feat(snowflake): add type annotation for DECODE function support #6199
Conversation
fivetran-ashashankar
changed the title
|
Decode function support is already available . added support to return type annotation for DECODE function. |
The annotation now determines the output type based solely on the return values (ret1, ret2, ..., default), ignoring comparison values.
sqlglot/dialects/snowflake.py
Outdated
| if len(expressions) % 2 == 1: | ||
| return_types.append(expressions[-1].type) |
There was a problem hiding this comment.
That'd be if the total number of expressions is even, right? Because:
- One argument is the input
<expr> - N pairs of
<search1> , <result1> - [Optional] The
<default>value
So if we don't have (3) there'd be 2*N + 1 arguments
There was a problem hiding this comment.
agree. Thank you. DECODE(x, 1, 'a', 2, 'b', 'default'). length 6.
in case of even length , return types can be obtained from the default. Fixed the code to check for even length
| DECODE(x, 1, 100, 2, 200, 0); | ||
| INT; |
There was a problem hiding this comment.
Let's also add test cases with mixed types e.g integers and floats:
snowflake> WITH t0 AS (SELECT DECODE(100, 100, 1, 90, 2, 5.5) AS col) SELECT SYSTEM$TYPEOF(col), col FROM t0;
COL_TYPE | COL
NUMBER(2,1) | 1.0
There was a problem hiding this comment.
added tests
sqlglot/dialects/snowflake.py
Outdated
| return_types = [] | ||
| for i in range(2, len(expressions), 2): | ||
| return_types.append(expressions[i].type) |
There was a problem hiding this comment.
[Nit] We can make this a one-liner with list comprehension
There was a problem hiding this comment.
updated to use a one-liner list comprehension.
…implemented. Fixed the annotate_decode_case function to identify the returntype based on default param and added more tests
fivetran-ashashankar
requested review from
VaggelisD,
fivetran-BradfordPaskewitz and
georgesittas
No description provided.