staran 0.4.0__py3-none-any.whl → 0.4.2__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- staran/schemas/aum/__init__.py +160 -168
- staran/schemas/document_generator.py +26 -13
- {staran-0.4.0.dist-info → staran-0.4.2.dist-info}/METADATA +6 -6
- {staran-0.4.0.dist-info → staran-0.4.2.dist-info}/RECORD +7 -7
- {staran-0.4.0.dist-info → staran-0.4.2.dist-info}/WHEEL +0 -0
- {staran-0.4.0.dist-info → staran-0.4.2.dist-info}/licenses/LICENSE +0 -0
- {staran-0.4.0.dist-info → staran-0.4.2.dist-info}/top_level.txt +0 -0
staran/schemas/aum/__init__.py
CHANGED
|
@@ -20,84 +20,61 @@ from ..document_generator import SchemaDocumentGenerator
|
|
|
20
20
|
|
|
21
21
|
|
|
22
22
|
class AUMBehaviorSchema:
|
|
23
|
-
"""AUM客户行为特征表 - A
|
|
23
|
+
"""AUM客户行为特征表 - A表结构定义（严格按照已提供给行方的字段）"""
|
|
24
|
+
|
|
25
|
+
# 统计指标定义 - 与原始定义完全一致
|
|
26
|
+
_STATS = [
|
|
27
|
+
("max", "最大值"),
|
|
28
|
+
("min", "最小值"),
|
|
29
|
+
("sum", "总和"),
|
|
30
|
+
("avg", "均值"),
|
|
31
|
+
("var", "方差"),
|
|
32
|
+
("std", "标准差"),
|
|
33
|
+
("rng", "极差"),
|
|
34
|
+
("med", "中位数"),
|
|
35
|
+
]
|
|
24
36
|
|
|
25
37
|
@staticmethod
|
|
26
38
|
def create() -> TableSchema:
|
|
27
|
-
"""创建客户行为特征表结构"""
|
|
39
|
+
"""创建客户行为特征表结构 - 严格按照已提供给行方的字段定义"""
|
|
28
40
|
schema = TableSchema('bi_hlwj_dfcw_f1_f4_wy')
|
|
29
41
|
schema.add_primary_key('party_id', 'string')
|
|
30
42
|
schema.add_date_field('data_dt', 'string')
|
|
31
43
|
|
|
32
|
-
#
|
|
33
|
-
schema.add_field("
|
|
34
|
-
schema.add_field("
|
|
35
|
-
schema.add_field("
|
|
36
|
-
schema.add_field("
|
|
37
|
-
schema.add_field("
|
|
38
|
-
schema.add_field("
|
|
39
|
-
schema.add_field("
|
|
40
|
-
|
|
41
|
-
|
|
42
|
-
|
|
43
|
-
|
|
44
|
-
|
|
45
|
-
|
|
46
|
-
|
|
47
|
-
|
|
48
|
-
|
|
49
|
-
|
|
50
|
-
|
|
51
|
-
|
|
52
|
-
|
|
53
|
-
|
|
54
|
-
|
|
55
|
-
|
|
56
|
-
|
|
57
|
-
|
|
58
|
-
|
|
59
|
-
|
|
60
|
-
schema.add_field("
|
|
61
|
-
schema.add_field("
|
|
62
|
-
|
|
63
|
-
# 产品持有情况
|
|
64
|
-
schema.add_field("WEALTH_PROD_COUNT", "int", comment="理财产品数量", aggregatable=True)
|
|
65
|
-
schema.add_field("FUND_PROD_COUNT", "int", comment="基金产品数量", aggregatable=True)
|
|
66
|
-
schema.add_field("INSURANCE_PROD_COUNT", "int", comment="保险产品数量", aggregatable=True)
|
|
67
|
-
schema.add_field("GOLD_PROD_COUNT", "int", comment="黄金产品数量", aggregatable=True)
|
|
68
|
-
|
|
69
|
-
# 风险评级相关
|
|
70
|
-
schema.add_field("RISK_LEVEL", "string", comment="风险等级")
|
|
71
|
-
schema.add_field("RISK_APPETITE", "string", comment="风险偏好")
|
|
72
|
-
schema.add_field("INVESTMENT_EXPERIENCE", "string", comment="投资经验")
|
|
73
|
-
|
|
74
|
-
# 服务渠道偏好
|
|
75
|
-
schema.add_field("PREFER_CHANNEL", "string", comment="偏好服务渠道")
|
|
76
|
-
schema.add_field("CONTACT_TIME_PREFER", "string", comment="联系时间偏好")
|
|
77
|
-
|
|
78
|
-
# 客户活跃度指标
|
|
79
|
-
