chore: add get_season_index utils && comment code

.gitignore

@@ -6,6 +6,7 @@ code-record/*
 *$py.class
 .vscode/
 *fund_morning_star.csv
+abnormal
 
 
 # C extensions

src/acquire_fund_season.py

@@ -19,7 +19,7 @@ from pprint import pprint
 import pymysql
 import pandas
 connect = pymysql.connect(host='127.0.0.1', user='root',
-                          password='rootroot', db='fund_work', charset='utf8')
+                          password='xxxxx', db='fund_work', charset='utf8')
 cursor = connect.cursor()
 lock = Lock()
 
@@ -123,6 +123,7 @@ if __name__ == '__main__':
                 each_fund = FundSpider(
                     record[0], record[1], record[2], chrome_driver, morning_cookies)
                 is_normal = each_fund.go_fund_url()
+                # 是否能正常跳转到基金详情页，没有的话，写入csv,退出当前循环
                 if is_normal == False:
                     lock.acquire()
                     error_funds.append(each_fund.fund_code)
@@ -139,8 +140,10 @@ if __name__ == '__main__':
                 each_fund.get_fund_manager_info()  # 基金经理模块
                 each_fund.get_fund_morning_rating()  # 基金晨星评级
                 each_fund.get_fund_qt_rating()  # 基金风险评级
+                # 判断是否有股票持仓，有则爬取
                 if each_fund.stock_position['total'] != '0.00' and each_fund.total_asset != None:
                     each_fund.get_asset_composition_info()
+                # 爬取过程中是否有异常
                 if each_fund._is_trigger_catch == True:
                     lock.acquire()
                     fund_infos = [each_fund.fund_code, each_fund.morning_star_code,
@@ -171,6 +174,7 @@ if __name__ == '__main__':
                                    tuple(manager_dict.values()))
                     connect.commit()
                     lock.release()
+                # 季度信息  TODO: 对比数据更新时间field
                 season_dict = {
                     'id': snow_flake_id,
                     'season_number': each_fund.season_number,

src/fund_info_crawler.py

@@ -18,7 +18,7 @@ from selenium.common.exceptions import NoSuchElementException
 class FundSpider:
     # 初始化定义，利用基金代码、基金名称进行唯一化
     def __init__(self, code, namecode, name,  chrome_driver, morning_cookies):
-        self.season_number = '2021-s1'
+        self.season_number = '2021-s1'  # TODO: get season_number by current time
         self.fund_code = code  # 基金代码，需要初始化赋值
         self.fund_name = name  # 基金名称，需要初始化赋值
         self.morning_star_code = namecode  # 基金编码，晨星网特有，需要建立索引表
@@ -247,7 +247,7 @@ class FundSpider:
             file_name = './abnormal/qt_rating-' + self.fund_code + "-no_such_element.png"
 
     def get_fund_season_info(self):
-        # 总资产
+        # 总资产  TODO: 增加一个数据更新时间field
         self.total_asset = self.get_element_text_by_class_name(
             "asset", 'qt_base')
         # 投资风格
@@ -259,7 +259,7 @@ class FundSpider:
         # 最差六个月回报
         self.june_month_retracement = self.get_element_text_by_class_name(
             "r6", 'qt_worst')
-        # 获取股票总仓位、前十大持仓、债券总仓位、前五大持仓
+        # 获取股票总仓位、前十大持仓、债券总仓位、前五大持仓 TODO: 增加一个数据更新时间field
         total = self.get_element_text_by_class_name(
             "stock", 'qt_asset')
         self.stock_position["total"] = total if total != None else '0.00'
@@ -291,6 +291,7 @@ class FundSpider:
         # 获取标准差
         # standard_deviation = self._chrome_driver.find_element_by_id(
         #     "qt_risk").find_element_by_xpath('li[16]').text
+        #  TODO: 增加一个数据更新时间field
         standard_deviation = self.get_element_text_by_xpath(
             'li[16]', 'qt_risk')
         if standard_deviation != None:

src/utils.py

@@ -1,6 +1,7 @@
 
 from urllib import parse
 import time
+import datetime
 import os
 
 
@@ -151,3 +152,20 @@ def parse_csv(datafile):
             counter += 1
 
     return data
+
+
+def get_season_index(input_date):
+    year = time.strftime("%Y", time.localtime())
+    boundary_date_list = ['03-31', '06-30', '09-30', '12-31']
+
+    input_date_strptime = datetime.datetime.strptime(
+        year + '-' + input_date, '%Y-%m-%d')
+    index = 1
+    for idx in range(len(boundary_date_list)):
+        join_date = year + '-' + boundary_date_list[idx]
+        season_date_strptime = datetime.datetime.strptime(
+            join_date, '%Y-%m-%d')
+        if input_date_strptime <= season_date_strptime:
+            index = idx + 1
+            break
+    return index