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
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
"""
author: 方星钧(ffman)
description: 基于dim_st_asin_base_info等表,计算出search_term和asin维度的基础信息表(包括预估销量)
table_read_name: dim_st_asin_info, ods_rank_flow
table_save_name: dwd_st_asin_info
table_save_level: dwd
version: 3.0
created_date: 2022-05-12
updated_date: 2022-12-15
"""
import os
import sys
from pyspark.storagelevel import StorageLevel
sys.path.append(os.path.dirname(sys.path[0])) # 上级目录
from utils.templates import Templates
# from ..utils.templates import Templates
#from AmazonSpider.pyspark_job.utils.templates import Templates
# 分组排序的udf窗口函数
from pyspark.sql.window import Window
from pyspark.sql import functions as F
from pyspark.sql.types import StringType, IntegerType
class DwdStAsinInfo(Templates):
def __init__(self, site_name="us", date_type="week", date_info="2022-1"):
super().__init__()
self.site_name = site_name
self.date_type = date_type
self.date_info = date_info
self.db_save = f"dwd_st_asin_info"
self.spark = self.create_spark_object(app_name=f"{self.db_save} {self.site_name}, {self.date_info}")
self.df_date = self.get_year_week_tuple()
self.df_save = self.spark.sql(f"select 1+1;")
# self.df_save_std = self.spark.sql(f"select * from {self.db_save} limit 0;")
self.df_st_asin = self.spark.sql(f"select 1+1;")
self.df_rank_flow = self.spark.sql(f"select 1+1;")
self.df_st = self.spark.sql(f"select 1+1;")
self.df_zr_page1_counts = self.spark.sql(f"select 1+1;")
self.week_counts = 1 if self.date_type == 'week' else len(self.year_week_tuple)
self.partitions_by = ['site_name', 'date_type', 'date_info']
if self.date_type in ["week"]:
self.reset_partitions(200)
elif self.date_type in ["month", "4_week"]:
self.reset_partitions(400)
elif self.date_type in ["quarter"]:
self.reset_partitions(600)
self.u_week_counts_flag = self.spark.udf.register('u_week_counts_flag', self.udf_week_counts_flag, IntegerType())
self.u_year_week = self.spark.udf.register('u_year_week', self.udf_year_week, StringType())
self.u_st_type = self.spark.udf.register('u_st_type', self.udf_st_type, StringType())
@staticmethod
def udf_week_counts_flag(zr_page1_counts, week_counts):
if zr_page1_counts == week_counts:
return 1
else:
return 0
@staticmethod
def udf_year_week(dt):
year, week = dt.split("-")[0], dt.split("-")[1]
if int(week) < 10:
return f"{year}-0{week}"
else:
return f"{year}-{week}"
@staticmethod
def udf_st_type(st_asin_zr_rate, zr_page1_flag, st_search_num, st_click_share1, st_conversion_share1, st_click_share2, st_conversion_share2, st_click_share3, st_conversion_share3):
st_click_share_sum = st_click_share1 + st_click_share2 + st_click_share3
st_conversion_share_sum = st_conversion_share1 + st_conversion_share2 + st_conversion_share3
st_type_list = []
if st_asin_zr_rate >= 0.05:
st_type_list.append('1') # 主要流量词
if zr_page1_flag:
if st_search_num < 10000:
st_type_list.append('2') # 精准长尾词
else:
st_type_list.append('3') # 精准流量词
if (st_conversion_share_sum - st_click_share_sum) / st_click_share_sum >= 0.2:
st_type_list.append('4') # 转化优质词
else:
st_type_list.append('5') # 转化平稳词
if (st_click_share_sum - st_conversion_share_sum) / st_click_share_sum >= 0.2:
st_type_list.append('6') # 转化流失词
if st_conversion_share_sum > 0:
st_type_list.append('7') # 出单词
if st_click_share_sum > 0 and st_conversion_share_sum == 0:
st_type_list.append('8') # 无效曝光词
return ",".join(st_type_list) if st_type_list else ''
def read_data(self):
print("1.1 读取dim_st_asin_info表")
sql = f"select * from dim_st_asin_info where site_name='{self.site_name}' and date_type='week' and date_info in {self.year_week_tuple}"
self.df_st_asin = self.spark.sql(sqlQuery=sql).cache()
self.df_st_asin = self.df_st_asin.withColumnRenamed("updated_time", "updated_at")
self.df_st_asin.show(10, truncate=False)
print("1.2 读取ods_rank_flow表")
sql = f"select rank as st_asin_zr_page_rank, rank as st_asin_sp_page_rank, flow as st_asin_zr_rate, flow as st_asin_sp_rate from ods_rank_flow " \
f"where site_name='{self.site_name}'"
