#!/usr/bin/python # Copyright (c) 2007 Heikki Hokkanen # 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(""" StatGit """) format = '%Y-%m-%d %H:%m:%S' f.write('

StatGit

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

Authors

') f.write('') f.write('') for author in data.getAuthors(): info = data.getAuthorInfo(author) f.write('' % (author, info['commits'], info['commits_frac'], info['date_first'], info['date_last'])) f.write('
AuthorCommits (%)First commitLast commit
%s%d (%.2f%%)%s%s
') f.write('

Tags

') f.write('') f.write('') for tag in data.getTags(): f.write('' % tag) f.write('
NameDateDevelopers
%s
') f.write('\n'); f.close() pass usage = """ Usage: statgit [options] 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)