from __future__ import absolute_import, print_function
from binascii import hexlify
-import glob, os, math, resource, struct, sys, tempfile
+import glob, os, math, resource, struct, sys
from bup import options, git, midx, _helpers, xstat
-from bup.compat import argv_bytes, hexstr, range
+from bup.compat import ExitStack, argv_bytes, hexstr, range
from bup.helpers import (Sha1, add_error, atomically_replaced_file, debug1, fdatasync,
- handle_ctrl_c, log, mmap_readwrite, qprogress,
+ log, mmap_readwrite, qprogress,
saved_errors, unlink)
from bup.io import byte_stream, path_msg
except git.GitError as e:
add_error('%s: %s' % (path_msg(name), e))
return
- for count,subname in enumerate(ix.idxnames):
- sub = git.open_idx(os.path.join(os.path.dirname(name), subname))
- for ecount,e in enumerate(sub):
+ with ix:
+ for count,subname in enumerate(ix.idxnames):
+ with git.open_idx(os.path.join(os.path.dirname(name), subname)) \
+ as sub:
+ for ecount,e in enumerate(sub):
+ if not (ecount % 1234):
+ qprogress(' %d/%d: %s %d/%d\r'
+ % (count, len(ix.idxnames),
+ git.shorten_hash(subname).decode('ascii'),
+ ecount, len(sub)))
+ if not sub.exists(e):
+ add_error("%s: %s: %s missing from idx"
+ % (path_msg(nicename),
+ git.shorten_hash(subname).decode('ascii'),
+ hexstr(e)))
+ if not ix.exists(e):
+ add_error("%s: %s: %s missing from midx"
+ % (path_msg(nicename),
+ git.shorten_hash(subname).decode('ascii'),
+ hexstr(e)))
+ prev = None
+ for ecount,e in enumerate(ix):
if not (ecount % 1234):
- qprogress(' %d/%d: %s %d/%d\r'
- % (count, len(ix.idxnames),
- git.shorten_hash(subname).decode('ascii'),
- ecount, len(sub)))
- if not sub.exists(e):
- add_error("%s: %s: %s missing from idx"
- % (path_msg(nicename),
- git.shorten_hash(subname).decode('ascii'),
- hexstr(e)))
- if not ix.exists(e):
- add_error("%s: %s: %s missing from midx"
- % (path_msg(nicename),
- git.shorten_hash(subname).decode('ascii'),
- hexstr(e)))
- prev = None
- for ecount,e in enumerate(ix):
- if not (ecount % 1234):
- qprogress(' Ordering: %d/%d\r' % (ecount, len(ix)))
- if e and prev and not e >= prev:
- add_error('%s: ordering error: %s < %s'
- % (nicename, hexstr(e), hexstr(prev)))
- prev = e
+ qprogress(' Ordering: %d/%d\r' % (ecount, len(ix)))
+ if e and prev and not e >= prev:
+ add_error('%s: ordering error: %s < %s'
+ % (nicename, hexstr(e), hexstr(prev)))
+ prev = e
_first = None
assert(outdir)
sum = hexlify(Sha1(b'\0'.join(infilenames)).digest())
outfilename = b'%s/midx-%s.midx' % (outdir, sum)
-
+
inp = []
total = 0
allfilenames = []
- midxs = []
- try:
+ with ExitStack() as contexts:
for name in infilenames:
ix = git.open_idx(name)
- midxs.append(ix)
+ contexts.enter_context(ix)
inp.append((
ix.map,
len(ix),
or ((auto or force) and len(infilenames) < 2) \
or (force and not total):
debug1('midx: nothing to do.\n')
- return
+ return None
pages = int(total/SHA_PER_PAGE) or 1
bits = int(math.ceil(math.log(pages, 2)))
f.flush()
fdatasync(f.fileno())
- fmap = mmap_readwrite(f, close=False)
- count = merge_into(fmap, bits, total, inp)
- del fmap # Assume this calls msync() now.
