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 # TODO getdictkeyssortedbyvaluekey(dict, key) - eg. dict['author'] = { 'commits' : 512 } - ...key(dict, 'commits')
  19
  20 class DataCollector:
  21         def __init__(self):
  22                 pass
  23
  24         ##
  25         # This should be the main function to extract data from the repository.
  26         def collect(self, dir):
  27                 self.dir = dir
  28
  29         ##
  30         # : get a dictionary of author
  31         def getAuthorInfo(self, author):
  32                 return None
  33
  34         def getActivityByDayOfWeek(self):
  35                 return {}
  36
  37         def getActivityByHourOfDay(self):
  38                 return {}
  39
  40         ##
  41         # Get a list of authors
  42         def getAuthors(self):
  43                 return []
  44
  45         def getFirstCommitDate(self):
  46                 return datetime.datetime.now()
  47
  48         def getLastCommitDate(self):
  49                 return datetime.datetime.now()
  50
  51         def getTags(self):
  52                 return []
  53
  54         def getTotalAuthors(self):
  55                 return -1
  56
  57         def getTotalCommits(self):
  58                 return -1
  59
  60         def getTotalFiles(self):
  61                 return -1
  62
  63         def getTotalLOC(self):
  64                 return -1
  65
  66 class GitDataCollector(DataCollector):
  67         def collect(self, dir):
  68                 DataCollector.collect(self, dir)
  69
  70                 self.total_authors = int(getoutput('git-log |git-shortlog -s |wc -l'))
  71                 self.total_commits = int(getoutput('git-rev-list HEAD |wc -l'))
  72                 self.total_files = int(getoutput('git-ls-files |wc -l'))
  73                 self.total_lines = int(getoutput('git-ls-files |xargs cat |wc -l'))
  74
  75                 self.activity_by_hour_of_day = {} # hour -> commits
  76                 self.activity_by_day_of_week = {} # day -> commits
  77
  78                 self.authors = {} # name -> {commits, first_commit_stamp, last_commit_stamp}
  79
  80                 # author of the month
  81                 self.author_of_month = {} # month -> author -> commits
  82                 self.author_of_year = {} # year -> author -> commits
  83                 self.commits_by_month = {} # month -> commits
  84                 self.commits_by_year = {} # year -> commits
  85                 self.first_commit_stamp = 0
  86                 self.last_commit_stamp = 0
  87
  88                 # tags
  89                 self.tags = {}
  90                 lines = getoutput('git-show-ref --tags').split('\n')
  91                 for line in lines:
  92                         (hash, tag) = line.split(' ')
  93                         tag = tag.replace('refs/tags/', '')
  94                         output = getoutput('git-log "%s" --pretty=format:"%%at %%an" -n 1' % hash)
  95                         if len(output) > 0:
  96                                 parts = output.split(' ')
  97                                 stamp = 0
  98                                 try:
  99                                         stamp = int(parts[0])
 100                                 except ValueError:
 101                                         stamp = 0
 102                                 self.tags[tag] = { 'stamp': stamp, 'hash' : hash, 'date' : datetime.datetime.fromtimestamp(stamp).strftime('%Y-%m-%d') }
 103                         pass
 104
 105                 # TODO also collect statistics for "last 30 days"/"last 12 months"
 106                 lines = getoutput('git-rev-list --pretty=format:"%at %an" HEAD |grep -v ^commit').split('\n')
 107                 for line in lines:
 108                         # linux-2.6 says "<unknown>" for one line O_o
 109                         parts = line.split(' ')
 110                         author = ''
 111                         try:
 112                                 stamp = int(parts[0])
 113                         except ValueError:
 114                                 stamp = 0
 115                         if len(parts) > 1:
 116                                 author = ' '.join(parts[1:])
 117                         date = datetime.datetime.fromtimestamp(float(stamp))
 118
 119                         # First and last commit stamp
 120                         if self.last_commit_stamp == 0:
 121                                 self.last_commit_stamp = stamp
 122                         self.first_commit_stamp = stamp
 123
 124                         # activity
 125                         # hour
 126                         hour = date.hour
 127                         if hour in self.activity_by_hour_of_day:
 128                                 self.activity_by_hour_of_day[hour] += 1
