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                 # TODO
  77                 self.authors = {} # name -> {commits, first_commit_stamp, last_commit_stamp}
  78
  79                 # activity
  80                 lines = getoutput('git-rev-list HEAD --pretty=format:%at |grep -v ^commit').split('\n')
  81                 for stamp in lines:
  82                         date = datetime.datetime.fromtimestamp(float(stamp))
  83
  84                         # hour
  85                         hour = date.hour
  86                         if hour in self.activity_by_hour_of_day:
  87                                 self.activity_by_hour_of_day[hour] += 1
  88                         else:
  89                                 self.activity_by_hour_of_day[hour] = 1
  90
  91                         # day
  92                         day = date.weekday()
  93                         if day in self.activity_by_day_of_week:
  94                                 self.activity_by_day_of_week[day] += 1
  95                         else:
  96                                 self.activity_by_day_of_week[day] = 1
  97
  98                 # TODO author of the month
  99                 self.author_of_month = {} # month -> author -> commits
 100                 self.author_of_year = {} # year -> author -> commits
 101                 self.commits_by_month = {} # month -> commits
 102                 self.first_commit_stamp = 0
 103                 self.last_commit_stamp = 0
 104
 105                 lines = getoutput('git-rev-list --pretty=format:"%at %an" HEAD |grep -v ^commit').split('\n')
 106                 for line in lines:
 107                         parts = line.split(' ')
 108                         stamp = int(parts[0])
 109                         author = ' '.join(parts[1:])
 110
 111                         # First and last commit stamp
 112                         if self.last_commit_stamp == 0:
 113                                 self.last_commit_stamp = stamp
 114                         self.first_commit_stamp = stamp
 115
 116                         # author stats
 117                         if author not in self.authors:
 118                                 self.authors[author] = {}
 119                         # TODO commits
 120                         if 'last_commit_stamp' not in self.authors[author]:
 121                                 self.authors[author]['last_commit_stamp'] = stamp
 122                         self.authors[author]['first_commit_stamp'] = stamp
 123                         if 'commits' in self.authors[author]:
 124                                 self.authors[author]['commits'] += 1
 125                         else:
 126                                 self.authors[author]['commits'] = 1
 127
 128                         # author of the month/year
 129                         yymm = datetime.datetime.fromtimestamp(stamp).strftime('%Y-%m')
 130                         if yymm in self.author_of_month:
 131                                 if author in self.author_of_month[yymm]:
 132                                         self.author_of_month[yymm][author] += 1
 133                                 else:
 134                                         self.author_of_month[yymm][author] = 1
 135                         else:
 136                                 self.author_of_month[yymm] = {}
 137                                 self.author_of_month[yymm][author] = 1
 138                         if yymm in self.commits_by_month:
 139                                 self.commits_by_month[yymm] += 1
 140                         else:
 141                                 self.commits_by_month[yymm] = 1
 142
 143                         yy = datetime.datetime.fromtimestamp(stamp).year
 144                         if yy in self.author_of_year:
 145                                 if author in self.author_of_year[yy]:
 146                                         self.author_of_year[yy][author] += 1
 147                                 else:
 148                                         self.author_of_year[yy][author] = 1
 149                         else:
 150                                 self.author_of_year[yy] = {}
 151                                 self.author_of_year[yy][author] = 1
 152
 153         def getActivityByDayOfWeek(self):
 154                 return self.activity_by_day_of_week
 155
 156         def getActivityByHourOfDay(self):
 157                 return self.activity_by_hour_of_day
 158
 159         def getAuthorInfo(self, author):
 160                 a = self.authors[author]
 161
 162                 commits = a['commits']
 163                 commits_frac = (100 * float(commits)) / self.getTotalCommits()
 164                 date_first = datetime.datetime.fromtimestamp(a['first_commit_stamp']).strftime('%Y-%m-%d')
 165                 date_last = datetime.datetime.fromtimestamp(a['last_commit_stamp']).strftime('%Y-%m-%d')
