pyspark --- 空串替换为None

df = ss.createDataFrame([{
    'time_h': '', 'city': '北京', 'model': '华为'}, 
                   {
    'time_h': '', 'city': '北京', 'model': '华为'},
                   {
    'time_h': '', 'city': '', 'model': ''},
                   {
    'time_h': None, 'city': None, 'model': None},
                   {
    'time_h': None, 'city': '青岛', 'model': '华为'},
                   {
    'time_h': '20', 'city': '青岛', 'model': '华为'}])\
                .select(*['city', 'model', 'time_h'])
df.show()

# # 方法1
# for column in df.columns:
# trimmed = fn.trim(fn.col(column))
# df = df.withColumn(column, fn.when(fn.length(trimmed) != 0, trimmed).otherwise(None))

# 方法2
df = df.replace(to_replace='', value=None, subset=['time_h', 'model', 'city'])

df.show()

+----+-----+------+
|city|model|time_h|
+----+-----+------+
|北京| 华为|      |
|北京| 华为|      |
|    |     |      |
|null| null|  null|
|青岛| 华为|  null|
|青岛| 华为|    20|
+----+-----+------+

+----+-----+------+
|city|model|time_h|
+----+-----+------+
|北京| 华为|  null|
|北京| 华为|  null|
|null| null|  null|
|null| null|  null|
|青岛| 华为|  null|
|青岛| 华为|    20|
+----+-----+------+