annotate mercurial/revlog.py @ 5451:0a43875677b1

revlog: break up compression of large deltas Python's zlib apparently makes an internal copy of strings passed to compress(). To avoid this, compress strings 1M at a time, then join them at the end if the result would be smaller than the original. For initial commits of large but compressible files, this cuts peak memory usage nearly in half.
author Matt Mackall <mpm@selenic.com>
date Thu, 11 Oct 2007 00:46:54 -0500
parents c728424d44c6
children 9d77f2b47eb7
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1 """
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
2 revlog.py - storage back-end for mercurial
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
3
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
4 This provides efficient delta storage with O(1) retrieve and append
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
5 and O(changes) merge between branches
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
6
4635
63b9d2deed48 Updated copyright notices and add "and others" to "hg version"
Thomas Arendsen Hein <thomas@intevation.de>
parents: 4497
diff changeset
7 Copyright 2005-2007 Matt Mackall <mpm@selenic.com>
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
8
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
9 This software may be used and distributed according to the terms
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
10 of the GNU General Public License, incorporated herein by reference.
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
11 """
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
12
1089
142b5d5ec9cc Break apart hg.py
mpm@selenic.com
parents: 1083
diff changeset
13 from node import *
3893
6b4127c7d52a Simplify i18n imports
Matt Mackall <mpm@selenic.com>
parents: 3886
diff changeset
14 from i18n import _
3886
abaee83ce0a6 Replace demandload with new demandimport
Matt Mackall <mpm@selenic.com>
parents: 3755
diff changeset
15 import binascii, changegroup, errno, ancestor, mdiff, os
abaee83ce0a6 Replace demandload with new demandimport
Matt Mackall <mpm@selenic.com>
parents: 3755
diff changeset
16 import sha, struct, util, zlib
36
da28286bf6b7 Add smart node lookup by substring or by rev number
mpm@selenic.com
parents: 26
diff changeset
17
5008
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
18 _pack = struct.pack
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
19 _unpack = struct.unpack
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
20 _compress = zlib.compress
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
21 _decompress = zlib.decompress
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
22 _sha = sha.new
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
23
4987
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
24 # revlog flags
2072
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
25 REVLOGV0 = 0
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
26 REVLOGNG = 1
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
27 REVLOGNGINLINEDATA = (1 << 16)
2222
c9e264b115e6 Use revlogng and inlined data files by default
mason@suse.com
parents: 2177
diff changeset
28 REVLOG_DEFAULT_FLAGS = REVLOGNGINLINEDATA
c9e264b115e6 Use revlogng and inlined data files by default
mason@suse.com
parents: 2177
diff changeset
29 REVLOG_DEFAULT_FORMAT = REVLOGNG
c9e264b115e6 Use revlogng and inlined data files by default
mason@suse.com
parents: 2177
diff changeset
30 REVLOG_DEFAULT_VERSION = REVLOG_DEFAULT_FORMAT | REVLOG_DEFAULT_FLAGS
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
31
4987
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
32 class RevlogError(Exception):
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
33 pass
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
34 class LookupError(RevlogError):
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
35 pass
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
36
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
37 def getoffset(q):
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
38 return int(q >> 16)
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
39
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
40 def gettype(q):
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
41 return int(q & 0xFFFF)
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
42
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
43 def offset_type(offset, type):
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
44 return long(long(offset) << 16 | type)
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
45
1091
d62130f99a73 Move hash function back to revlog from node
mpm@selenic.com
parents: 1089
diff changeset
46 def hash(text, p1, p2):
d62130f99a73 Move hash function back to revlog from node
mpm@selenic.com
parents: 1089
diff changeset
47 """generate a hash from the given text and its parent hashes
d62130f99a73 Move hash function back to revlog from node
mpm@selenic.com
parents: 1089
diff changeset
48
d62130f99a73 Move hash function back to revlog from node
mpm@selenic.com
parents: 1089
diff changeset
49 This hash combines both the current file contents and its history
d62130f99a73 Move hash function back to revlog from node
mpm@selenic.com
parents: 1089
diff changeset
50 in a manner that makes it easy to distinguish nodes with the same
d62130f99a73 Move hash function back to revlog from node
mpm@selenic.com
parents: 1089
diff changeset
51 content in the revision graph.
d62130f99a73 Move hash function back to revlog from node
mpm@selenic.com
parents: 1089
diff changeset
52 """
d62130f99a73 Move hash function back to revlog from node
mpm@selenic.com
parents: 1089
diff changeset
53 l = [p1, p2]
d62130f99a73 Move hash function back to revlog from node
mpm@selenic.com
parents: 1089
diff changeset
54 l.sort()
5008
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
55 s = _sha(l[0])
1091
d62130f99a73 Move hash function back to revlog from node
mpm@selenic.com
parents: 1089
diff changeset
56 s.update(l[1])
d62130f99a73 Move hash function back to revlog from node
mpm@selenic.com
parents: 1089
diff changeset
57 s.update(text)
d62130f99a73 Move hash function back to revlog from node
mpm@selenic.com
parents: 1089
diff changeset
58 return s.digest()
d62130f99a73 Move hash function back to revlog from node
mpm@selenic.com
parents: 1089
diff changeset
59
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
60 def compress(text):
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
61 """ generate a possibly-compressed representation of text """
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
62 if not text:
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
63 return ("", text)
5451
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
64 l = len(text)
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
65 if l < 44:
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
66 if text[0] == '\0':
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
67 return ("", text)
1533
3d11f81c9145 Reduce string duplication in compression code
mason@suse.com
parents: 1509
diff changeset
68 return ('u', text)
5451
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
69 elif l > 1000000:
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
70 # zlib makes an internal copy, thus doubling memory usage for
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
71 # large files, so lets do this in pieces
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
72 z = zlib.compressobj()
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
73 p = []
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
74 pos = 0
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
75 while pos < l:
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
76 pos2 = pos + 2**20
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
77 p.append(z.compress(text[pos:pos2]))
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
78 pos = pos2
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
79 p.append(z.flush())
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
80 if sum(map(len, p)) < l:
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
81 bin = "".join(p)
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
82 else:
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
83 bin = _compress(text)
0a43875677b1 revlog: break up compression of large deltas
Matt Mackall <mpm@selenic.com>
parents: 5450
diff changeset
84 if len(bin) > l:
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
85 if text[0] == '\0':
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
86 return ("", text)
1533
3d11f81c9145 Reduce string duplication in compression code
mason@suse.com
parents: 1509
diff changeset
87 return ('u', text)
3d11f81c9145 Reduce string duplication in compression code
mason@suse.com
parents: 1509
diff changeset
88 return ("", bin)
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
89
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
90 def decompress(bin):
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
91 """ decompress the given input """
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
92 if not bin:
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
93 return bin
112
aea6562add6c Make compression more intelligent:
mpm@selenic.com
parents: 98
diff changeset
94 t = bin[0]
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
95 if t == '\0':
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
96 return bin
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
97 if t == 'x':
5008
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
98 return _decompress(bin)
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
99 if t == 'u':
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
100 return bin[1:]
1853
5ac811b720de Fix some problems when working on broken repositories:
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1784
diff changeset
101 raise RevlogError(_("unknown compression type %r") % t)
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
102
1559
59b3639df0a9 Convert all classes to new-style classes by deriving them from object.
