1 # pack.py -- For dealing wih packed git objects.
2 # Copyright (C) 2007 James Westby <jw+debian@jameswestby.net>
3 # Copryight (C) 2008 Jelmer Vernooij <jelmer@samba.org>
4 # The code is loosely based on that in the sha1_file.c file from git itself,
5 # which is Copyright (C) Linus Torvalds, 2005 and distributed under the
8 # This program is free software; you can redistribute it and/or
9 # modify it under the terms of the GNU General Public License
10 # as published by the Free Software Foundation; version 2
13 # This program is distributed in the hope that it will be useful,
14 # but WITHOUT ANY WARRANTY; without even the implied warranty of
15 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 # GNU General Public License for more details.
18 # You should have received a copy of the GNU General Public License
19 # along with this program; if not, write to the Free Software
20 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston,
23 """Classes for dealing with packed git objects.
25 A pack is a compact representation of a bunch of objects, stored
26 using deltas where possible.
28 They have two parts, the pack file, which stores the data, and an index
29 that tells you where the data is.
31 To find an object you look in all of the index files 'til you find a
32 match for the object name. You then use the pointer got from this as
33 a pointer in to the corresponding packfile.
36 from collections import defaultdict
38 from itertools import imap, izip
49 from errors import ApplyDeltaError
51 supports_mmap_offset = (sys.version_info[0] >= 3 or
52 (sys.version_info[0] == 2 and sys.version_info[1] >= 6))
55 def take_msb_bytes(map, offset):
57 while len(ret) == 0 or ret[-1] & 0x80:
58 ret.append(ord(map[offset]))
63 def read_zlib(data, offset, dec_size):
64 obj = zlib.decompressobj()
67 while obj.unused_data == "":
69 add = data[base:base+1024]
71 x += obj.decompress(add)
72 assert len(x) == dec_size
73 comp_len = fed-len(obj.unused_data)
81 return sha.hexdigest()
85 """Convert a hex string to a binary sha string."""
87 for i in range(0, len(hex), 2):
88 ret += chr(int(hex[i:i+2], 16))
93 """Convert a binary sha string to a hex sha string."""
96 ret += "%02x" % ord(i)
100 MAX_MMAP_SIZE = 256 * 1024 * 1024
102 def simple_mmap(f, offset, size, access=mmap.ACCESS_READ):
103 """Simple wrapper for mmap() which always supports the offset parameter.
105 :param f: File object.
106 :param offset: Offset in the file, from the beginning of the file.
107 :param size: Size of the mmap'ed area
108 :param access: Access mechanism.
109 :return: MMAP'd area.
111 if offset+size > MAX_MMAP_SIZE and not supports_mmap_offset:
112 raise AssertionError("%s is larger than 256 meg, and this version "
113 "of Python does not support the offset argument to mmap().")
114 if supports_mmap_offset:
115 return mmap.mmap(f.fileno(), size, access=access, offset=offset)
117 class ArraySkipper(object):
119 def __init__(self, array, offset):
123 def __getslice__(self, i, j):
124 return self.array[i+self.offset:j+self.offset]
126 def __getitem__(self, i):
127 return self.array[i+self.offset]
130 return len(self.array) - self.offset
133 return str(self.array[self.offset:])
135 mem = mmap.mmap(f.fileno(), size+offset, access=access)
138 return ArraySkipper(mem, offset)
141 def resolve_object(offset, type, obj, get_ref, get_offset):
142 """Resolve an object, possibly resolving deltas when necessary."""
143 if not type in (6, 7): # Not a delta
146 if type == 6: # offset delta
147 (delta_offset, delta) = obj
148 assert isinstance(delta_offset, int)
149 assert isinstance(delta, str)
150 offset = offset-delta_offset
151 type, base_obj = get_offset(offset)
152 assert isinstance(type, int)
153 elif type == 7: # ref delta
154 (basename, delta) = obj
155 assert isinstance(basename, str) and len(basename) == 20
156 assert isinstance(delta, str)
157 type, base_obj = get_ref(basename)
158 assert isinstance(type, int)
159 type, base_text = resolve_object(offset, type, base_obj, get_ref, get_offset)
160 return type, apply_delta(base_text, delta)
163 class PackIndex(object):
164 """An index in to a packfile.
166 Given a sha id of an object a pack index can tell you the location in the
167 packfile of that object if it has it.
