shell/bin/git-evtag-compute-py

   1 #!/usr/bin/env python
   2 #
   3 # An implementation of Git-EVTag in a mixture of Python and
   4 # git-as-subprocess.  Slower than C, but easier to understand
   5 # and test.
   6 #
   7 # For correct submodule handling, requires the working directory have
   8 # submodule checkouts.
   9
  10 from __future__ import print_function
  11
  12 import os
  13 import sys
  14 import argparse
  15 import subprocess
  16 import hashlib
  17
  18 try:
  19     from subprocess import DEVNULL # pylint: disable=no-name-in-module
  20 except ImportError:
  21     import os
  22     DEVNULL = open(os.devnull, 'wb')
  23
  24 parser = argparse.ArgumentParser(description="Compute Git-EVTag checksum")
  25 parser.add_argument('rev', help='Revision to checksum')
  26 opts = parser.parse_args()
  27
  28 csum = hashlib.sha512()
  29
  30 stats = {'commit': 0,
  31          'blob': 0,
  32          'tree': 0,
  33          'commitbytes': 0,
  34          'blobbytes': 0,
  35          'treebytes': 0}
  36
  37 def checksum_bytes(otype, buf):
  38     blen = len(buf)
  39     csum.update(buf)
  40     stats[otype + 'bytes'] += blen
  41     return blen
  42
  43 def checksum_object(repo, objid):
  44     p = subprocess.Popen(['git', 'cat-file', '--batch'],
  45                          stdin=subprocess.PIPE,
  46                          stdout=subprocess.PIPE,
  47                          close_fds=True,
  48                          cwd=repo)
  49     p.stdin.write(objid + '\n')
  50     p.stdin.close()
  51     (objid,objtype,lenstr) = p.stdout.readline().split(None, 2)
  52     olen = int(lenstr)
  53     lenstr = lenstr.strip()
  54     buf = "{0} {1}\000".format(objtype, lenstr)
  55     checksum_bytes(objtype, buf)
  56
  57     stats[objtype] += 1
  58
  59     if objtype == 'commit':
  60         buf = p.stdout.readline()
  61         olen -= checksum_bytes(objtype, buf)
  62         (treestr, treeobjid) = buf.split(None, 1)
  63         treeobjid = treeobjid.strip()
  64         assert treestr == 'tree'
  65     else:
  66         treeobjid = None
  67
  68     while olen > 0:
  69         b = p.stdout.read(min(8192, olen))
  70         bytes_read = checksum_bytes(objtype, b)
  71         olen -= bytes_read
  72     if olen > 0:
  73         raise ValueError("Failed to read {0} bytes from object".format(olen))
  74     p.wait()
  75     if p.returncode != 0:
  76         raise subprocess.CalledProcessError(p.returncode, 'git cat-file')
  77     return treeobjid
  78
  79 def checksum_tree(repo, objid):
  80     checksum_object(repo, objid)
  81     p = subprocess.Popen(['git', 'ls-tree', objid],
  82                          stdin=DEVNULL,
  83                          stdout=subprocess.PIPE,
  84                          close_fds=True,
  85                          cwd=repo)
  86     for line in p.stdout:
  87         (mode, otype, subid, fname) = line.split(None, 3)
  88         fname = fname.strip()
  89         if otype == 'blob':
  90             checksum_object(repo, subid)
  91         elif otype == 'tree':
  92             checksum_tree(repo, subid)
  93         elif otype == 'commit':
  94             checksum_repo(os.path.join(repo, fname), subid)
  95         else:
  96             assert False
  97     p.wait()
  98     if p.returncode != 0:
  99         raise subprocess.CalledProcessError(p.returncode, 'git ls-tree')
 100
 101 def checksum_repo(repo, objid):
 102     treeid = checksum_object(repo, objid)
 103     checksum_tree(repo, treeid)
 104
 105 checksum_repo('.', opts.rev)
 106
 107 print("# git-evtag comment: submodules={0} commits={1} ({2}) trees={3} ({4}) blobs={5} ({6})".format(stats['commit']-1, stats['commit'], stats['commitbytes'], stats['tree'], stats['treebytes'], stats['blob'], stats['blobbytes']))
 108 print("Git-EVTag-v0-SHA512: {0}".format(csum.hexdigest()))