Eric Hopper <hopper@omnifarious.org>
parents: 1551
diff changeset
103 class lazyparser(object):
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
104 """
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
105 this class avoids the need to parse the entirety of large indices
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
106 """
2250
45aef5ddcdbe windows: revlog.lazyparser not always safe to use.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2222
diff changeset
107
45aef5ddcdbe windows: revlog.lazyparser not always safe to use.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2222
diff changeset
108 # lazyparser is not safe to use on windows if win32 extensions not
45aef5ddcdbe windows: revlog.lazyparser not always safe to use.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2222
diff changeset
109 # available. it keeps file handle open, which make it not possible
45aef5ddcdbe windows: revlog.lazyparser not always safe to use.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2222
diff changeset
110 # to break hardlinks on local cloned repos.
45aef5ddcdbe windows: revlog.lazyparser not always safe to use.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2222
diff changeset
111 safe_to_use = os.name != 'nt' or (not util.is_win_9x() and
45aef5ddcdbe windows: revlog.lazyparser not always safe to use.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2222
diff changeset
112 hasattr(util, 'win32api'))
45aef5ddcdbe windows: revlog.lazyparser not always safe to use.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2222
diff changeset
113
4976
79c39cc9ff69 revlog: only allow lazy parsing with revlogng files
Matt Mackall <mpm@selenic.com>
parents: 4975
diff changeset
114 def __init__(self, dataf, size):
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
115 self.dataf = dataf
4976
79c39cc9ff69 revlog: only allow lazy parsing with revlogng files
Matt Mackall <mpm@selenic.com>
parents: 4975
diff changeset
116 self.s = struct.calcsize(indexformatng)
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
117 self.datasize = size
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
118 self.l = size/self.s
76
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
119 self.index = [None] * self.l
3578
3b4e00cba57a Define and use nullrev (revision of nullid) instead of -1.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3505
diff changeset
120 self.map = {nullid: nullrev}
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
121 self.allmap = 0
323
c6f0673ab7e9 lazyparser speed ups
mpm@selenic.com
parents: 306
diff changeset
122 self.all = 0
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
123 self.mapfind_count = 0
76
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
124
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
125 def loadmap(self):
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
126 """
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
127 during a commit, we need to make sure the rev being added is
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
128 not a duplicate. This requires loading the entire index,
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
129 which is fairly slow. loadmap can load up just the node map,
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
130 which takes much less time.
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
131 """
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
132 if self.allmap:
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
133 return
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
134 end = self.datasize
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
135 self.allmap = 1
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
136 cur = 0
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
137 count = 0
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
138 blocksize = self.s * 256
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
139 self.dataf.seek(0)
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
140 while cur < end:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
141 data = self.dataf.read(blocksize)
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
142 off = 0
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
143 for x in xrange(256):
4976
79c39cc9ff69 revlog: only allow lazy parsing with revlogng files
Matt Mackall <mpm@selenic.com>
parents: 4975
diff changeset
144 n = data[off + ngshaoffset:off + ngshaoffset + 20]
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
145 self.map[n] = count
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
146 count += 1
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
147 if count >= self.l:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
148 break
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
149 off += self.s
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
150 cur += blocksize
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
151
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
152 def loadblock(self, blockstart, blocksize, data=None):
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
153 if self.all:
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
154 return
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
155 if data is None:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
156 self.dataf.seek(blockstart)
3075
baa3873eb387 don't let lazyparser read more data than it can handle
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 2910
diff changeset
157 if blockstart + blocksize > self.datasize:
baa3873eb387 don't let lazyparser read more data than it can handle
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 2910
diff changeset
158 # the revlog may have grown since we've started running,
baa3873eb387 don't let lazyparser read more data than it can handle
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 2910
diff changeset
159 # but we don't have space in self.index for more entries.
baa3873eb387 don't let lazyparser read more data than it can handle
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 2910
diff changeset
160 # limit blocksize so that we don't get too much data.
3086
e7fc04dc6349 Avoid negative block sizes in lazyparser.
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3075
diff changeset
161 blocksize = max(self.datasize - blockstart, 0)
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
162 data = self.dataf.read(blocksize)
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
163 lend = len(data) / self.s
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
164 i = blockstart / self.s
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
165 off = 0
4061
40030c1b6bc6 lazyindex: handle __delitem__ in loadblock
Brendan Cully <brendan@kublai.com>
parents: 3930
diff changeset
166 # lazyindex supports __delitem__
40030c1b6bc6 lazyindex: handle __delitem__ in loadblock
Brendan Cully <brendan@kublai.com>
parents: 3930
diff changeset
167 if lend > len(self.index) - i:
40030c1b6bc6 lazyindex: handle __delitem__ in loadblock
Brendan Cully <brendan@kublai.com>
parents: 3930
diff changeset
168 lend = len(self.index) - i
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
169 for x in xrange(lend):
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
170 if self.index[i + x] == None:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
171 b = data[off : off + self.s]
2080
1cbb14c048cb Reduce index memory usage by storing the bare string instead of tuples
mason@suse.com
parents: 2079
diff changeset
172 self.index[i + x] = b
4976
79c39cc9ff69 revlog: only allow lazy parsing with revlogng files
Matt Mackall <mpm@selenic.com>
parents: 4975
diff changeset
173 n = b[ngshaoffset:ngshaoffset + 20]
2080
1cbb14c048cb Reduce index memory usage by storing the bare string instead of tuples
mason@suse.com
parents: 2079
diff changeset
174 self.map[n] = i + x
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
175 off += self.s
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
176
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
177 def findnode(self, node):
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
178 """search backwards through the index file for a specific node"""
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
179 if self.allmap:
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
180 return None
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
181
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
182 # hg log will cause many many searches for the manifest
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
183 # nodes. After we get called a few times, just load the whole
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
184 # thing.
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
185 if self.mapfind_count > 8:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
186 self.loadmap()
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
187 if node in self.map:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
188 return node
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
189 return None
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
190 self.mapfind_count += 1
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
191 last = self.l - 1
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
192 while self.index[last] != None:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
193 if last == 0:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
194 self.all = 1
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
195 self.allmap = 1
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
196 return None
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
197 last -= 1
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
198 end = (last + 1) * self.s
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
199 blocksize = self.s * 256
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
200 while end >= 0:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
201 start = max(end - blocksize, 0)
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
202 self.dataf.seek(start)
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
203 data = self.dataf.read(end - start)
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
204 findend = end - start
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
205 while True:
4994
d36310dd51d7 lazyparser.findnode: fix typo and s/rfind/find/
Matt Mackall <mpm@selenic.com>
parents: 4993
diff changeset
206 # we're searching backwards, so we have to make sure
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
207 # we don't find a changeset where this node is a parent
4994
d36310dd51d7 lazyparser.findnode: fix typo and s/rfind/find/
Matt Mackall <mpm@selenic.com>
parents: 4993
diff changeset
208 off = data.find(node, 0, findend)
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
209 findend = off
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
210 if off >= 0:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
211 i = off / self.s
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
212 off = i * self.s
4976
79c39cc9ff69 revlog: only allow lazy parsing with revlogng files
Matt Mackall <mpm@selenic.com>
parents: 4975
diff changeset
213 n = data[off + ngshaoffset:off + ngshaoffset + 20]
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
214 if n == node:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
215 self.map[n] = i + start / self.s
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
216 return node
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
217 else:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
218 break
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
219 end -= blocksize
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
220 return None
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
221
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
222 def loadindex(self, i=None, end=None):
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
223 if self.all:
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
224 return
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
225 all = False
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
226 if i == None:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
227 blockstart = 0
4992
0a676643687b lazyparser: up the blocksize from 512 bytes to 64k
Matt Mackall <mpm@selenic.com>
parents: 4991
diff changeset
228 blocksize = (65536 / self.s) * self.s
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
229 end = self.datasize
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
230 all = True
323
c6f0673ab7e9 lazyparser speed ups
mpm@selenic.com
parents: 306
diff changeset
231 else:
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
232 if end:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
233 blockstart = i * self.s
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
234 end = end * self.s
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
235 blocksize = end - blockstart
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
236 else:
4992
0a676643687b lazyparser: up the blocksize from 512 bytes to 64k
Matt Mackall <mpm@selenic.com>
parents: 4991
diff changeset
237 blockstart = (i & ~1023) * self.s
0a676643687b lazyparser: up the blocksize from 512 bytes to 64k
Matt Mackall <mpm@selenic.com>
parents: 4991
diff changeset
238 blocksize = self.s * 1024
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
239 end = blockstart + blocksize
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
240 while blockstart < end:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
241 self.loadblock(blockstart, blocksize)
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
242 blockstart += blocksize
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
243 if all:
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
244 self.all = True
515
03f27b1381f9 Whitespace cleanups
mpm@selenic.com
parents: 484
diff changeset
245
1559
59b3639df0a9 Convert all classes to new-style classes by deriving them from object.
