How to create new columns from key: value pairs in big query?

Question

I have a table A with the structure:

id	key	value
1	"abc"	"123"
1	"def"	"346"
1	"xyz"	"789"
2	"abc"	"462"
2	"def"	"464"
3	"abc"	"362"

However this representation proves to be a bit difficult to process in our infra pipelines, so I'm trying to create a new table/materialized view in big query with the following structure:

id	"abc"	"def"	"xyz"
1	"123"	"346"	"789"
2	"462"	"464"	NULL
3	"362"	NULL	NULL

The number of keys is guaranteed to be small (<100). What's the best way to do this using only SQL/JavaScript UDFs in Big Query?

Answer 1

Consider below approach

execute immediate format('''
select * from your_table
pivot (any_value(value) for key in (%s))
''', (select string_agg(distinct "'" || key || "'") from your_table))           

if applied to sample data in your question - output is