wip: add fund_code pool

src/fund_info/statistic.py

@@ -22,7 +22,7 @@ class FundStatistic:
     def __init__(self):
       # 统计上一个季度
         last_quarter_time = time.localtime(time.time() - 3 * 30 * 24 * 3600)
-        time.strftime("%m-%d", last_quarter_time)
+        # time.strftime("%m-%d", last_quarter_time)
         year = time.strftime("%Y", last_quarter_time)
         date = time.strftime("%m-%d", last_quarter_time)
         index = get_season_index(date)
@@ -72,3 +72,15 @@ class FundStatistic:
     # 分组查询特定股票的每个季度基金持有总数
     def item_stock_fund_count(self, stock_name, fund_code_pool=None):
         return self.each_query.select_special_stock_fund_count(stock_name, fund_code_pool)
+
+    def select_fund_pool(self, *, morning_star_rating_5="", morning_star_rating_3="", **args):
+        # print("morning_star_rating_5", morning_star_rating_5)
+        # print("morning_star_rating_3", morning_star_rating_3)
+        # print("args", args)
+        # morning_star_rating_5 = 5
+
+        return self.each_query.select_certain_condition_funds(
+            morning_star_rating_5=morning_star_rating_5,
+            morning_star_rating_3=morning_star_rating_3,
+            **args
+        )

src/fund_statistic.py

@@ -9,15 +9,10 @@ Author: luxuemin2108@gmail.com
 Copyright (c) 2020 Camel Lu
 '''
 
-import pymysql
 from pprint import pprint
-from db.connect import connect
 import pandas as pd
-from sql_model.fund_query import FundQuery
 from fund_info.statistic import FundStatistic
 
-# cursor = connect().cursor()
-
 
 def stocks_compare(stock_list, fund_code_pool=None):
     each_statistic = FundStatistic()
@@ -64,25 +59,42 @@ def stocks_compare(stock_list, fund_code_pool=None):
 if __name__ == '__main__':
     each_statistic = FundStatistic()
 
-    fund_code_pool = ['000001', '160133', '360014', '420002',
-                      '420102', '000409', '000418', '000746',
-                      '000751', '000884', '000991', '001043',
-                      '001054', '001104', '001410', '001473',
-                      '519714', '000003', '000011', '000029']
+    # fund_code_pool = ['000001', '160133', '360014', '420002',
+    #                   '420102', '000409', '000418', '000746',
+    #                   '000751', '000884', '000991', '001043',
+    #                   '001054', '001104', '001410', '001473',
+    #                   '519714', '000003', '000011', '000029']
+    morning_star_rating_5_condition = {
+        'value': 4,
+        'operator': '>='
+    }
+    morning_star_rating_3_condition = {
+        'value': 5,
+        'operator': '='
+    }
+    condition_dict = {
+        'morning_star_rating_5': morning_star_rating_5_condition,
+        'morning_star_rating_3': morning_star_rating_3_condition,
+        'manager_start_date': '2020-05-25'
+    }
+    fund_code_pool = each_statistic.select_fund_pool(
+        **condition_dict,
+    )
+    # pprint(fund_code_pool)
     stock_top_list = each_statistic.all_stock_fund_count(
         quarter_index="2021-Q1",
-        fund_code_pool=None,
-        filter_count=90)
+        fund_code_pool=fund_code_pool,
+        filter_count=5)
     # print('2020-Q4 top 100 股票')
-    # pprint(stock_top_list)
-    # print(len(stock_top_list))
+    pprint(stock_top_list)
+    print(len(stock_top_list))
 
-    filter_list = stocks_compare(stock_top_list)
+    filter_list = stocks_compare(stock_top_list, fund_code_pool)
 
-    # pprint(filter_list)
-    # pprint(len(filter_list))
-    df_filter_list = pd.DataFrame(filter_list, columns=[
-        '名称', '2021-Q1持有数量（只）', '2020-Q4持有数量（只）', '环比', '环比百分比', '升Or降'])
-    print(df_filter_list)
-    df_filter_list.to_excel(
-        './output/xlsx/top100-2021_q1_vs_2020_q4.xlsx', sheet_name='top100')
+    pprint(filter_list)
+    pprint(len(filter_list))
+    # df_filter_list = pd.DataFrame(filter_list, columns=[
+    #     '名称', '2021-Q1持有数量（只）', '2020-Q4持有数量（只）', '环比', '环比百分比', '升Or降'])
+    # print(df_filter_list)
+    # df_filter_list.to_excel(
+    #     './output/xlsx/top100-2021_q1_vs_2020_q4.xlsx', sheet_name='top100')