 166
 167                 res = { 'commits': commits, 'commits_frac': commits_frac, 'date_first': date_first, 'date_last': date_last }
 168                 return res
 169
 170         def getAuthors(self):
 171                 lines = getoutput('git-rev-list --all --pretty=format:%an |grep -v ^commit |sort |uniq')
 172                 return lines.split('\n')
 173
 174         def getFirstCommitDate(self):
 175                 return datetime.datetime.fromtimestamp(self.first_commit_stamp)
 176
 177         def getLastCommitDate(self):
 178                 return datetime.datetime.fromtimestamp(self.last_commit_stamp)
 179
 180         def getTags(self):
 181                 lines = getoutput('git-show-ref --tags |cut -d/ -f3')
 182                 return lines.split('\n')
 183
 184         def getTagDate(self, tag):
 185                 return self.revToDate('tags/' + tag)
 186
 187         def getTotalAuthors(self):
 188                 return self.total_authors
 189
 190         def getTotalCommits(self):
 191                 return self.total_commits
 192
 193         def getTotalFiles(self):
 194                 return self.total_files
 195
 196         def getTotalLOC(self):
 197                 return self.total_lines
 198
 199         def revToDate(self, rev):
 200                 stamp = int(getoutput('git-log --pretty=format:%%at "%s" -n 1' % rev))
 201                 return datetime.datetime.fromtimestamp(stamp).strftime('%Y-%m-%d')
 202
 203 class ReportCreator:
 204         def __init__(self):
 205                 pass
 206
 207         def create(self, data, path):
 208                 self.data = data
 209                 self.path = path
 210
 211 class HTMLReportCreator(ReportCreator):
 212         def create(self, data, path):
 213                 ReportCreator.create(self, data, path)
 214
 215                 f = open(path + "/index.html", 'w')
 216                 format = '%Y-%m-%d %H:%m:%S'
 217                 self.printHeader(f)
 218
 219                 f.write('<h1>StatGit</h1>')
 220
 221                 f.write('<dl>');
 222                 f.write('<dt>Generated</dt><dd>%s</dd>' % datetime.datetime.now().strftime(format));
 223                 f.write('<dt>Report Period</dt><dd>%s to %s</dd>' % (data.getFirstCommitDate().strftime(format), data.getLastCommitDate().strftime(format)))
 224                 f.write('<dt>Total Files</dt><dd>%s</dd>' % data.getTotalFiles())
 225                 f.write('<dt>Total Lines of Code</dt><dd>%s</dd>' % data.getTotalLOC())
 226                 f.write('<dt>Total Commits</dt><dd>%s</dd>' % data.getTotalCommits())
 227                 f.write('<dt>Authors</dt><dd>%s</dd>' % data.getTotalAuthors())
 228                 f.write('</dl>');
 229
 230                 f.write("""<ul>
 231 <li><a href="activity.html">Activity</a></li>
 232 <li><a href="authors.html">Authors</a></li>
 233 <li><a href="files.html">Files</a></li>
 234 <li><a href="lines.html">Lines</a></li>
 235 </ul>
 236 """)
 237
 238                 f.write('<h2>Tags</h2>')
 239                 f.write('<table>')
 240                 f.write('<tr><th>Name</th><th>Date</th><th>Developers</th></tr>')
 241                 for tag in data.getTags():
 242                         f.write('<tr><td>%s</td><td></td></tr>' % tag)
 243                 f.write('</table>')
 244
 245                 f.write('</body>\n</html>');
 246                 f.close()
 247
 248                 # activity.html
 249                 f = open(path + '/activity.html', 'w')
 250                 self.printHeader(f)
 251                 f.write('<h1>Activity</h1>')
 252
 253                 f.write('<h2>Last 30 days</h2>')
 254
 255                 f.write('<h2>Last 12 months</h2>')
 256
 257                 f.write('\n<h2>Hour of Day</h2>\n\n')
 258                 hour_of_day = data.getActivityByHourOfDay()
 259                 f.write('<table><tr><th>Hour</th>')
 260                 for i in range(1, 25):
 261                         f.write('<th>%d</th>' % i)
 262                 f.write('</tr>\n<tr><th>Commits</th>')
 263                 for i in range(0, 24):
 264                         if i in hour_of_day:
 265                                 f.write('<td>%d</td>' % hour_of_day[i])
 266                         else:
 267                                 f.write('<td>0</td>')
 268                 f.write('</tr>\n<tr><th>%</th>')
 269                 totalcommits = data.getTotalCommits()
 270                 for i in range(0, 24):
 271                         if i in hour_of_day:
 272                                 f.write('<td>%.2f</td>' % ((100.0 * hour_of_day[i]) / totalcommits))
 273                         else:
 274                                 f.write('<td>0.00</td>')
 275                 f.write('</tr></table>')
 276
 277                 ### Day of Week
 278                 # TODO show also by hour of weekday?
