Support Cluster By in table definition

[pythonruss]

There needs to be a dialect construct for adding a cluster by to a table definition

Here's what they did for the snowflake sqlalchemy package link

[pythonruss]

BTW anyone who runs into this as a workaround:

from sqlalchemy_bigquery import dialect as bq_dialect
from sqlalchemy import Table
from sqlalchemy.schema import CreateTable
from sqlalchemy.ext.compiler import compiles

@compiles(CreateTable, "bigquery")
def compile_create_table(create, compiler, **kw):
    preparer = compiler.preparer
    stmt = compiler.visit_create_table(create, **kw)
    bigquery_opts = create.element.dialect_options["bigquery"]
    cluster_by = bigquery_opts.get("clusterby")
    if cluster_by:
        columns = ", ".join([
            compiler.sql_compiler.process(col, include_table=False, **kw)
            for col in cluster_by
        ])
        stmt = f"{stmt} CLUSTER BY {columns}\n\n"
    return stmt


    dialect = bq_dialect
    dialect.construct_arguments = [
        (Table, {"clusterby": None } )
        ]

[jlynchMicron]

Possibly duplicate of: #395

[leblancfg]

To anyone coming here trying to extend the dialect to support partitioning and/or clustering: it looks like @pythonruss' code doesn't work as a workaround and made other parts of my existing code break. Would make a good forked branch or PR, but doesn't extend the dialect.