 129                         else:
 130                                 self.activity_by_hour_of_day[hour] = 1
 131
 132                         # day
 133                         day = date.weekday()
 134                         if day in self.activity_by_day_of_week:
 135                                 self.activity_by_day_of_week[day] += 1
 136                         else:
 137                                 self.activity_by_day_of_week[day] = 1
 138
 139                         # author stats
 140                         if author not in self.authors:
 141                                 self.authors[author] = {}
 142                         # TODO commits
 143                         if 'last_commit_stamp' not in self.authors[author]:
 144                                 self.authors[author]['last_commit_stamp'] = stamp
 145                         self.authors[author]['first_commit_stamp'] = stamp
 146                         if 'commits' in self.authors[author]:
 147                                 self.authors[author]['commits'] += 1
 148                         else:
 149                                 self.authors[author]['commits'] = 1
 150
 151                         # author of the month/year
 152                         yymm = datetime.datetime.fromtimestamp(stamp).strftime('%Y-%m')
 153                         if yymm in self.author_of_month:
 154                                 if author in self.author_of_month[yymm]:
 155                                         self.author_of_month[yymm][author] += 1
 156                                 else:
 157                                         self.author_of_month[yymm][author] = 1
 158                         else:
 159                                 self.author_of_month[yymm] = {}
 160                                 self.author_of_month[yymm][author] = 1
 161                         if yymm in self.commits_by_month:
 162                                 self.commits_by_month[yymm] += 1
 163                         else:
 164                                 self.commits_by_month[yymm] = 1
 165
 166                         yy = datetime.datetime.fromtimestamp(stamp).year
 167                         if yy in self.author_of_year:
 168                                 if author in self.author_of_year[yy]:
 169                                         self.author_of_year[yy][author] += 1
 170                                 else:
 171                                         self.author_of_year[yy][author] = 1
 172                         else:
 173                                 self.author_of_year[yy] = {}
 174                                 self.author_of_year[yy][author] = 1
 175                         if yy in self.commits_by_year:
 176                                 self.commits_by_year[yy] += 1
 177                         else:
 178                                 self.commits_by_year[yy] = 1
 179
 180         def getActivityByDayOfWeek(self):
 181                 return self.activity_by_day_of_week
 182
 183         def getActivityByHourOfDay(self):
 184                 return self.activity_by_hour_of_day
 185
 186         def getAuthorInfo(self, author):
 187                 a = self.authors[author]
 188
 189                 commits = a['commits']
 190                 commits_frac = (100 * float(commits)) / self.getTotalCommits()
 191                 date_first = datetime.datetime.fromtimestamp(a['first_commit_stamp']).strftime('%Y-%m-%d')
 192                 date_last = datetime.datetime.fromtimestamp(a['last_commit_stamp']).strftime('%Y-%m-%d')
 193
 194                 res = { 'commits': commits, 'commits_frac': commits_frac, 'date_first': date_first, 'date_last': date_last }
 195                 return res
 196
 197         def getAuthors(self):
 198                 return self.authors.keys()
 199
 200         def getFirstCommitDate(self):
 201                 return datetime.datetime.fromtimestamp(self.first_commit_stamp)
 202
 203         def getLastCommitDate(self):
 204                 return datetime.datetime.fromtimestamp(self.last_commit_stamp)
 205
 206         def getTags(self):
 207                 lines = getoutput('git-show-ref --tags |cut -d/ -f3')
 208                 return lines.split('\n')
 209
 210         def getTagDate(self, tag):
 211                 return self.revToDate('tags/' + tag)
 212
 213         def getTotalAuthors(self):
 214                 return self.total_authors
 215
 216         def getTotalCommits(self):
 217                 return self.total_commits
 218
 219         def getTotalFiles(self):
 220                 return self.total_files
 221
 222         def getTotalLOC(self):
 223                 return self.total_lines
 224
 225         def revToDate(self, rev):
 226                 stamp = int(getoutput('git-log --pretty=format:%%at "%s" -n 1' % rev))