 279                 f.write('\n<h2>Day of Week</h2>\n\n')
 280                 day_of_week = data.getActivityByDayOfWeek()
 281                 f.write('<table>')
 282                 f.write('<tr><th>Day</th><th>Total (%)</th></tr>')
 283                 for d in range(0, 7):
 284                         f.write('<tr>')
 285                         f.write('<th>%d</th>' % (d + 1))
 286                         if d in day_of_week:
 287                                 f.write('<td>%d (%.2f%%)</td>' % (day_of_week[d], (100.0 * day_of_week[d]) / totalcommits))
 288                         else:
 289                                 f.write('<td>0</td>')
 290                         f.write('</tr>')
 291                 f.write('</table>')
 292
 293                 f.close()
 294
 295                 # authors.html
 296                 f = open(path + '/authors.html', 'w')
 297                 self.printHeader(f)
 298
 299                 f.write('<h1>Authors</h1>')
 300
 301                 f.write('\n<h2>List of authors</h2>\n\n')
 302
 303                 f.write('<table class="authors">')
 304                 f.write('<tr><th>Author</th><th>Commits (%)</th><th>First commit</th><th>Last commit</th></tr>')
 305                 for author in data.getAuthors():
 306                         info = data.getAuthorInfo(author)
 307                         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']))
 308                 f.write('</table>')
 309
 310                 f.write('\n<h2>Author of Month</h2>\n\n')
 311                 f.write('<table>')
 312                 f.write('<tr><th>Month</th><th>Author</th><th>Commits (%)</th></tr>')
 313                 for yymm in reversed(sorted(data.author_of_month.keys())):
 314                         authordict = data.author_of_month[yymm]
 315                         authors = getkeyssortedbyvalues(authordict)
 316                         authors.reverse()
 317                         commits = data.author_of_month[yymm][authors[0]]
 318                         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]))
 319
 320                 f.write('</table>')
 321
 322                 f.write('\n<h2>Author of Year</h2>\n\n')
 323                 f.write('<table><tr><th>Year</th><th>Author</th><th>Commits (%)</th></tr>')
 324                 for yy in reversed(sorted(data.author_of_year.keys())):
 325                         authordict = data.author_of_year[yy]
 326                         authors = getkeyssortedbyvalues(authordict)
 327                         authors.reverse()
 328                         commits = data.author_of_year[yy][authors[0]]
 329                         f.write('<tr><td>%s</td><td>%s</td><td>%d</td></tr>' % (yy, authors[0], commits))
 330                 f.write('</table>')
 331
 332                 f.write('</body></html>')
 333                 f.close()
 334         pass
 335
 336         def printHeader(self, f):
 337                 f.write("""<html>
 338 <head>
 339         <title>StatGit</title>
 340         <link rel="stylesheet" href="statgit.css" type="text/css" />
 341 </head>
 342 <body>
 343 """)
 344
 345
 346 usage = """
 347 Usage: statgit [options] <gitpath> <outputpath>
 348
 349 Options:
 350   -o html
 351 """
 352
 353 if len(sys.argv) <  3:
 354         print usage
 355         sys.exit(0)
 356
 357 gitpath = sys.argv[1]
 358 outputpath = sys.argv[2]
 359
 360 print 'Git path: %s' % gitpath
 361 print 'Output path: %s' % outputpath
 362
 363 os.chdir(gitpath)
 364
 365 print 'Collecting data...'
 366 data = GitDataCollector()
 367 data.collect(gitpath)
 368
 369 print 'Generating report...'
 370 report = HTMLReportCreator()
 371 report.create(data, outputpath)
 372
 373