可能使用pandas/sqlalchemy將數組插入到sql數據庫中？ （Postgres的）

Question

下列要求：

engine = sqlalchemy.create_engine(url) 

df = pd.DataFrame({ 
    "eid": [1,2], 
    "f_i": [123, 1231], 
    "f_i_arr": [[123], [0]], 
    "f_53": ["2013/12/1","2013/12/1",], 
    "f_53a": [["2013/12/1"], ["2013/12/1"],], 
}) 

with engine.connect() as con: 
    con.execute(""" 
     DROP TABLE IF EXISTS public.test; 
     CREATE TABLE public.test 
     (
      eid integer NOT NULL, 
      f_i INTEGER NULL, 
      f_i_arr INTEGER NULL, 
      f_53 DATE NULL, 
      f_53a DATE[] NULL, 
      PRIMARY KEY(eid) 
     );; 
     """) 
    df.to_sql("test", con, if_exists='append') 

如果我嘗試插入僅列 「f_53」（一個date）成功爲止。

如果我嘗試添加列「f_53a」（一date[]）它失敗：

          ^
sqlalchemy.exc.ProgrammingError: (psycopg2.ProgrammingError) column "f_53a" is of type date[] but expression is of type text[] 
LINE 1: ..._53, f_53a, f_i, f_i_arr) VALUES (1, '2013/12/1', ARRAY['201... 
                  ^
HINT: You will need to rewrite or cast the expression. 
[SQL: 'INSERT INTO test (eid, f_53, f_53a, f_i, f_i_arr) VALUES (%(eid)s, %(f_53)s, %(f_53a)s, %(f_i)s, %(f_i_arr)s)'] [parameters: ({'f_53': '2013/12/1', 'f_53a': ['2013/12/1', '2013/12/1'], 'f_i_arr': [123], 'eid': 1, 'f_i': 123}, {'f_53': '2013/12/1', 'f_53a': ['2013/12/1', '2013/12/1'], 'f_i_arr': [0], 'eid': 2, 'f_i': 1231})] 

Answer 1

是 - 可以從一個數據幀插入[]和[][]類型向Postgres形成一個數據幀。

與平面日期類型不同，它們可能會被sql正確解析，因此需要先將DATE[]和DATE[][]轉換爲日期時間對象。像這樣。

with engine.connect() as con: 
    con.execute(""" 
     DROP TABLE IF EXISTS public.test; 
     CREATE TABLE public.test 
     (
      eid integer NOT NULL, 
      f_i INTEGER NULL, 
      f_ia INTEGER[] NULL, 
      f_iaa INTEGER[][] NULL, 
      f_d DATE NULL, 
      f_da DATE[] NULL, 
      f_daa DATE[][] NULL, 
      PRIMARY KEY(eid) 
     ); 
     """) 


    d = pd.to_datetime("2013/12/1") 
    i = 99 
    df = pd.DataFrame({ 
     "eid": [1,2], 

     "f_i": [i,i], 
     "f_ia": [None, [i,i]], 
     "f_iaa": [[[i,i],[i,i]], None], 

     "f_d": [d,d], 
     "f_da": [[d,d],None], 
     "f_daa": [[[d,d],[d,d]],None], 
    }) 
    df.to_sql("test", con, if_exists='append', index=None)