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 class DataCollector:
  16         def __init__(self):
  17                 pass
  18
  19         ##
  20         # This should be the main function to extract data from the repository.
  21         def collect(self, dir):
  22                 self.dir = dir
  23
  24         ##
  25         # : get a dictionary of author
  26         def getAuthorInfo(self, author):
  27                 return None
  28
  29         def getActivityByHourOfDay(self):
  30                 return {}
  31
  32         ##
  33         # Get a list of authors
  34         def getAuthors(self):
  35                 return []
  36
  37         def getFirstCommitDate(self):
  38                 return datetime.datetime.now()
  39
  40         def getLastCommitDate(self):
  41                 return datetime.datetime.now()
  42
  43         def getTags(self):
  44                 return []
  45
  46         def getTotalAuthors(self):
  47                 return -1
  48
  49         def getTotalCommits(self):
  50                 return -1
  51
  52         def getTotalFiles(self):
  53                 return -1
  54
  55         def getTotalLOC(self):
  56                 return -1
  57
  58 class GitDataCollector(DataCollector):
  59         def collect(self, dir):
  60                 DataCollector.collect(self, dir)
  61
  62                 self.total_authors = int(getoutput('git-log |git-shortlog -s |wc -l'))
  63                 self.total_commits = int(getoutput('git-rev-list --all |wc -l'))
  64                 self.total_files = int(getoutput('git-ls-files |wc -l'))
  65                 self.total_lines = int(getoutput('git-ls-files |xargs cat |wc -l'))
  66
  67                 self.activity_by_hour_of_day = {} # hour -> commits
  68                 self.activity_by_day_of_week = {} # day -> commits
  69
  70                 # activity
  71                 lines = getoutput('git-rev-list --all --pretty=format:%at |grep -v ^commit').split('\n')
  72                 for stamp in lines:
  73                         date = datetime.datetime.fromtimestamp(float(stamp))
  74
  75                         # hour
  76                         hour = date.hour
  77                         if hour in self.activity_by_hour_of_day:
  78                                 self.activity_by_hour_of_day[hour] += 1
  79                         else:
  80                                 self.activity_by_hour_of_day[hour] = 1
  81
  82                         # day
  83                         day = date.weekday()
  84                         if day in self.activity_by_day_of_week:
  85                                 self.activity_by_day_of_week[day] += 1
  86                         else:
  87                                 self.activity_by_day_of_week[day] = 1
  88
  89
  90         def getActivityByHourOfDay(self):
  91                 return self.activity_by_hour_of_day
  92
  93         def getAuthorInfo(self, author):
  94                 commits = int(getoutput('git-rev-list --all --author="%s" |wc -l' % author))
  95                 commits_frac = (100 * float(commits)) / self.getTotalCommits()
  96                 date_first = '0000-00-00'
  97                 date_last = '0000-00-00'
  98                 rev_last = getoutput('git-rev-list --all --author="%s" -n 1' % author)
  99                 rev_first = getoutput('git-rev-list --all --author="%s" |tail -n 1' % author)
 100                 date_first = self.revToDate(rev_first)
 101                 date_last = self.revToDate(rev_last)
 102
 103                 res = { 'commits': commits, 'commits_frac': commits_frac, 'date_first': date_first, 'date_last': date_last }
 104                 return res
 105
 106         def getAuthors(self):
 107                 lines = getoutput('git-rev-list --all --pretty=format:%an |grep -v ^commit |sort |uniq')
 108                 return lines.split('\n')
 109
 110         def getTags(self):
 111                 lines = getoutput('git-show-ref --tags |cut -d/ -f3')
 112                 return lines.split('\n')
 113
 114         def getTagDate(self, tag):
 115                 return self.revToDate('tags/' + tag)
 116
 117         def getTotalAuthors(self):
 118                 return self.total_authors
 119
 120         def getTotalCommits(self):
 121                 return self.total_commits
 122
 123         def getTotalFiles(self):
 124                 return self.total_files
 125
 126         def getTotalLOC(self):
 127                 return self.total_lines
 128
 129         def revToDate(self, rev):
 130                 stamp = int(getoutput('git-log --pretty=format:%%at "%s" -n 1' % rev))
 131                 return datetime.datetime.fromtimestamp(stamp).strftime('%Y-%m-%d')
 132
