hoxu
/
gitstats
Mirror von https://github.com/hoxu/gitstats


			
				
					
						
						
							123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191
							#!/usr/bin/python
# Copyright (c) 2007 Heikki Hokkanen <hoxu@users.sf.net>
# GPLv2
import commands
import datetime
import os
import re
import sys

def getoutput(cmd):
	print '>> %s' % cmd
	output = commands.getoutput(cmd)
	return output

class DataCollector:
	def __init__(self):
		pass
	
	##
	# This should be the main function to extract data from the repository.
	def collect(self, dir):
		self.dir = dir
	
	##
	# : get a dictionary of author
	def getAuthorInfo(self, author):
		return None
	
	##
	# Get a list of authors
	def getAuthors(self):
		return []
	
	def getFirstCommitDate(self):
		return datetime.datetime.now()
	
	def getLastCommitDate(self):
		return datetime.datetime.now()
	
	def getTags(self):
		return []
	
	def getTotalAuthors(self):
		return -1
	
	def getTotalCommits(self):
		return -1
		
	def getTotalFiles(self):
		return -1
	
	def getTotalLOC(self):
		return -1

class GitDataCollector(DataCollector):
	def collect(self, dir):
		DataCollector.collect(self, dir)

		self.total_authors = int(getoutput('git-log |git-shortlog -s |wc -l'))
		self.total_commits = int(getoutput('git-rev-list --all |wc -l'))
		self.total_files = int(getoutput('git-ls-files |wc -l'))
		self.total_lines = int(getoutput('git-ls-files |xargs cat |wc -l'))
	
	def getAuthorInfo(self, author):
		commits = int(getoutput('git-rev-list --all --author="%s" |wc -l' % author))
		commits_frac = (100 * float(commits)) / self.getTotalCommits()
		date_first = '0000-00-00'
		date_last = '0000-00-00'
		rev_last = getoutput('git-rev-list --all --author="%s" -n 1' % author)
		rev_first = getoutput('git-rev-list --all --author="%s" |tail -n 1' % author)
		date_first = self.revToDate(rev_first)
		date_last = self.revToDate(rev_last)

		res = { 'commits': commits, 'commits_frac': commits_frac, 'date_first': date_first, 'date_last': date_last }
		return res
	
	def getAuthors(self):
		lines = getoutput('git-rev-list --all --pretty=format:%an |grep -v ^commit |sort |uniq')
		return lines.split('\n')
	
	def getTags(self):
		lines = getoutput('git-show-ref --tags |cut -d/ -f3')
		return lines.split('\n')
	
	def getTagDate(self, tag):
		return self.revToDate('tags/' + tag)
	
	def getTotalAuthors(self):
		return self.total_authors
	
	def getTotalCommits(self):
		return self.total_commits

	def getTotalFiles(self):
		return self.total_files
	
	def getTotalLOC(self):
		return self.total_lines
	
	def revToDate(self, rev):
		stamp = int(getoutput('git-log --pretty=format:%%at "%s" -n 1' % rev))
		return datetime.datetime.fromtimestamp(stamp).strftime('%Y-%m-%d')

class ReportCreator:
	def __init__(self):
		pass
	
	def create(self, data, path):
		self.data = data
		self.path = path

class HTMLReportCreator(ReportCreator):
	def create(self, data, path):
		ReportCreator.create(self, data, path)

		f = open(path + "/index.html", 'w')
		f.write("""<html>
<head>
	<title>StatGit</title>
	<link rel="stylesheet" href="statgit.css" type="text/css" />
</head>
<body>
""")
		format = '%Y-%m-%d %H:%m:%S'

		f.write('<h1>StatGit</h1>')

		f.write('<dl>');
		f.write('<dt>Generated</dt><dd>%s</dd>' % datetime.datetime.now().strftime(format));
		f.write('<dt>Report Period</dt><dd>%s to %s</dd>' % (data.getFirstCommitDate().strftime(format), data.getLastCommitDate().strftime(format)))
		f.write('<dt>Total Files</dt><dd>%s</dd>' % data.getTotalFiles())
		f.write('<dt>Total Lines of Code</dt><dd>%s</dd>' % data.getTotalLOC())
		f.write('<dt>Total Commits</dt><dd>%s</dd>' % data.getTotalCommits())
		f.write('<dt>Authors</dt><dd>%s</dd>' % data.getTotalAuthors())
		f.write('</dl>');

		f.write("""<ul>
<li><a href="activity.html">Activity</a></li>
<li><a href="authors.html">Authors</a></li>
<li><a href="files.html">Files</a></li>
</ul>
""")

		f.write('<h2>Authors</h2>')

		f.write('<table class="authors">')
		f.write('<tr><th>Author</th><th>Commits (%)</th><th>First commit</th><th>Last commit</th></tr>')
		for author in data.getAuthors():
			info = data.getAuthorInfo(author)
			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']))
		f.write('</table>')

		f.write('<h2>Tags</h2>')
		f.write('<table>')
		f.write('<tr><th>Name</th><th>Date</th><th>Developers</th></tr>')
		for tag in data.getTags():
			f.write('<tr><td>%s</td><td></td></tr>' % tag)
		f.write('</table>')

		f.write('</body>\n</html>');
		f.close()
	pass

usage = """
Usage: statgit [options] <gitpath> <outputpath>

Options:
  -o html
"""

if len(sys.argv) <  3:
	print usage
	sys.exit(0)

gitpath = sys.argv[1]
outputpath = sys.argv[2]

print 'Git path: %s' % gitpath
print 'Output path: %s' % outputpath

os.chdir(gitpath)

print 'Collecting data...'
data = GitDataCollector()
data.collect(gitpath)

print 'Generating report...'
report = HTMLReportCreator()
report.create(data, outputpath)