Eric Hopper <hopper@omnifarious.org>
parents: 1551
diff changeset
246 class lazyindex(object):
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
247 """a lazy version of the index array"""
76
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
248 def __init__(self, parser):
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
249 self.p = parser
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
250 def __len__(self):
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
251 return len(self.p.index)
115
39b438eeb25a Make lazyindex load slightly faster
mpm@selenic.com
parents: 112
diff changeset
252 def load(self, pos):
1403
bc3e66edb04c lazyindex fix, make load handle negative indexes properly.
Eric Hopper <hopper@omnifarious.org>
parents: 1402
diff changeset
253 if pos < 0:
bc3e66edb04c lazyindex fix, make load handle negative indexes properly.
Eric Hopper <hopper@omnifarious.org>
parents: 1402
diff changeset
254 pos += len(self.p.index)
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
255 self.p.loadindex(pos)
115
39b438eeb25a Make lazyindex load slightly faster
mpm@selenic.com
parents: 112
diff changeset
256 return self.p.index[pos]
76
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
257 def __getitem__(self, pos):
5008
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
258 return _unpack(indexformatng, self.p.index[pos] or self.load(pos))
2072
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
259 def __setitem__(self, pos, item):
5008
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
260 self.p.index[pos] = _pack(indexformatng, *item)
1535
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
261 def __delitem__(self, pos):
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
262 del self.p.index[pos]
4979
06abdaf78788 revlog: add a magic null revision to our index
Matt Mackall <mpm@selenic.com>
parents: 4978
diff changeset
263 def insert(self, pos, e):
5008
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
264 self.p.index.insert(pos, _pack(indexformatng, *e))
76
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
265 def append(self, e):
5008
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
266 self.p.index.append(_pack(indexformatng, *e))
515
03f27b1381f9 Whitespace cleanups
mpm@selenic.com
parents: 484
diff changeset
267
1559
59b3639df0a9 Convert all classes to new-style classes by deriving them from object.
Eric Hopper <hopper@omnifarious.org>
parents: 1551
diff changeset
268 class lazymap(object):
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
269 """a lazy version of the node map"""
76
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
270 def __init__(self, parser):
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
271 self.p = parser
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
272 def load(self, key):
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
273 n = self.p.findnode(key)
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
274 if n == None:
1214
34706a835d4a Smarter handling of revlog key errors
mpm@selenic.com
parents: 1201
diff changeset
275 raise KeyError(key)
76
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
276 def __contains__(self, key):
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
277 if key in self.p.map:
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
278 return True
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
279 self.p.loadmap()
323
c6f0673ab7e9 lazyparser speed ups
mpm@selenic.com
parents: 306
diff changeset
280 return key in self.p.map
97
7a2abee6b0c2 Add iterator to the lazymap code
mpm@selenic.com
parents: 94
diff changeset
281 def __iter__(self):
469
e205194ca7ef Various node id lookup tweaks
mpm@selenic.com
parents: 451
diff changeset
282 yield nullid
97
7a2abee6b0c2 Add iterator to the lazymap code
mpm@selenic.com
parents: 94
diff changeset
283 for i in xrange(self.p.l):
2080
1cbb14c048cb Reduce index memory usage by storing the bare string instead of tuples
mason@suse.com
parents: 2079
diff changeset
284 ret = self.p.index[i]
1cbb14c048cb Reduce index memory usage by storing the bare string instead of tuples
mason@suse.com
parents: 2079
diff changeset
285 if not ret:
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
286 self.p.loadindex(i)
2080
1cbb14c048cb Reduce index memory usage by storing the bare string instead of tuples
mason@suse.com
parents: 2079
diff changeset
287 ret = self.p.index[i]
1cbb14c048cb Reduce index memory usage by storing the bare string instead of tuples
mason@suse.com
parents: 2079
diff changeset
288 if isinstance(ret, str):
5008
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
289 ret = _unpack(indexformatng, ret)
4978
93d48a8fa496 revlog: change accesses to index entry elements to use positive offsets
Matt Mackall <mpm@selenic.com>
parents: 4977
diff changeset
290 yield ret[7]
76
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
291 def __getitem__(self, key):
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
292 try:
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
293 return self.p.map[key]
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
294 except KeyError:
86
1b945e8ba67b Friendlier exceptions for unknown node errors
mpm@selenic.com
parents: 84
diff changeset
295 try:
1b945e8ba67b Friendlier exceptions for unknown node errors
mpm@selenic.com
parents: 84
diff changeset
296 self.load(key)
1b945e8ba67b Friendlier exceptions for unknown node errors
mpm@selenic.com
parents: 84
diff changeset
297 return self.p.map[key]
1b945e8ba67b Friendlier exceptions for unknown node errors
mpm@selenic.com
parents: 84
diff changeset
298 except KeyError:
1b945e8ba67b Friendlier exceptions for unknown node errors
mpm@selenic.com
parents: 84
diff changeset
299 raise KeyError("node " + hex(key))
76
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
300 def __setitem__(self, key, val):
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
301 self.p.map[key] = val
1535
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
302 def __delitem__(self, key):
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
303 del self.p.map[key]
76
d993ebd69d28 Add lazy{parser,index,map} to speed up processing of index files
mpm@selenic.com
parents: 73
diff changeset
304
4987
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
305 indexformatv0 = ">4l20s20s20s"
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
306 v0shaoffset = 56
4962
e017d3a82e1d revlog: raise offset/type helpers to global scope
Matt Mackall <mpm@selenic.com>
parents: 4781
diff changeset
307
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
308 class revlogoldio(object):
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
309 def __init__(self):
4977
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
310 self.size = struct.calcsize(indexformatv0)
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
311
4985
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
312 def parseindex(self, fp, inline):
4977
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
313 s = self.size
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
314 index = []
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
315 nodemap = {nullid: nullrev}
4973
a386a6e4fe46 revlog: simplify the v0 parser
Matt Mackall <mpm@selenic.com>
parents: 4972
diff changeset
316 n = off = 0
a386a6e4fe46 revlog: simplify the v0 parser
Matt Mackall <mpm@selenic.com>
parents: 4972
diff changeset
317 data = fp.read()
a386a6e4fe46 revlog: simplify the v0 parser
Matt Mackall <mpm@selenic.com>
parents: 4972
diff changeset
318 l = len(data)
a386a6e4fe46 revlog: simplify the v0 parser
Matt Mackall <mpm@selenic.com>
parents: 4972
diff changeset
319 while off + s <= l:
a386a6e4fe46 revlog: simplify the v0 parser
Matt Mackall <mpm@selenic.com>
parents: 4972
diff changeset
320 cur = data[off:off + s]
a386a6e4fe46 revlog: simplify the v0 parser
Matt Mackall <mpm@selenic.com>
parents: 4972
diff changeset
321 off += s
5008
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
322 e = _unpack(indexformatv0, cur)
4977
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
323 # transform to revlogv1 format
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
324 e2 = (offset_type(e[0], 0), e[1], -1, e[2], e[3],
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
325 nodemap[e[4]], nodemap[e[5]], e[6])
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
326 index.append(e2)
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
327 nodemap[e[6]] = n
4973
a386a6e4fe46 revlog: simplify the v0 parser
Matt Mackall <mpm@selenic.com>
parents: 4972
