interact with the Git data structures.
"""
+from __future__ import absolute_import
import errno, os, sys, zlib, time, subprocess, struct, stat, re, tempfile, glob
from collections import namedtuple
from itertools import islice
from numbers import Integral
from bup import _helpers, compat, hashsplit, path, midx, bloom, xstat
+from bup.compat import range
from bup.helpers import (Sha1, add_error, chunkyreader, debug1, debug2,
fdatasync,
hostname, localtime, log, merge_iter,
mmap_read, mmap_readwrite,
parse_num,
- progress, qprogress, stat_if_exists,
+ progress, qprogress, shstr, stat_if_exists,
unlink, username, userfullname,
utc_offset_str)
def _git_wait(cmd, p):
rv = p.wait()
if rv != 0:
- raise GitError('%s returned %d' % (cmd, rv))
+ raise GitError('%s returned %d' % (shstr(cmd), rv))
def _git_capture(argv):
p = subprocess.Popen(argv, stdout=subprocess.PIPE, preexec_fn = _gitenv())
_start_end_char, _content_char, _start_end_char)
_tz_rx = r'[-+]\d\d[0-5]\d'
_parent_rx = r'(?:parent [abcdefABCDEF0123456789]{40}\n)'
+# Assumes every following line starting with a space is part of the
+# mergetag. Is there a formal commit blob spec?
+_mergetag_rx = r'(?:\nmergetag object [abcdefABCDEF0123456789]{40}(?:\n [^\0\n]*)*)'
_commit_rx = re.compile(r'''tree (?P<tree>[abcdefABCDEF0123456789]{40})
(?P<parents>%s*)author (?P<author_name>%s) <(?P<author_mail>%s)> (?P<asec>\d+) (?P<atz>%s)
-committer (?P<committer_name>%s) <(?P<committer_mail>%s)> (?P<csec>\d+) (?P<ctz>%s)
+committer (?P<committer_name>%s) <(?P<committer_mail>%s)> (?P<csec>\d+) (?P<ctz>%s)(?P<mergetag>%s?)
(?P<message>(?:.|\n)*)''' % (_parent_rx,
_safe_str_rx, _safe_str_rx, _tz_rx,
- _safe_str_rx, _safe_str_rx, _tz_rx))
+ _safe_str_rx, _safe_str_rx, _tz_rx,
+ _mergetag_rx))
_parent_hash_rx = re.compile(r'\s*parent ([abcdefABCDEF0123456789]{40})\s*')
-
-# Note that the author_sec and committer_sec values are (UTC) epoch seconds.
+# Note that the author_sec and committer_sec values are (UTC) epoch
+# seconds, and for now the mergetag is not included.
CommitInfo = namedtuple('CommitInfo', ['tree', 'parents',
'author_name', 'author_mail',
'author_sec', 'author_offset',
message=matches['message'])
-def get_commit_items(id, cp):
- commit_it = cp.get(id)
- _, typ, _ = next(commit_it)
- assert(typ == 'commit')
- commit_content = ''.join(commit_it)
- return parse_commit(commit_content)
+def get_cat_data(cat_iterator, expected_type):
+ _, kind, _ = next(cat_iterator)
+ if kind != expected_type:
+ raise Exception('expected %r, saw %r' % (expected_type, kind))
+ return ''.join(cat_iterator)
+def get_commit_items(id, cp):
+ return parse_commit(get_cat_data(cp.get(id), 'commit'))
def _local_git_date_str(epoch_sec):
return '%d %s' % (epoch_sec, utc_offset_str(epoch_sec))
def repo(sub = '', repo_dir=None):
"""Get the path to the git repository or one of its subdirectories."""
- global repodir
repo_dir = repo_dir or repodir
if not repo_dir:
raise GitError('You should call check_repo_or_die()')
# If there's a .git subdirectory, then the actual repo is in there.
gd = os.path.join(repo_dir, '.git')
if os.path.exists(gd):
- repodir = gd
+ repo_dir = gd
return os.path.join(repo_dir, sub)
return str(self.shatable[idx*24+4 : idx*24+24])
def __iter__(self):
- for i in xrange(self.fanout[255]):
+ for i in range(self.fanout[255]):
yield buffer(self.map, 256*4 + 24*i + 4, 20)
return str(self.shatable[idx*20:(idx+1)*20])
def __iter__(self):
- for i in xrange(self.fanout[255]):
+ for i in range(self.fanout[255]):
yield buffer(self.map, 8 + 256*4 + 20*i, 20)
if self.bloom is None and os.path.exists(bfull):
self.bloom = bloom.ShaBloom(bfull)
self.packs = list(set(d.values()))
- self.packs.sort(lambda x,y: -cmp(len(x),len(y)))
+ self.packs.sort(reverse=True, key=lambda x: len(x))
if self.bloom and self.bloom.valid() and len(self.bloom) >= len(self):
self.do_bloom = True
else:
"""Writes Git objects inside a pack file."""
