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                 # activity
  77                 lines = getoutput('git-rev-list HEAD --pretty=format:%at |grep -v ^commit').split('\n')
  78                 for stamp in lines:
  79                         date = datetime.datetime.fromtimestamp(float(stamp))
  80
  81                         # hour
  82                         hour = date.hour
  83                         if hour in self.activity_by_hour_of_day:
  84                                 self.activity_by_hour_of_day[hour] += 1
  85                         else:
  86                                 self.activity_by_hour_of_day[hour] = 1
  87
  88                         # day
  89                         day = date.weekday()
  90                         if day in self.activity_by_day_of_week:
  91                                 self.activity_by_day_of_week[day] += 1
  92                         else:
  93                                 self.activity_by_day_of_week[day] = 1
  94
  95                 # TODO author of the month
  96                 self.author_of_month = {} # month -> author -> commits
  97                 self.author_of_year = {} # year -> author -> commits
  98                 self.commits_by_month = {} # month -> commits
  99
 100                 lines = getoutput('git-rev-list --pretty=format:"%at %an" HEAD |grep -v ^commit').split('\n')
 101                 for line in lines:
 102                         parts = line.split(' ')
 103                         stamp = int(parts[0])
 104                         author = ' '.join(parts[1:])
 105
 106                         yymm = datetime.datetime.fromtimestamp(stamp).strftime('%Y-%m')
 107                         if yymm in self.author_of_month:
 108                                 if author in self.author_of_month[yymm]:
 109                                         self.author_of_month[yymm][author] += 1
 110                                 else:
 111                                         self.author_of_month[yymm][author] = 1
 112                         else:
 113                                 self.author_of_month[yymm] = {}
 114                                 self.author_of_month[yymm][author] = 1
 115                         if yymm in self.commits_by_month:
 116                                 self.commits_by_month[yymm] += 1
 117                         else:
 118                                 self.commits_by_month[yymm] = 1
 119
 120                         yy = datetime.datetime.fromtimestamp(stamp).year
 121                         if yy in self.author_of_year:
 122                                 if author in self.author_of_year[yy]:
 123                                         self.author_of_year[yy][author] += 1
 124                                 else:
 125                                         self.author_of_year[yy][author] = 1
 126                         else:
 127                                 self.author_of_year[yy] = {}
 128                                 self.author_of_year[yy][author] = 1
 129
 130                 print self.author_of_month
 131
 132         def getActivityByDayOfWeek(self):
 133                 return self.activity_by_day_of_week
 134
 135         def getActivityByHourOfDay(self):
 136                 return self.activity_by_hour_of_day
 137
 138         def getAuthorInfo(self, author):
 139                 commits = int(getoutput('git-rev-list HEAD --author="%s" |wc -l' % author))
 140                 commits_frac = (100 * float(commits)) / self.getTotalCommits()
 141                 date_first = '0000-00-00'
 142                 date_last = '0000-00-00'
 143                 rev_last = getoutput('git-rev-list --all --author="%s" -n 1' % author)
 144                 rev_first = getoutput('git-rev-list --all --author="%s" |tail -n 1' % author)
 145                 date_first = self.revToDate(rev_first)
 146                 date_last = self.revToDate(rev_last)
 147
 148                 res = { 'commits': commits, 'commits_frac': commits_frac, 'date_first': date_first, 'date_last': date_last }
 149                 return res
 150
 151         def getAuthors(self):
 152                 lines = getoutput('git-rev-list --all --pretty=format:%an |grep -v ^commit |sort |uniq')
 153                 return lines.split('\n')
 154
 155         def getTags(self):
 156                 lines = getoutput('git-show-ref --tags |cut -d/ -f3')
 157                 return lines.split('\n')
 158
 159         def getTagDate(self, tag):
 160                 return self.revToDate('tags/' + tag)
 161
 162         def getTotalAuthors(self):
 163                 return self.total_authors
 164
 165         def getTotalCommits(self):
 166                 return self.total_commits
 167
 168         def getTotalFiles(self):
 169                 return self.total_files
 170
 171         def getTotalLOC(self):
 172                 return self.total_lines
 173
 174         def revToDate(self, rev):
 175                 stamp = int(getoutput('git-log --pretty=format:%%at "%s" -n 1' % rev))
 176                 return datetime.datetime.fromtimestamp(stamp).strftime('%Y-%m-%d')
 177
 178 class ReportCreator:
 179         def __init__(self):
 180                 pass
 181
 182         def create(self, data, path):
 183                 self.data = data
 184                 self.path = path
 185
 186 class HTMLReportCreator(ReportCreator):
 187         def create(self, data, path):
 188                 ReportCreator.create(self, data, path)
 189
 190                 f = open(path + "/index.html", 'w')
 191                 format = '%Y-%m-%d %H:%m:%S'
 192                 self.printHeader(f)
 193
 194                 f.write('<h1>StatGit</h1>')
 195
 196                 f.write('<dl>');
 197                 f.write('<dt>Generated</dt><dd>%s</dd>' % datetime.datetime.now().strftime(format));
 198                 f.write('<dt>Report Period</dt><dd>%s to %s</dd>' % (data.getFirstCommitDate().strftime(format), data.getLastCommitDate().strftime(format)))
 199                 f.write('<dt>Total Files</dt><dd>%s</dd>' % data.getTotalFiles())
