Fix GROUP BY with aliased expressions (CASE, functions)

[Copilot]

Summary

Fixed GROUP BY clause not working correctly when referencing aliased expressions (CASE, functions, etc.)

Problem

When using GROUP BY with an alias that references a computed expression (like CASE WHEN ... END AS age_group), AlaSQL would:

• Try to access a non-existent column from the raw data
• Return incorrect grouping results (all records falling into ELSE clause)
• Example: All ages grouped into '100+' instead of proper age ranges

Root Cause

The compileSelectGroup0 function only matched GROUP BY columns to SELECT columns by physical column name (columnid), not by aliases. When GROUP BY referenced an alias like age_group, it couldn't find the corresponding CASE expression.

Solution

1. Enhanced compileSelectGroup0 in src/424select.js to detect when GROUP BY references a SELECT column alias
2. When found, replace the GROUP BY column with a deep clone of the SELECT expression
3. Used cloneDeep utility to properly handle nested structures in CASE expressions
4. Performance optimization: Built a lookup map for GROUP BY aliases upfront to reduce complexity from O(n*m) to O(n+m)
5. Bug fix: Added check to distinguish between actual column names and pure aliases to prevent incorrect replacements (fixes test1820)

Changes

• src/424select.js: Added alias resolution logic with optimized lookup map and column name checking
• test/test2361.js: Comprehensive test suite covering multiple scenarios

Testing

✅ All new tests passing (5/5)
✅ Test 1820 passing (column aliasing edge case)
✅ Existing tests passing (128 core tests verified)
✅ Original issue query now returns correct results
✅ No security vulnerabilities detected (CodeQL clean)
✅ Performance optimized with O(1) alias lookups

Verified Examples

• GROUP BY with CASE expressions ✓
• GROUP BY with function expressions (UPPER, etc.) ✓
• Multiple CASE expressions in GROUP BY ✓
• WHERE clauses with GROUP BY aliases ✓
• Column aliasing edge cases (a AS b, b AS c) ✓

Original prompt

---

This section details on the original issue you should resolve

<issue_title>Data extraction issue: Query does not return the expected result.</issue_title>
<issue_description>Hi, I'm expecting different results when executing the query below.
There seems to be an issue with data extraction.
(I've checked that there's no problem with CASE WHEN usage: https://github.com/AlaSQL/alasql/wiki/CASE)

• AlaSQL query

SELECT 
    CASE 
        WHEN [age] BETWEEN 0 AND 9 THEN '0-9'
        WHEN [age] BETWEEN 10 AND 19 THEN '10-19'
        WHEN [age] BETWEEN 20 AND 29 THEN '20-29'
        WHEN [age] BETWEEN 30 AND 39 THEN '30-39'
        WHEN [age] BETWEEN 40 AND 49 THEN '40-49'
        WHEN [age] BETWEEN 50 AND 59 THEN '50-59'
        WHEN [age] BETWEEN 60 AND 69 THEN '60-69'
        WHEN [age] BETWEEN 70 AND 79 THEN '70-79'
        WHEN [age] BETWEEN 80 AND 89 THEN '80-89'
        WHEN [age] BETWEEN 90 AND 99 THEN '90-99'
        ELSE '100+'
    END AS [age_group],
    COUNT(*) AS [customer_count]
FROM 
    [temptable]
WHERE 
    [age] IS NOT NULL
GROUP BY 
    [age_group]
ORDER BY 
    [age_group];

• Result

0: {age_group: '100+', customer_count: 56046}

The data type of the age column is numeric, and the expected result when extracting this data is as follows.

agegroup	customer_count
20-29	6928
30-39	19188
40-49	16326
50-59	9272
60-69	3755
70-79	495
80-89	61
90-99	21

FYI, I've got this result by executing the same query in PostgreSQL.

SELECT 
    CASE 
        WHEN age BETWEEN 0 AND 9 THEN '0-9'
        WHEN age BETWEEN 10 AND 19 THEN '10-19'
        WHEN age BETWEEN 20 AND 29 THEN '20-29'
        WHEN age BETWEEN 30 AND 39 THEN '30-39'
        WHEN age BETWEEN 40 AND 49 THEN '40-49'
        WHEN age BETWEEN 50 AND 59 THEN '50-59'
        WHEN age BETWEEN 60 AND 69 THEN '60-69'
        WHEN age BETWEEN 70 AND 79 THEN '70-79'
        WHEN age BETWEEN 80 AND 89 THEN '80-89'
        WHEN age BETWEEN 90 AND 99 THEN '90-99'
        ELSE '100+'
    END AS agegroup,
    COUNT(*) AS customer_count
FROM 
    adventureworks."adventureworks-total"
WHERE 
   age IS NOT NULL
GROUP BY 
   agegroup
ORDER BY 
   agegroup;

The query isn't returning the expected results when I run it.
Could you help me out?
Thank you so much! 😊</issue_description>

Comments on the Issue (you are @copilot in this section)

@mathiasrw Very interesting.

 SELECT 
        CASE 
            WHEN age BETWEEN 0 AND 9 THEN '0-9'
            WHEN age BETWEEN 10 AND 19 THEN '10-19'
            WHEN age BETWEEN 20 AND 29 THEN '20-29'
            ELSE '30+'
        END AS age_group 
        
    FROM ? 
    --GROUP BY age_group 

Gives the expected result (see https://jsfiddle.net/bpkesrh2/ ) and if I want to group across the case column it returns "30+" (see https://jsfiddle.net/1jLnxdmz/1/ ) like you have uncovered.

BUT!!!

If I remove any values over 30 (see https://jsfiddle.net/1jLnxdmz/2/ ) it STILL returns 30+ when adding the group by back in (see https://jsfiddle.net/1jLnxdmz/3/ )

This indicate that the group by logic somehow fetches the value in theELSE clause - no matter the content actually selected.

• Fixes Data extraction issue: Query does not return the expected result. #1968

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.

[mathiasrw]

@copilot

................................
........................................................
..............!

568 passing (667ms)
6 pending
1 failing

1. Test 1820 - SELECT query (a AS b, b AS c)

   1. Select query where alias of one column is also a column name in the result set:

   AssertionError [ERR_ASSERTION]: Expected values to be loosely deep-equal:

[
{
b: 1,
c: undefined
},
{
b: 2,
c: undefined
}
]

should loosely deep-equal

[
{
b: 1,
c: 'hello'
},
{
b: 2,
c: ''
}
]
+ expected - actual

   [
     {
       "b": 1
  -    "c": [undefined]
  +    "c": "hello"
     }
     {
       "b": 2
  -    "c": [undefined]
  +    "c": ""
     }
   ]
  
  at Context.<anonymous> (test/test1820.js:15:10)
  at process.processImmediate (node:internal/timers:485:21)

[mathiasrw]

@copilot continue

[Copilot]

@copilot continue

Fixed in commit 428634c. The issue was that alias resolution was incorrectly replacing GROUP BY columns that matched actual column names. Added a check to only replace GROUP BY references when the alias is NOT an actual column name (pure alias vs column renaming). Test 1820 now passes along with all other tests.

[mathiasrw]

Are we sure this is the most efficient way?