shell/bin/git-evtag-compute-py

   1 #!/usr/bin/env python3
   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     assert objid is not None
  45     p = subprocess.Popen(['git', 'cat-file', '--batch'],
  46                          stdin=subprocess.PIPE,
  47                          stdout=subprocess.PIPE,
  48                          close_fds=True,
  49                          cwd=repo)
  50     p.stdin.write(objid.encode('ascii') + b'\n')
  51     p.stdin.close()
  52     (objid,objtype,lenstr) = p.stdout.readline().decode('ascii').split(None, 2)
  53     olen = int(lenstr)
  54     lenstr = lenstr.strip()
  55     buf = "{0} {1}\000".format(objtype, lenstr).encode('ascii')
  56     checksum_bytes(objtype, buf)
  57
  58     stats[objtype] += 1
  59
  60     if objtype == 'commit':
  61         buf = p.stdout.readline()
  62         olen -= checksum_bytes(objtype, buf)
  63         (treestr, treeobjid) = buf.decode('ascii').split(None, 1)
  64         treeobjid = treeobjid.strip()
  65         assert treestr == 'tree'
  66     else:
  67         treeobjid = None
  68
  69     while olen > 0:
  70         b = p.stdout.read(min(8192, olen))
  71         bytes_read = checksum_bytes(objtype, b)
  72         olen -= bytes_read
  73     if olen > 0:
  74         raise ValueError("Failed to read {0} bytes from object".format(olen))
  75     p.wait()
  76     if p.returncode != 0:
  77         raise subprocess.CalledProcessError(p.returncode, 'git cat-file')
  78     return treeobjid
  79
  80 def checksum_tree(repo, path, objid):
  81     checksum_object(repo, objid)
  82     p = subprocess.Popen(['git', 'ls-tree', objid],
  83                          stdin=DEVNULL,
  84                          stdout=subprocess.PIPE,
  85                          close_fds=True,
  86                          cwd=repo)
  87     for line in p.stdout:
  88         (mode, otype, subid, fname) = line.decode('ascii').split(None, 3)
  89         fname = fname.strip()
  90         if otype == 'blob':
  91             checksum_object(repo, subid)
  92         elif otype == 'tree':
  93             checksum_tree(repo, os.path.join(path, fname), subid)
  94         elif otype == 'commit':
  95             checksum_repo(os.path.join(repo, path, fname), subid)
  96         else:
  97             assert False
  98     p.wait()
  99     if p.returncode != 0:
 100         raise subprocess.CalledProcessError(p.returncode, 'git ls-tree')
 101
 102 def checksum_repo(repo, objid):
 103     treeid = checksum_object(repo, objid)
 104     checksum_tree(repo, '.', treeid)
 105
 106 checksum_repo('.', opts.rev)
 107
 108 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']))
 109 print("Git-EVTag-v0-SHA512: {0}".format(csum.hexdigest()))