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