 200                 f.write('<dt>Total Lines of Code</dt><dd>%s</dd>' % data.getTotalLOC())
 201                 f.write('<dt>Total Commits</dt><dd>%s</dd>' % data.getTotalCommits())
 202                 f.write('<dt>Authors</dt><dd>%s</dd>' % data.getTotalAuthors())
 203                 f.write('</dl>');
 204
 205                 f.write("""<ul>
 206 <li><a href="activity.html">Activity</a></li>
 207 <li><a href="authors.html">Authors</a></li>
 208 <li><a href="files.html">Files</a></li>
 209 <li><a href="lines.html">Lines</a></li>
 210 </ul>
 211 """)
 212
 213                 f.write('<h2>Tags</h2>')
 214                 f.write('<table>')
 215                 f.write('<tr><th>Name</th><th>Date</th><th>Developers</th></tr>')
 216                 for tag in data.getTags():
 217                         f.write('<tr><td>%s</td><td></td></tr>' % tag)
 218                 f.write('</table>')
 219
 220                 f.write('</body>\n</html>');
 221                 f.close()
 222
 223                 # activity.html
 224                 f = open(path + '/activity.html', 'w')
 225                 self.printHeader(f)
 226                 f.write('<h1>Activity</h1>')
 227
 228                 f.write('<h2>Last 30 days</h2>')
 229
 230                 f.write('<h2>Last 12 months</h2>')
 231
 232                 f.write('\n<h2>Hour of Day</h2>\n\n')
 233                 hour_of_day = data.getActivityByHourOfDay()
 234                 f.write('<table><tr><th>Hour</th>')
 235                 for i in range(1, 25):
 236                         f.write('<th>%d</th>' % i)
 237                 f.write('</tr>\n<tr><th>Commits</th>')
 238                 for i in range(0, 24):
 239                         if i in hour_of_day:
 240                                 f.write('<td>%d</td>' % hour_of_day[i])
 241                         else:
 242                                 f.write('<td>0</td>')
 243                 f.write('</tr>\n<tr><th>%</th>')
 244                 totalcommits = data.getTotalCommits()
 245                 for i in range(0, 24):
 246                         if i in hour_of_day:
 247                                 f.write('<td>%.2f</td>' % ((100.0 * hour_of_day[i]) / totalcommits))
 248                         else:
 249                                 f.write('<td>0.00</td>')
 250                 f.write('</tr></table>')
 251
 252                 ### Day of Week
 253                 # TODO show also by hour of weekday?
 254                 f.write('\n<h2>Day of Week</h2>\n\n')
 255                 day_of_week = data.getActivityByDayOfWeek()
 256                 f.write('<table>')
 257                 f.write('<tr><th>Day</th><th>Total (%)</th></tr>')
 258                 for d in range(0, 7):
 259                         f.write('<tr>')
 260                         f.write('<th>%d</th>' % (d + 1))
 261                         if d in day_of_week:
 262                                 f.write('<td>%d (%.2f%%)</td>' % (day_of_week[d], (100.0 * day_of_week[d]) / totalcommits))
 263                         else:
 264                                 f.write('<td>0</td>')
 265                         f.write('</tr>')
 266                 f.write('</table>')
 267
 268                 f.close()
 269
 270                 # authors.html
 271                 f = open(path + '/authors.html', 'w')
 272                 self.printHeader(f)
 273
 274                 f.write('<h1>Authors</h1>')
 275
 276                 f.write('\n<h2>List of authors</h2>\n\n')
 277
 278                 f.write('<table class="authors">')
 279                 f.write('<tr><th>Author</th><th>Commits (%)</th><th>First commit</th><th>Last commit</th></tr>')
 280                 for author in data.getAuthors():
 281                         info = data.getAuthorInfo(author)
 282                         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']))
 283                 f.write('</table>')
 284
 285                 f.write('\n<h2>Author of Month</h2>\n\n')
 286                 f.write('<table>')
 287                 f.write('<tr><th>Month</th><th>Author</th><th>Commits (%)</th></tr>')
 288                 for yymm in reversed(sorted(data.author_of_month.keys())):
 289                         authordict = data.author_of_month[yymm]
 290                         authors = getkeyssortedbyvalues(authordict)
 291                         authors.reverse()
 292                         commits = data.author_of_month[yymm][authors[0]]
 293                         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]))
 294
 295                 f.write('</table>')
 296
 297                 f.write('\n<h2>Author of Year</h2>\n\n')
 298                 f.write('<table><tr><th>Year</th><th>Author</th><th>Commits (%)</th></tr>')
 299                 for yy in reversed(sorted(data.author_of_year.keys())):
 300                         authordict = data.author_of_year[yy]
 301                         authors = getkeyssortedbyvalues(authordict)
 302                         authors.reverse()
 303                         commits = data.author_of_year[yy][authors[0]]
 304                         f.write('<tr><td>%s</td><td>%s</td><td>%d</td></tr>' % (yy, authors[0], commits))
 305                 f.write('</table>')
 306
 307                 f.write('</body></html>')
 308                 f.close()
 309         pass
 310
 311         def printHeader(self, f):
 312                 f.write("""<html>
 313 <head>
 314         <title>StatGit</title>
 315         <link rel="stylesheet" href="statgit.css" type="text/css" />
 316 </head>
 317 <body>
 318 """)
 319
 320
 321 usage = """
 322 Usage: statgit [options] <gitpath> <outputpath>
 323
 324 Options:
 325   -o html
 326 """
 327
 328 if len(sys.argv) <  3:
 329         print usage
 330         sys.exit(0)
 331
 332 gitpath = sys.argv[1]
 333 outputpath = sys.argv[2]
 334
 335 print 'Git path: %s' % gitpath
 336 print 'Output path: %s' % outputpath
 337
 338 os.chdir(gitpath)
 339
 340 print 'Collecting data...'
 341 data = GitDataCollector()
 342 data.collect(gitpath)
 343
 344 print 'Generating report...'
 345 report = HTMLReportCreator()
 346 report.create(data, outputpath)
 347
 348