如何通过在python的同一psql表中添加2列来获得结果

Question

I recently started learning about databases. Now I am testing my skills on a database backed application in python. I have created 2 tables in database something like this

CREATE TABLE players
(
    id     SERIAL PRIMARY KEY,
    name   TEXT
);

CREATE TABLE matches 
(
    game  SERIAL,
    winner SERIAL REFERENCES players(id),
    loser SERIAL REFERENCES players(id)
);

I am tying to make a query to those to get results as playerid, name wins, matches in order something like (id, name, wins, matches) .

I am using this code to do that

SELECT id, name, count(matches.winner) as wins, 
 (count(matches.winner) + count(matches.loser)) as match 
  FROM players LEFT JOIN matches ON players.id = matches.winner 
  GROUP BY players.id ORDER BY wins

it seems it is not working please help me to solve this.

Thanks.

I have tried this code below

    DB = connect()
    c = DB.cursor()
    c.execute("""WITH
                   results AS (
                            select P.id,
                                   P.name,
                                   1     win
                            from   players P inner join matches M
                            on M.winner = P.id
                            union all
                            select P.id,
                                   P.name,
                                   0    win
                            from   players P inner join matches M
                            on M.loser = P.id)
                select id,
                       name,
                       sum(win)   wins,
                       count(*)   matches
                from   results
                group by id, name """)
    return c.fetchall()
    DB.close()

This code is not getting any data from the database instead it is returning just []. I have also tried to replacing my query string with

query = oldstring.replace(" ","_")
c.execute(query)

but it is also not worked. Please help me to solve this.

Answer 1

First, you probably don't want to use SERIAL data type for FK columns, as it is an autoincrementing integer type which is good for PK, but not suitable for FK, better stick with INTEGER .

Second, consider some constraints to enforce what we call business-rules. Are there always two players? Is it realistic for match to have no winner or loser? What if it ends with a tie?

Anyway, let's keep it simple for now and assume that any time two players face each other one of them emerges the winner while the other one loses. So the total number of matches is wins + losses.

WITH
  results AS (
    select P.id,
           P.name,
           1     win
    from   players P inner join matches M
    on M.winner = P.id
    union all
    select P.id,
           P.name,
           0    win
    from   players P inner join matches M
    on M.loser = P.id)
select id,
       name,
       sum(win)   wins,
       count(*)   matches
from   results
group by id, name

So, what is going on here:

• WITH is a Common Table Expression . It is basically a named subquery. I used it here for readability. I could put subquery inside the FROM clause, but it looks much better this way.
• You need to join players and matches somehow, but each row in matches references two players. Ok, then, we will join them twice.
• When you join on matches.winner you get one row for every match, and there is the winner in this row. This is already enough to count wins, but there is absolutely no info on losses.
• Second query adds one more row for every game, but this time there is loser in this row.
• UNION ALL combines results of two queries in one resultset.
• But we need to be able to distinguish between rows with different outcomes. The simpliest way is to sum numbers, so I set 1 for wins and 0 for losses.
• Finally group rows now using GROUP BY and aggregate.

You can count losses in similar fashion.