schema.add_field("LOGIN_DAYS_MON3", "int", comment="近3月登录天数", aggregatable=True)
|
|
80
|
-
schema.add_field("LOGIN_DAYS_MON6", "int", comment="近6月登录天数", aggregatable=True)
|
|
81
|
-
schema.add_field("LAST_LOGIN_DAYS", "int", comment="最后登录距今天数", aggregatable=True)
|
|
82
|
-
schema.add_field("ACTIVE_LEVEL", "string", comment="活跃度等级")
|
|
83
|
-
|
|
84
|
-
# 客户价值指标
|
|
85
|
-
schema.add_field("CUSTOMER_VALUE_SCORE", "decimal", comment="客户价值评分", aggregatable=True)
|
|
86
|
-
schema.add_field("POTENTIAL_VALUE_SCORE", "decimal", comment="潜在价值评分", aggregatable=True)
|
|
87
|
-
schema.add_field("RETENTION_SCORE", "decimal", comment="留存倾向评分", aggregatable=True)
|
|
88
|
-
|
|
89
|
-
# 营销响应历史
|
|
90
|
-
schema.add_field("CAMPAIGN_RESPONSE_RATE", "decimal", comment="营销响应率", aggregatable=True)
|
|
91
|
-
schema.add_field("LAST_CAMPAIGN_RESPONSE", "string", comment="最近营销响应")
|
|
92
|
-
schema.add_field("PRODUCT_CROSS_SELL_COUNT", "int", comment="交叉销售产品数", aggregatable=True)
|
|
93
|
-
|
|
94
|
-
# 投诉与满意度
|
|
95
|
-
schema.add_field("COMPLAINT_COUNT_YEAR1", "int", comment="近1年投诉次数", aggregatable=True)
|
|
96
|
-
schema.add_field("SATISFACTION_SCORE", "decimal", comment="满意度评分", aggregatable=True)
|
|
97
|
-
schema.add_field("NPS_SCORE", "decimal", comment="净推荐值", aggregatable=True)
|
|
44
|
+
# 基础字段 - 严格按照原始定义
|
|
45
|
+
schema.add_field("buy_ct", "int", comment="购买次数", aggregatable=True)
|
|
46
|
+
schema.add_field("recency", "int", comment="最近一次购买距今天数", aggregatable=True)
|
|
47
|
+
schema.add_field("tenure", "int", comment="客户关系持续时间", aggregatable=True)
|
|
48
|
+
schema.add_field("window1", "string", comment="时间窗口标记")
|
|
49
|
+
schema.add_field("freq", "float", comment="总购买频率", aggregatable=True)
|
|
50
|
+
schema.add_field("freq1", "float", comment="最近时间段购买频率", aggregatable=True)
|
|
51
|
+
schema.add_field("productidcount", "int", comment="产品种类数", aggregatable=True)
|
|
52
|
+
schema.add_field("orderidcount", "int", comment="订单数", aggregatable=True)
|
|
53
|
+
schema.add_field("label", "float", comment="标签值（如是否购买）", aggregatable=True)
|
|
54
|
+
|
|
55
|
+
# productamount and m1 ~ m4 的含义描述 - 严格按照原始定义
|
|
56
|
+
m_fields = {
|
|
57
|
+
"productamount": "购买金额",
|
|
58
|
+
"m1": "去重订单数",
|
|
59
|
+
"m2": "去重商品数",
|
|
60
|
+
"m3": "去重渠道数",
|
|
61
|
+
"m4": "去重产品品类数",
|
|
62
|
+
}
|
|
63
|
+
|
|
64
|
+
# 使用循环注册 productamount and m1~m4 各统计字段 - 严格按照原始逻辑
|
|
65
|
+
for prefix, meaning in m_fields.items():
|
|
66
|
+
for stat_key, stat_desc in AUMBehaviorSchema._STATS:
|
|
67
|
+
field_name = f"{prefix}_{stat_key}"
|
|
68
|
+
description = f"{meaning}的{stat_desc}"
|
|
69
|
+
schema.add_field(field_name, "float", comment=description, aggregatable=True)
|
|
70
|
+
|
|
71
|
+
# 其他字段 - 严格按照原始定义
|
|
72
|
+
schema.add_field("life_day", "float", comment="客户生命周期天数", aggregatable=True)