src/sql_model/fund_query.py

@@ -12,6 +12,26 @@ from threading import Lock
 from db.connect import connect
 
 
+def format_sql(table_name, field_dict, prefix="AND"):
+    print("field_dict", field_dict['name'])
+    sql_str = ''
+    field_name = field_dict.get('name')
+    field_value = field_dict.get('value')
+    operator = field_dict.get('operator')
+    if not field_name or not field_value or not operator:
+        return sql_str
+    format_dict = {
+        "prefix": prefix,
+        'table_name': table_name,
+        'field_name': field_name,
+        "operator": operator,
+        "value": field_value
+    }
+    sql_str = '{prefix} {table_name}.{field_name} {operator} {value}'.format(
+        **format_dict)
+    return sql_str
+
+
 class FundQuery:
 
     def __init__(self):
@@ -86,20 +106,40 @@ class FundQuery:
         self.lock.release()
         return results
 
-    def select_certain_condition_funds(self, *, quarter_index=None, morning_star_rating_5=None, morning_star_rating_3=None):
+    def select_certain_condition_funds(self, *, quarter_index=None, morning_star_rating_5=None, morning_star_rating_3=None, manager_start_date=None):
         if quarter_index == None:
             quarter_index = self.quarter_index
-        print("condition", quarter_index, morning_star_rating_5)
-        sql = "SELECT a.fund_code, b.fund_name, a.quarter_index, a.total_asset , a.manager_start_date, \
-            a.investname_style, a.three_month_retracement, a.june_month_retracement, a.risk_assessment_sharpby,\
-            a.risk_statistics_alpha, a.risk_statistics_beta, a.risk_statistics_r_square, a.risk_assessment_standard_deviation,\
-            a.risk_assessment_risk_coefficient, a.risk_rating_2, a.risk_rating_3, a.risk_rating_5, a.morning_star_rating_5,\
-            a.morning_star_rating_3, a.stock_position_total, a.stock_position_ten FROM fund_morning_quarter as a \
+        last_year_time = time.localtime(time.time() - 365 * 24 * 3600)
+        last_year_date = time.strftime('%Y-%m-%d', last_year_time)
+        morning_star_rating_3_sql = format_sql('a', {
+            'name': 'morning_star_rating_3',
+            **morning_star_rating_3
+        })
+        # site = {"name": "菜鸟教程", "url": ""}
+        # print("网站名：{name}, 地址 {url}".format(**site))
+        # if morning_star_rating_5 not None:
+
+        sql = "SELECT a.fund_code FROM fund_morning_quarter as a \
             LEFT JOIN fund_morning_base AS b ON a.fund_code = b.fund_code \
-            WHERE b.fund_name NOT LIKE '%%C' AND b.fund_name NOT LIKE '%%E'  AND a.quarter_index = '2021-Q1' AND \
-            a.morning_star_rating_5 >= 3 AND a.morning_star_rating_3 = 5 AND a.stock_position_total >= 50 AND a.stock_position_ten <= 60 \
-            AND a.risk_assessment_sharpby >1 AND a.risk_rating_2 > 1 AND a.risk_rating_3 > 1 AND a.risk_rating_5 > 1 AND a.manager_start_date < '2020-05-24'"
-        # TODO: finish
+            WHERE a.quarter_index = '{quarter_index}' AND b.fund_name NOT LIKE '%%C' AND b.fund_name NOT LIKE '%%E' {morning_star_rating_5} {morning_star_rating_3} AND a.stock_position_total >= 50 AND a.stock_position_ten <= 60 \
+            AND a.risk_assessment_sharpby >1 AND a.risk_rating_2 > 1 AND a.risk_rating_3 > 1 AND a.risk_rating_5 > 1 AND a.manager_start_date < '{manager_start_date}'"
+
+        format_dict = {
+            'quarter_index': quarter_index,
+            'morning_star_rating_5': "AND a.morning_star_rating_5 >= 3",
+            'morning_star_rating_3': morning_star_rating_3_sql,
+            'manager_start_date': str(last_year_date)
+        }
+        sql_format = sql.format(**format_dict)
+        self.lock.acquire()
+        self.cursor.execute(sql_format)    # 执行sql语句
+        results = self.cursor.fetchall()    # 获取查询的所有记录
+        self.lock.release()
+        fund_pool = []
+        for item in results:
+            fund_code = item[0]
+            fund_pool.append(fund_code)
+        return fund_pool
 
     # 筛选同类基金，除了A类
     def select_similar_fund(self, similar_name):