#!/usr/bin/env python
-import sys, os, subprocess, errno, zlib, time
-import hashsplit
-from sha import sha
-
-# FIXME: duplicated in C module. This shouldn't really be here at all...
-BLOBBITS = 14
-BLOBSIZE = 1 << (BLOBBITS-1)
-
-
-def log(s):
- sys.stderr.write('%s\n' % s)
-
-
-class Buf:
- def __init__(self):
- self.data = ''
- self.start = 0
-
- def put(self, s):
- #log('oldsize=%d+%d adding=%d' % (len(self.data), self.start, len(s)))
- if s:
- self.data = buffer(self.data, self.start) + s
- self.start = 0
-
- def peek(self, count):
- return buffer(self.data, self.start, count)
-
- def eat(self, count):
- self.start += count
-
- def get(self, count):
- v = buffer(self.data, self.start, count)
- self.start += count
- return v
-
- def used(self):
- return len(self.data) - self.start
-
-
-def splitbuf(buf):
- #return buf.get(BLOBSIZE)
- b = buf.peek(buf.used())
- ofs = hashsplit.splitbuf(b)
- if ofs:
- buf.eat(ofs)
- return buffer(b, 0, ofs)
- return None
-
-
-ocache = {}
-def save_blob(blob):
- header = 'blob %d\0' % len(blob)
- sum = sha(header)
- sum.update(blob)
- hex = sum.hexdigest()
- dir = '.git/objects/%s' % hex[0:2]
- fn = '%s/%s' % (dir, hex[2:])
- if not ocache.get(hex) and not os.path.exists(fn):
- #log('creating %s' % fn)
- try:
- os.mkdir(dir)
- except OSError, e:
- if e.errno != errno.EEXIST:
- raise
- tfn = '%s.%d' % (fn, os.getpid())
- f = open(tfn, 'w')
- z = zlib.compressobj(1)
- f.write(z.compress(header))
- f.write(z.compress(blob))
- f.write(z.flush())
- f.close()
- os.rename(tfn, fn)
+import sys, time, re
+import hashsplit, git, options
+from helpers import *
+
+optspec = """
+bup split [-tcb] [-n name] [--bench] [filenames...]
+--
+r,remote= remote repository path
+b,blobs output a series of blob ids
+t,tree output a tree id
+c,commit output a commit id
+n,name= name of backup set to update (if any)
+v,verbose increase log output (can be used more than once)
+bench print benchmark timings to stderr
+"""
+o = options.Options('bup split', optspec)
+(opt, flags, extra) = o.parse(sys.argv[1:])
+
+git.check_repo_or_die()
+if not (opt.blobs or opt.tree or opt.commit or opt.name):
+ log("bup split: use one or more of -b, -t, -c, -n\n")
+ o.usage()
+
+hashsplit.split_verbosely = opt.verbose
+if opt.verbose >= 2:
+ git.verbose = opt.verbose - 1
+ opt.bench = 1
+
+start_time = time.time()
+
+def server_connect(remote):
+ rs = remote.split(':', 1)
+ if len(rs) == 1:
+ p = subprocess.Popen(['bup', 'server', '-d', opt.remote],
+ stdin=subprocess.PIPE, stdout=subprocess.PIPE)
else:
- #log('exists %s' % fn)
- pass
- ocache[hex] = 1
- print hex
- return hex
-
-
-def do_main():
- start_time = time.time()
- ofs = 0
- buf = Buf()
- blob = 1
-
- eof = 0
- lv = 0
- while blob or not eof:
- if not eof and (buf.used() < BLOBSIZE*2 or not blob):
- bnew = sys.stdin.read(1024*1024)
- if not len(bnew): eof = 1
- #log('got %d, total %d' % (len(bnew), buf.used()))
- buf.put(bnew)
-
- blob = splitbuf(buf)
- if eof and not blob:
- blob = buf.get(buf.used())
- if not blob and buf.used() >= BLOBSIZE*8:
- blob = buf.get(BLOBSIZE*4) # limit max blob size
- if not blob and not eof:
- continue
-
- if blob:
- ofs += len(blob)
- #log('SPLIT @ %-8d size=%-8d (blobsize=%d)'
- # % (ofs, len(blob), BLOBSIZE))
- save_blob(blob)
-
- nv = (ofs + buf.used())/1000000
- if nv != lv:
- log(nv)
- lv = nv
- secs = time.time() - start_time
- log('\n%.2fkbytes in %.2f secs = %.2f kbytes/sec'
- % (ofs/1024., secs, ofs/1024./secs))
-
-
-assert(BLOBSIZE >= 32)
-do_main()
+ (host, dir) = rs
+ p = subprocess.Popen(['ssh', host, '--', 'bup', 'server'],
+ stdin=subprocess.PIPE, stdout=subprocess.PIPE)
+ dir = re.sub(r'[\r\n]', ' ', dir)
+ p.stdin.write('set-dir\n%s\n' % dir)
+ return p
+
+if opt.remote:
+ p = server_connect(opt.remote)
+ p.stdin.write('receive-objects\n')
+ w = git.PackWriter_Remote(p.stdin)
+else:
+ w = git.PackWriter()
+
+(shalist,tree) = hashsplit.split_to_tree(w, hashsplit.autofiles(extra))
+
+if opt.verbose:
+ log('\n')
+if opt.blobs:
+ for (mode,name,bin) in shalist:
+ print bin.encode('hex')
+if opt.tree:
+ print tree.encode('hex')
+if opt.commit or opt.name:
+ msg = 'bup split\n\nGenerated by command:\n%r' % sys.argv
+ ref = opt.name and ('refs/heads/%s' % opt.name) or None
+ commit = w.new_commit(ref, tree, msg)
+ if opt.commit:
+ print commit.encode('hex')
+
+if opt.remote:
+ w.close()
+ p.stdin.write('quit\n')
+ p.wait()
+
+secs = time.time() - start_time
+size = hashsplit.total_split
+if opt.bench:
+ log('\nbup: %.2fkbytes in %.2f secs = %.2f kbytes/sec\n'
+ % (size/1024., secs, size/1024./secs))