diff changeset
328 n += 1
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
329
4983
4dbcfc6e359e revlog: pull chunkcache back into revlog
Matt Mackall <mpm@selenic.com>
parents: 4982
diff changeset
330 return index, nodemap, None
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
331
5338
f87685355c9c revlog: fix revlogio.packentry corner case
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 5325
diff changeset
332 def packentry(self, entry, node, version, rev):
4986
58cc017ec7e0 revlog: abstract out index entry packing
Matt Mackall <mpm@selenic.com>
parents: 4985
diff changeset
333 e2 = (getoffset(entry[0]), entry[1], entry[3], entry[4],
58cc017ec7e0 revlog: abstract out index entry packing
Matt Mackall <mpm@selenic.com>
parents: 4985
diff changeset
334 node(entry[5]), node(entry[6]), entry[7])
5008
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
335 return _pack(indexformatv0, *e2)
4986
58cc017ec7e0 revlog: abstract out index entry packing
Matt Mackall <mpm@selenic.com>
parents: 4985
diff changeset
336
4987
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
337 # index ng:
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
338 # 6 bytes offset
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
339 # 2 bytes flags
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
340 # 4 bytes compressed length
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
341 # 4 bytes uncompressed length
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
342 # 4 bytes: base rev
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
343 # 4 bytes link rev
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
344 # 4 bytes parent 1 rev
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
345 # 4 bytes parent 2 rev
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
346 # 32 bytes: nodeid
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
347 indexformatng = ">Qiiiiii20s12x"
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
348 ngshaoffset = 32
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
349 versionformat = ">I"
8d30004ada40 revlog: some basic code reordering
Matt Mackall <mpm@selenic.com>
parents: 4986
diff changeset
350
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
351 class revlogio(object):
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
352 def __init__(self):
4977
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
353 self.size = struct.calcsize(indexformatng)
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
354
4985
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
355 def parseindex(self, fp, inline):
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
356 try:
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
357 size = util.fstat(fp).st_size
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
358 except AttributeError:
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
359 size = 0
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
360
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
361 if lazyparser.safe_to_use and not inline and size > 1000000:
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
362 # big index, let's parse it on demand
4985
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
363 parser = lazyparser(fp, size)
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
364 index = lazyindex(parser)
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
365 nodemap = lazymap(parser)
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
366 e = list(index[0])
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
367 type = gettype(e[0])
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
368 e[0] = offset_type(0, type)
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
369 index[0] = e
4983
4dbcfc6e359e revlog: pull chunkcache back into revlog
Matt Mackall <mpm@selenic.com>
parents: 4982
diff changeset
370 return index, nodemap, None
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
371
4977
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
372 s = self.size
4983
4dbcfc6e359e revlog: pull chunkcache back into revlog
Matt Mackall <mpm@selenic.com>
parents: 4982
diff changeset
373 cache = None
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
374 index = []
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
375 nodemap = {nullid: nullrev}
4975
8b7e480a7603 revlog: simplify the v1 immediate parser
Matt Mackall <mpm@selenic.com>
parents: 4974
diff changeset
376 n = off = 0
8b7e480a7603 revlog: simplify the v1 immediate parser
Matt Mackall <mpm@selenic.com>
parents: 4974
diff changeset
377 # if we're not using lazymap, always read the whole index
8b7e480a7603 revlog: simplify the v1 immediate parser
Matt Mackall <mpm@selenic.com>
parents: 4974
diff changeset
378 data = fp.read()
4990
4491125c0f21 revlogio: speed up parsing
Matt Mackall <mpm@selenic.com>
parents: 4989
diff changeset
379 l = len(data) - s
4491125c0f21 revlogio: speed up parsing
Matt Mackall <mpm@selenic.com>
parents: 4989
diff changeset
380 append = index.append
4975
8b7e480a7603 revlog: simplify the v1 immediate parser
Matt Mackall <mpm@selenic.com>
parents: 4974
diff changeset
381 if inline:
4983
4dbcfc6e359e revlog: pull chunkcache back into revlog
Matt Mackall <mpm@selenic.com>
parents: 4982
diff changeset
382 cache = (0, data)
4990
4491125c0f21 revlogio: speed up parsing
Matt Mackall <mpm@selenic.com>
parents: 4989
diff changeset
383 while off <= l:
5008
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
384 e = _unpack(indexformatng, data[off:off + s])
4990
4491125c0f21 revlogio: speed up parsing
Matt Mackall <mpm@selenic.com>
parents: 4989
diff changeset
385 nodemap[e[7]] = n
4491125c0f21 revlogio: speed up parsing
Matt Mackall <mpm@selenic.com>
parents: 4989
diff changeset
386 append(e)
4491125c0f21 revlogio: speed up parsing
Matt Mackall <mpm@selenic.com>
parents: 4989
diff changeset
387 n += 1
4975
8b7e480a7603 revlog: simplify the v1 immediate parser
Matt Mackall <mpm@selenic.com>
parents: 4974
diff changeset
388 if e[1] < 0:
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
389 break
4990
4491125c0f21 revlogio: speed up parsing
Matt Mackall <mpm@selenic.com>
parents: 4989
diff changeset
390 off += e[1] + s
4491125c0f21 revlogio: speed up parsing
Matt Mackall <mpm@selenic.com>
parents: 4989
diff changeset
391 else:
4491125c0f21 revlogio: speed up parsing
Matt Mackall <mpm@selenic.com>
parents: 4989
diff changeset
392 while off <= l:
5008
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
393 e = _unpack(indexformatng, data[off:off + s])
4990
4491125c0f21 revlogio: speed up parsing
Matt Mackall <mpm@selenic.com>
parents: 4989
diff changeset
394 nodemap[e[7]] = n
4491125c0f21 revlogio: speed up parsing
Matt Mackall <mpm@selenic.com>
parents: 4989
diff changeset
395 append(e)
4491125c0f21 revlogio: speed up parsing
Matt Mackall <mpm@selenic.com>
parents: 4989
diff changeset
396 n += 1
4491125c0f21 revlogio: speed up parsing
Matt Mackall <mpm@selenic.com>
parents: 4989
diff changeset
397 off += s
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
398
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
399 e = list(index[0])
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
400 type = gettype(e[0])
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
401 e[0] = offset_type(0, type)
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
402 index[0] = e
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
403
4983
4dbcfc6e359e revlog: pull chunkcache back into revlog
Matt Mackall <mpm@selenic.com>
parents: 4982
diff changeset
404 return index, nodemap, cache
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
405
5338
f87685355c9c revlog: fix revlogio.packentry corner case
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 5325
diff changeset
406 def packentry(self, entry, node, version, rev):
5008
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
407 p = _pack(indexformatng, *entry)
5338
f87685355c9c revlog: fix revlogio.packentry corner case
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 5325
diff changeset
408 if rev == 0:
5008
3addf4531643 revlog: localize some fastpath functions
Matt Mackall <mpm@selenic.com>
parents: 5007
diff changeset
409 p = _pack(versionformat, version) + p[4:]
4986
58cc017ec7e0 revlog: abstract out index entry packing
Matt Mackall <mpm@selenic.com>
parents: 4985
diff changeset
410 return p
58cc017ec7e0 revlog: abstract out index entry packing
Matt Mackall <mpm@selenic.com>
parents: 4985
diff changeset
411
1559
59b3639df0a9 Convert all classes to new-style classes by deriving them from object.