self.df_rank_flow = self.spark.sql(sql).cache()
self.df_rank_flow.show(10, truncate=False)
print("1.3 读取dim_st_info表")
# sql = f"select search_term, st_rank, st_rank_avg, st_search_num, st_search_rate, st_search_sum from dim_st_info " \
# f"where site_name='{self.site_name}' and date_type='{self.date_type}' and date_info in {self.date_info_tuple};"
sql = f"select search_term, st_rank, st_rank as st_rank_avg, st_search_num, st_search_rate, st_search_sum, " \
f"st_click_share1, st_conversion_share1, st_click_share2, st_conversion_share2, st_click_share3, st_conversion_share3 from dim_st_detail " \
f"where site_name='{self.site_name}' and date_type='{self.date_type}' and date_info = '{self.date_info}';"
print("sql:", sql)
self.df_st = self.spark.sql(sql).cache()
self.df_st = self.df_st.fillna(0)
self.df_st.show(10, truncate=False)
def handle_data(self):
self.handle_st_zr_page1_counts()
self.handle_st_duplicated()
self.handle_st_asin_pivot()
self.handle_st_asin_orders()
self.handle_st_type()
self.handle_st_dtypes()
self.df_save = self.df_save.withColumn("site_name", F.lit(self.site_name))
self.df_save = self.df_save.withColumn("date_type", F.lit(self.date_type))
self.df_save = self.df_save.withColumn("date_info", F.lit(self.date_info))
# self.df_save.show(20, truncate=False)
def handle_st_zr_page1_counts(self):
print("2.1 计算zr类型下,关键词对应的asin在选择的历史周中page=1出现的次数,如果全出现=1,否则=0")
self.df_zr_page1_counts = self.df_st_asin.filter("data_type='zr' and page=1").\
groupby(['search_term', 'asin']).\
agg(F.count_distinct("date_info").alias("zr_page1_counts"))
self.df_zr_page1_counts = self.df_zr_page1_counts.withColumn("week_counts", F.lit(self.week_counts))
self.df_zr_page1_counts = self.df_zr_page1_counts.withColumn(
"zr_page1_flag", self.u_week_counts_flag("zr_page1_counts", "zr_page1_counts")
)
# self.df_zr_page1_counts.show(10, truncate=False)
def handle_st_duplicated(self):
print("2.2 根据search_term,asin,data_type进行去重, page_rank选择最小值")
window = Window.partitionBy(['search_term', 'asin', 'data_type']).orderBy(
self.df_st_asin.page_rank.asc(),
self.df_st_asin.date_info.desc(),
)
self.df_st_asin = self.df_st_asin. \
withColumn("page_rank_top", F.row_number().over(window=window))
# print("self.df_st_asin, 开窗去重前:", self.df_st_asin_base_info.count())
self.df_st_asin = self.df_st_asin.filter("page_rank_top=1")
# print("self.df_st_asin, 开窗去重后:", self.df_st_asin.count())
self.df_st_asin = self.df_st_asin.cache()
# self.df_st_asin_base_info.show(10, truncate=False)
def handle_st_asin_pivot(self):
print(f"2.3 根据search_term和asin进行透视表")
self.df_st_asin = self.df_st_asin. \
withColumn("updated_at_data_type",
F.concat(F.lit("st_asin_"), self.df_st_asin.data_type, F.lit("_updated_at"))). \
withColumn("page_data_type",
F.concat(F.lit("st_asin_"), self.df_st_asin.data_type, F.lit("_page"))). \
withColumn("page_row_data_type",
F.concat(F.lit("st_asin_"), self.df_st_asin.data_type, F.lit("_page_row"))). \
withColumn("page_rank_data_type",
F.concat(F.lit("st_asin_"), self.df_st_asin.data_type, F.lit("_page_rank")))
df1 = self.df_st_asin.select("search_term", "asin", "updated_at_data_type", "updated_at"). \
withColumnRenamed("updated_at_data_type", "pivot_key"). \
withColumnRenamed("updated_at", "pivot_value")
df2 = self.df_st_asin.select("search_term", "asin", "page_data_type", "page")
# page_row和page_rank: 只有zr,sp才有
self.df_st_asin = self.df_st_asin.filter("data_type in ('zr', 'sp')")
df3 = self.df_st_asin.select("search_term", "asin", "page_row_data_type", "page_row")
df4 = self.df_st_asin.select("search_term", "asin", "page_rank_data_type", "page_rank")
self.df_save = df1.union(df2).union(df3).union(df4)
self.df_save = self.df_save.groupby(["search_term", "asin"]). \
pivot(f"pivot_key").agg(F.min(f"pivot_value")). \
join(self.df_zr_page1_counts, on=["search_term", "asin"], how="left"). \
join(self.df_rank_flow.select("st_asin_zr_page_rank", "st_asin_zr_rate"), on=["st_asin_zr_page_rank"], how="left"). \
