linetags

   1 #!/usr/bin/python
   2 #
   3 # Find out how many lines were introduced in each major release.
   4 #
   5 # linetags <directory>
   6 #
   7 # This code is part of the LWN git data miner.
   8 #
   9 # Copyright 2007-11 Eklektix, Inc.
  10 # Copyright 2007-11 Jonathan Corbet <corbet@lwn.net>
  11 #
  12 # This file may be distributed under the terms of the GNU General
  13 # Public License, version 2.
  14 #
  15 import sys, re, os, pickle, subprocess
  16
  17 CommitLines = { }
  18
  19 commitpat = re.compile(r'^([\da-f][\da-f]+) ')
  20
  21 def GetCommitLines(file):
  22     print(file)
  23     # blame = os.popen('git blame -p ' + file, 'rb')
  24     with subprocess.Popen(['/usr/bin/git', 'blame', '-p', file],
  25                           stdout = subprocess.PIPE) as blame:
  26         for line in blame.stdout.readlines():
  27             try:
  28                 line = line.decode('utf8')
  29             except UnicodeDecodeError:
  30                 print('Bad line in %s:\n\t %s' % (file, line.decode('latin1')))
  31                 continue
  32             m = commitpat.search(line)
  33             #
  34             # All-zero commits mean we got fed a file that git doesn't
  35             # know about.  We could throw an exception and abort processing
  36             # now, or we can just silently ignore it...
  37             #
  38             if not m or m.group(1) == '0000000000000000000000000000000000000000':
  39                 continue
  40             try:
  41                 CommitLines[m.group(1)] += 1
  42             except KeyError:
  43                 CommitLines[m.group(1)] = 1
  44
  45 #
  46 # Try to figure out which tag is the first to contain each commit.
  47 #
  48 def CommitToTag(commit):
  49     try:
  50         return DB[commit]
  51     except KeyError:
  52         print('Missing commit %s' % (commit))
  53         return 'WTF?'
  54
  55 TagLines = { }
  56 def MapCommits():
  57     print('Mapping tags...')
  58     for commit in CommitLines.keys():
  59         tag = CommitToTag(commit)
  60         try:
  61             TagLines[tag] += CommitLines[commit]
  62         except KeyError:
  63             TagLines[tag] = CommitLines[commit]
  64
  65 #
  66 # Here we just plow through all the files.
  67 #
  68 if len(sys.argv) != 2:
  69     sys.stderr.write('Usage: linetags directory\n')
  70     sys.exit(1)
  71 #
  72 # Grab the tags/version database.
  73 #
  74 dbf = open('committags.db', 'rb')
  75 DB = pickle.load(dbf)
  76 dbf.close()
  77
  78 out = open('linetags.out', 'w')
  79 os.chdir(sys.argv[1])
  80 files = os.popen('/usr/bin/find . -type f', 'r')
  81 for file in files.readlines():
  82     if file.find('.git/') < 0:
  83         GetCommitLines(file[:-1])
  84 MapCommits()
  85 tags = sorted(TagLines.keys())
  86 for tag in tags:
  87     out.write('%s %d\n' % (tag, TagLines[tag]))
  88 out.close()