Eric Hopper <hopper@omnifarious.org>
parents: 1551
diff changeset
412 class revlog(object):
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
413 """
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
414 the underlying revision storage object
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
415
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
416 A revlog consists of two parts, an index and the revision data.
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
417
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
418 The index is a file with a fixed record size containing
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
419 information on each revision, includings its nodeid (hash), the
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
420 nodeids of its parents, the position and offset of its data within
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
421 the data file, and the revision it's based on. Finally, each entry
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
422 contains a linkrev entry that can serve as a pointer to external
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
423 data.
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
424
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
425 The revision data itself is a linear collection of data chunks.
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
426 Each chunk represents a revision and is usually represented as a
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
427 delta against the previous chunk. To bound lookup time, runs of
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
428 deltas are limited to about 2 times the length of the original
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
429 version data. This makes retrieval of a version proportional to
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
430 its size, or O(1) relative to the number of revisions.
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
431
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
432 Both pieces of the revlog are written to in an append-only
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
433 fashion, which means we never need to rewrite a file to insert or
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
434 remove data, and can use some simple techniques to avoid the need
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
435 for locking while reading.
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
436 """
4267
b11a2fb59cf5 revlog: simplify revlog version handling
Matt Mackall <mpm@selenic.com>
parents: 4266
diff changeset
437 def __init__(self, opener, indexfile):
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
438 """
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
439 create a revlog object
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
440
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
441 opener is a function that abstracts the file opening operation
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
442 and can be used to implement COW semantics or the like.
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
443 """
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
444 self.indexfile = indexfile
4266
1b5c38e9d7aa revlog: don't pass datafile as an argument
Matt Mackall <mpm@selenic.com>
parents: 4224
diff changeset
445 self.datafile = indexfile[:-2] + ".d"
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
446 self.opener = opener
4984
b4066fcbd6ba revlog: mark cache private
Matt Mackall <mpm@selenic.com>
parents: 4983
diff changeset
447 self._cache = None
4983
4dbcfc6e359e revlog: pull chunkcache back into revlog
Matt Mackall <mpm@selenic.com>
parents: 4982
diff changeset
448 self._chunkcache = None
4985
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
449 self.nodemap = {nullid: nullrev}
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
450 self.index = []
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
451
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
452 v = REVLOG_DEFAULT_VERSION
4267
b11a2fb59cf5 revlog: simplify revlog version handling
Matt Mackall <mpm@selenic.com>
parents: 4266
diff changeset
453 if hasattr(opener, "defversion"):
4985
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
454 v = opener.defversion
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
455 if v & REVLOGNG:
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
456 v |= REVLOGNGINLINEDATA
116
e484cd5ec282 Only use lazy indexing for big indices and avoid the overhead of the
mpm@selenic.com
parents: 115
diff changeset
457
4985
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
458 i = ""
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
459 try:
1784
2e0a288ca93e revalidate revlog data after locking the repo (issue132)
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1749
diff changeset
460 f = self.opener(self.indexfile)
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
461 i = f.read(4)
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
462 f.seek(0)
4962
e017d3a82e1d revlog: raise offset/type helpers to global scope
Matt Mackall <mpm@selenic.com>
parents: 4781
diff changeset
463 if len(i) > 0:
e017d3a82e1d revlog: raise offset/type helpers to global scope
Matt Mackall <mpm@selenic.com>
parents: 4781
diff changeset
464 v = struct.unpack(versionformat, i)[0]
1322
b3d44e9b3092 Make revlog constructor more discerning in its treatment of errors.
Bryan O'Sullivan <bos@serpentine.com>
parents: 1232
diff changeset
465 except IOError, inst:
b3d44e9b3092 Make revlog constructor more discerning in its treatment of errors.
Bryan O'Sullivan <bos@serpentine.com>
parents: 1232
diff changeset
466 if inst.errno != errno.ENOENT:
b3d44e9b3092 Make revlog constructor more discerning in its treatment of errors.
Bryan O'Sullivan <bos@serpentine.com>
parents: 1232
diff changeset
467 raise
4985
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
468
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
469 self.version = v
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
470 self._inline = v & REVLOGNGINLINEDATA
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
471 flags = v & ~0xFFFF
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
472 fmt = v & 0xFFFF
4985
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
473 if fmt == REVLOGV0 and flags:
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
474 raise RevlogError(_("index %s unknown flags %#04x for format v0")
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
475 % (self.indexfile, flags >> 16))
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
476 elif fmt == REVLOGNG and flags & ~REVLOGNGINLINEDATA:
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
477 raise RevlogError(_("index %s unknown flags %#04x for revlogng")
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
478 % (self.indexfile, flags >> 16))
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
479 elif fmt > REVLOGNG:
3744
3a099154b110 Make revlog error slightly less scary
Matt Mackall <mpm@selenic.com>
parents: 3683
diff changeset
480 raise RevlogError(_("index %s unknown format %d")
3680
69cf255a55a1 Indentation cleanups for 2956948b81f3.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3679
diff changeset
481 % (self.indexfile, fmt))
4985
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
482
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
483 self._io = revlogio()
4971
3e6dae278c99 revlog: regroup parsing code
Matt Mackall <mpm@selenic.com>
parents: 4964
diff changeset
484 if self.version == REVLOGV0:
4972
8d0cf46e0dc6 revlog: add revlogio interface
Matt Mackall <mpm@selenic.com>
parents: 4971
diff changeset
485 self._io = revlogoldio()
2072
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
486 if i:
4985
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
487 d = self._io.parseindex(f, self._inline)
4983
4dbcfc6e359e revlog: pull chunkcache back into revlog
Matt Mackall <mpm@selenic.com>
parents: 4982
diff changeset
488 self.index, self.nodemap, self._chunkcache = d
4985
e6525e459157 revlog: simplify revlog.__init__
Matt Mackall <mpm@selenic.com>
parents: 4984
diff changeset
489
4979
06abdaf78788 revlog: add a magic null revision to our index
Matt Mackall <mpm@selenic.com>
parents: 4978
diff changeset
490 # add the magic null revision at -1
06abdaf78788 revlog: add a magic null revision to our index
Matt Mackall <mpm@selenic.com>
parents: 4978
diff changeset
491 self.index.append((0, 0, 0, -1, -1, -1, -1, nullid))
116
e484cd5ec282 Only use lazy indexing for big indices and avoid the overhead of the
mpm@selenic.com
parents: 115
diff changeset
492
4964
ee983d0dbea8 revlog: privatize some methods
Matt Mackall <mpm@selenic.com>
parents: 4963
diff changeset
493 def _loadindex(self, start, end):
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
494 """load a block of indexes all at once from the lazy parser"""
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
495 if isinstance(self.index, lazyindex):
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
496 self.index.p.loadindex(start, end)
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
497
4964
ee983d0dbea8 revlog: privatize some methods
Matt Mackall <mpm@selenic.com>
parents: 4963
diff changeset
498 def _loadindexmap(self):
2072
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
499 """loads both the map and the index from the lazy parser"""
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
500 if isinstance(self.index, lazyindex):
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
501 p = self.index.p
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
502 p.loadindex()
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
503 self.nodemap = p.map
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
504
4964
ee983d0dbea8 revlog: privatize some methods
Matt Mackall <mpm@selenic.com>
parents: 4963
diff changeset
505 def _loadmap(self):
2079
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
506 """loads the map from the lazy parser"""
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
507 if isinstance(self.nodemap, lazymap):
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
508 self.nodemap.p.loadmap()
ee96ca273f32 New lazy index code for revlogs.