|
|
73
|
+
schema.add_field("gender", "float", comment="性别（编码）", aggregatable=True)
|
|
74
|
+
schema.add_field("open_day", "float", comment="开户天数", aggregatable=True)
|
|
98
75
|
|
|
99
|
-
#
|
|
100
|
-
schema
|
|
76
|
+
schema.set_monthly_unique(False) # A表每人每日记录
|
|
77
|
+
return schema
|
|
101
78
|
schema.add_field("FREQ_BRANCH_CODE", "string", comment="常用网点代码")
|
|
102
79
|
schema.add_field("CROSS_REGION_TXN", "string", comment="跨地区交易情况")
|
|
103
80
|
|
|
@@ -106,161 +83,176 @@ class AUMBehaviorSchema:
|
|
|
106
83
|
|
|
107
84
|
|
|
108
85
|
class AUMAssetAvgSchema:
|
|
109
|
-
"""AUM资产平均值表 - B
|
|
86
|
+
"""AUM资产平均值表 - B表结构定义（严格按照已提供给行方的字段）"""
|
|
110
87
|
|
|
111
88
|
@staticmethod
|
|
112
89
|
def create() -> TableSchema:
|
|
113
|
-
"""创建资产平均值表结构"""
|
|
90
|
+
"""创建资产平均值表结构 - 严格按照原始定义"""
|
|
114
91
|
schema = TableSchema('bi_hlwj_zi_chan_avg_wy')
|
|
115
92
|
schema.add_primary_key('party_id', 'string')
|
|
116
93
|
schema.add_date_field('data_dt', 'string')
|
|
117
94
|
|
|
118
|
-
#
|
|
119
|
-
schema.add_field("
|
|
120
|
-
schema.add_field("
|
|
121
|
-
schema.add_field("
|
|
122
|
-
schema.add_field("
|
|
123
|
-
schema.add_field("
|
|
124
|
-
schema.add_field("
|
|
125
|
-
|
|
126
|
-
|
|
127
|
-
|
|
128
|
-
|
|
129
|
-
|
|
130
|
-
|
|
131
|
-
|
|
132
|
-
|
|
133
|
-
|
|
134
|
-
|
|
135
|
-
schema.add_field("OTHER_LOAN_AVG", "decimal", comment="其他贷款平均余额", aggregatable=True)
|
|
95
|
+
# 基础余额字段 - 严格按照原始定义
|
|
96
|
+
schema.add_field("asset_total_bal", "decimal", comment="总资产余额", aggregatable=True)
|
|
97
|
+
schema.add_field("liab_total_bal", "decimal", comment="总负债余额", aggregatable=True)
|
|
98
|
+
schema.add_field("dpsit_total_bal", "decimal", comment="存款总余额", aggregatable=True)
|
|
99
|
+
schema.add_field("loan_total_bal", "decimal", comment="贷款总余额", aggregatable=True)
|
|
100
|
+
schema.add_field("card_total_bal", "decimal", comment="信用卡总余额", aggregatable=True)
|
|
101
|
+
schema.add_field("mid_busi_total_bal", "decimal", comment="中间业务总余额", aggregatable=True)
|
|
102
|
+
|
|
103
|
+
# Register average balance fields - 严格按照原始逻辑
|
|
104
|
+
for period in ["month", "year", "3", "6", "12"]:
|
|
105
|
+
schema.add_field(
|
|
106
|
+
f"avg_asset_bal_{period}", "decimal", comment=f"平均资产余额 ({period}期)", aggregatable=True
|
|
107
|
+
)
|
|
108
|
+
for period in ["3", "12"]:
|
|
109
|
+
schema.add_field(
|
|
110
|
+
f"avg_dpsit_bal_{period}", "decimal", comment=f"平均存款余额 ({period}期)", aggregatable=True
|
|
111
|
+
)
|
|
136
112
|
|
|
137
113
|
schema.set_monthly_unique(True) # B表每人每月唯一
|
|
138
114
|
return schema
|
|
139
115
|
|
|
140
116
|
|
|
141
117
|
class AUMAssetConfigSchema:
|
|
142
|
-
"""AUM资产配置表 - C
|
|
118
|
+
"""AUM资产配置表 - C表结构定义（严格按照已提供给行方的字段）"""
|
|
143
119
|
|
|
144
120
|
@staticmethod
|
|
145
121
|
def create() -> TableSchema:
