doc/user/cwl/federated/extract.py

   1 import csv
   2 import sys
   3
   4 select_column = sys.argv[1]
   5 select_values = sys.argv[2]
   6 dataset = sys.argv[3]
   7 cluster = sys.argv[4]
   8
   9 sv = open(select_values, "rt")
  10 selectvals = [s.strip() for s in sv]
  11
  12 print("selectvals", selectvals)
  13
  14 ds = csv.reader(open(dataset, "rt"))
  15 header = next(ds)
  16 print("header is", header)
  17 columnindex = None
  18 for i,v in enumerate(header):
  19     if v == select_column:
  20         columnindex = i
  21 if columnindex is None:
  22     raise Exception("Column %s not found" % select_column)
  23
  24 print("column index", columnindex)
  25
  26 ex = csv.writer(open("extracted.csv", "wt"))
  27 ex.writerow(["cluster"]+list(header))
  28
  29 for row in ds:
  30     if row[columnindex] in selectvals:
  31         ex.writerow([cluster]+list(row))