Python 列增量计数器并根据其他列重置

可以使用GroupBy.cumcount对连续值进行分组，最后设置0以Series.mask对status=True个值进行分组:

s = df['status'].eq(True)
df['count'] = (df.groupby(['car', s.ne(s.shift()).cumsum()])
                 .cumcount()
                 .add(1)
                 .mask(df['status'], 0))

print(df)
      car  status  count
0    audi   False      1
1    audi   False      2
2    audi   False      3
3    audi    True      0
4     bmw   False      1
5     bmw   False      2
6     bmw   False      3
7     bmw    True      0
8     bmw   False      1
9   lexus    True      0
10  lexus    True      0
11  lexus   False      1

替代解决方案，仅适用于布尔列:

s = df['status'].eq(True)
df['count'] = (df.groupby(['car', (~s & s.shift(fill_value=False)).cumsum()])
                 .cumcount()
                 .add(1)
                 .mask(df['status'], 0))

编辑:

data = [['audi', False],        
        ['audi', False],         
        ['bmw', False],        
        ['bmw', False],        
        ['bmw', ''],         
        ['bmw', ''],         
        ['bmw', False],         
        ['bmw', True],         
        ['bmw', False],         
        ['lexus', True],         
        ['lexus', True],         
        ['lexus', False]]
df = pd.DataFrame(data=data, columns=['car', 'status'])

m = df['status'].isin([True, False])
df1 = df[m].copy()
df.loc[m, 'count'] = (df1.groupby(['car', df1['status'].ne(df1['status'].shift()).cumsum()])
                        .cumcount())

df['count'] = df.groupby('car')['count'].bfill().fillna(0).astype('int')
print(df)
      car status  count
0    audi  False      0
1    audi  False      1
2     bmw  False      0
3     bmw  False      1
4     bmw             2
5     bmw             2
6     bmw  False      2
7     bmw   True      0
8     bmw  False      0
9   lexus   True      0
10  lexus   True      1
11  lexus  False      0