 227                 return datetime.datetime.fromtimestamp(stamp).strftime('%Y-%m-%d')
 228
 229 class ReportCreator:
 230         def __init__(self):
 231                 pass
 232
 233         def create(self, data, path):
 234                 self.data = data
 235                 self.path = path
 236
 237 class HTMLReportCreator(ReportCreator):
 238         def create(self, data, path):
 239                 ReportCreator.create(self, data, path)
 240
 241                 f = open(path + "/index.html", 'w')
 242                 format = '%Y-%m-%d %H:%m:%S'
 243                 self.printHeader(f)
 244
 245                 f.write('<h1>StatGit</h1>')
 246
 247                 self.printNav(f)
 248
 249                 f.write('<dl>');
 250                 f.write('<dt>Generated</dt><dd>%s</dd>' % datetime.datetime.now().strftime(format));
 251                 f.write('<dt>Report Period</dt><dd>%s to %s</dd>' % (data.getFirstCommitDate().strftime(format), data.getLastCommitDate().strftime(format)))
 252                 f.write('<dt>Total Files</dt><dd>%s</dd>' % data.getTotalFiles())
 253                 f.write('<dt>Total Lines of Code</dt><dd>%s</dd>' % data.getTotalLOC())
 254                 f.write('<dt>Total Commits</dt><dd>%s</dd>' % data.getTotalCommits())
 255                 f.write('<dt>Authors</dt><dd>%s</dd>' % data.getTotalAuthors())
 256                 f.write('</dl>');
 257
 258                 f.write('</body>\n</html>');
 259                 f.close()
 260
 261                 ###
 262                 # Activity
 263                 f = open(path + '/activity.html', 'w')
 264                 self.printHeader(f)
 265                 f.write('<h1>Activity</h1>')
 266                 self.printNav(f)
 267
 268                 f.write('<h2>Last 30 days</h2>')
 269
 270                 f.write('<h2>Last 12 months</h2>')
 271
 272                 # Hour of Day
 273                 f.write('\n<h2>Hour of Day</h2>\n\n')
 274                 hour_of_day = data.getActivityByHourOfDay()
 275                 f.write('<table><tr><th>Hour</th>')
 276                 for i in range(1, 25):
 277                         f.write('<th>%d</th>' % i)
 278                 f.write('</tr>\n<tr><th>Commits</th>')
 279                 fp = open(path + '/hour_of_day.dat', 'w')
 280                 for i in range(0, 24):
 281                         if i in hour_of_day:
 282                                 f.write('<td>%d</td>' % hour_of_day[i])
 283                                 fp.write('%d %d\n' % (i, hour_of_day[i]))
 284                         else:
 285                                 f.write('<td>0</td>')
 286                                 fp.write('%d 0\n' % i)
 287                 fp.close()
 288                 f.write('</tr>\n<tr><th>%</th>')
 289                 totalcommits = data.getTotalCommits()
 290                 for i in range(0, 24):
 291                         if i in hour_of_day:
 292                                 f.write('<td>%.2f</td>' % ((100.0 * hour_of_day[i]) / totalcommits))
 293                         else:
 294                                 f.write('<td>0.00</td>')
 295                 f.write('</tr></table>')
 296
 297                 # Day of Week
 298                 # TODO show also by hour of weekday?
 299                 f.write('\n<h2>Day of Week</h2>\n\n')
 300                 day_of_week = data.getActivityByDayOfWeek()
 301                 f.write('<table>')
 302                 f.write('<tr><th>Day</th><th>Total (%)</th></tr>')
 303                 fp = open(path + '/day_of_week.dat', 'w')
 304                 for d in range(0, 7):
 305                         fp.write('%d %d\n' % (d + 1, day_of_week[d]))
 306                         f.write('<tr>')
 307                         f.write('<th>%d</th>' % (d + 1))
 308                         if d in day_of_week:
 309                                 f.write('<td>%d (%.2f%%)</td>' % (day_of_week[d], (100.0 * day_of_week[d]) / totalcommits))
 310                         else:
 311                                 f.write('<td>0</td>')
 312                         f.write('</tr>')
 313                 f.write('</table>')
 314                 fp.close()
 315
 316                 # Commits by year/month
 317                 f.write('<h2>Commits by year/month</h2>')
 318                 f.write('<table><tr><th>Month</th><th>Commits</th></tr>')
 319                 for yymm in reversed(sorted(data.commits_by_month.keys())):
 320                         f.write('<tr><td>%s</td><td>%d</td></tr>' % (yymm, data.commits_by_month[yymm]))
 321                 f.write('</table>')
 322
 323                 # Commits by year
 324                 f.write('<h2>Commits by year</h2>')
 325                 f.write('<table><tr><th>Year</th><th>Commits (% of all)</th></tr>')
 326                 for yy in reversed(sorted(data.commits_by_year.keys())):
