tools/bdb/erase-all-text-data.py

   1 #!/usr/bin/env python
   2 #
   3 # Erases the text of every file in a BDB repository
   4 #
   5
   6 import sys, os
   7 import skel, svnfs
   8
   9 def main():
  10   if len(sys.argv) == 2:
  11     dbhome = os.path.join(sys.argv[1], 'db')
  12     for i in ('', 'uuids', 'revisions', 'transactions', 'representations',
  13         'strings', 'changes', 'copies', 'nodes'):
  14       if not os.path.exists(os.path.join(dbhome, i)):
  15         sys.stderr.write("%s: '%s' is not a valid bdb svn repository\n" %
  16             (sys.argv[0], sys.argv[1]))
  17         sys.exit(1)
  18   else:
  19     sys.stderr.write("Usage: %s <bdb-svn-repository>\n" % sys.argv[0])
  20     sys.exit(1)
  21
  22   print "WARNING!: This program will destroy all text data in the subversion"
  23   print "repository '%s'" % sys.argv[1]
  24   print "Do not proceed unless this is a *COPY* of your real repository"
  25   print "If this is really what you want to do, " \
  26       "type 'YESERASE' and press Return"
  27   confirmation = raw_input("Confirmation string> ")
  28   if confirmation != "YESERASE":
  29     print "Cancelled - confirmation string not matched"
  30     sys.exit(0)
  31   print "Opening database environment..."
  32   cur = None
  33   ctx = svnfs.Ctx(dbhome)
  34   try:
  35     cur = ctx.nodes_db.cursor()
  36     nodecount = 0
  37     newrep = skel.Rep()
  38     newrep.str = "empty"
  39     empty_fulltext_rep_skel = newrep.unparse()
  40     del newrep
  41     ctx.strings_db['empty'] = ""
  42     rec = cur.first()
  43     while rec:
  44       if rec[0] != "next-key":
  45         if (nodecount % 10000 == 0) and nodecount != 0:
  46           print "Processed %d nodes..." % nodecount
  47         nodecount += 1
  48         node = skel.Node(rec[1])
  49         if node.kind == "file":
  50           rep = skel.Rep(ctx.reps_db[node.datarep])
  51           if rep.kind == "fulltext":
  52             if ctx.strings_db.has_key(rep.str):
  53               del ctx.strings_db[rep.str]
  54             ctx.reps_db[node.datarep] = empty_fulltext_rep_skel
  55           else:
  56             for w in rep.windows:
  57               if ctx.strings_db.has_key(w.str):
  58                 del ctx.strings_db[w.str]
  59             ctx.reps_db[node.datarep] = empty_fulltext_rep_skel
  60       rec = cur.next()
  61     print "Processed %d nodes" % nodecount
  62   finally:
  63     if cur:
  64       cur.close()
  65     ctx.close()
  66   print "Done"
  67
  68 if __name__ == '__main__':
  69   main()