|
|
146
|
-
"""创建资产配置表结构"""
|
|
122
|
+
"""创建资产配置表结构 - 严格按照原始定义"""
|
|
147
123
|
schema = TableSchema('bi_hlwj_zi_chang_month_total_zb')
|
|
148
124
|
schema.add_primary_key('party_id', 'string')
|
|
149
125
|
schema.add_date_field('data_dt', 'string')
|
|
150
126
|
|
|
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
|
-
schema.add_field("CONFIG_CHANGE_FREQ", "int", comment="配置调整频率", aggregatable=True)
|
|
189
|
-
schema.add_field("LAST_REBALANCE_DAYS", "int", comment="最后再平衡距今天数", aggregatable=True)
|
|
190
|
-
|
|
191
|
-
# 配置绩效相关
|
|
192
|
-
schema.add_field("CONFIG_RETURN_RATE", "decimal", comment="配置收益率", aggregatable=True)
|
|
193
|
-
schema.add_field("RISK_ADJUSTED_RETURN", "decimal", comment="风险调整收益", aggregatable=True)
|
|
194
|
-
schema.add_field("SHARPE_RATIO", "decimal", comment="夏普比率", aggregatable=True)
|
|
195
|
-
|
|
196
|
-
# 配置建议相关
|
|
197
|
-
schema.add_field("OPTIMAL_CONFIG_SCORE", "decimal", comment="最优配置评分", aggregatable=True)
|
|
198
|
-
schema.add_field("CONFIG_IMPROVEMENT_POTENTIAL", "decimal", comment="配置优化潜力", aggregatable=True)
|
|
127
|
+
# asset_fields - 严格按照原始定义
|
|
128
|
+
asset_fields = [
|
|
129
|
+
("seg_asset_total", "总资产余额"),
|
|
130
|
+
("indv_consm_loan_amt", "个人消费贷款余额"),
|
|
131
|
+
("indv_house_loan_amt", "个人住房贷款余额"),
|
|
132
|
+
("indv_oper_loan_amt", "个人经营贷款余额"),
|
|
133
|
+
("dpsit_bal", "存款余额"),
|
|
134
|
+
("tbond_bal", "国债余额"),
|
|
135
|
+
("fund_bal", "基金余额"),
|
|
136
|
+
("bond_bal", "债券余额"),
|
|
137
|
+
("gold_bal", "黄金余额"),
|
|
138
|
+
("wcurr_chrem_bal", "外币现钞余额"),
|
|
139
|
+
("presv_mgmt_secu_bal", "保值管理证券余额"),
|
|
140
|
+
("insure_form_bal", "保险单余额"),
|
|
141
|
+
("crdt_card_od_bal", "信用卡透支余额"),
|
|
142
|
+
("crdt_card_con_amt", "信用卡消费金额"),
|
|
143
|
+
("semi_crdt_card_od_bal", "准贷记卡透支余额"),
|
|
144
|
+
("semi_crdt_card_con_amt", "准贷记卡消费金额"),
|
|
145
|
+
("inter_card_con_amt", "国际卡消费金额"),
|
|
146
|
+
("inter_card_od_bal", "国际卡透支余额"),
|
|
147
|
+
("crdt_card_dpsit_bal", "信用卡存款余额"),
|
|
148
|
+
("semi_crdt_card_dpsit_bal", "准贷记卡存款余额"),
|
|
149
|
+
("inter_card_dpsit_bal", "国际卡存款余额"),
|
|
150
|
+
("silver_bal", "白银余额"),
|
|
151
|
+
("agent_solid_silver_bal", "代发实物白银余额"),
|
|
152
|
+
("pt_bal", "个人养老金余额"),
|
|
153
|
+
("pd_bal", "个人养老金存款余额"),
|
|
154
|
+
("other_metal_bal", "其他金属余额"),
|
|
155
|
+
("curr_dpsit_bal", "活期存款余额"),
|
|
156
|
+
("time_dpsit_bal", "定期存款余额"),
|
|
157
|
+
("oil_bal", "石油余额"),
|
|
158
|
+
("fx_bal", "外汇余额"),
|
|
159
|
+
]
|
|
160
|
+
|
|
161
|
+
# 严格按照原始循环逻辑注册字段
|
|
162
|
+
for field_name, desc in asset_fields:
|
|
163
|
+
schema.add_field(field_name, "decimal", comment=desc, aggregatable=True)
|
|
199
164
|
|
|
200
165
|
schema.set_monthly_unique(True) # C表每人每月唯一
|
|
201
166
|
return schema
|
|
202
167
|
|
|
203
168
|
|
|
204
169
|
class AUMMonthlyStatSchema:
|
