show-odd-cache.py

"""
show-odd-cache.py
-----------------

Usage:

python show-odd-cache.py <repository> [<column to sort on>] [<number of entries to show>]

Prints a list of the twenty cache entries with the highest reference count.
This makes certain kinds of cache corruption (by bad drives, RAM etc) easy.

Columns: **refcount**, size (payload size), csize (size in repo)

Prepend a '-' to see the smallest values.

Tested on Borg 1.0.x, Borg 1.1.x; works with remote repos.
"""

import heapq
import sys
from binascii import hexlify

from borg.archiver import Archiver, with_repository, UMASK_DEFAULT
from borg.helpers import location_validator


class XArchiver(Archiver):
    @with_repository(cache=True)
    def show_odd_cache_entries(_, args, repository, cache, manifest, key, count, column):
        fun = heapq.nsmallest if column.startswith('-') else heapq.nlargest
        column = column.strip('-')
        index = {
            'refcount': 0,
            'size': 1,
            'csize': 2,
        }[column]
        contents = fun(count, cache.chunks.iteritems(), key=lambda k_r_s_c: k_r_s_c[1][index])
        fmt = '%-64s %-16s %-16s %-16s'
        print(fmt % ('Object ID', 'refcount', 'size', 'csize'))
        for key, (refcount, size, csize) in contents:
            print(fmt % (hexlify(key).decode(), refcount, size, csize))


class Args:
    location = location_validator(archive=False)(sys.argv[1])
    umask = UMASK_DEFAULT
    remote_path = 'borg'


if __name__ == '__main__':
    try:
        column = sys.argv[2]
    except IndexError:
        column = 'refcount'
    try:
        count = int(sys.argv[3])
    except IndexError:
        count = 20
    XArchiver().show_odd_cache_entries(Args, count=count, column=column)