github.com/pachyderm/pachyderm@v1.13.4/examples/spouts/spout101/src/processor/main.py (about)

     1  #!/usr/local/bin/python3
     2  import glob
     3  import os
     4  from datetime import datetime
     5  
     6  file_paths = glob.glob(os.path.join("/pfs/spout", "*.txt"))
     7  
     8  # logs the files in the input repo based on their size.
     9  for file_path in file_paths:
    10      size = os.stat(file_path).st_size
    11      now = datetime.now().time()
    12      if (size == 1024):
    13          with open("/pfs/out/"+"1K"+".txt", 'w') as onek_files:
    14              onek_files.write(str(now) + " " + file_path + "\n")
    15      elif (size == 2048):
    16          with open("/pfs/out/"+"2K"+".txt", 'w') as twok_files:
    17              twok_files.write(str(now) + " " + file_path + "\n")
    18      else:
    19          print("Not matching size" + size + " " + file_path)