169 To do the loop it opens the file, and indexes first 256 4 byte groups
170 with the first byte of the sha id. The value in the four byte group indexed
171 is the end of the group that shares the same starting byte. Subtract one
172 from the starting byte and index again to find the start of the group.
173 The values are sorted by sha id within the group, so do the math to find
174 the start and end offset and then bisect in to find if the value is present.
177 def __init__(self, filename):
178 """Create a pack index object.
180 Provide it with the name of the index file to consider, and it will map
181 it whenever required.
183 self._filename = filename
184 # Take the size now, so it can be checked each time we map the file to
185 # ensure that it hasn't changed.
186 self._size = os.path.getsize(filename)
187 self._file = open(filename, 'r')
188 self._contents = simple_mmap(self._file, 0, self._size)
189 if self._contents[:4] != '\377tOc':
191 self._fan_out_table = self._read_fan_out_table(0)
193 (self.version, ) = struct.unpack_from(">L", self._contents, 4)
194 assert self.version in (2,), "Version was %d" % self.version
195 self._fan_out_table = self._read_fan_out_table(8)
196 self._name_table_offset = 8 + 0x100 * 4
197 self._crc32_table_offset = self._name_table_offset + 20 * len(self)
198 self._pack_offset_table_offset = self._crc32_table_offset + 4 * len(self)
200 def __eq__(self, other):
201 if type(self) != type(other):
204 if self._fan_out_table != other._fan_out_table:
207 for (name1, _, _), (name2, _, _) in izip(self.iterentries(), other.iterentries()):
216 """Return the number of entries in this pack index."""
217 return self._fan_out_table[-1]
219 def _unpack_entry(self, i):
220 """Unpack the i-th entry in the index file.
222 :return: Tuple with object name (SHA), offset in pack file and
223 CRC32 checksum (if known)."""
224 if self.version == 1:
225 (offset, name) = struct.unpack_from(">L20s", self._contents,
226 (0x100 * 4) + (i * 24))
227 return (name, offset, None)
229 return (self._unpack_name(i), self._unpack_offset(i),
230 self._unpack_crc32_checksum(i))
232 def _unpack_name(self, i):
233 if self.version == 1:
234 return self._unpack_entry(i)[0]
236 return struct.unpack_from("20s", self._contents,
237 self._name_table_offset + i * 20)[0]
239 def _unpack_offset(self, i):
240 if self.version == 1:
241 return self._unpack_entry(i)[1]
243 return struct.unpack_from(">L", self._contents,
244 self._pack_offset_table_offset + i * 4)[0]
246 def _unpack_crc32_checksum(self, i):
247 if self.version == 1:
250 return struct.unpack_from(">L", self._contents,
251 self._crc32_table_offset + i * 4)[0]
254 return imap(sha_to_hex, self._itersha())
257 for i in range(len(self)):
258 yield self._unpack_name(i)
260 def objects_sha1(self):
261 return iter_sha1(self._itersha())
263 def iterentries(self):
264 """Iterate over the entries in this pack index.
266 Will yield tuples with object name, offset in packfile and crc32 checksum.
268 for i in range(len(self)):
269 yield self._unpack_entry(i)
271 def _read_fan_out_table(self, start_offset):
273 for i in range(0x100):
274 ret.append(struct.unpack(">L", self._contents[start_offset+i*4:start_offset+(i+1)*4])[0])
278 """Check that the stored checksum matches the actual checksum."""
279 return self.calculate_checksum() == self.get_stored_checksums()[1]
281 def calculate_checksum(self):
282 f = open(self._filename, 'r')
284 return hashlib.sha1(self._contents[:-20]).digest()
288 def get_stored_checksums(self):
289 """Return the SHA1 checksums stored for the corresponding packfile and
290 this header file itself."""
291 return str(self._contents[-40:-20]), str(self._contents[-20:])
293 def object_index(self, sha):
294 """Return the index in to the corresponding packfile for the object.
296 Given the name of an object it will return the offset that object lives
297 at within the corresponding pack file. If the pack file doesn't have the
298 object then None will be returned.
300 size = os.path.getsize(self._filename)
301 assert size == self._size, "Pack index %s has changed size, I don't " \
302 "like that" % self._filename
304 sha = hex_to_sha(sha)
305 return self._object_index(sha)
307 def _object_index(self, sha):
308 """See object_index"""
313 start = self._fan_out_table[idx-1]
314 end = self._fan_out_table[idx]
318 file_sha = self._unpack_name(i)
324 return self._unpack_offset(i)
328 class PackData(object):
329 """The data contained in a packfile.
