数据

bbztx/get_data.py

@@ -2,9 +2,6 @@
 # -*- coding:utf-8 -*-
 from util import pgsql_util
 import json
-import sys
-reload(sys)
-sys.setdefaultencoding('utf8')
 
 query_articles_sql = '''
 select a.aid,a.title,a.cities,a.cid,a.other_info,"array_agg"(t.tag_value),days,a.recom,a."rank",read_count
@@ -18,39 +15,40 @@ limit 100 OFFSET 0
 
 query_sell_main_sql = '''
 select count(*)
-FROM order 
-where aid in (select aid from articles where stock_aid = %s)
+FROM zhima.order 
+where aid in (select aid from articles where stock_aid = '%s')
 and pay_status = 7
 '''
 
 query_stock_main_sql = '''
 select sum(maxt::int)
 FROM zm_alipay 
-where aid in (select aid from articles where stock_aid = %s)
+where aid in (select aid from articles where stock_aid = '%s')
 and status in (1,2)
 '''
 
 query_sell_main_sql_2 = '''
 select count(*)
-FROM order 
-where aid = %s
+FROM zhima.order 
+where aid = '%s'
 and pay_status = 7
 '''
 
 query_stock_main_sql_2 = '''
 select sum(maxt::int)
 FROM zm_alipay 
-where aid = %s
+where aid = '%s'
 and status in (1,2)
 '''
 
+
 def get_articles():
     rows = pgsql_util.get_rows(query_articles_sql)
 
     attr_list = []
     for row in rows:
         attr = {}
-        attr['city'] = row['cities'].encode('utf-8')
+        attr['city'] = row['cities']
         attr['cid'] = row['cid']
 
         other_info = json.loads(row['other_info'])
@@ -64,13 +62,89 @@ def get_articles():
         attr['rank'] = row['rank']
         attr['read_count'] = row['read_count']
 
-
-
+        if row['cities'] == '株洲':
+            sell_count = pgsql_util.get_rows(query_sell_main_sql % (row['aid']))[0]['count']
+            stock_count = pgsql_util.get_rows(query_stock_main_sql % (row['aid']))[0]['sum']
+        else:
+            sell_count = pgsql_util.get_rows(query_sell_main_sql_2 % (row['aid']))[0]['count']
+            stock_count = pgsql_util.get_rows(query_stock_main_sql_2 % (row['aid']))[0]['sum']
+        attr['sell_count'] = sell_count
+        attr['stock_count'] = stock_count
 
         attr_list.append(attr)
 
-        print attr
+        print(attr)
+    return attr_list
+
+
+def to_a1(attr_list):
+    for attr in attr_list:
+        if attr['recruit'] is None:
+            attr['recruit'] = 0
+
+        if attr['stock_count'] is None:
+            attr['stock_count'] = 0
+
+    new_attr_list = []
+    for attr in attr_list:
+        new_attr = []
+        city = attr['city']
+        cid = attr['cid']
+        dtype = attr['dtype']
+        recruit = attr['recruit']
+        country = attr['country']
+        price = attr['price']
+        days = attr['days']
+        recom = attr['recom']
+        rank = attr['rank']
+        read_count = attr['read_count']
+        sell_count = attr['sell_count']
+        stock_count = attr['stock_count']
+
+        new_attr.extend(to_list_attr(city, open_city_list))
+        new_attr.extend(to_list_attr(cid, open_cid_list))
+        new_attr.extend(to_list_attr(dtype, open_dtype_list))
+        new_attr.extend(to_list_attr(recruit, open_recruit_list))
+        new_attr.extend(to_list_attr(country, open_country_list))
+        new_attr.append(float(price))
+        new_attr.append(float(days))
+        new_attr.extend(to_list_attr(recom, open_recom_list))
+        new_attr.append(rank)
+        new_attr.append(read_count)
+        new_attr.append(stock_count)
+        new_attr.append(sell_count)
+
+        print(new_attr)
+
+    return new_attr_list
+
+
+open_city_list = ['北京', '长沙', '成都', '重庆', '佛山', '福州', '广州', '杭州', '合肥', '济南',
+                  '南京', '宁波', '青岛', '上海', '深圳', '苏州', '天津', '武汉', '厦门', '西安',
+                  '郑州', '株洲']
+open_cid_list = [23,24,25,26,27,28,29]
+open_dtype_list = [0,1,2,3,4,5]
+open_recruit_list = [0,1,2,3,4,5,6,7,8,9]
+open_country_list = ['港澳台', '国内', '新加坡', '英国', '日本', '泰国', '美国']
+open_recom_list = ['yes', 'no']
+
+def city_attr(city):
+    return to_list_attr(city, open_city_list)
+
+def cid_attr(cid):
+    return to_list_attr(cid, open_cid_list)
+
+
+def to_list_attr(item, a_list):
+    c_list = [0 for i in a_list]
+    try:
+        idx = a_list.index(item)
+    except:
+        idx = 0
+    c_list[idx] = 1
+    return c_list
 
 
 if __name__ == '__main__':
-    get_articles()
+    attr_list = get_articles()
+    new_attr_list = to_a1(attr_list)