+ with mmap_readwrite(f, close=False) as fmap:
+ count = merge_into(fmap, bits, total, inp)
f.seek(0, os.SEEK_END)
f.write(b'\0'.join(allfilenames))
- finally:
- for ix in midxs:
- if isinstance(ix, midx.PackMidx):
- ix.close()
- midxs = None
- inp = None
-
# This is just for testing (if you enable this, don't clear inp above)
- if 0:
- p = midx.PackMidx(outfilename)
- assert(len(p.idxnames) == len(infilenames))
- log(repr(p.idxnames) + '\n')
- assert(len(p) == total)
- for pe, e in p, git.idxmerge(inp, final_progress=False):
- pin = next(pi)
- assert(i == pin)
- assert(p.exists(i))
+ # if 0:
+ # p = midx.PackMidx(outfilename)
+ # assert(len(p.idxnames) == len(infilenames))
+ # log(repr(p.idxnames) + '\n')
+ # assert(len(p) == total)
+ # for pe, e in p, git.idxmerge(inp, final_progress=False):
+ # pin = next(pi)
+ # assert(i == pin)
+ # assert(p.exists(i))
return total, outfilename
midxs = glob.glob(b'%s/*.midx' % path)
contents = {}
for mname in midxs:
- m = git.open_idx(mname)
- contents[mname] = [(b'%s/%s' % (path,i)) for i in m.idxnames]
- sizes[mname] = len(m)
-
+ with git.open_idx(mname) as m:
+ contents[mname] = [(b'%s/%s' % (path,i)) for i in m.idxnames]
+ sizes[mname] = len(m)
+
# sort the biggest+newest midxes first, so that we can eliminate
# smaller (or older) redundant ones that come later in the list
midxs.sort(key=lambda ix: (-sizes[ix], -xstat.stat(ix).st_mtime))
-
+
for mname in midxs:
any = 0
for iname in contents[mname]:
idxs = [k for k in glob.glob(b'%s/*.idx' % path) if not already.get(k)]
for iname in idxs:
- i = git.open_idx(iname)
- sizes[iname] = len(i)
+ with git.open_idx(iname) as i:
+ sizes[iname] = len(i)
all = [(sizes[n],n) for n in (midxs + idxs)]
-
+
# FIXME: what are the optimal values? Does this make sense?
DESIRED_HWM = force and 1 or 5
DESIRED_LWM = force and 1 or 2
existed = dict((name,1) for sz,name in all)
- debug1('midx: %d indexes; want no more than %d.\n'
+ debug1('midx: %d indexes; want no more than %d.\n'
% (len(all), DESIRED_HWM))
if len(all) <= DESIRED_HWM:
debug1('midx: nothing to do.\n')
def main(argv):
o = options.Options(optspec)
opt, flags, extra = o.parse_bytes(argv[1:])
- opt.dir = argv_bytes(opt.dir) if opt.dir else None
opt.output = argv_bytes(opt.output) if opt.output else None
if extra and (opt.auto or opt.force):
opt.max_files = max_files()
assert(opt.max_files >= 5)
+ path = opt.dir and argv_bytes(opt.dir) or git.repo(b'objects/pack')
+
extra = [argv_bytes(x) for x in extra]
if opt.check:
if extra:
midxes = extra
else:
- path = opt.dir or git.repo(b'objects/pack')
debug1('midx: scanning %s\n' % path)
midxes = glob.glob(os.path.join(path, b'*.midx'))
for name in midxes:
else:
if extra:
sys.stdout.flush()
- do_midx(git.repo(b'objects/pack'), opt.output, extra, b'',
+ do_midx(path, opt.output, extra, b'',
byte_stream(sys.stdout), auto=opt.auto, force=opt.force,
print_names=opt.print)
elif opt.auto or opt.force:
sys.stdout.flush()
- path = opt.dir or git.repo(b'objects/pack')
debug1('midx: scanning %s\n' % path_msg(path))
do_midx_dir(path, opt.output, byte_stream(sys.stdout),
auto=opt.auto, force=opt.force,
projects / bup.git / blobdiff