|
205
|
-
"""AUM月度统计表 - D
|
|
170
|
+
"""AUM月度统计表 - D表结构定义（严格按照已提供给行方的字段）"""
|
|
206
171
|
|
|
207
172
|
@staticmethod
|
|
208
173
|
def create() -> TableSchema:
|
|
209
|
-
"""创建月度统计表结构"""
|
|
174
|
+
"""创建月度统计表结构 - 严格按照原始定义"""
|
|
210
175
|
schema = TableSchema('bi_hlwj_realy_month_stat_wy')
|
|
211
|
-
schema.add_primary_key('
|
|
176
|
+
schema.add_primary_key('party_id', 'string') # 修正主键名称
|
|
212
177
|
schema.add_date_field('data_dt', 'string')
|
|
213
178
|
|
|
214
|
-
#
|
|
179
|
+
# channels字典 - 严格按照原始定义
|
|
215
180
|
channels = {
|
|
216
181
|
"CASH_DEPIST": "现金",
|
|
217
|
-
"REMIT": "汇款",
|
|
182
|
+
"REMIT": "汇款",
|
|
218
183
|
"YY": "邮政储蓄",
|
|
219
184
|
"UNIONPAY": "银联",
|
|
220
185
|
"FIN_ASSET": "理财产品",
|
|
221
|
-
"CORP_ACCT": "对公账户"
|
|
186
|
+
"CORP_ACCT": "对公账户",
|
|
222
187
|
}
|
|
223
|
-
|
|
188
|
+
|
|
189
|
+
# 注册存入和取出字段 - 严格按照原始逻辑
|
|
224
190
|
for prefix, desc in channels.items():
|
|
225
191
|
schema.add_field(f"{prefix}_IN", "decimal", comment=f"{desc}存入金额", aggregatable=True)
|
|
226
192
|
schema.add_field(f"{prefix}_OUT", "decimal", comment=f"{desc}取出金额", aggregatable=True)
|
|
227
|
-
|
|
228
|
-
#
|
|
193
|
+
|
|
194
|
+
# 其他特定字段 - 严格按照原始定义
|
|
229
195
|
schema.add_field("AGENT_SALARY_IN", "decimal", comment="代发工资存入金额", aggregatable=True)
|
|
230
196
|
schema.add_field("CREDIT_CARD_OUT", "decimal", comment="信用卡取出金额", aggregatable=True)
|
|
231
197
|
schema.add_field("DEBIT_CARD_OUT", "decimal", comment="借记卡取出金额", aggregatable=True)
|
|
232
198
|
schema.add_field("BATCH_DEDUCT_OUT", "decimal", comment="批量扣款金额", aggregatable=True)
|
|
233
|
-
|
|
234
|
-
#
|
|
199
|
+
|
|
200
|
+
# 定义字段结构：交易渠道、指标、时间范围、描述前缀 - 严格按照原始定义
|
|
235
201
|
fields = [
|
|
236
202
|
("DEBIT_CARD", "借记卡", "MON3"),
|
|
237
203
|
("CREDIT_CARD", "信用卡", "MON3"),
|
|
238
204
|
("THIRD_PAYMENT", "第三方支付", "MON3"),
|
|
239
205
|
("MOBBANK", "手机银行", "MON12"),
|
|
240
|
-
("TELBANK", "电话银行", "MON12")
|
|
206
|
+
("TELBANK", "电话银行", "MON12"),
|
|
241
207
|
]
|
|
242
|
-
|
|
243
|
-
|
|
244
|
-
|
|
208
|
+
|
|
209
|
+
# 定义交易指标 - 严格按照原始定义
|
|
210
|
+
metrics = [
|
|
211
|
+
("TX_CNT", "交易次数"),
|
|
212
|
+
("TX_AMT", "交易金额"),
|
|
213
|
+
]
|
|
214
|
+
|
|
215
|
+
# 自动注册 - 严格按照原始逻辑
|
|
245
216
|
for channel, desc, period in fields:
|
|
246
217
|
for metric_code, metric_desc in metrics:
|
|
247
218
|
field_name = f"{channel}_{metric_code}_{period}"
|
|
248
219
|
description = f"{desc}{metric_desc}（近{period[-2:]}个月）"
|
|
249
220
|
schema.add_field(field_name, "decimal", comment=description, aggregatable=True)
|
|
250
|
-
|
|
251
|
-
#
|
|
252
|
-
schema.add_field(
|
|
253
|
-
|
|
254
|
-
|
|
255
|
-
|
|
221
|
+
|
|
222
|
+
# 其他固定字段 - 严格按照原始定义
|
|
223
|
+
schema.add_field(
|
|
224
|
+
"COUNTER_TX_CNT_MON12", "int", comment="柜台交易次数（近12个月）", aggregatable=True
|
|
225