mason@suse.com
parents: 2078
diff changeset
509 self.nodemap = self.nodemap.p.map
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
510
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
511 def tip(self):
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
512 return self.node(len(self.index) - 2)
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
513 def count(self):
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
514 return len(self.index) - 1
4964
ee983d0dbea8 revlog: privatize some methods
Matt Mackall <mpm@selenic.com>
parents: 4963
diff changeset
515
1201
59bfbdbc38f6 revlog: raise informative exception if file is missing.
Bryan O'Sullivan <bos@serpentine.com>
parents: 1099
diff changeset
516 def rev(self, node):
59bfbdbc38f6 revlog: raise informative exception if file is missing.
Bryan O'Sullivan <bos@serpentine.com>
parents: 1099
diff changeset
517 try:
59bfbdbc38f6 revlog: raise informative exception if file is missing.
Bryan O'Sullivan <bos@serpentine.com>
parents: 1099
diff changeset
518 return self.nodemap[node]
59bfbdbc38f6 revlog: raise informative exception if file is missing.
Bryan O'Sullivan <bos@serpentine.com>
parents: 1099
diff changeset
519 except KeyError:
3930
01d98d68d697 Add revlog.LookupError exception, and use it instead of RevlogError.
Brendan Cully <brendan@kublai.com>
parents: 3928
diff changeset
520 raise LookupError(_('%s: no node %s') % (self.indexfile, hex(node)))
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
521 def node(self, rev):
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
522 return self.index[rev][7]
2651
6414ee2eb688 correct the handling of linkrev with nullid
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 2650
diff changeset
523 def linkrev(self, node):
4979
06abdaf78788 revlog: add a magic null revision to our index
Matt Mackall <mpm@selenic.com>
parents: 4978
diff changeset
524 return self.index[self.rev(node)][4]
2
ecf3fd948051 Handle nullid better for ancestor
mpm@selenic.com
parents: 0
diff changeset
525 def parents(self, node):
4979
06abdaf78788 revlog: add a magic null revision to our index
Matt Mackall <mpm@selenic.com>
parents: 4978
diff changeset
526 d = self.index[self.rev(node)][5:7]
3505
0aef94f45ebf revlog.py: always return tuples from parents and parentrevs
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3499
diff changeset
527 return (self.node(d[0]), self.node(d[1]))
2489
568e58eed096 Add revlog.parentrevs function.
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 2354
diff changeset
528 def parentrevs(self, rev):
4979
06abdaf78788 revlog: add a magic null revision to our index
Matt Mackall <mpm@selenic.com>
parents: 4978
diff changeset
529 return self.index[rev][5:7]
2072
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
530 def start(self, rev):
5007
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
531 return int(self.index[rev][0] >> 16)
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
532 def end(self, rev):
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
533 return self.start(rev) + self.length(rev)
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
534 def length(self, rev):
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
535 return self.index[rev][1]
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
536 def base(self, rev):
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
537 return self.index[rev][3]
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
538
2078
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
539 def size(self, rev):
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
540 """return the length of the uncompressed text for a given revision"""
4977
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
541 l = self.index[rev][2]
2078
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
542 if l >= 0:
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
543 return l
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
544
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
545 t = self.revision(self.node(rev))
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
546 return len(t)
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
547
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
548 # alternate implementation, The advantage to this code is it
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
549 # will be faster for a single revision. But, the results are not
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
550 # cached, so finding the size of every revision will be slower.
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
551 """
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
552 if self.cache and self.cache[1] == rev:
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
553 return len(self.cache[2])
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
554
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
555 base = self.base(rev)
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
556 if self.cache and self.cache[1] >= base and self.cache[1] < rev:
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
557 base = self.cache[1]
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
558 text = self.cache[2]
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
559 else:
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
560 text = self.revision(self.node(base))
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
561
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
562 l = len(text)
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
563 for x in xrange(base + 1, rev + 1):
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
564 l = mdiff.patchedsize(l, self.chunk(x))
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
565 return l
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
566 """
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
567
3633
508036290b00 revlog: reachable actually takes a node
Matt Mackall <mpm@selenic.com>
parents: 3585
diff changeset
568 def reachable(self, node, stop=None):
3683
b0fd43effdb3 add docstring to reachable
Matt Mackall <mpm@selenic.com>
parents: 3680
diff changeset
569 """return a hash of all nodes ancestral to a given node, including
b0fd43effdb3 add docstring to reachable
Matt Mackall <mpm@selenic.com>
parents: 3680
diff changeset
570 the node itself, stopping when stop is matched"""
1074
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
571 reachable = {}
3633
508036290b00 revlog: reachable actually takes a node
Matt Mackall <mpm@selenic.com>
parents: 3585
diff changeset
572 visit = [node]
508036290b00 revlog: reachable actually takes a node
Matt Mackall <mpm@selenic.com>
parents: 3585
diff changeset
573 reachable[node] = 1
1074
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
574 if stop:
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
575 stopn = self.rev(stop)
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
576 else:
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
577 stopn = 0
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
578 while visit:
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
579 n = visit.pop(0)
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
580 if n == stop:
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
581 continue
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
582 if n == nullid:
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
583 continue
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
584 for p in self.parents(n):
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
585 if self.rev(p) < stopn:
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
586 continue
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
587 if p not in reachable:
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
588 reachable[p] = 1
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
589 visit.append(p)
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
590 return reachable
55bf5cfde69e Add revlog.reachable to find a graph of ancestors for a given rev
mason@suse.com
parents: 1073
diff changeset
591
1457
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
592 def nodesbetween(self, roots=None, heads=None):
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
593 """Return a tuple containing three elements. Elements 1 and 2 contain
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
594 a final list bases and heads after all the unreachable ones have been
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
595 pruned. Element 0 contains a topologically sorted list of all
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
596
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
597 nodes that satisfy these constraints:
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
598 1. All nodes must be descended from a node in roots (the nodes on
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
599 roots are considered descended from themselves).
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
600 2. All nodes must also be ancestors of a node in heads (the nodes in
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
601 heads are considered to be their own ancestors).
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
602
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
603 If roots is unspecified, nullid is assumed as the only root.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
604 If heads is unspecified, it is taken to be the output of the
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
605 heads method (i.e. a list of all nodes in the repository that
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
606 have no children)."""
1463
26e73acc0cdf Fix to handle case of empty list for roots or heads in nodesbetween.
Eric Hopper <hopper@omnifarious.org>
parents: 1459
diff changeset
607 nonodes = ([], [], [])
1457
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
608 if roots is not None:
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
609 roots = list(roots)
1463
26e73acc0cdf Fix to handle case of empty list for roots or heads in nodesbetween.
