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