|
+
)
|
|
226
|
+
schema.add_field(
|
|
227
|
+
"WEBBANK_TX_CNT_MON12", "int", comment="网银交易次数（近12个月）", aggregatable=True
|
|
228
|
+
)
|
|
229
|
+
|
|
230
|
+
# 编号国家（1~5） - 严格按照原始循环逻辑
|
|
256
231
|
for i in range(1, 6):
|
|
257
|
-
schema.add_field(
|
|
258
|
-
|
|
259
|
-
|
|
260
|
-
schema.add_field(
|
|
261
|
-
|
|
262
|
-
|
|
263
|
-
|
|
232
|
+
schema.add_field(
|
|
233
|
+
f"Y1_OVERS_CTY{i}_CNT", "int", comment=f"近一年境外国家{i}的交易次数", aggregatable=True
|
|
234
|
+
)
|
|
235
|
+
schema.add_field(
|
|
236
|
+
f"Y1_OVERS_CNT_CTY{i}_CD",
|
|
237
|
+
"string",
|
|
238
|
+
comment=f"近一年境外国家{i}的交易次数（编码）",
|
|
239
|
+
)
|
|
240
|
+
schema.add_field(
|
|
241
|
+
f"Y1_OVERS_CTY{i}_AMT", "decimal", comment=f"近一年境外国家{i}的交易金额", aggregatable=True
|
|
242
|
+
)
|
|
243
|
+
schema.add_field(
|
|
244
|
+
f"Y1_OVERS_AMT_CTY{i}_CD",
|
|
245
|
+
"string",
|
|
246
|
+
comment=f"近一年境外国家{i}的交易金额（编码）",
|
|
247
|
+
)
|
|
248
|
+
|
|
249
|
+
# 其他国家 - 严格按照原始定义
|
|
250
|
+
schema.add_field(
|
|
251
|
+
"Y1_OVERS_OTHER_CTY_CNT", "int", comment="近一年其他境外国家的交易次数", aggregatable=True
|
|
252
|
+
)
|
|
253
|
+
schema.add_field(
|
|
254
|
+
"Y1_OVERS_OTHER_CTY_AMT", "decimal", comment="近一年其他境外国家的交易金额", aggregatable=True
|
|
255
|
+
)
|
|
264
256
|
|
|
265
257
|
schema.set_monthly_unique(True) # D表每人每月唯一
|
|
266
258
|
return schema
|
|
@@ -101,8 +101,10 @@ class SchemaDocumentGenerator:
|
|
|
101
101
|
content = []
|
|
102
102
|
|
|
103
103
|
# 文档头部
|
|
104
|
-
content.append(f"# {business_domain} - {table_type.upper()}
|
|
105
|
-
content.append(
|
|
104
|
+
content.append(f"# {business_domain} - {table_type.upper()}表结构文档")
|
|
105
|
+
content.append("")
|
|
106
|
+
content.append(f"## 基本信息")
|
|
107
|
+
content.append("")
|
|
106
108
|
content.append(f"- **表名**: `{schema.table_name}`")
|
|
107
109
|
content.append(f"- **业务域**: {business_domain}")
|
|
108
110
|
content.append(f"- **表类型**: {table_type}")
|
|
@@ -111,30 +113,37 @@ class SchemaDocumentGenerator:
|
|
|
111
113
|
content.append("")
|
|
112
114
|
|
|
113
115
|
# 表结构说明
|
|
114
|
-
content.append("##
|
|
116
|
+
content.append("## 表结构说明")
|
|
117
|
+
content.append("")
|
|
115
118
|
if hasattr(schema, 'description'):
|
|
116
|
-
content.append(f"{schema.description}
|
|
119
|
+
content.append(f"{schema.description}")
|
|
120
|
+
content.append("")
|
|
117
121
|
|
|
118
122
|
# 主键信息
|
|
119
123
|
if hasattr(schema, 'primary_key') and schema.primary_key:
|
|
120
|
-
content.append("###
|
|
124
|
+
content.append("### 主键字段")
|
|
125
|
+
content.append("")
|
|
121
126
|
content.append(f"- `{schema.primary_key}` (主键)")
|
|
122
127
|
content.append("")
|
|
123
128
|
|
|
124
129
|
# 日期字段
|
|
125
130
|
if hasattr(schema, 'date_field') and schema.date_field:
|
|
126
|
-
content.append("###
|
|
131
|
+
content.append("### 日期字段")
|
