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         ##
  30         # Get a list of authors
  31         def getAuthors(self):
  32                 return []
  33
  34         def getFirstCommitDate(self):
  35                 return datetime.datetime.now()
  36
  37         def getLastCommitDate(self):
  38                 return datetime.datetime.now()
  39
  40         def getTags(self):
  41                 return []
  42
  43         def getTotalAuthors(self):
  44                 return -1
  45
  46         def getTotalCommits(self):
  47                 return -1
  48
  49         def getTotalFiles(self):
  50                 return -1
  51
  52         def getTotalLOC(self):
  53                 return -1
  54
  55 class GitDataCollector(DataCollector):
  56         def collect(self, dir):
  57                 DataCollector.collect(self, dir)
  58
  59                 self.total_authors = int(getoutput('git-log |git-shortlog -s |wc -l'))
  60                 self.total_commits = int(getoutput('git-rev-list --all |wc -l'))
  61                 self.total_files = int(getoutput('git-ls-files |wc -l'))
  62                 self.total_lines = int(getoutput('git-ls-files |xargs cat |wc -l'))
  63
  64         def getAuthorInfo(self, author):
  65                 commits = int(getoutput('git-rev-list --all --author="%s" |wc -l' % author))
  66                 commits_frac = (100 * float(commits)) / self.getTotalCommits()
  67                 date_first = '0000-00-00'
  68                 date_last = '0000-00-00'
  69                 rev_last = getoutput('git-rev-list --all --author="%s" -n 1' % author)
  70                 rev_first = getoutput('git-rev-list --all --author="%s" |tail -n 1' % author)
  71                 date_first = self.revToDate(rev_first)
  72                 date_last = self.revToDate(rev_last)
  73
  74                 res = { 'commits': commits, 'commits_frac': commits_frac, 'date_first': date_first, 'date_last': date_last }
  75                 return res
  76
  77         def getAuthors(self):
  78                 lines = getoutput('git-rev-list --all --pretty=format:%an |grep -v ^commit |sort |uniq')
  79                 return lines.split('\n')
  80
  81         def getTags(self):
  82                 lines = getoutput('git-show-ref --tags |cut -d/ -f3')
  83                 return lines.split('\n')
  84
  85         def getTagDate(self, tag):
  86                 return self.revToDate('tags/' + tag)
  87
  88         def getTotalAuthors(self):
  89                 return self.total_authors
  90
  91         def getTotalCommits(self):
  92                 return self.total_commits
  93
  94         def getTotalFiles(self):
  95                 return self.total_files
  96
  97         def getTotalLOC(self):
  98                 return self.total_lines
  99
 100         def revToDate(self, rev):
 101                 stamp = int(getoutput('git-log --pretty=format:%%at "%s" -n 1' % rev))
 102                 return datetime.datetime.fromtimestamp(stamp).strftime('%Y-%m-%d')
 103
 104 class ReportCreator:
 105         def __init__(self):
 106                 pass
 107
 108         def create(self, data, path):
 109                 self.data = data
 110                 self.path = path
 111
 112 class HTMLReportCreator(ReportCreator):
 113         def create(self, data, path):
 114                 ReportCreator.create(self, data, path)
 115
 116                 f = open(path + "/index.html", 'w')
 117                 f.write("""<html>
 118 <head>
 119         <title>StatGit</title>
 120         <link rel="stylesheet" href="statgit.css" type="text/css" />
 121 </head>
 122 <body>
 123 """)
 124                 format = '%Y-%m-%d %H:%m:%S'
 125
 126                 f.write('<h1>StatGit</h1>')
 127
 128                 f.write('<dl>');
 129                 f.write('<dt>Generated</dt><dd>%s</dd>' % datetime.datetime.now().strftime(format));
 130                 f.write('<dt>Report Period</dt><dd>%s to %s</dd>' % (data.getFirstCommitDate().strftime(format), data.getLastCommitDate().strftime(format)))
 131                 f.write('<dt>Total Files</dt><dd>%s</dd>' % data.getTotalFiles())
 132                 f.write('<dt>Total Lines of Code</dt><dd>%s</dd>' % data.getTotalLOC())
 133                 f.write('<dt>Total Commits</dt><dd>%s</dd>' % data.getTotalCommits())
 134                 f.write('<dt>Authors</dt><dd>%s</dd>' % data.getTotalAuthors())
 135                 f.write('</dl>');
 136
 137                 f.write("""<ul>
 138 <li><a href="activity.html">Activity</a></li>
 139 <li><a href="authors.html">Authors</a></li>
 140 <li><a href="files.html">Files</a></li>
 141 </ul>
 142 """)
 143
 144                 f.write('<h2>Authors</h2>')
 145
 146                 f.write('<table class="authors">')
 147                 f.write('<tr><th>Author</th><th>Commits (%)</th><th>First commit</th><th>Last commit</th></tr>')
 148                 for author in data.getAuthors():
 149                         info = data.getAuthorInfo(author)
 150                         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']))
 151                 f.write('</table>')
 152
 153                 f.write('<h2>Tags</h2>')
 154                 f.write('<table>')
 155                 f.write('<tr><th>Name</th><th>Date</th><th>Developers</th></tr>')
 156                 for tag in data.getTags():
 157                         f.write('<tr><td>%s</td><td></td></tr>' % tag)
 158                 f.write('</table>')
 159
 160                 f.write('</body>\n</html>');
 161                 f.close()
 162         pass
 163
 164 usage = """
 165 Usage: statgit [options] <gitpath> <outputpath>
 166
 167 Options:
 168   -o html
 169 """
 170
 171 if len(sys.argv) <  3:
 172         print usage
 173         sys.exit(0)
 174
 175 gitpath = sys.argv[1]
 176 outputpath = sys.argv[2]
 177
 178 print 'Git path: %s' % gitpath
 179 print 'Output path: %s' % outputpath
 180
 181 os.chdir(gitpath)
 182
 183 print 'Collecting data...'
 184 data = GitDataCollector()
 185 data.collect(gitpath)
 186
 187 print 'Generating report...'
 188 report = HTMLReportCreator()
 189 report.create(data, outputpath)
 190
 191