 133 class ReportCreator:
 134         def __init__(self):
 135                 pass
 136
 137         def create(self, data, path):
 138                 self.data = data
 139                 self.path = path
 140
 141 class HTMLReportCreator(ReportCreator):
 142         def create(self, data, path):
 143                 ReportCreator.create(self, data, path)
 144
 145                 f = open(path + "/index.html", 'w')
 146                 format = '%Y-%m-%d %H:%m:%S'
 147                 self.printHeader(f)
 148
 149                 f.write('<h1>StatGit</h1>')
 150
 151                 f.write('<dl>');
 152                 f.write('<dt>Generated</dt><dd>%s</dd>' % datetime.datetime.now().strftime(format));
 153                 f.write('<dt>Report Period</dt><dd>%s to %s</dd>' % (data.getFirstCommitDate().strftime(format), data.getLastCommitDate().strftime(format)))
 154                 f.write('<dt>Total Files</dt><dd>%s</dd>' % data.getTotalFiles())
 155                 f.write('<dt>Total Lines of Code</dt><dd>%s</dd>' % data.getTotalLOC())
 156                 f.write('<dt>Total Commits</dt><dd>%s</dd>' % data.getTotalCommits())
 157                 f.write('<dt>Authors</dt><dd>%s</dd>' % data.getTotalAuthors())
 158                 f.write('</dl>');
 159
 160                 f.write("""<ul>
 161 <li><a href="activity.html">Activity</a></li>
 162 <li><a href="authors.html">Authors</a></li>
 163 <li><a href="files.html">Files</a></li>
 164 <li><a href="lines.html">Lines</a></li>
 165 </ul>
 166 """)
 167
 168                 f.write('<h2>Authors</h2>')
 169
 170                 f.write('<table class="authors">')
 171                 f.write('<tr><th>Author</th><th>Commits (%)</th><th>First commit</th><th>Last commit</th></tr>')
 172                 for author in data.getAuthors():
 173                         info = data.getAuthorInfo(author)
 174                         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']))
 175                 f.write('</table>')
 176
 177                 f.write('<h2>Tags</h2>')
 178                 f.write('<table>')
 179                 f.write('<tr><th>Name</th><th>Date</th><th>Developers</th></tr>')
 180                 for tag in data.getTags():
 181                         f.write('<tr><td>%s</td><td></td></tr>' % tag)
 182                 f.write('</table>')
 183
 184                 f.write('</body>\n</html>');
 185                 f.close()
 186
 187                 # activity.html
 188                 f = open(path + '/activity.html', 'w')
 189                 self.printHeader(f)
 190                 f.write('<h1>Activity</h1>')
 191
 192                 f.write('<h2>Last 30 days</h2>')
 193
 194                 f.write('<h2>Last 12 months</h2>')
 195
 196                 f.write('\n<h2>Hour of Day</h2>\n\n')
 197                 hour_of_day = data.getActivityByHourOfDay()
 198                 f.write('<table><tr><th>Hour</th>')
 199                 for i in range(1, 25):
 200                         f.write('<th>%d</th>' % i)
 201                 f.write('</tr>\n<tr><th>Commits</th>')
 202                 for i in range(1, 25):
 203                         if i in hour_of_day:
 204                                 f.write('<td>%d</td>' % hour_of_day[i])
 205                         else:
 206                                 f.write('<td>0</td>')
 207                 f.write('</tr>\n<tr><th>%</th>')
 208                 totalcommits = data.getTotalCommits()
 209                 for i in range(1, 25):
 210                         if i in hour_of_day:
 211                                 f.write('<td>%.2f</td>' % ((100.0 * hour_of_day[i]) / totalcommits))
 212                         else:
 213                                 f.write('<td>0.00</td>')
 214                 f.write('</tr></table>')
 215
 216                 f.write('\n\n<h2>Day of Week</h2>')
 217                 # TODO 7x(24+1)
 218
 219                 f.close()
 220         pass
 221
 222         def printHeader(self, f):
 223                 f.write("""<html>
 224 <head>
 225         <title>StatGit</title>
 226         <link rel="stylesheet" href="statgit.css" type="text/css" />
 227 </head>
 228 <body>
 229 """)
 230
 231
 232 usage = """
 233 Usage: statgit [options] <gitpath> <outputpath>
 234
 235 Options:
 236   -o html
 237 """
 238
 239 if len(sys.argv) <  3:
 240         print usage
 241         sys.exit(0)
 242
 243 gitpath = sys.argv[1]
 244 outputpath = sys.argv[2]
 245
 246 print 'Git path: %s' % gitpath
 247 print 'Output path: %s' % outputpath
 248
 249 os.chdir(gitpath)
 250
 251 print 'Collecting data...'
 252 data = GitDataCollector()
 253 data.collect(gitpath)
 254
 255 print 'Generating report...'
 256 report = HTMLReportCreator()
 257 report.create(data, outputpath)
 258
 259