zcatfollow

   1 #!/usr/bin/python
   2 #
   3 # Will uncompress a gzipped file in "streaming" mode,
   4 # like `tail -b +0 -f ` would do.
   5 #
   6 # If you pass -n <int> parameter it will try to act like
   7 # a smart "tail" command: reads until the end of the current
   8 # file, shows you the last <int> lines, then "follows" the file.
   9 #
  10 # If the writer is writing faster than we read, then it may never
  11 # "detect" the "end of the file"...
  12 #
  13 # License: GPLv2
  14 # Author: Martin Langhoff <martin.langhoff@gmail.com>
  15 # Copyright: Remote Learner US - http://www.remote-learner.net/
  16 #
  17 import sys
  18 import gzip
  19 import zlib
  20 import time
  21 import getopt
  22 from collections import deque
  23 import signal
  24
  25 try:
  26     optslist, args = getopt.getopt(sys.argv[1:], "tn:")
  27     opts={}
  28     for k,v in optslist:
  29         opts[k] = v
  30
  31 except getopt.GetoptError as err:
  32     # print help information and exit:
  33     print(err) # will print something like "option -a not recognized"
  34     #usage()
  35     sys.exit(2)
  36
  37 # lookbehind buffer, defined as a list
  38 lbbuf=deque([])
  39 buffering=False
  40 bufferfull=False
  41
  42 if '-n' in opts:
  43         buffering=True
  44         lbbufsize=int(opts['-n'])
  45
  46 # tidy up - no errors on ctrl-C, etc
  47 def sig_exit(signum, frame):
  48         # this will prompt fh cleanup, etc
  49         sys.exit(0)
  50 signal.signal(signal.SIGHUP,  sig_exit)
  51 signal.signal(signal.SIGINT,  sig_exit)
  52 signal.signal(signal.SIGTERM, sig_exit)
  53
  54 f = gzip.open(args[0], 'r')
  55
  56 # while / readline() sidesteps magic input buffering
  57 while 1:
  58         try:
  59                 line = f.readline()
  60         except IOError:
  61                 ## we ignore  IOError("CRC check failed")
  62                 ## because we are reading an "unfinished"
  63                 ## gzip stream...
  64                 # line won't get unset if this triggers
  65                 line = None
  66
  67         if not line:
  68                 try:
  69                         if f.closed:
  70                                 break
  71                 except AttributeError:
  72                         # old gzip does not have "closed"
  73                         pass
  74                 if buffering:
  75                         # got to the current tail of it
  76                         # dump our lookbehind buffer, stop buffering
  77                         for buf_line in lbbuf:
  78                                 sys.stdout.write(buf_line)
  79                         buffering=False
  80                 # internal gzip API use?
  81                 # f.decompress.flush()
  82                 time.sleep(0.1)
  83         else:
  84                 if not buffering:
  85                         sys.stdout.write(line)
  86                 else:
  87                         if bufferfull:
  88                                 lbbuf.popleft()
  89                                 lbbuf.append(line)
  90                         else:
  91                                 lbbuf.append(line)
  92                                 if len(lbbuf) >= lbbufsize:
  93                                         bufferfull=True
  94
  95