Eric Hopper <hopper@omnifarious.org>
parents: 1459
diff changeset
610 if not roots:
26e73acc0cdf Fix to handle case of empty list for roots or heads in nodesbetween.
Eric Hopper <hopper@omnifarious.org>
parents: 1459
diff changeset
611 return nonodes
1457
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
612 lowestrev = min([self.rev(n) for n in roots])
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
613 else:
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
614 roots = [nullid] # Everybody's a descendent of nullid
3578
3b4e00cba57a Define and use nullrev (revision of nullid) instead of -1.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3505
diff changeset
615 lowestrev = nullrev
3b4e00cba57a Define and use nullrev (revision of nullid) instead of -1.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3505
diff changeset
616 if (lowestrev == nullrev) and (heads is None):
1457
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
617 # We want _all_ the nodes!
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
618 return ([self.node(r) for r in xrange(0, self.count())],
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
619 [nullid], list(self.heads()))
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
620 if heads is None:
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
621 # All nodes are ancestors, so the latest ancestor is the last
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
622 # node.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
623 highestrev = self.count() - 1
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
624 # Set ancestors to None to signal that every node is an ancestor.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
625 ancestors = None
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
626 # Set heads to an empty dictionary for later discovery of heads
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
627 heads = {}
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
628 else:
1463
26e73acc0cdf Fix to handle case of empty list for roots or heads in nodesbetween.
Eric Hopper <hopper@omnifarious.org>
parents: 1459
diff changeset
629 heads = list(heads)
26e73acc0cdf Fix to handle case of empty list for roots or heads in nodesbetween.
Eric Hopper <hopper@omnifarious.org>
parents: 1459
diff changeset
630 if not heads:
26e73acc0cdf Fix to handle case of empty list for roots or heads in nodesbetween.
Eric Hopper <hopper@omnifarious.org>
parents: 1459
diff changeset
631 return nonodes
1457
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
632 ancestors = {}
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
633 # Turn heads into a dictionary so we can remove 'fake' heads.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
634 # Also, later we will be using it to filter out the heads we can't
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
635 # find from roots.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
636 heads = dict.fromkeys(heads, 0)
3360
ef8307585b41 nodesbetween: fix a bug with duplicate heads
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3333
diff changeset
637 # Start at the top and keep marking parents until we're done.
ef8307585b41 nodesbetween: fix a bug with duplicate heads
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3333
diff changeset
638 nodestotag = heads.keys()
1457
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
639 # Remember where the top was so we can use it as a limit later.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
640 highestrev = max([self.rev(n) for n in nodestotag])
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
641 while nodestotag:
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
642 # grab a node to tag
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
643 n = nodestotag.pop()
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
644 # Never tag nullid
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
645 if n == nullid:
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
646 continue
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
647 # A node's revision number represents its place in a
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
648 # topologically sorted list of nodes.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
649 r = self.rev(n)
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
650 if r >= lowestrev:
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
651 if n not in ancestors:
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
652 # If we are possibly a descendent of one of the roots
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
653 # and we haven't already been marked as an ancestor
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
654 ancestors[n] = 1 # Mark as ancestor
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
655 # Add non-nullid parents to list of nodes to tag.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
656 nodestotag.extend([p for p in self.parents(n) if
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
657 p != nullid])
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
658 elif n in heads: # We've seen it before, is it a fake head?
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
659 # So it is, real heads should not be the ancestors of
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
660 # any other heads.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
661 heads.pop(n)
1459
106fdec8e1fb Fix small bug in nodesbetween if heads is [nullid].
Eric Hopper <hopper@omnifarious.org>
parents: 1458
diff changeset
662 if not ancestors:
1463
26e73acc0cdf Fix to handle case of empty list for roots or heads in nodesbetween.
Eric Hopper <hopper@omnifarious.org>
parents: 1459
diff changeset
663 return nonodes
1457
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
664 # Now that we have our set of ancestors, we want to remove any
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
665 # roots that are not ancestors.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
666
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
667 # If one of the roots was nullid, everything is included anyway.
3578
3b4e00cba57a Define and use nullrev (revision of nullid) instead of -1.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3505
diff changeset
668 if lowestrev > nullrev:
1457
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
669 # But, since we weren't, let's recompute the lowest rev to not
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
670 # include roots that aren't ancestors.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
671
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
672 # Filter out roots that aren't ancestors of heads
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
673 roots = [n for n in roots if n in ancestors]
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
674 # Recompute the lowest revision
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
675 if roots:
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
676 lowestrev = min([self.rev(n) for n in roots])
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
677 else:
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
678 # No more roots? Return empty list
1463
26e73acc0cdf Fix to handle case of empty list for roots or heads in nodesbetween.
Eric Hopper <hopper@omnifarious.org>
parents: 1459
diff changeset
679 return nonodes
1457
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
680 else:
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
681 # We are descending from nullid, and don't need to care about
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
682 # any other roots.
3578
3b4e00cba57a Define and use nullrev (revision of nullid) instead of -1.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3505
diff changeset
683 lowestrev = nullrev
1457
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
684 roots = [nullid]
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
685 # Transform our roots list into a 'set' (i.e. a dictionary where the
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
686 # values don't matter.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
687 descendents = dict.fromkeys(roots, 1)
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
688 # Also, keep the original roots so we can filter out roots that aren't
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
689 # 'real' roots (i.e. are descended from other roots).
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
690 roots = descendents.copy()
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
691 # Our topologically sorted list of output nodes.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
692 orderedout = []
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
693 # Don't start at nullid since we don't want nullid in our output list,
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
694 # and if nullid shows up in descedents, empty parents will look like
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
695 # they're descendents.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
696 for r in xrange(max(lowestrev, 0), highestrev + 1):
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
697 n = self.node(r)
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
698 isdescendent = False
3578
3b4e00cba57a Define and use nullrev (revision of nullid) instead of -1.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3505
diff changeset
699 if lowestrev == nullrev: # Everybody is a descendent of nullid
1457
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
700 isdescendent = True
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
701 elif n in descendents:
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
702 # n is already a descendent
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
703 isdescendent = True
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
704 # This check only needs to be done here because all the roots
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
705 # will start being marked is descendents before the loop.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
706 if n in roots:
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
707 # If n was a root, check if it's a 'real' root.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
708 p = tuple(self.parents(n))
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
709 # If any of its parents are descendents, it's not a root.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
710 if (p[0] in descendents) or (p[1] in descendents):
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
711 roots.pop(n)
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
712 else:
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
713 p = tuple(self.parents(n))
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
714 # A node is a descendent if either of its parents are
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
715 # descendents. (We seeded the dependents list with the roots
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
716 # up there, remember?)
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
717 if (p[0] in descendents) or (p[1] in descendents):
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
718 descendents[n] = 1
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
719 isdescendent = True
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
720 if isdescendent and ((ancestors is None) or (n in ancestors)):
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
721 # Only include nodes that are both descendents and ancestors.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
722 orderedout.append(n)
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
723 if (ancestors is not None) and (n in heads):
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
724 # We're trying to figure out which heads are reachable
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
725 # from roots.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
726 # Mark this head as having been reached
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
727 heads[n] = 1
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
728 elif ancestors is None:
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
729 # Otherwise, we're trying to discover the heads.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
730 # Assume this is a head because if it isn't, the next step
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
731 # will eventually remove it.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
732 heads[n] = 1
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
733 # But, obviously its parents aren't.
