tpch_tables_scale_1 / prepare_large_files.py
kunishou's picture
Upload 284 files
2eae90c verified
import sys
import polars as pl
scale_fac = int(sys.argv[1])
h_nation = """
n_nationkey
n_name
n_regionkey
n_comment""".split(
"\n"
)
h_region = """
r_regionkey
r_name
r_comment""".split(
"\n"
)
h_part = """
p_partkey
p_name
p_mfgr
p_brand
p_type
p_size
p_container
p_retailprice
p_comment""".split(
"\n"
)
h_supplier = """
s_suppkey
s_name
s_address
s_nationkey
s_phone
s_acctbal
s_comment""".split(
"\n"
)
h_partsupp = """
ps_partkey
ps_suppkey
ps_availqty
ps_supplycost
ps_comment""".split(
"\n"
)
h_customer = """
c_custkey
c_name
c_address
c_nationkey
c_phone
c_acctbal
c_mktsegment
c_comment""".split(
"\n"
)
h_orders = """
o_orderkey
o_custkey
o_orderstatus
o_totalprice
o_orderdate
o_orderpriority
o_clerk
o_shippriority
o_comment""".split(
"\n"
)
h_lineitem = """
l_orderkey
l_partkey
l_suppkey
l_linenumber
l_quantity
l_extendedprice
l_discount
l_tax
l_returnflag
l_linestatus
l_shipdate
l_commitdate
l_receiptdate
l_shipinstruct
l_shipmode
comments""".split(
"\n"
)
for name in [
"nation",
"region",
"part",
"supplier",
"partsupp",
"customer",
"orders",
"lineitem",
]:
print("process table:", name)
df = pl.scan_csv(
f"tables_scale_{scale_fac}/{name}.tbl",
has_header=False,
separator="|",
try_parse_dates=True,
with_column_names=lambda _: eval(f"h_{name}"),
)
df = df.with_columns([pl.col(pl.Date).cast(pl.Datetime)])
df.sink_parquet(f"tables_scale_{scale_fac}/{name}.parquet")