BigQuery parsing produces incorrect AST for DATE_TRUNC where unit is WEEK(WEEKDAY)

[pmsanford]

I've tested this on the latest released version as well as on main and it is the same both places

Fully reproducible code snippet

parse_one("DATE_TRUNC(tbl.col, WEEK(MONDAY))", read="bigquery").sql("bigquery")
# produces 'DATE_TRUNC(tbl.col, )'

Official Documentation
Here are the BigQuery docs for DATE_TRUNC showing WEEK(WEEKDAY) is a valid option: https://cloud.google.com/bigquery/docs/reference/standard-sql/date_functions#date_trunc

I believe this happens because at this point:

sqlglot/sqlglot/dialects/bigquery.py

Line 166 in fb819f0

unit=exp.Literal.string(seq_get(args, 1).name), # type: ignore

the return value of seq_get(args, 1) for a WEEK (or QUARTER or other date part) is an exp.Column, but WEEK(MONDAY) is an exp.Week (which of course doesn't have a name property). I hacked together the following custom dialect, which fixes the issue, but I'm guessing isn't going to cut it as a real fix:

class NewBigquery(BigQuery):
    class Parser(BigQuery.Parser):
        FUNCTIONS = {
            **BigQuery.Parser.FUNCTIONS,
            "DATE_TRUNC": lambda args: exp.DateTrunc(
                unit=exp.Literal.string(seq_get(args, 1).sql()),  # type: ignore
                this=seq_get(args, 0),
                )
            }

I hacked around a little in bigquery.py to try and put together a PR but I wasn't able to quite figure out the right way to do it. I'm happy to contribute a PR if someone can point me in the right direction.

[tobymao]

thanks for the detailed report. i look forward to your future prs!