feat: add MapColumns method

[molon]

• Do only one thing
• Non breaking API changes
• Tested

What did this pull request do?

Add MapColumns method: modify the column names in the query results to facilitate align to the corresponding structural fields

User Case Description

I recently used gorm to do aggregation queries based on timescaledb. For example, the following query:

SELECT
    symbol,
    '5min' AS timeframe,
    TRUE AS adj,
    time_bucket('5 minute', t, 'America/New_York') AS bucket,
    time_bucket('5 minute', t, 'America/New_York') AS t,
    FIRST(o, t) AS o,
    MAX(h) AS h,
    MIN(l) AS l,
    LAST(c, t) AS c,
    SUM(v) AS v,
    SUM(n) AS n,
    SUM(v*c)/SUM(v) AS vw
FROM
    market.candle
WHERE
    adj = TRUE AND
    timeframe = '1min' AND 
    symbol = 'AAPL' AND
    time_bucket('5 minute', t, 'America/New_York') > '2023-03-03 17:01:00+08'
GROUP BY
    symbol, bucket
ORDER BY
    symbol, bucket ASC
LIMIT 10000;

bucketWidth := timeFrameToBucketWidth(req.TimeFrame)
db = db.Select(`
	symbol,
	? AS timeframe,
	TRUE AS adj,
	time_bucket(?, t, ?) AS bucket,
	time_bucket(?, t, ?) AS t,
	FIRST(o, t) AS o,
	MAX(h) AS h,
	MIN(l) AS l,
	LAST(c, t) AS c,
	SUM(v) AS v,
	SUM(n) AS n,
	SUM(v*c)/SUM(v) AS vw
`, req.TimeFrame.String(), bucketWidth, s.locationName, bucketWidth, s.locationName)
db = db.Where(`symbol IN ? AND timeframe = '1min' AND adj = ?`, req.Symbols, req.Adjustment)
if !req.Start.IsZero() {
	db = db.Where(fmt.Sprintf("time_bucket(?, t, ?) %s ?", startCompare), bucketWidth, s.locationName, req.Start)
}
if !req.End.IsZero() {
	db = db.Where("time_bucket(?, t, ?) < ?", bucketWidth, s.locationName, req.End)
}
db = db.Group("symbol, bucket")
db = db.Order("symbol ASC, bucket ASC")

The result will be an error ERROR: column "candle.t" must appear in the GROUP BY clause or be used in an aggregate function (SQLSTATE 42803).
However, I cannot put t in Group, which will affect the correct result. So I tried using db = db.Group("symbol, time_bucket('5 minute', t, 'America/New_York')") which unfortunately returns the same error.
So I can only give up returning the t column in the result set, but then I need to customize a nested structure to supplement a bucket column, and then manually assign it to MarketCandle.T. This is very inconvenient.
So I added the MapColumns method. I think this may be a requirement that also exists in other scenarios, especially when GROUP BY is used.

[jinzhu]

Hi @molon

Could you please include some tests to ensure it functions correctly?

[molon]

Hi @molon

Could you please include some tests to ensure it functions correctly?

When I tried to write test cases, I found that my current implementation was unable to meet the scenario where Embedded existed.
Because Embedded will cause duplicate column names to appear.

Do you have any suggestions for this?

[jinzhu]

Hi @molon
Could you please include some tests to ensure it functions correctly?

When I tried to write test cases, I found that my current implementation was unable to meet the scenario where Embedded existed. Because Embedded will cause duplicate column names to appear.

Do you have any suggestions for this?

Embedded fields do not necessarily lead to duplicate field names, but I believe we should be able to disregard such issues, right?

[molon]

Agreed, already added it.

[shayan0v0n]

thanks for adding this