|
132
|
+
content.append("")
|
|
127
133
|
content.append(f"- `{schema.date_field}` (日期字段)")
|
|
128
134
|
content.append("")
|
|
129
135
|
|
|
130
136
|
# 字段详情表格
|
|
131
|
-
content.append("##
|
|
137
|
+
content.append("## 字段详情")
|
|
138
|
+
content.append("")
|
|
132
139
|
content.append("| 字段名 | 数据类型 | 业务含义 | 可聚合 | 备注 |")
|
|
133
140
|
content.append("|--------|----------|----------|--------|------|")
|
|
134
141
|
|
|
135
142
|
if hasattr(schema, 'fields'):
|
|
136
143
|
for field_name, field in schema.fields.items():
|
|
137
|
-
|
|
144
|
+
# 简化数据类型显示
|
|
145
|
+
field_type_str = str(field.field_type) if hasattr(field, 'field_type') else 'string'
|
|
146
|
+
field_type = field_type_str.replace('FieldType.', '').lower()
|
|
138
147
|
comment = field.comment if hasattr(field, 'comment') else ''
|
|
139
148
|
aggregatable = '是' if getattr(field, 'aggregatable', False) else '否'
|
|
140
149
|
remarks = '' # 可以从其他地方获取备注
|
|
@@ -144,9 +153,10 @@ class SchemaDocumentGenerator:
|
|
|
144
153
|
content.append("")
|
|
145
154
|
|
|
146
155
|
# 业务规则说明
|
|
147
|
-
content.append("##
|
|
156
|
+
content.append("## 业务规则")
|
|
157
|
+
content.append("")
|
|
148
158
|
content.append("### 数据更新规则")
|
|
149
|
-
if getattr(schema, '
|
|
159
|
+
if getattr(schema, 'is_monthly_unique', False):
|
|
150
160
|
content.append("- 每人每月一条记录")
|
|
151
161
|
content.append("- 月末批量更新")
|
|
152
162
|
else:
|
|
@@ -162,7 +172,8 @@ class SchemaDocumentGenerator:
|
|
|
162
172
|
content.append("")
|
|
163
173
|
|
|
164
174
|
# 使用说明
|
|
165
|
-
content.append("##
|
|
175
|
+
content.append("## 使用说明")
|
|
176
|
+
content.append("")
|
|
166
177
|
content.append("### 特征工程配置")
|
|
167
178
|
if table_type == 'behavior':
|
|
168
179
|
content.append("- 生成原始拷贝特征")
|
|
@@ -187,7 +198,7 @@ class SchemaDocumentGenerator:
|
|
|
187
198
|
content.append("---")
|
|
188
199
|
content.append("*本文档由Staran Schema自动生成*")
|
|
189
200
|
|
|
190
|
-
return "
|
|
201
|
+
return "\n".join(content)
|
|
191
202
|
|
|
192
203
|
def _generate_html_content(self, schema, business_domain: str, table_type: str) -> str:
|
|
193
204
|
"""生成HTML格式内容"""
|
|
@@ -238,7 +249,9 @@ class SchemaDocumentGenerator:
|
|
|
238
249
|
# 添加字段行
|
|
239
250
|
if hasattr(schema, 'fields'):
|
|
240
251
|
for field_name, field in schema.fields.items():
|
|
241
|
-
|
|
252
|
+
# 简化数据类型显示
|
|
253
|
+
field_type_str = str(field.field_type) if hasattr(field, 'field_type') else 'string'
|
|
254
|
+
field_type = field_type_str.replace('FieldType.', '').lower()
|
|
242
255
|
comment = field.comment if hasattr(field, 'comment') else ''
|
|
243
256
|
aggregatable = '是' if getattr(field, 'aggregatable', False) else '否'
|
|
244
257
|
remarks = '' # 可以从其他地方获取备注
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: staran
|
|
3
|
-
Version: 0.4.
