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