 327                         f.write('<tr><td>%s</td><td>%d (%.2f%%)</td></tr>' % (yy, data.commits_by_year[yy], (100.0 * data.commits_by_year[yy]) / data.getTotalCommits()))
 328                 f.write('</table>')
 329
 330                 f.write('</body></html>')
 331                 f.close()
 332
 333                 ###
 334                 # Authors
 335                 f = open(path + '/authors.html', 'w')
 336                 self.printHeader(f)
 337
 338                 f.write('<h1>Authors</h1>')
 339                 self.printNav(f)
 340
 341                 f.write('\n<h2>List of authors</h2>\n\n')
 342
 343                 f.write('<table class="authors">')
 344                 f.write('<tr><th>Author</th><th>Commits (%)</th><th>First commit</th><th>Last commit</th></tr>')
 345                 for author in sorted(data.getAuthors()):
 346                         info = data.getAuthorInfo(author)
 347                         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']))
 348                 f.write('</table>')
 349
 350                 f.write('\n<h2>Author of Month</h2>\n\n')
 351                 f.write('<table>')
 352                 f.write('<tr><th>Month</th><th>Author</th><th>Commits (%)</th></tr>')
 353                 for yymm in reversed(sorted(data.author_of_month.keys())):
 354                         authordict = data.author_of_month[yymm]
 355                         authors = getkeyssortedbyvalues(authordict)
 356                         authors.reverse()
 357                         commits = data.author_of_month[yymm][authors[0]]
 358                         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]))
 359
 360                 f.write('</table>')
 361
 362                 f.write('\n<h2>Author of Year</h2>\n\n')
 363                 f.write('<table><tr><th>Year</th><th>Author</th><th>Commits (%)</th></tr>')
 364                 for yy in reversed(sorted(data.author_of_year.keys())):
 365                         authordict = data.author_of_year[yy]
 366                         authors = getkeyssortedbyvalues(authordict)
 367                         authors.reverse()
 368                         commits = data.author_of_year[yy][authors[0]]
 369                         f.write('<tr><td>%s</td><td>%s</td><td>%d (%.2f%% of %d)</td></tr>' % (yy, authors[0], commits, (100 * commits) / data.commits_by_year[yy], data.commits_by_year[yy]))
 370                 f.write('</table>')
 371
 372                 f.write('</body></html>')
 373                 f.close()
 374
 375                 ###
 376                 # tags.html
 377                 f = open(path + '/tags.html', 'w')
 378                 self.printHeader(f)
 379                 f.write('<h1>Tags</h1>')
 380                 self.printNav(f)
 381
 382                 f.write('<dl>')
 383                 f.write('<dt>Total tags</dt><dd>%d</dd>' % len(data.tags))
 384                 f.write('<dt>Average commits per tag</dt><dd>%.2f</dd>' % (data.getTotalCommits() / len(data.tags)))
 385                 f.write('</dl>')
 386
 387                 f.write('<table>')
 388                 f.write('<tr><th>Name</th><th>Date</th></tr>')
 389                 # sort the tags by date desc
 390                 tags_sorted_by_date_desc = map(lambda el : el[1], reversed(sorted(map(lambda el : (el[1]['date'], el[0]), data.tags.items()))))
 391                 for tag in tags_sorted_by_date_desc:
 392                         f.write('<tr><td>%s</td><td>%s</td></tr>' % (tag, data.tags[tag]['date']))
 393                 f.write('</table>')
 394
 395                 f.write('</body></html>')
 396                 f.close()
 397         pass
 398
 399         def printHeader(self, f):
 400                 f.write("""<html>
 401 <head>
 402         <title>StatGit</title>
 403         <link rel="stylesheet" href="statgit.css" type="text/css" />
 404 </head>
 405 <body>
 406 """)
 407
 408         def printNav(self, f):
 409                 f.write("""
 410 <div class="nav">
 411 <li><a href="index.html">General</a></li>
 412 <li><a href="activity.html">Activity</a></li>
 413 <li><a href="authors.html">Authors</a></li>
 414 <li><a href="files.html">Files</a></li>
 415 <li><a href="lines.html">Lines</a></li>
 416 <li><a href="tags.html">Tags</a></li>
 417 </ul>
 418 </div>
 419 """)
 420
 421
 422 usage = """
 423 Usage: statgit [options] <gitpath> <outputpath>
 424
 425 Options:
 426   -o html
 427 """
 428
 429 if len(sys.argv) <  3:
 430         print usage
 431         sys.exit(0)
 432
 433 gitpath = sys.argv[1]
 434 outputpath = os.path.abspath(sys.argv[2])
 435
 436 print 'Git path: %s' % gitpath
 437 print 'Output path: %s' % outputpath
 438
 439 os.chdir(gitpath)
 440
 441 print 'Collecting data...'
 442 data = GitDataCollector()
 443 data.collect(gitpath)
 444
 445 print 'Generating report...'
 446 report = HTMLReportCreator()
 447 report.create(data, outputpath)
 448
 449