我怎样才能使这个 SQL 更容易？

Question

date    customer_name   service_name    price_paid
2021-01-01  Andrew      Cable TV        5000
2021-02-02  Brad        cabletv         5000
2021-03-03  Charlie     Cable TV        5000
2021-02-05  Dan         ISP             6000
2021-02-18  Eric        ISP             6000
2021-10-09  Felix       ISP             6000
2021-09-10  Gerald      isp             6000
2022-03-10  Hubert      Cable TV        5000
2022-04-12  Isaac       i.s.p           6000
2022-04-15  Jason       ISp 
2022-05-23  Karen       Cable T.V   
2022-06-23  Leah        ISP             6000
2022-05-17  Marie       Cable TV        5000
2022-06-11  Norman      ISP             6000

So i got this SQL output, as you can see my service_name is really messy. I wanna make it clean and i found this example

SELECT name, 
    CASE WHEN department == 'Math' THEN 'Math'
    ELSE upper(replace(replace(department, 'Information Technology', 'I.T'), 'it', 'I.T')) 
    END as department
FROM messy_df)

but when we use large data i think it will more harder becase we need replace one by one, can we make this more easier?

this is my code

sql_query = """
SELECT 
  strftime('%Y', date) 'year',
  strftime('%m', date) 'month',
  strftime('%d', date) 'date',
  customer_name, 
  replace(replace(replace(replace(replace(service_name, 'cabletv', 'Cable TV'), 'Cable T.V', 'Cable TV'), 'i.s.p', 'ISP'), 'isp', 'ISP'),'ISp', 'ISP') service_name,
  COALESCE(price_paid, (SELECT avg(price_paid) FROM df)) as 'price_paid'
FROM df
"""

sql_run(sql_query)

Answer 1

What you're doing is fine, given the state of the data.

I would create a view

CREATE VIEW vdf as SELECT *, CASE ... FROM df

that at least includes, for each "munged" column, the original, so you can see the transformation. The user/application will often want to see the unmodified form, too.

The best solution to this kind of problem (as with many problems) is prevention . The service_name column needs a constraint restricting it to a set of uniform values, so there's only one form of, say, "ISP". This is usually achieved via a lookup table and maybe a drop-down box in the application. Alternatively, your CASE clauses can be applied on input as part of INSERT.