Quick hack to fix my git-annex repo

annex-breakage.py

#! /usr/bin/python
# Quick hack to find and rearrange files from a broken git-annex.
#
# One of my drives with a git annex suffered major filesystem
# corruption, and pretty much all the files got dumped into
# lost+found. Fortunately git-annex symlinks contain all the
# information you need to find your files and see if they're still OK
# -- to wit, the filesize and a SHA.

import os
import stat
import shutil
import argparse
import subprocess
import sys

ALL_READ = stat.S_IRUSR | stat.S_IRGRP | stat.S_IROTH
ALL_EXC  = stat.S_IXUSR | stat.S_IXGRP | stat.S_IXOTH

class Main():
    def __init__(self, argv=None):
        self.parser = argparse.ArgumentParser(
            description='find the files pointed to by broken git-annex links')
        self.parser.add_argument('linksdir', type=str, help='directory where all the broken links are')
        self.parser.add_argument('filesdir', type=str, help='directory where all the lost+found files are')
        self.options = self.parser.parse_args(argv)

    def find_broken_links(self):
        find_output = subprocess.check_output(['find', '-L', self.options.linksdir,
                                               '-type', 'l', '-print0'])
        return find_output.strip('\0').split('\0')

    def parse_annex_symlink(self, link):
        target = os.readlink(link)
        basename = os.path.basename(target)
        # Not completely sure about the format of git annex
        # links. This might break.
        backend_size, hash = basename.split('--')
        info = {'hash': hash}
        backend, size = backend_size.split('-')
        info['backend'] = backend
        info['size'] = size.strip('s')
        return info

    def matchup_source(self, link):
        info = self.parse_annex_symlink(link)
        possibles = subprocess.check_output(['find', self.options.filesdir, '-type', 'f',
                                             '-size', info['size'] + 'c', '-print0'])
        possibles = possibles.strip('\0').split('\0')
        if len(possibles) != 1:
            print "WARNING: two matches for {0} with size {1}: {2}".format(
                link, info['size'], possibles)
        # FIXME: try all candidates
        found = None
        for possible in possibles:
            hash = subprocess.check_output(['sha256sum', possible])
            hash, _ = hash.split()
            if hash == info['hash']:
                found = possible
            else:
                print "WARNING: hash doesn't match up for {0} ({1}) and {2} ({3})".format(
                    link, info['hash'], possible, hash)
        if not found:
            print "No match found for {0}".format(link)
        else:
            dest = os.path.abspath(os.path.join(os.path.dirname(link), os.readlink(link)))
            if not os.path.exists(os.path.dirname(dest)):
                os.makedirs(os.path.dirname(dest))
            # Unlock both source and dest, so that moving it doesn't
            # cause problems
            os.chmod(os.path.dirname(dest), 0777)
            os.chmod(os.path.dirname(found), 0777)
            subprocess.check_call(['mv', found, dest])
            os.chmod(dest, ALL_READ)
            os.chmod(os.path.dirname(dest), ALL_READ | ALL_EXC)

    def run(self):
        links = self.find_broken_links()
        for link in links:
            self.matchup_source(link)

if __name__ == '__main__':
    m = Main()
    m.run()