statgit

   1 #!/usr/bin/python
   2 # Copyright (c) 2007 Heikki Hokkanen <hoxu@users.sf.net>
   3 # GPLv2
   4 import commands
   5 import datetime
   6 import os
   7 import re
   8 import sys
   9
  10 def getoutput(cmd):
  11         print '>> %s' % cmd
  12         output = commands.getoutput(cmd)
  13         return output
  14
  15 def getkeyssortedbyvalues(dict):
  16         return map(lambda el : el[1], sorted(map(lambda el : (el[1], el[0]), dict.items())))
  17
  18 class DataCollector:
  19         def __init__(self):
  20                 pass
  21
  22         ##
  23         # This should be the main function to extract data from the repository.
  24         def collect(self, dir):
  25                 self.dir = dir
  26
  27         ##
  28         # : get a dictionary of author
  29         def getAuthorInfo(self, author):
  30                 return None
  31
  32         def getActivityByDayOfWeek(self):
  33                 return {}
  34
  35         def getActivityByHourOfDay(self):
  36                 return {}
  37
  38         ##
  39         # Get a list of authors
  40         def getAuthors(self):
  41                 return []
  42
  43         def getFirstCommitDate(self):
  44                 return datetime.datetime.now()
  45
  46         def getLastCommitDate(self):
  47                 return datetime.datetime.now()
  48
  49         def getTags(self):
  50                 return []
  51
  52         def getTotalAuthors(self):
  53                 return -1
  54
  55         def getTotalCommits(self):
  56                 return -1
  57
  58         def getTotalFiles(self):
  59                 return -1
  60
  61         def getTotalLOC(self):
  62                 return -1
  63
  64 class GitDataCollector(DataCollector):
  65         def collect(self, dir):
  66                 DataCollector.collect(self, dir)
  67
  68                 self.total_authors = int(getoutput('git-log |git-shortlog -s |wc -l'))
  69                 self.total_commits = int(getoutput('git-rev-list HEAD |wc -l'))
  70                 self.total_files = int(getoutput('git-ls-files |wc -l'))
  71                 self.total_lines = int(getoutput('git-ls-files |xargs cat |wc -l'))
  72
  73                 self.activity_by_hour_of_day = {} # hour -> commits
  74                 self.activity_by_day_of_week = {} # day -> commits
  75
  76                 self.authors = {} # name -> {commits, first_commit_stamp, last_commit_stamp}
  77
  78                 # author of the month
  79                 self.author_of_month = {} # month -> author -> commits
  80                 self.author_of_year = {} # year -> author -> commits
  81                 self.commits_by_month = {} # month -> commits
  82                 self.first_commit_stamp = 0
  83                 self.last_commit_stamp = 0
  84
  85                 # TODO also collect statistics for "last 30 days"/"last 12 months"
  86                 lines = getoutput('git-rev-list --pretty=format:"%at %an" HEAD |grep -v ^commit').split('\n')
  87                 for line in lines:
  88                         # linux-2.6 says "<unknown>" for one line O_o
  89                         parts = line.split(' ')
  90                         author = ''
  91                         try:
  92                                 stamp = int(parts[0])
  93                         except ValueError:
  94                                 stamp = 0
  95                         if len(parts) > 1:
  96                                 author = ' '.join(parts[1:])
  97                         date = datetime.datetime.fromtimestamp(float(stamp))
  98
  99                         # First and last commit stamp
 100                         if self.last_commit_stamp == 0:
 101                                 self.last_commit_stamp = stamp
 102                         self.first_commit_stamp = stamp
 103
 104                         # activity
 105                         # hour
 106                         hour = date.hour
 107                         if hour in self.activity_by_hour_of_day:
 108                                 self.activity_by_hour_of_day[hour] += 1
 109                         else:
 110                                 self.activity_by_hour_of_day[hour] = 1
 111
 112                         # day
 113                         day = date.weekday()
 114                         if day in self.activity_by_day_of_week:
 115                                 self.activity_by_day_of_week[day] += 1
 116                         else:
 117                                 self.activity_by_day_of_week[day] = 1
 118
 119                         # author stats
 120                         if author not in self.authors:
 121                                 self.authors[author] = {}
 122                         # TODO commits
 123                         if 'last_commit_stamp' not in self.authors[author]:
 124                                 self.authors[author]['last_commit_stamp'] = stamp
 125                         self.authors[author]['first_commit_stamp'] = stamp
 126                         if 'commits' in self.authors[author]:
 127                                 self.authors[author]['commits'] += 1
 128                         else:
 129                                 self.authors[author]['commits'] = 1
 130
 131                         # author of the month/year
 132                         yymm = datetime.datetime.fromtimestamp(stamp).strftime('%Y-%m')
 133                         if yymm in self.author_of_month:
 134                                 if author in self.author_of_month[yymm]:
 135                                         self.author_of_month[yymm][author] += 1
 136                                 else:
 137                                         self.author_of_month[yymm][author] = 1
 138                         else:
 139                                 self.author_of_month[yymm] = {}
 140                                 self.author_of_month[yymm][author] = 1
 141                         if yymm in self.commits_by_month:
