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