bin/bin/mail_dup_hdrs_remove.py

   1 #!/usr/bin/env python3.6
   2
   3 import sys
   4
   5 fn = sys.argv[1]
   6 out = sys.argv[2]
   7 with open(fn, "rb") as fd:
   8     lines = fd.read().split(b"\n")
   9
  10 def dup(lines, what):
  11     idx = []
  12     met = False
  13     for i, line in enumerate(lines):
  14         if line == "":
  15             break
  16         if met:
  17             if line.startswith(b" "):
  18                 idx.append(i)
  19                 continue
  20             else:
  21                 met = False
  22         if line.startswith(what):
  23             if len(idx) == 0:
  24                 idx.append(i)
  25                 met = True
  26             else:
  27                 return [l for n, l in enumerate(lines) if n not in idx]
  28     return lines
  29
  30 lines = dup(lines, b"To")
  31 lines = dup(lines, b"From")
  32 lines = dup(lines, b"Subject")
  33 with open(out, "wb") as fd:
  34     fd.write(b"\n".join(lines))