331 Pack files can be accessed both sequentially for exploding a pack, and
332 directly with the help of an index to retrieve a specific object.
334 The objects within are either complete or a delta aginst another.
336 The header is variable length. If the MSB of each byte is set then it
337 indicates that the subsequent byte is still part of the header.
338 For the first byte the next MS bits are the type, which tells you the type
339 of object, and whether it is a delta. The LS byte is the lowest bits of the
340 size. For each subsequent byte the LS 7 bits are the next MS bits of the
341 size, i.e. the last byte of the header contains the MS bits of the size.
343 For the complete objects the data is stored as zlib deflated data.
344 The size in the header is the uncompressed object size, so to uncompress
345 you need to just keep feeding data to zlib until you get an object back,
346 or it errors on bad data. This is done here by just giving the complete
347 buffer from the start of the deflated object on. This is bad, but until I
348 get mmap sorted out it will have to do.
350 Currently there are no integrity checks done. Also no attempt is made to try
351 and detect the delta case, or a request for an object at the wrong position.
352 It will all just throw a zlib or KeyError.
355 def __init__(self, filename):
356 """Create a PackData object that represents the pack in the given filename.
358 The file must exist and stay readable until the object is disposed of. It
359 must also stay the same size. It will be mapped whenever needed.
361 Currently there is a restriction on the size of the pack as the python
362 mmap implementation is flawed.
364 self._filename = filename
365 assert os.path.exists(filename), "%s is not a packfile" % filename
366 self._size = os.path.getsize(filename)
367 self._header_size = 12
368 assert self._size >= self._header_size, "%s is too small for a packfile" % filename
371 def _read_header(self):
372 f = open(self._filename, 'rb')
375 f.seek(self._size-20)
376 self._stored_checksum = f.read(20)
379 assert header[:4] == "PACK"
380 (version,) = struct.unpack_from(">L", header, 4)
381 assert version in (2, 3), "Version was %d" % version
382 (self._num_objects,) = struct.unpack_from(">L", header, 8)
385 """Returns the number of objects in this pack."""
386 return self._num_objects
388 def calculate_checksum(self):
389 f = open(self._filename, 'rb')
391 map = simple_mmap(f, 0, self._size)
392 return hashlib.sha1(map[:-20]).digest()
396 def iterobjects(self):
397 offset = self._header_size
398 f = open(self._filename, 'rb')
399 for i in range(len(self)):
400 map = simple_mmap(f, offset, self._size-offset)
401 (type, obj, total_size) = self._unpack_object(map)
402 yield offset, type, obj
406 def iterentries(self, ext_resolve_ref=None):
409 postponed = defaultdict(list)
410 class Postpone(Exception):
411 """Raised to postpone delta resolving."""
413 def get_ref_text(sha):
418 return ext_resolve_ref(sha)
421 raise Postpone, (sha, )
422 todo = list(self.iterobjects())
424 (offset, type, obj) = todo.pop(0)
425 at[offset] = (type, obj)
426 assert isinstance(offset, int)
427 assert isinstance(type, int)
428 assert isinstance(obj, tuple) or isinstance(obj, str)
430 type, obj = resolve_object(offset, type, obj, get_ref_text,
432 except Postpone, (sha, ):
433 postponed[sha].append((offset, type, obj))
435 shafile = ShaFile.from_raw_string(type, obj)
436 sha = shafile.sha().digest()
437 found[sha] = (type, obj)
438 yield sha, offset, shafile.crc32()
439 todo += postponed.get(sha, [])
441 raise KeyError([sha_to_hex(h) for h in postponed.keys()])
443 def sorted_entries(self, resolve_ext_ref=None):
444 ret = list(self.iterentries(resolve_ext_ref))
448 def create_index_v1(self, filename):
449 entries = self.sorted_entries()
450 write_pack_index_v1(filename, entries, self.calculate_checksum())
452 def create_index_v2(self, filename):
453 entries = self.sorted_entries()
454 write_pack_index_v2(filename, entries, self.calculate_checksum())
456 def get_stored_checksum(self):
457 return self._stored_checksum
460 return (self.calculate_checksum() == self.get_stored_checksum())
462 def get_object_at(self, offset):
463 """Given an offset in to the packfile return the object that is there.