def __init__(self, objcache_maker=_make_objcache, compression_level=1,
run_midx=True, on_pack_finish=None,
- max_pack_size=None, max_pack_objects=None):
- self.repo_dir = repo()
+ max_pack_size=None, max_pack_objects=None, repo_dir=None):
+ self.repo_dir = repo_dir or repo()
self.file = None
self.parentfd = None
self.count = 0
def __del__(self):
self.close()
+ def __enter__(self):
+ return self
+
+ def __exit__(self, type, value, traceback):
+ self.close()
+
def _open(self):
if not self.file:
objdir = dir = os.path.join(self.repo_dir, 'objects')
return self.objcache.exists(id, want_source=want_source)
def just_write(self, sha, type, content):
- """Write an object to the pack file, bypassing the objcache. Fails if
- sha exists()."""
+ """Write an object to the pack file without checking for duplication."""
self._write(sha, type, content)
+ # If nothing else, gc doesn't have/want an objcache
+ if self.objcache is not None:
+ self.objcache.add(sha)
def maybe_write(self, type, content):
"""Write an object to the pack file if not present and return its id."""
sha = calc_hash(type, content)
if not self.exists(sha):
- self.just_write(sha, type, content)
self._require_objcache()
- self.objcache.add(sha)
+ self.just_write(sha, type, content)
return sha
def new_blob(self, blob):
return None
-def rev_list(ref_or_refs, count=None, parse=None, format=None, repo_dir=None):
- """Yield information about commits as per "git rev-list". If a format
- is not provided, yield one hex hash at a time. If a format is
- provided, pass it to rev-list and call parse(git_stdout) for each
- commit with the stream positioned just after the rev-list "commit
- HASH" header line. When a format is provided yield (oidx,
- parse(git_stdout)) for each commit.
-
- """
- assert bool(parse) == bool(format)
+def rev_list_invocation(ref_or_refs, count=None, format=None):
if isinstance(ref_or_refs, compat.str_type):
refs = (ref_or_refs,)
else:
argv = ['git', 'rev-list']
if isinstance(count, Integral):
argv.extend(['-n', str(count)])
- else:
- assert not count
+ elif count:
+ raise ValueError('unexpected count argument %r' % count)
+
if format:
argv.append('--pretty=format:' + format)
for ref in refs:
assert not ref.startswith('-')
argv.append(ref)
argv.append('--')
- p = subprocess.Popen(argv,
+ return argv
+
+
+def rev_list(ref_or_refs, count=None, parse=None, format=None, repo_dir=None):
+ """Yield information about commits as per "git rev-list". If a format
+ is not provided, yield one hex hash at a time. If a format is
+ provided, pass it to rev-list and call parse(git_stdout) for each
+ commit with the stream positioned just after the rev-list "commit
+ HASH" header line. When a format is provided yield (oidx,
+ parse(git_stdout)) for each commit.
+
+ """
+ assert bool(parse) == bool(format)
+ p = subprocess.Popen(rev_list_invocation(ref_or_refs, count=count,
+ format=format),
preexec_fn = _gitenv(repo_dir),
stdout = subprocess.PIPE)
if not format:
return
info = hdr.split(' ')
if len(info) != 3 or len(info[0]) != 40:
- raise GitError('expected object (id, type, size), got %r' % spl)
+ raise GitError('expected object (id, type, size), got %r' % info)
oidx, typ, size = info
size = int(size)
it = _AbortableIter(chunkyreader(self.p.stdout, size),
# ...
-def walk_object(cat_pipe, oidx,
- stop_at=None,
- include_data=None):
- """Yield everything reachable from oidx via cat_pipe as a WalkItem,
- stopping whenever stop_at(oidx) returns true. Throw MissingObject
- if a hash encountered is missing from the repository, and don't
- read or return blob content in the data field unless include_data
- is set.
+def walk_object(get_ref, oidx, stop_at=None, include_data=None):
+ """Yield everything reachable from oidx via get_ref (which must behave
+ like CatPipe get) as a WalkItem, stopping whenever stop_at(oidx)
+ returns true. Throw MissingObject if a hash encountered is
+ missing from the repository, and don't read or return blob content
+ in the data field unless include_data is set.
+
"""
# Maintain the pending stack on the heap to avoid stack overflow
pending = [(oidx, [], [], None)]
data=None)
continue
- item_it = cat_pipe.get(oidx)
+ item_it = get_ref(oidx)
get_oidx, typ, _ = next(item_it)
if not get_oidx:
raise MissingObject(oidx.decode('hex'))