|
|
3
|
+
Version: 0.4.2
|
|
4
4
|
Summary: staran - 高性能Python工具库
|
|
5
5
|
Home-page: https://github.com/starlxa/staran
|
|
6
6
|
Author: StarAn
|
|
@@ -24,7 +24,7 @@ Dynamic: requires-dist
|
|
|
24
24
|
Dynamic: requires-python
|
|
25
25
|
Dynamic: summary
|
|
26
26
|
|
|
27
|
-
# Star## ✨ v0.4.
|
|
27
|
+
# Star## ✨ v0.4.2 新特性
|
|
28
28
|
|
|
29
29
|
- 📋 **独立Schema模块** - 专门的表结构定义和管理模块
|
|
30
30
|
- 📄 **文档自动生成** - 支持Markdown/PDF/HTML格式的技术文档生成
|
|
@@ -42,7 +42,7 @@ Dynamic: summary
|
|
|
42
42
|
|
|
43
43
|
Staran是一个强大的特征工程和数据处理工具包，提供从数据到模型的完整解决方案。特别针对工银图灵平台优化，让特征工程和模型训练变得前所未有的简单。
|
|
44
44
|
|
|
45
|
-
## ✨ v0.4.
|
|
45
|
+
## ✨ v0.4.2 新特性
|
|
46
46
|
|
|
47
47
|
- �️ **模块化引擎架构** - 独立的引擎模块，支持Spark、Hive、图灵平台
|
|
48
48
|
- 🔧 **统一接口设计** - 所有引擎提供一致的SQL生成、执行和下载接口
|
|
@@ -361,7 +361,7 @@ tomorrow = date.add_days(1) # 202504 (智能处理)
|
|
|
361
361
|
|
|
362
362
|
```
|
|
363
363
|
staran/
|
|
364
|
-
├── __init__.py # 主包入口，v0.4.
|
|
364
|
+
├── __init__.py # 主包入口，v0.4.2功能导出
|
|
365
365
|
├── schemas/ # 🆕 表结构定义与文档生成模块
|
|
366
366
|
│ ├── __init__.py # Schema模块入口
|
|
367
367
|
│ ├── document_generator.py # 文档生成器 (MD/PDF/HTML)
|
|
@@ -385,7 +385,7 @@ staran/
|
|
|
385
385
|
│ ├── __init__.py # 工具模块
|
|
386
386
|
│ └── date.py # Date类实现
|
|
387
387
|
├── setup.py # 安装配置
|
|
388
|
-
├── README.md # 本文档 v0.4.
|
|
388
|
+
├── README.md # 本文档 v0.4.2
|
|
389
389
|
└── quick-upload.sh # 快速部署脚本
|
|
390
390
|
```
|
|
391
391
|
|
|
@@ -561,4 +561,4 @@ MIT License
|
|
|
561
561
|
|
|
562
562
|
---
|
|
563
563
|
|
|
564
|
-
**Staran v0.4.
|
|
564
|
+
**Staran v0.4.2** - 模块化引擎架构，让机器学习特征工程变得前所未有的简单 🌟
|
|
@@ -13,12 +13,12 @@ staran/features/generator.py,sha256=CI1F_PshOvokQJelsqSaVp-SNQpMc-WVmjMQKzgdeLw,
|
|
|
13
13
|
staran/features/manager.py,sha256=2-3Hc3qthtyzwiuQy5QTz6RfhKK3szoylconzI3moc4,5201
|
|
14
14
|
staran/features/schema.py,sha256=FwOfpTcxq4K8zkO3MFNqKPQBp_e8qY-N6gazqm9_lAQ,6067
|
|
15
15
|
staran/schemas/__init__.py,sha256=2RkcWCaIkrOHd37zzRCla0-jNg4cPnc6BGmmW5Vha0Y,652
|
|
16
|
-
staran/schemas/document_generator.py,sha256=
|
|
17
|
-
staran/schemas/aum/__init__.py,sha256=
|
|
16
|
+
staran/schemas/document_generator.py,sha256=Mr7TjmKwspqxXnp9DhzZxsRx0l2Bo7MOI8mOxRtgwxU,13600
|
|
17
|
+
staran/schemas/aum/__init__.py,sha256=jVkmJdhHGHdGE4rJ605zsRU2zIQMEHWnlgW2ZQk8AdU,13082
|
|
18
18
|
staran/tools/__init__.py,sha256=KtudrYnxKD9HZEL4H-mrWlKrmsI3rYjJrLeC9YDTpG4,1054
|
|
19
19
|
staran/tools/date.py,sha256=-QyEMWVx6czMuOIwcV7kR3gBMRVOwb5qevo7GEFSJKE,10488
|
|
20
|
-
staran-0.4.
|
|
21
|
-
staran-0.4.
|
|
22
|
-
staran-0.4.
|
|
23
|
-
staran-0.4.
|
|
24
|
-
staran-0.4.
|
|
20
|
+
staran-0.4.2.dist-info/licenses/LICENSE,sha256=2EmsBIyDCono4iVXNpv5_px9qt2b7hfPq1WuyGVMNP4,1361
|
|
21
|
+
staran-0.4.2.dist-info/METADATA,sha256=u8_IU7kGJqbFbKZLXUIgACqklcrSp56UWcawoWAtGtY,18809
|
|
22
|
+
staran-0.4.2.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
|
|
23
|
+
staran-0.4.2.dist-info/top_level.txt,sha256=NOUZtXSh5oSIEjHrC0lQ9WmoKtD010Q00dghWyag-Zs,7
|
|
24
|
+
staran-0.4.2.dist-info/RECORD,,
|
|
File without changes
|
|
File without changes
|
|
File without changes
|