465 Using the associated index the location of an object can be looked up, and
466 then the packfile can be asked directly for that object using this
469 assert isinstance(offset, long) or isinstance(offset, int),\
470 "offset was %r" % offset
471 assert offset >= self._header_size
472 size = os.path.getsize(self._filename)
473 assert size == self._size, "Pack data %s has changed size, I don't " \
474 "like that" % self._filename
475 f = open(self._filename, 'rb')
477 map = simple_mmap(f, offset, size-offset)
478 return self._unpack_object(map)[:2]
482 def _unpack_object(self, map):
483 bytes = take_msb_bytes(map, 0)
484 type = (bytes[0] >> 4) & 0x07
485 size = bytes[0] & 0x0f
486 for i, byte in enumerate(bytes[1:]):
487 size += (byte & 0x7f) << ((i * 7) + 4)
488 raw_base = len(bytes)
489 if type == 6: # offset delta
490 bytes = take_msb_bytes(map, raw_base)
491 assert not (bytes[-1] & 0x80)
492 delta_base_offset = bytes[0] & 0x7f
493 for byte in bytes[1:]:
494 delta_base_offset += 1
495 delta_base_offset <<= 7
496 delta_base_offset += (byte & 0x7f)
498 uncomp, comp_len = read_zlib(map, raw_base, size)
499 assert size == len(uncomp)
500 return type, (delta_base_offset, uncomp), comp_len+raw_base
501 elif type == 7: # ref delta
502 basename = map[raw_base:raw_base+20]
503 uncomp, comp_len = read_zlib(map, raw_base+20, size)
504 assert size == len(uncomp)
505 return type, (basename, uncomp), comp_len+raw_base+20
507 uncomp, comp_len = read_zlib(map, raw_base, size)
508 assert len(uncomp) == size
509 return type, uncomp, comp_len+raw_base
512 class SHA1Writer(object):
514 def __init__(self, f):
516 self.sha1 = hashlib.sha1("")
518 def write(self, data):
519 self.sha1.update(data)
523 sha = self.sha1.digest()
524 assert len(sha) == 20
529 sha = self.write_sha()
537 def write_pack_object(f, type, object):
538 """Write pack object to a file.
540 :param f: File to write to
541 :param o: Object to write
544 if type == 6: # ref delta
545 (delta_base_offset, object) = object
546 elif type == 7: # offset delta
547 (basename, object) = object
549 c = (type << 4) | (size & 15)
552 f.write(chr(c | 0x80))
556 if type == 6: # offset delta
557 ret = [delta_base_offset & 0x7f]
558 delta_base_offset >>= 7
559 while delta_base_offset:
560 delta_base_offset -= 1
561 ret.insert(0, 0x80 | (delta_base_offset & 0x7f))
562 delta_base_offset >>= 7
563 f.write("".join([chr(x) for x in ret]))
564 elif type == 7: # ref delta
565 assert len(basename) == 20
567 f.write(zlib.compress(object))
571 def write_pack(filename, objects, num_objects):
572 f = open(filename + ".pack", 'w')
574 entries, data_sum = write_pack_data(f, objects, num_objects)
578 write_pack_index_v2(filename + ".idx", entries, data_sum)
581 def write_pack_data(f, objects, num_objects):
582 """Write a new pack file.
584 :param filename: The filename of the new pack file.
585 :param objects: List of objects to write.
586 :return: List with (name, offset, crc32 checksum) entries, pack checksum
590 f.write("PACK") # Pack header
591 f.write(struct.pack(">L", 2)) # Pack version
592 f.write(struct.pack(">L", num_objects)) # Number of objects in pack
594 sha1 = o.sha().digest()
597 t, o = o.as_raw_string()
598 offset = write_pack_object(f, t, o)
599 entries.append((sha1, offset, crc32))
600 return entries, f.write_sha()
603 def write_pack_index_v1(filename, entries, pack_checksum):
604 """Write a new pack index file.
606 :param filename: The filename of the new pack index file.
607 :param entries: List of tuples with object name (sha), offset_in_pack, and
609 :param pack_checksum: Checksum of the pack file.
611 f = open(filename, 'w')
613 fan_out_table = defaultdict(lambda: 0)
614 for (name, offset, entry_checksum) in entries:
615 fan_out_table[ord(name[0])] += 1
617 for i in range(0x100):
618 f.write(struct.pack(">L", fan_out_table[i]))
619 fan_out_table[i+1] += fan_out_table[i]
620 for (name, offset, entry_checksum) in entries:
621 f.write(struct.pack(">L20s", offset, name))
622 assert len(pack_checksum) == 20
623 f.write(pack_checksum)
627 def apply_delta(src_buf, delta):
628 """Based on the similar function in git's patch-delta.c."""
