CODEWITHSUNDEEP
pythonmysqlpandassqlalchemy
ostrokach
ostrokach

Reputation: 19932

How to get autoincrement values for a column after uploading a Pandas dataframe to a MySQL database

I have a Pandas DataFrame (called df), which I would like to upload to a MySql database. The dataframe has columns [A, B, C] and the table in the database has columns [ID, A, B, C]. The ID column in the database is the auto-incrementing primary key.

I can upload the dataframe to the database using the df.to_sql('table_name', engine) command. However, this does not give me any information about the values that the database assigned to the ID column of the incoming data. The only way I have of getting this information is by querying the database using the values for columns A, B, C:

select 
ID, A, B, C 
from db_table 
where (A, B, C) in ((x1, y1, z1), (x2, y2, z2), ...)

However, this query takes a very long time when I am inserting a lot of data.

Is there a simpler and quicker way of getting the values that the database assigned to the ID column of the incoming data?

Edit 1: I can assign the ID column myself, as per user3364098's answer below. However, my job is part of a pipeline that is ran in parallel. If I assign the ID column myself, there is a chance that I may assign the same id values to different dataframes that are uploaded at the same time. This is why I would like to relegate the ID assignment task to the database.

Solution: I ended up assigning the ID column myself, and issuing a lock on the table while uploading the data in order to guarantee that no other process uploads data with the same id value. Basically:

try:
    engine.execute('lock tables `table_name` write')
    max_id_query = 'select max(ID) FROM `table_name`'
    max_id = int(pd.read_sql_query(max_id_query, engine).values)
    df['ID'] = range(max_id + 1, max_id + len(df) + 1)
    df.to_sql('table_name', engine, if_exists='append', index=False)
finally:
    engine.execute('unlock tables')

Upvotes: 14

Views: 7373

Answers (2)

trader
trader

Reputation: 65

import pandas as pd
df['ID'] = pd.read_sql_query('select MAX(ID)+1 from db_table',cnx).iloc[0,0] + range(len(df))

Upvotes: -1

hvedrung
hvedrung

Reputation: 467

You can assign id by yourself:

import pandas as pd
df['ID'] = pd.read_sql_query('select ifnull(max(id),0)+1 from db_table',cnx).iloc[0,0]+range(len(df))

where cnx is your connection and then upload your df.

Upvotes: 11

Related Questions