 142                                 self.commits_by_month[yymm] += 1
 143                         else:
 144                                 self.commits_by_month[yymm] = 1
 145
 146                         yy = datetime.datetime.fromtimestamp(stamp).year
 147                         if yy in self.author_of_year:
 148                                 if author in self.author_of_year[yy]:
 149                                         self.author_of_year[yy][author] += 1
 150                                 else:
 151                                         self.author_of_year[yy][author] = 1
 152                         else:
 153                                 self.author_of_year[yy] = {}
 154                                 self.author_of_year[yy][author] = 1
 155
 156         def getActivityByDayOfWeek(self):
 157                 return self.activity_by_day_of_week
 158
 159         def getActivityByHourOfDay(self):
 160                 return self.activity_by_hour_of_day
 161
 162         def getAuthorInfo(self, author):
 163                 a = self.authors[author]
 164
 165                 commits = a['commits']
 166                 commits_frac = (100 * float(commits)) / self.getTotalCommits()
 167                 date_first = datetime.datetime.fromtimestamp(a['first_commit_stamp']).strftime('%Y-%m-%d')
 168                 date_last = datetime.datetime.fromtimestamp(a['last_commit_stamp']).strftime('%Y-%m-%d')
 169
 170                 res = { 'commits': commits, 'commits_frac': commits_frac, 'date_first': date_first, 'date_last': date_last }
 171                 return res
 172
 173         def getAuthors(self):
 174                 return self.authors.keys()
 175
 176         def getFirstCommitDate(self):
 177                 return datetime.datetime.fromtimestamp(self.first_commit_stamp)
 178
 179         def getLastCommitDate(self):
 180                 return datetime.datetime.fromtimestamp(self.last_commit_stamp)
 181
 182         def getTags(self):
 183                 lines = getoutput('git-show-ref --tags |cut -d/ -f3')
 184                 return lines.split('\n')
 185
 186         def getTagDate(self, tag):
 187                 return self.revToDate('tags/' + tag)
 188
 189         def getTotalAuthors(self):
 190                 return self.total_authors
 191
 192         def getTotalCommits(self):
 193                 return self.total_commits
 194
 195         def getTotalFiles(self):
 196                 return self.total_files
 197
 198         def getTotalLOC(self):
 199                 return self.total_lines
 200
 201         def revToDate(self, rev):
 202                 stamp = int(getoutput('git-log --pretty=format:%%at "%s" -n 1' % rev))
 203                 return datetime.datetime.fromtimestamp(stamp).strftime('%Y-%m-%d')
 204
 205 class ReportCreator:
 206         def __init__(self):
 207                 pass
 208
 209         def create(self, data, path):
 210                 self.data = data
 211                 self.path = path
 212
 213 class HTMLReportCreator(ReportCreator):
 214         def create(self, data, path):
 215                 ReportCreator.create(self, data, path)
 216
 217                 f = open(path + "/index.html", 'w')
 218                 format = '%Y-%m-%d %H:%m:%S'
 219                 self.printHeader(f)
 220
 221                 f.write('<h1>StatGit</h1>')
 222
 223                 self.printNav(f)
 224
 225                 f.write('<dl>');
 226                 f.write('<dt>Generated</dt><dd>%s</dd>' % datetime.datetime.now().strftime(format));
 227                 f.write('<dt>Report Period</dt><dd>%s to %s</dd>' % (data.getFirstCommitDate().strftime(format), data.getLastCommitDate().strftime(format)))
 228                 f.write('<dt>Total Files</dt><dd>%s</dd>' % data.getTotalFiles())
 229                 f.write('<dt>Total Lines of Code</dt><dd>%s</dd>' % data.getTotalLOC())
 230                 f.write('<dt>Total Commits</dt><dd>%s</dd>' % data.getTotalCommits())
 231                 f.write('<dt>Authors</dt><dd>%s</dd>' % data.getTotalAuthors())
 232                 f.write('</dl>');
 233
 234
 235                 #f.write('<h2>Tags</h2>')
 236                 #f.write('<table>')
 237                 #f.write('<tr><th>Name</th><th>Date</th><th>Developers</th></tr>')
 238                 #for tag in data.getTags():
 239                 #       f.write('<tr><td>%s</td><td></td></tr>' % tag)
 240                 #f.write('</table>')
 241
 242                 f.write('</body>\n</html>');
 243                 f.close()
 244
 245                 # activity.html
 246                 f = open(path + '/activity.html', 'w')
 247                 self.printHeader(f)
 248                 f.write('<h1>Activity</h1>')
 249                 self.printNav(f)
 250
 251                 f.write('<h2>Last 30 days</h2>')
 252
 253                 f.write('<h2>Last 12 months</h2>')
 254
 255                 f.write('\n<h2>Hour of Day</h2>\n\n')
 256                 hour_of_day = data.getActivityByHourOfDay()
 257                 f.write('<table><tr><th>Hour</th>')
 258                 for i in range(1, 25):
 259                         f.write('<th>%d</th>' % i)
 260                 f.write('</tr>\n<tr><th>Commits</th>')
 261                 for i in range(0, 24):
 262                         if i in hour_of_day:
 263                                 f.write('<td>%d</td>' % hour_of_day[i])
 264                         else:
 265                                 f.write('<td>0</td>')
 266                 f.write('</tr>\n<tr><th>%</th>')
 267                 totalcommits = data.getTotalCommits()
 268                 for i in range(0, 24):
 269                         if i in hour_of_day:
 270                                 f.write('<td>%.2f</td>' % ((100.0 * hour_of_day[i]) / totalcommits))
 271                         else:
 272                                 f.write('<td>0.00</td>')
 273                 f.write('</tr></table>')
 274
 275                 ### Day of Week
 276                 # TODO show also by hour of weekday?
