1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
import os
import re
import sys
os.environ["PYARROW_IGNORE_TIMEZONE"] = "1"
sys.path.append(os.path.dirname(sys.path[0])) # 上级目录
from utils.templates import Templates
# from ..utils.templates import Templates
from pyspark.sql.types import StringType, BooleanType, StructType, StructField, DoubleType, FloatType
# 分组排序的udf窗口函数
from pyspark.sql.window import Window
from pyspark.sql import functions as F
from yswg_utils.common_udf import parse_weight_str
class TmpSelfAsinDetail(Templates):
def __init__(self, site_name='us'):
super().__init__()
self.site_name = site_name
self.db_save = f'dim_asin_stable_info'
self.spark = self.create_spark_object(app_name=f"{self.db_save}: {self.site_name}")
self.df_asin_detail = self.spark.sql(f"select 1+1;")
def read_data(self):
sql = "select * from ods_self_asin_detail where site_name='us';" # limit 1000000
self.df_asin_detail = self.spark.sql(sql).cache()
self.df_asin_detail.show()
def handle_data(self):
window = Window.partitionBy(['asin']).orderBy(F.desc("date_info")) # 按照 date_info 列进行分区，并按照 date 列进行排序
self.df_asin_detail = self.df_asin_detail.withColumn('row_number', F.row_number().over(window)) # 使用窗口函数为每个分区的行编号
self.df_asin_detail = self.df_asin_detail.filter(self.df_asin_detail.row_number == 1).drop('row_number') # 只保留每个分区中 row_number 最大的行，并删除 row_number 列
def save_data(self):
df_save = self.df_asin_detail.toPandas()
print(df_save.shape)
df_save.to_csv("/root/tmp_self_asin_detail.csv", index=False)
if __name__ == '__main__':
site_name = 'us'
handle_obj = TmpSelfAsinDetail(site_name=site_name)
handle_obj.run()