Dump Subversion log and diff

svnlog.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

"""%prog [options] svn-repository [svn-repository [ ... ]]

"""

import argparse
import codecs
import datetime
import logging
import json
import os
import subprocess
from StringIO import StringIO
from xml.etree.ElementTree import parse


class SvnLog(object):

    def __init__(self, directory_or_url):
        logging.info("Base directory: %s", directory_or_url)
        if subprocess.call(['svn', 'info', directory_or_url]):
            raise ValueError("%s is not SVN directory." % (directory_or_url,))
        self.url = directory_or_url

    def dump(self, outdir, start, days):
        end = start + datetime.timedelta(days=days)
        revisions = "{%s}:{%s}" % (start.strftime("%Y-%m-%d"), end.strftime("%Y-%m-%d"))
        logging.debug("Range: %s - %s, Days: %d", start, end, days)
        if not os.path.exists(outdir):
            os.mkdir(outdir)
            logging.info("%s does not exist, then is created.", outdir)
        entries = []
        for e in self.logentry(revisions):
            r = int(e.get('revision'))
            io = self.parse_logentry(e)
            fname = 'r%d.txt' % (r, )
            out = os.path.join(outdir, fname)
            with open(out, 'wb') as w:
                w.write(io.getvalue())
            logging.info("Wrote SVN DIFF into %s", out)
            entries.append({
                "revision": r,
                "author": e.find('author').text,
                "date": e.find('date').text,
                "message": e.find('msg').text,
                "paths": [p.text for p in e.findall('paths/path')],
                "out": out
            })
        return entries

    def logentry(self, revisions):
        cmd = ['svn', 'log', '-v', '--xml', '--stop-on-copy', '-r', revisions, self.url]
        logging.info(" ".join(cmd))
        p = subprocess.Popen(cmd, stdout=subprocess.PIPE)
        e = parse(p.stdout)
        entries = e.findall('logentry')
        if entries:
            for log in entries:
                yield log

    def parse_logentry(self, e):
        io = StringIO()
        r = int(e.get('revision'))
        revisions = "%d:%d" % (r - 1, r)
        cmd = ['svn', 'diff', '-r', revisions, self.url]
        logging.info(" ".join(cmd))
        p = subprocess.Popen(cmd, stdout=subprocess.PIPE)
        io.write("r%d - Author: %s\n" % (r, e.find('author').text))
        io.write("Date: %s\n" % (e.find('date').text,))
        io.write('\n')
        if e.find('msg').text:
            io.write(e.find('msg').text.encode('utf-8'))
            io.write('\n\n')
        io.write('Changed:\n')
        for path in e.findall('paths/path'):
            io.write("[%s] %s\n" % (path.get('action'), path.text.encode('utf-8')))
        io.write('\n')
        for l in p.stdout:
            io.write(l)
        return io


def main():
    parser = argparse.ArgumentParser(__doc__, version='1.0')
    parser.add_argument('--start', type=str, help='Start of day. (YYYY-MM-DD)')
    parser.add_argument('--days', type=int, default=1, help='Amount of days.')
    parser.add_argument('--output', default=os.getcwd(), help='Output directory.')
    parser.add_argument('dirs', nargs='+', help='Subversion workspace or URL.')
    parser.add_argument('--create-newdir', dest="newdir",
        default=False, action="store_true", help="create new directory")
    parser.add_argument('--verbose', default=False, action="store_true",
        help="verbose mode for logging")
    parser.add_argument('--quiet', default=False, action="store_true",
        help="quiet mode for logging")

    args = parser.parse_args()

    # Set logging verbosity from command line option.
    if args.verbose:
        logging.basicConfig(level=logging.DEBUG)
    elif not args.quiet:
        logging.basicConfig(level=logging.INFO)

    # Set start date.
    if args.start:
        start = datetime.datetime.strptime(args.start, "%Y-%m-%d")
    else:
        start = datetime.date.today() - datetime.timedelta(days=1)

    outdir = args.output
    if args.newdir:
        outdir = os.path.join(outdir, start.strftime('%Y-%m-%d'))

    entries = []
    for d in args.dirs:
        try:
            sl = SvnLog(d)
            e = sl.dump(outdir, start, args.days)
            entries.extend(e)
        except ValueError, e:
            logging.error(e)
    with open(os.path.join(outdir, 'index.json'), 'w') as w:
        json.dump(entries, w)


def test():
    logging.basicConfig(level=logging.DEBUG)
    url = "http://svn.apache.org/repos/asf/lucene/dev/trunk"
    start = datetime.date.today() - datetime.timedelta(days=1)
    sl = SvnLog(url)
    entries = sl.dump('_test', start, 1)
    with open(os.path.join('_test', 'index.json'), 'w') as w:
        json.dump(entries, w)


if __name__ == '__main__':
    main()
    #test()

# vim: set et ts=4 sw=4 cindent fileencoding=utf-8 :