cvs2svn_lib/stats_keeper.py

   1 # (Be in -*- python -*- mode.)
   2 #
   3 # ====================================================================
   4 # Copyright (c) 2000-2008 CollabNet.  All rights reserved.
   5 #
   6 # This software is licensed as described in the file COPYING, which
   7 # you should have received as part of this distribution.  The terms
   8 # are also available at http://subversion.tigris.org/license-1.html.
   9 # If newer versions of this license are posted there, you may use a
  10 # newer version instead, at your option.
  11 #
  12 # This software consists of voluntary contributions made by many
  13 # individuals.  For exact contribution history, see the revision
  14 # history and logs, available at http://cvs2svn.tigris.org/.
  15 # ====================================================================
  16
  17 """This module contains the StatsKeeper class.
  18
  19 A StatsKeeper can pickle itself to a STATISTICS_FILE.  This module
  20 also includes a function to read a StatsKeeper from a STATISTICS_FILE."""
  21
  22
  23 import time
  24 import cPickle
  25 from cStringIO import StringIO
  26
  27 from cvs2svn_lib.cvs_item import CVSRevision
  28 from cvs2svn_lib.cvs_item import CVSBranch
  29 from cvs2svn_lib.cvs_item import CVSTag
  30
  31
  32 class StatsKeeper:
  33   def __init__(self):
  34     self._svn_rev_count = None
  35     self._first_rev_date = 1L<<32
  36     self._last_rev_date = 0
  37     self._pass_timings = { }
  38     self._stats_reflect_exclude = False
  39     self.reset_cvs_rev_info()
  40
  41   def log_duration_for_pass(self, duration, pass_num, pass_name):
  42     self._pass_timings[pass_num] = (pass_name, duration,)
  43
  44   def set_stats_reflect_exclude(self, value):
  45     self._stats_reflect_exclude = value
  46
  47   def reset_cvs_rev_info(self):
  48     self._repos_file_count = 0
  49     self._repos_size = 0
  50     self._cvs_revs_count = 0
  51     self._cvs_branches_count = 0
  52     self._cvs_tags_count = 0
  53
  54     # A set of tag_ids seen:
  55     self._tag_ids = set()
  56
  57     # A set of branch_ids seen:
  58     self._branch_ids = set()
  59
  60   def record_cvs_file(self, cvs_file):
  61     self._repos_file_count += 1
  62     self._repos_size += cvs_file.file_size
  63
  64   def _record_cvs_rev(self, cvs_rev):
  65     self._cvs_revs_count += 1
  66
  67     if cvs_rev.timestamp < self._first_rev_date:
  68       self._first_rev_date = cvs_rev.timestamp
  69
  70     if cvs_rev.timestamp > self._last_rev_date:
  71       self._last_rev_date = cvs_rev.timestamp
  72
  73   def _record_cvs_branch(self, cvs_branch):
  74     self._cvs_branches_count += 1
  75     self._branch_ids.add(cvs_branch.symbol.id)
  76
  77   def _record_cvs_tag(self, cvs_tag):
  78     self._cvs_tags_count += 1
  79     self._tag_ids.add(cvs_tag.symbol.id)
  80
  81   def record_cvs_item(self, cvs_item):
  82     if isinstance(cvs_item, CVSRevision):
  83       self._record_cvs_rev(cvs_item)
  84     elif isinstance(cvs_item, CVSBranch):
  85       self._record_cvs_branch(cvs_item)
  86     elif isinstance(cvs_item, CVSTag):
  87       self._record_cvs_tag(cvs_item)
  88     else:
  89       raise RuntimeError('Unknown CVSItem type')
  90
  91   def set_svn_rev_count(self, count):
  92     self._svn_rev_count = count
  93
  94   def svn_rev_count(self):
  95     return self._svn_rev_count
  96
  97   def __getstate__(self):
  98     state = self.__dict__.copy()
  99     # This can get kinda large, so we don't store it:
 100     return state
 101
 102   def archive(self, filename):
 103     f = open(filename, 'wb')
 104     cPickle.dump(self, f)
 105     f.close()
 106
 107   def __str__(self):
 108     f = StringIO()
 109     f.write('\n')
 110     f.write('cvs2svn Statistics:\n')
 111     f.write('------------------\n')
 112     f.write('Total CVS Files:        %10i\n' % (self._repos_file_count,))
 113     f.write('Total CVS Revisions:    %10i\n' % (self._cvs_revs_count,))
 114     f.write('Total CVS Branches:     %10i\n' % (self._cvs_branches_count,))
 115     f.write('Total CVS Tags:         %10i\n' % (self._cvs_tags_count,))
 116     f.write('Total Unique Tags:      %10i\n' % (len(self._tag_ids),))
 117     f.write('Total Unique Branches:  %10i\n' % (len(self._branch_ids),))
 118     f.write('CVS Repos Size in KB:   %10i\n' % ((self._repos_size / 1024),))
 119
 120     if self._svn_rev_count is not None:
 121       f.write('Total SVN Commits:      %10i\n' % self._svn_rev_count)
 122
 123     f.write(
 124         'First Revision Date:    %s\n' % (time.ctime(self._first_rev_date),)
 125         )
 126     f.write(
 127         'Last Revision Date:     %s\n' % (time.ctime(self._last_rev_date),)
 128         )
 129     f.write('------------------')
 130
 131     if not self._stats_reflect_exclude:
 132       f.write(
 133           '\n'
 134           '(These are unaltered CVS repository stats and do not\n'
 135           ' reflect tags or branches excluded via --exclude)\n'
 136           )
 137
 138     return f.getvalue()
 139
 140   @staticmethod
 141   def _get_timing_format(value):
 142     # Output times with up to 3 decimal places:
 143     decimals = max(0, 4 - len('%d' % int(value)))
 144     length = len(('%%.%df' % decimals) % value)
 145     return '%%%d.%df' % (length, decimals,)
 146
 147   def single_pass_timing(self, pass_num):
 148     (pass_name, duration,) = self._pass_timings[pass_num]
 149     format = self._get_timing_format(duration)
 150     time_string = format % (duration,)
 151     return (
 152         'Time for pass%d (%s): %s seconds.'
 153         % (pass_num, pass_name, time_string,)
 154         )
 155
 156   def timings(self):
 157     passes = self._pass_timings.keys()
 158     passes.sort()
 159     f = StringIO()
 160     f.write('Timings (seconds):\n')
 161     f.write('------------------\n')
 162
 163     total = 0.0
 164     for pass_num in passes:
 165       (pass_name, duration,) = self._pass_timings[pass_num]
 166       total += duration
 167
 168     format = self._get_timing_format(total)
 169
 170     for pass_num in passes:
 171       (pass_name, duration,) = self._pass_timings[pass_num]
 172       f.write(
 173           (format + '   pass%-2d   %s\n') % (duration, pass_num, pass_name,)
 174           )
 175
 176     f.write((format + '   total') % total)
 177     return f.getvalue()
 178
 179
 180 def read_stats_keeper(filename):
 181   """Factory function: Return a _StatsKeeper instance.
 182
 183   Read the instance from FILENAME as written by StatsKeeper.archive()."""
 184
 185   f = open(filename, 'rb')
 186   retval = cPickle.load(f)
 187   f.close()
 188   return retval
 189