join(self.df_rank_flow.select("st_asin_sp_page_rank", "st_asin_sp_rate"), on=["st_asin_sp_page_rank"], how="left"). \
join(self.df_st_info, on=["search_term"], how="inner") # ["search_term", "dt"]
self.df_save = self.df_save.fillna(
{
"st_asin_zr_rate": 0,
"st_asin_sp_rate": 0
}
)
def handle_st_asin_orders(self):
print("2.4 计算zr, sp预估销量")
self.df_save = self.df_save.withColumn(
"st_asin_zr_orders", F.ceil(self.df_save.st_asin_zr_rate * self.df_save.st_search_sum)
).withColumn(
"st_asin_sp_orders", F.ceil(self.df_save.st_asin_sp_rate * self.df_save.st_search_sum)
)
self.df_save = self.df_save.withColumn(
"asin_st_zr_orders", self.df_save.st_asin_zr_orders
).withColumn(
"asin_st_sp_orders", self.df_save.st_asin_sp_orders
)
df_asin_st_zr_orders_sum = self.df_save.groupby(['asin']). \
agg({"st_asin_zr_orders": "sum"})
df_asin_st_sp_orders_sum = self.df_save.groupby(['asin']). \
agg({"st_asin_sp_orders": "sum"})
df_asin_st_zr_orders_sum = df_asin_st_zr_orders_sum.withColumnRenamed("sum(st_asin_zr_orders)", "asin_st_zr_orders_sum")
df_asin_st_sp_orders_sum = df_asin_st_sp_orders_sum.withColumnRenamed("sum(st_asin_sp_orders)", "asin_st_sp_orders_sum")
df_asin_st_zr_orders_sum = df_asin_st_zr_orders_sum.withColumn(f"is_zr_flag", F.lit(1))
df_asin_st_sp_orders_sum = df_asin_st_sp_orders_sum.withColumn(f"is_sp_flag", F.lit(1))
df_st_asin_zr_orders_sum = self.df_save.groupby(['search_term']). \
agg({"st_asin_zr_orders": "sum"})
df_st_asin_zr_orders_sum = df_st_asin_zr_orders_sum.withColumnRenamed("sum(st_asin_zr_orders)", "st_asin_zr_orders_sum")
df_st_asin_zr_orders_sum = df_st_asin_zr_orders_sum.withColumn(f"is_zr_flag", F.lit(1))
df_st_asin_sp_orders_sum = self.df_save.groupby(['search_term']). \
agg({"st_asin_sp_orders": "sum"})
df_st_asin_sp_orders_sum = df_st_asin_sp_orders_sum.withColumnRenamed("sum(st_asin_sp_orders)", "st_asin_sp_orders_sum")
df_st_asin_sp_orders_sum = df_st_asin_sp_orders_sum.withColumn(f"is_sp_flag", F.lit(1))
self.df_save = self.df_save.withColumn("is_zr_flag", F.when(self.df_save.st_asin_zr_page > 0, 1))
self.df_save = self.df_save.withColumn("is_sp_flag", F.when(self.df_save.st_asin_sp_page > 0, 1))
self.df_save = self.df_save. \
join(df_asin_st_zr_orders_sum, on=['asin', "is_zr_flag"], how='left'). \
join(df_asin_st_sp_orders_sum, on=['asin', "is_sp_flag"], how='left'). \
join(df_st_asin_zr_orders_sum, on=['search_term', "is_zr_flag"], how='left'). \
join(df_st_asin_sp_orders_sum, on=['search_term', "is_sp_flag"], how='left')
self.df_save = self.df_save.withColumn(
"st_asin_zr_flow", self.df_save.st_asin_zr_orders / self.df_save.st_asin_zr_orders_sum
)
self.df_save = self.df_save.withColumn(
"st_asin_sp_flow", self.df_save.st_asin_sp_orders / self.df_save.st_asin_sp_orders_sum
)
self.df_save = self.df_save.withColumn(
"asin_st_zr_flow", self.df_save.asin_st_zr_orders / self.df_save.asin_st_zr_orders_sum
)
self.df_save = self.df_save.withColumn(
"asin_st_sp_flow", self.df_save.asin_st_sp_orders / self.df_save.asin_st_sp_orders_sum
)
self.df_save = self.df_save.drop("is_zr_flag", "is_sp_flag")
print("self.df_save.columns:", self.df_save.columns)
# self.df_save.show(10, truncate=False)
def handle_st_type(self):
print("2.5 根据search_term,asin等信息进行计算关键词的分类情况")
self.df_save = self.df_save.withColumn(
"st_type", self.u_st_type(
"st_asin_zr_rate", "zr_page1_flag", "st_search_num", "st_click_share1", "st_conversion_share1",
"st_click_share2", "st_conversion_share2", "st_click_share3", "st_conversion_share3"
)
)
def handle_st_dtypes(self):
print("2.5 更改pivot之后的列的数据类型, 保持和hive的数据类型一致")
for col in self.df_save.columns:
if ("_page" in col) or ("_page_row" in col) or ("_page_rank" in col):
print("col:", col)
self.df_save = self.df_save.withColumn(col, self.df_save[f'{col}'].cast("int"))
if __name__ == '__main__':
site_name = sys.argv[1] # 参数1:站点
date_type = sys.argv[2] # 参数2:类型:week/4_week/month/quarter
date_info = sys.argv[3] # 参数3:年-周/年-月/年-季, 比如: 2022-01
handle_obj = DwdStAsinInfo(site_name=site_name, date_type=date_type, date_info=date_info)
handle_obj.run()