 277                 f.write('\n<h2>Day of Week</h2>\n\n')
 278                 day_of_week = data.getActivityByDayOfWeek()
 279                 f.write('<table>')
 280                 f.write('<tr><th>Day</th><th>Total (%)</th></tr>')
 281                 for d in range(0, 7):
 282                         f.write('<tr>')
 283                         f.write('<th>%d</th>' % (d + 1))
 284                         if d in day_of_week:
 285                                 f.write('<td>%d (%.2f%%)</td>' % (day_of_week[d], (100.0 * day_of_week[d]) / totalcommits))
 286                         else:
 287                                 f.write('<td>0</td>')
 288                         f.write('</tr>')
 289                 f.write('</table>')
 290
 291                 f.close()
 292
 293                 # authors.html
 294                 f = open(path + '/authors.html', 'w')
 295                 self.printHeader(f)
 296
 297                 f.write('<h1>Authors</h1>')
 298                 self.printNav(f)
 299
 300                 f.write('\n<h2>List of authors</h2>\n\n')
 301
 302                 f.write('<table class="authors">')
 303                 f.write('<tr><th>Author</th><th>Commits (%)</th><th>First commit</th><th>Last commit</th></tr>')
 304                 for author in data.getAuthors():
 305                         info = data.getAuthorInfo(author)
 306                         f.write('<tr><td>%s</td><td>%d (%.2f%%)</td><td>%s</td><td>%s</td></tr>' % (author, info['commits'], info['commits_frac'], info['date_first'], info['date_last']))
 307                 f.write('</table>')
 308
 309                 f.write('\n<h2>Author of Month</h2>\n\n')
 310                 f.write('<table>')
 311                 f.write('<tr><th>Month</th><th>Author</th><th>Commits (%)</th></tr>')
 312                 for yymm in reversed(sorted(data.author_of_month.keys())):
 313                         authordict = data.author_of_month[yymm]
 314                         authors = getkeyssortedbyvalues(authordict)
 315                         authors.reverse()
 316                         commits = data.author_of_month[yymm][authors[0]]
 317                         f.write('<tr><td>%s</td><td>%s</td><td>%d (%.2f%% of %d)</td></tr>' % (yymm, authors[0], commits, (100 * commits) / data.commits_by_month[yymm], data.commits_by_month[yymm]))
 318
 319                 f.write('</table>')
 320
 321                 f.write('\n<h2>Author of Year</h2>\n\n')
 322                 f.write('<table><tr><th>Year</th><th>Author</th><th>Commits (%)</th></tr>')
 323                 for yy in reversed(sorted(data.author_of_year.keys())):
 324                         authordict = data.author_of_year[yy]
 325                         authors = getkeyssortedbyvalues(authordict)
 326                         authors.reverse()
 327                         commits = data.author_of_year[yy][authors[0]]
 328                         f.write('<tr><td>%s</td><td>%s</td><td>%d</td></tr>' % (yy, authors[0], commits))
 329                 f.write('</table>')
 330
 331                 f.write('</body></html>')
 332                 f.close()
 333         pass
 334
 335         def printHeader(self, f):
 336                 f.write("""<html>
 337 <head>
 338         <title>StatGit</title>
 339         <link rel="stylesheet" href="statgit.css" type="text/css" />
 340 </head>
 341 <body>
 342 """)
 343
 344         def printNav(self, f):
 345                 f.write("""
 346 <div class="nav">
 347 <li><a href="index.html">General</a></li>
 348 <li><a href="activity.html">Activity</a></li>
 349 <li><a href="authors.html">Authors</a></li>
 350 <li><a href="files.html">Files</a></li>
 351 <li><a href="lines.html">Lines</a></li>
 352 </ul>
 353 </div>
 354 """)
 355
 356
 357 usage = """
 358 Usage: statgit [options] <gitpath> <outputpath>
 359
 360 Options:
 361   -o html
 362 """
 363
 364 if len(sys.argv) <  3:
 365         print usage
 366         sys.exit(0)
 367
 368 gitpath = sys.argv[1]
 369 outputpath = sys.argv[2]
 370
 371 print 'Git path: %s' % gitpath
 372 print 'Output path: %s' % outputpath
 373
 374 os.chdir(gitpath)
 375
 376 print 'Collecting data...'
 377 data = GitDataCollector()
 378 data.collect(gitpath)
 379
 380 print 'Generating report...'
 381 report = HTMLReportCreator()
 382 report.create(data, outputpath)
 383
 384