tools/rcs_import.py

   1 #!/usr/bin/python
   2 """Read TWiki RCS files, and import the data into git.
   3
   4 cd path/for/new/git/repo
   5 git init --shared=all
   6 rcs_import.py path/to/twiki/data | git-fast-import
   7
   8 # now the data is in git but not visible yet
   9
  10 git checkout
  11
  12 # examine it with one of these:
  13
  14 gitk
  15 git log
  16 git gui
  17
  18 # to start again at any point:
  19
  20 rm -rf .git
  21 # and if you have the done `git checkout`:
  22 rm -r *
  23
  24 """
  25
  26 FORCE = False
  27
  28 import traceback
  29 import os, sys
  30
  31 from rcs.core import thoeny_filter, Version
  32
  33 def recursive_history_generator(path, rfilter=None):
  34     """The generator that yields available RCS histories form the
  35     given directory.  rfilter should return true if the revision is
  36     acceptable (based on metadata)"""
  37     versions = []
  38     os.chdir(path)
  39     for root, dirs, files in os.walk('.'):
  40         for f in files:
  41             if acceptable_file(f):
  42                 try:
  43                     vs = extract(os.path.join(root, f), rfilter)
  44                     yield vs
  45                 except Exception, e:
  46                     if not FORCE:
  47                         raise
  48                     traceback.print_exc()
  49                     print >> sys.stderr, "Continuing, but ignoring %r..." % f
  50
  51
  52 def sorted_history(path, rfilter=None):
  53     """All file changes in chronological order"""
  54     versions = []
  55     for vs in recursive_history_generator(path, rfilter):
  56         versions.extend(vs)
  57     _versions = [(int(v.date), v) for v in versions]
  58     _versions.sort()
  59     for d, v in _versions:
  60         v.to_git()
  61
  62
  63 def file_by_file_history(path, rfilter=None):
  64     """Complete each files history before beginning on the next.  (revisions in the order they fall out of """
  65     for vs in recursive_history_generator(path, rfilter):
  66         for v in reversed(vs):
  67             v.to_git()
  68
  69 def book_as_branch_history(path, rfilter=None):
  70     """Complete each files history before beginning on the next"""
  71     branch = None
  72     for vs in recursive_history_generator(path, rfilter):
  73         if not vs:
  74             continue
  75         d = os.path.dirname(vs[0].name)
  76         if d != branch:
  77             #new book, new branch
  78             branch = d
  79             print "reset %s" % branch
  80
  81         for v in reversed(vs):
  82             print v
  83             v.to_git(branch, strip_dir=True)
  84
  85
  86 SORT_MODES = {
  87     'sorted' : sorted_history,
  88     'by-file' : file_by_file_history,
  89     'branches' : book_as_branch_history,
  90     }
  91
  92
  93
  94 # TODO: try slow import (via working dir)
  95 # try one branch at a time import
  96
  97
  98
  99 if __name__ == '__main__':
 100     from optparse import OptionParser
 101     parser = OptionParser()
 102     parser.add_option("-t", "--no-thoeny", action="store_true",
 103                       help="ignore TWiki housekeeping commits", default=False)
 104     parser.add_option("-m", "--sort-mode", metavar="MODE",
 105                       help="use sort mode, MODE %s." % (SORT_MODES.keys(),),
 106                       default='sorted')
 107     parser.add_option("-r", "--use-rcs", action="store_true",
 108                       help="Use rcs subprocesses (slow, canonical).", default=False)
 109     parser.add_option("-f", "--force", action="store_true",
 110                       help="Don't give up on parsing errors.", default=False)
 111     parser.add_option("-w", "--working-tree", action="store_true",
 112                       help="Use working tree for commits.", default=False)
 113     options, dirs = parser.parse_args()
 114
 115     if options.use_rcs:
 116         from rcs.subprocess_parse import extract, acceptable_file
 117     else:
 118         from rcs.parse import extract, acceptable_file
 119
 120     FORCE = options.force
 121
 122     if options.no_thoeny:
 123         rfilter = thoeny_filter
 124     else:
 125         rfilter=None
 126
 127     if options.working_tree:
 128         Version.to_git = Version.to_git_slow
 129
 130
 131     sorter = {
 132         'sorted' : sorted_history,
 133         'by-file' : file_by_file_history,
 134         'branches' : book_as_branch_history,
 135         }[options.sort_mode]
 136
 137     for d in dirs:
 138         sorter(d, rfilter)
 139