StarSchemaBenchmark tbl2csv converter

ssb_tbl2csv.py

# Author: Vincenzo De Notaris (vdenotaris@google.com)
# aka a not so brilliant Python programmer!
# Reference: https://github.com/lemire/StarSchemaBenchmark

# Libs
import sys
import os

# Parsing function
def parseAndWrite(name, header):
    csv = open("".join([path, name, ".csv"]), "w+")
    csv.write(header + "\n")
    tbl = open("".join([path, name, ".tbl"]), "r")
    lines = tbl.readlines()
    for line in lines:
        length = len(line)
        line = line[:length - 2] + line[length-1:]
        line = line.replace(",","N")
        line = line.replace("|",",")
        csv.write(line)
    tbl.close()
    csv.close()
    print "- " + name + " [Done]"

# Let's do this dirty job
path = os.path.dirname(os.path.abspath(__file__)) + "/"

ssb = [
    ("customer","c_custkey,c_name,c_address,c_city,c_nation,c_region,c_phone,c_mktsegment"),
    ("date","d_datekey,d_date,d_dayofweek,d_month,d_year,d_yearmonthnum,d_yearmonth,d_daynuminweek,d_daynuminmonth,d_daynuminyear,d_monthnuminyear,d_weeknuminyear,d_sellingseason,d_lastdayinweekfl,d_lastdayinmonthfl,d_holidayfl,d_weekdayfl"),
    ("lineorder","lo_orderkey,lo_linenumber,lo_custkey,lo_partkey,lo_suppkey,lo_orderdate,lo_orderpriority,lo_shippriority,lo_quantity,lo_extendedprice,lo_ordertotalprice,lo_discount,lo_revenue,lo_supplycost,lo_tax,lo_commitdate,lo_shipmode"),
    ("part","p_partkey,p_name,p_mfgr,p_category,p_brand1,p_color,p_type,p_size,p_container"),
    ("supplier","s_suppkey,s_name,s_address,s_city,s_nation,s_region,s_phone")
]

print "----------------------"
print "StarSchemaBenchmark tbl2csv converter\nSee: https://github.com/lemire/StarSchemaBenchmark"
print "----------------------"
print "Converting..."
for t in ssb:
    parseAndWrite(t[0],t[1])
print "That's all folks!"
print "----------------------"

# Have fun!