52
52
is that sequence numbers are stable references. But not every
53
53
repository in the world will assign the same sequence numbers,
54
54
therefore the SHA-1 is the only universally unique reference.
56
This is meant to scale to hold 100,000 revisions of a single file, by
57
which time the index file will be ~4.8MB and a bit big to read
60
Some of the reserved fields could be used to implement a (semi?)
61
balanced tree indexed by SHA1 so we can much more efficiently find the
62
index associated with a particular hash. For 100,000 revs we would be
63
able to find it in about 17 random reads, which is not too bad.
65
This performs pretty well except when trying to calculate deltas of
66
really large files. For that the main thing would be to plug in
67
something faster than difflib, which is after all pure Python.
68
Another approach is to just store the gzipped full text of big files,
69
though perhaps that's too perverse?
71
55
The iter method here will generally read through the whole index file
72
56
in one go. With readahead in the kernel and python/libc (typically
73
57
128kB) this means that there should be no seeks and often only one
94
78
import sys, zlib, struct, mdiff, stat, os, sha
95
79
from binascii import hexlify, unhexlify
99
85
_HEADER = "bzr revfile v1\n"
241
227
return self._add_compressed(text_sha, text, _NO_RECORD, compress)
245
def _choose_base(self, seed, base):
247
if base == _NO_RECORD:
250
if idxrec[I_BASE] == _NO_RECORD:
253
base = idxrec[I_BASE]
256
return base # relative to this full text
260
230
def _add_delta(self, text, text_sha, base, compress):
261
231
"""Add a text stored relative to a previous text."""
262
232
self._check_index(base)
265
base_text = self.get(base, CHAIN_LIMIT)
235
base_text = self.get(base, recursion_limit=CHAIN_LIMIT)
266
236
except LimitHitException:
267
237
return self._add_full_text(text, text_sha, compress)
302
272
# it's the same, in case someone ever breaks SHA-1.
303
273
return idx # already present
305
# base = self._choose_base(ord(text_sha[0]), base)
307
275
if base == _NO_RECORD:
308
276
return self._add_full_text(text, text_sha, compress)
420
388
"""Read back all index records.
422
390
Do not seek the index file while this is underway!"""
423
## sys.stderr.write(" ** iter called ** \n")
391
sys.stderr.write(" ** iter called ** \n")
424
392
self._seek_index(0)
426
394
idxrec = self._read_next_index()
478
445
except IndexError:
479
sys.stderr.write("usage: revfile dump REVFILE\n"
480
" revfile add REVFILE < INPUT\n"
481
" revfile add-delta REVFILE BASE < INPUT\n"
482
" revfile add-series REVFILE BASE FILE...\n"
483
" revfile get REVFILE IDX\n"
484
" revfile find-sha REVFILE HEX\n"
485
" revfile total-text-size REVFILE\n"
486
" revfile last REVFILE\n")
446
sys.stderr.write("usage: revfile dump\n"
448
" revfile add-delta BASE\n"
450
" revfile find-sha HEX\n"
451
" revfile total-text-size\n"
490
return Revfile(filename, 'w')
456
return Revfile('testrev', 'w')
493
return Revfile(filename, 'r')
459
return Revfile('testrev', 'r')
496
462
print rw().add(sys.stdin.read())
497
463
elif cmd == 'add-delta':
498
print rw().add(sys.stdin.read(), int(argv[3]))
499
elif cmd == 'add-series':
504
rev = r.add(file(fn).read(), rev)
464
print rw().add(sys.stdin.read(), int(argv[2]))
505
465
elif cmd == 'dump':
507
467
elif cmd == 'get':
510
470
except IndexError:
511
sys.stderr.write("usage: revfile get FILE IDX\n")
471
sys.stderr.write("usage: revfile get IDX\n")
516
474
if idx < 0 or idx >= len(r):
517
475
sys.stderr.write("invalid index %r\n" % idx)
520
sys.stdout.write(r.get(idx))
478
sys.stdout.write(ro().get(idx))
521
479
elif cmd == 'find-sha':
523
s = unhexlify(argv[3])
481
s = unhexlify(argv[2])
524
482
except IndexError:
525
sys.stderr.write("usage: revfile find-sha FILE HEX\n")
483
sys.stderr.write("usage: revfile find-sha HEX\n")
528
486
idx = ro().find_sha(s)
added
removed
bzrlib/revfile.py
