C33761-0S   C33761-0S

多个条件

如果有多个条件，因为字符串操作不容易向量化，请使用循环，这里使用自定义函数

def clean(no, condition):
    out = []
    for n, c in zip(no, condition):
        if c == 1:
            n = n.removesuffix('-0S')
        elif c == 9:
            n = n.removesuffix('-0G')
        out.append(n)
    return pd.Series(out, index=no.index)

df['test'] = df['PM No_'].mask(df['PM No_'].eq(''), clean(df['No_'], df['Condition']))

或者，如果所有操作都包括删除后缀：

def clean(no, condition):
    suffixes = {1: '-0S', 9: '-0G'}
    return pd.Series([n.removesuffix(suffixes.get(c, ''))
                      for n, c in zip(no, condition)],
                     index=no.index)

df['test'] = df['PM No_'].mask(df['PM No_'].eq(''), clean(df['No_'], df['Condition']))

输出量：

PM No_  Condition           No_        test
0  GVSU10K20H          1    GVSU10K20H  GVSU10K20H
1    1Y002VMH          5   1Y002VMH-0H    1Y002VMH
2                      9  2NX00ESRU-0G   2NX00ESRU
3      20WNGE          2        20WNGE      20WNGE
4                      1     C33761-0S      C33761