3 bup_python="$(dirname "$0")/bup-python" || exit $?
4 exec "$bup_python" "$0" ${1+"$@"}
8 from __future__ import absolute_import, print_function
9 from collections import defaultdict
10 from itertools import groupby
11 from sys import stderr
12 from time import localtime, strftime, time
15 from bup import git, options
16 from bup.gc import bup_gc
17 from bup.helpers import die_if_errors, log, partition, period_as_secs
18 from bup.repo import LocalRepo
19 from bup.rm import bup_rm
22 def branches(refnames=()):
23 return ((name[11:], sha.encode('hex')) for (name,sha)
24 in git.list_refs(patterns=('refs/heads/' + n for n in refnames),
27 def save_name(branch, utc):
28 return branch + '/' + strftime('%Y-%m-%d-%H%M%S', localtime(utc))
30 def classify_saves(saves, period_start):
31 """For each (utc, id) in saves, yield (True, (utc, id)) if the save
32 should be kept and (False, (utc, id)) if the save should be removed.
33 The ids are binary hashes.
36 def retain_newest_in_region(region):
37 for save in region[0:1]:
39 for save in region[1:]:
42 matches, rest = partition(lambda s: s[0] >= period_start['all'], saves)
46 tm_ranges = ((period_start['dailies'], lambda s: localtime(s[0]).tm_yday),
47 (period_start['monthlies'], lambda s: localtime(s[0]).tm_mon),
48 (period_start['yearlies'], lambda s: localtime(s[0]).tm_year))
50 # Break the decreasing utc sorted saves up into the respective
51 # period ranges (dailies, monthlies, ...). Within each range,
52 # group the saves by the period scale (days, months, ...), and
53 # then yield a "keep" action (True, utc) for the newest save in
54 # each group, and a "drop" action (False, utc) for the rest.
55 for pstart, time_region_id in tm_ranges:
56 matches, rest = partition(lambda s: s[0] >= pstart, rest)
57 for region_id, region_saves in groupby(matches, time_region_id):
58 for action in retain_newest_in_region(list(region_saves)):
61 # Finally, drop any saves older than the specified periods
67 bup prune-older [options...] [BRANCH...]
69 keep-all-for= retain all saves within the PERIOD
70 keep-dailies-for= retain the newest save per day within the PERIOD
71 keep-monthlies-for= retain the newest save per month within the PERIOD
72 keep-yearlies-for= retain the newest save per year within the PERIOD
73 wrt= end all periods at this number of seconds since the epoch
74 pretend don't prune, just report intended actions to standard output
75 gc collect garbage after removals [1]
76 gc-threshold= only rewrite a packfile if it's over this percent garbage [10]
77 #,compress= set compression level to # (0-9, 9 is highest) [1]
78 v,verbose increase log output (can be used more than once)
79 unsafe use the command even though it may be DANGEROUS
82 o = options.Options(optspec)
83 opt, flags, roots = o.parse(sys.argv[1:])
86 o.fatal('refusing to run dangerous, experimental command without --unsafe')
88 now = int(time()) if not opt.wrt else opt.wrt
89 if not isinstance(now, (int, long)):
90 o.fatal('--wrt value ' + str(now) + ' is not an integer')
93 for period, extent in (('all', opt.keep_all_for),
94 ('dailies', opt.keep_dailies_for),
95 ('monthlies', opt.keep_monthlies_for),
96 ('yearlies', opt.keep_yearlies_for)):
98 secs = period_as_secs(extent)
100 o.fatal('%r is not a valid period' % extent)
101 period_start[period] = now - secs
104 o.fatal('at least one keep argument is required')
106 period_start = defaultdict(lambda: float('inf'), period_start)
109 epoch_ymd = strftime('%Y-%m-%d-%H%M%S', localtime(0))
110 for kind in ['all', 'dailies', 'monthlies', 'yearlies']:
111 period_utc = period_start[kind]
112 if period_utc != float('inf'):
113 if not (period_utc > float('-inf')):
114 log('keeping all ' + kind)
117 when = strftime('%Y-%m-%d-%H%M%S', localtime(period_utc))
118 log('keeping ' + kind + ' since ' + when + '\n')
119 except ValueError as ex:
121 log('keeping %s since %d seconds before %s\n'
122 %(kind, abs(period_utc), epoch_ymd))
124 log('keeping %s since %d seconds after %s\n'
125 %(kind, period_utc, epoch_ymd))
127 log('keeping %s since %s\n' % (kind, epoch_ymd))
129 git.check_repo_or_die()
131 # This could be more efficient, but for now just build the whole list
132 # in memory and let bup_rm() do some redundant work.
135 author_secs = f.readline().strip()
136 return int(author_secs)
139 for branch, branch_id in branches(roots):
141 saves = ((utc, oidx.decode('hex')) for (oidx, utc) in
142 git.rev_list(branch_id, format='%at', parse=parse_info))
143 for keep_save, (utc, id) in classify_saves(saves, period_start):
144 assert(keep_save in (False, True))
145 # FIXME: base removals on hashes
147 print('+' if keep_save else '-', save_name(branch, utc))
149 removals.append(save_name(branch, utc))
154 bup_rm(repo, removals, compression=opt.compress, verbosity=opt.verbose)
157 bup_gc(threshold=opt.gc_threshold,
158 compression=opt.compress,
159 verbosity=opt.verbose)