629 assert isinstance(src_buf, str), "was %r" % (src_buf,)
630 assert isinstance(delta, str)
635 return ord(ret), delta
636 def get_delta_header_size(delta):
640 cmd, delta = pop(delta)
641 size |= (cmd & ~0x80) << i
646 src_size, delta = get_delta_header_size(delta)
647 dest_size, delta = get_delta_header_size(delta)
648 assert src_size == len(src_buf)
650 cmd, delta = pop(delta)
655 x, delta = pop(delta)
656 cp_off |= x << (i * 8)
659 if cmd & (1 << (4+i)):
660 x, delta = pop(delta)
661 cp_size |= x << (i * 8)
664 if (cp_off + cp_size < cp_size or
665 cp_off + cp_size > src_size or
666 cp_size > dest_size):
668 out += src_buf[cp_off:cp_off+cp_size]
673 raise ApplyDeltaError("Invalid opcode 0")
676 raise ApplyDeltaError("delta not empty: %r" % delta)
678 if dest_size != len(out):
679 raise ApplyDeltaError("dest size incorrect")
684 def write_pack_index_v2(filename, entries, pack_checksum):
685 """Write a new pack index file.
687 :param filename: The filename of the new pack index file.
688 :param entries: List of tuples with object name (sha), offset_in_pack, and
690 :param pack_checksum: Checksum of the pack file.
692 f = open(filename, 'w')
694 f.write('\377tOc') # Magic!
695 f.write(struct.pack(">L", 2))
696 fan_out_table = defaultdict(lambda: 0)
697 for (name, offset, entry_checksum) in entries:
698 fan_out_table[ord(name[0])] += 1
700 for i in range(0x100):
701 f.write(struct.pack(">L", fan_out_table[i]))
702 fan_out_table[i+1] += fan_out_table[i]
703 for (name, offset, entry_checksum) in entries:
705 for (name, offset, entry_checksum) in entries:
706 f.write(struct.pack(">l", entry_checksum))
707 for (name, offset, entry_checksum) in entries:
708 # FIXME: handle if MSBit is set in offset
709 f.write(struct.pack(">L", offset))
710 # FIXME: handle table for pack files > 8 Gb
711 assert len(pack_checksum) == 20
712 f.write(pack_checksum)
718 def __init__(self, basename):
719 self._basename = basename
720 self._data_path = self._basename + ".pack"
721 self._idx_path = self._basename + ".idx"
726 return self.idx.objects_sha1()
730 if self._data is None:
731 self._data = PackData(self._data_path)
732 assert len(self.idx) == len(self._data)
733 assert self.idx.get_stored_checksums()[0] == self._data.get_stored_checksum()
738 if self._idx is None:
739 self._idx = PackIndex(self._idx_path)
743 if self._data is not None:
747 def __eq__(self, other):
748 return type(self) == type(other) and self.idx == other.idx
751 """Number of entries in this pack."""
755 return "Pack(%r)" % self._basename
758 """Iterate over all the sha1s of the objects in this pack."""
759 return iter(self.idx)
762 return self.idx.check() and self.data.check()
764 def get_stored_checksum(self):
765 return self.data.get_stored_checksum()
767 def __contains__(self, sha1):
768 """Check whether this pack contains a particular SHA1."""
769 return (self.idx.object_index(sha1) is not None)
771 def get_raw(self, sha1, resolve_ref=None):
772 if resolve_ref is None:
773 resolve_ref = self.get_raw
774 offset = self.idx.object_index(sha1)
778 type, obj = self.data.get_object_at(offset)
779 assert isinstance(offset, int)
780 return resolve_object(offset, type, obj, resolve_ref,
781 self.data.get_object_at)
783 def __getitem__(self, sha1):
784 """Retrieve the specified SHA1."""
785 type, uncomp = self.get_raw(sha1)
786 return ShaFile.from_raw_string(type, uncomp)
788 def iterobjects(self):
789 for offset, type, obj in self.data.iterobjects():
790 assert isinstance(offset, int)
791 yield ShaFile.from_raw_string(
792 *resolve_object(offset, type, obj, self.get_raw,
793 self.data.get_object_at))
796 def load_packs(path):
797 if not os.path.exists(path):
799 for name in os.listdir(path):
800 if name.startswith("pack-") and name.endswith(".pack"):
801 yield Pack(os.path.join(path, name[:-len(".pack")]))