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
734 for p in self.parents(n):
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
735 heads.pop(p, None)
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
736 heads = [n for n in heads.iterkeys() if heads[n] != 0]
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
737 roots = roots.keys()
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
738 assert orderedout
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
739 assert roots
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
740 assert heads
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
741 return (orderedout, roots, heads)
518da3c3b6ce This implements the nodesbetween method, and it removes the newer method
Eric Hopper <hopper@omnifarious.org>
parents: 1351
diff changeset
742
3925
27230c29bfec fix calculation of new heads added during push with -r
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3755
diff changeset
743 def heads(self, start=None, stop=None):
1550
ccb9b62de892 add a -r/--rev option to heads to show only heads descendant from rev
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1535
diff changeset
744 """return the list of all nodes that have no children
1551
e793cbc8be00 Fixes to "hg heads -r FOO":
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1550
diff changeset
745
e793cbc8be00 Fixes to "hg heads -r FOO":
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1550
diff changeset
746 if start is specified, only heads that are descendants of
e793cbc8be00 Fixes to "hg heads -r FOO":
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1550
diff changeset
747 start will be returned
3925
27230c29bfec fix calculation of new heads added during push with -r
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3755
diff changeset
748 if stop is specified, it will consider all the revs from stop
27230c29bfec fix calculation of new heads added during push with -r
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3755
diff changeset
749 as if they had no children
1551
e793cbc8be00 Fixes to "hg heads -r FOO":
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1550
diff changeset
750 """
4991
9c8c42bcf17a revlog: implement a fast path for heads
Matt Mackall <mpm@selenic.com>
parents: 4990
diff changeset
751 if start is None and stop is None:
9c8c42bcf17a revlog: implement a fast path for heads
Matt Mackall <mpm@selenic.com>
parents: 4990
diff changeset
752 count = self.count()
9c8c42bcf17a revlog: implement a fast path for heads
Matt Mackall <mpm@selenic.com>
parents: 4990
diff changeset
753 if not count:
9c8c42bcf17a revlog: implement a fast path for heads
Matt Mackall <mpm@selenic.com>
parents: 4990
diff changeset
754 return [nullid]
9c8c42bcf17a revlog: implement a fast path for heads
Matt Mackall <mpm@selenic.com>
parents: 4990
diff changeset
755 ishead = [1] * (count + 1)
9c8c42bcf17a revlog: implement a fast path for heads
Matt Mackall <mpm@selenic.com>
parents: 4990
diff changeset
756 index = self.index
9c8c42bcf17a revlog: implement a fast path for heads
Matt Mackall <mpm@selenic.com>
parents: 4990
diff changeset
757 for r in xrange(count):
9c8c42bcf17a revlog: implement a fast path for heads
Matt Mackall <mpm@selenic.com>
parents: 4990
diff changeset
758 e = index[r]
9c8c42bcf17a revlog: implement a fast path for heads
Matt Mackall <mpm@selenic.com>
parents: 4990
diff changeset
759 ishead[e[5]] = ishead[e[6]] = 0
9c8c42bcf17a revlog: implement a fast path for heads
Matt Mackall <mpm@selenic.com>
parents: 4990
diff changeset
760 return [self.node(r) for r in xrange(count) if ishead[r]]
9c8c42bcf17a revlog: implement a fast path for heads
Matt Mackall <mpm@selenic.com>
parents: 4990
diff changeset
761
1551
e793cbc8be00 Fixes to "hg heads -r FOO":
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1550
diff changeset
762 if start is None:
e793cbc8be00 Fixes to "hg heads -r FOO":
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1550
diff changeset
763 start = nullid
3925
27230c29bfec fix calculation of new heads added during push with -r
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3755
diff changeset
764 if stop is None:
27230c29bfec fix calculation of new heads added during push with -r
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3755
diff changeset
765 stop = []
27230c29bfec fix calculation of new heads added during push with -r
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3755
diff changeset
766 stoprevs = dict.fromkeys([self.rev(n) for n in stop])
1550
ccb9b62de892 add a -r/--rev option to heads to show only heads descendant from rev
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1535
diff changeset
767 startrev = self.rev(start)
2490
6ff82ec1f4b8 Change revlog.heads to walk the revision graph using revision numbers
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 2489
diff changeset
768 reachable = {startrev: 1}
6ff82ec1f4b8 Change revlog.heads to walk the revision graph using revision numbers
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 2489
diff changeset
769 heads = {startrev: 1}
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
770
2490
6ff82ec1f4b8 Change revlog.heads to walk the revision graph using revision numbers
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 2489
diff changeset
771 parentrevs = self.parentrevs
1550
ccb9b62de892 add a -r/--rev option to heads to show only heads descendant from rev
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1535
diff changeset
772 for r in xrange(startrev + 1, self.count()):
2490
6ff82ec1f4b8 Change revlog.heads to walk the revision graph using revision numbers
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 2489
diff changeset
773 for p in parentrevs(r):
6ff82ec1f4b8 Change revlog.heads to walk the revision graph using revision numbers
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 2489
diff changeset
774 if p in reachable:
3925
27230c29bfec fix calculation of new heads added during push with -r
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3755
diff changeset
775 if r not in stoprevs:
27230c29bfec fix calculation of new heads added during push with -r
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3755
diff changeset
776 reachable[r] = 1
2490
6ff82ec1f4b8 Change revlog.heads to walk the revision graph using revision numbers
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 2489
diff changeset
777 heads[r] = 1
3925
27230c29bfec fix calculation of new heads added during push with -r
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3755
diff changeset
778 if p in heads and p not in stoprevs:
2490
6ff82ec1f4b8 Change revlog.heads to walk the revision graph using revision numbers
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 2489
diff changeset
779 del heads[p]
3925
27230c29bfec fix calculation of new heads added during push with -r
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3755
diff changeset
780
2490
6ff82ec1f4b8 Change revlog.heads to walk the revision graph using revision numbers
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 2489
diff changeset
781 return [self.node(r) for r in heads]
370
c90385d82aec revlog: add a children function
mpm@selenic.com
parents: 330
diff changeset
782
c90385d82aec revlog: add a children function
mpm@selenic.com
parents: 330
diff changeset
783 def children(self, node):
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
784 """find the children of a given node"""
370
c90385d82aec revlog: add a children function
mpm@selenic.com
parents: 330
diff changeset
785 c = []
c90385d82aec revlog: add a children function
mpm@selenic.com
parents: 330
diff changeset
786 p = self.rev(node)
c90385d82aec revlog: add a children function
mpm@selenic.com
parents: 330
diff changeset
787 for r in range(p + 1, self.count()):
4781
62c56d8f368b Fix revlog.children so the real children of the null revision can be calculated.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 4635
diff changeset
788 prevs = [pr for pr in self.parentrevs(r) if pr != nullrev]
62c56d8f368b Fix revlog.children so the real children of the null revision can be calculated.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 4635
diff changeset
789 if prevs:
62c56d8f368b Fix revlog.children so the real children of the null revision can be calculated.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 4635
diff changeset
790 for pr in prevs:
62c56d8f368b Fix revlog.children so the real children of the null revision can be calculated.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 4635
diff changeset
791 if pr == p:
62c56d8f368b Fix revlog.children so the real children of the null revision can be calculated.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 4635
diff changeset
792 c.append(self.node(r))
62c56d8f368b Fix revlog.children so the real children of the null revision can be calculated.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 4635
diff changeset
793 elif p == nullrev:
62c56d8f368b Fix revlog.children so the real children of the null revision can be calculated.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 4635
diff changeset
794 c.append(self.node(r))
370
c90385d82aec revlog: add a children function
mpm@selenic.com
parents: 330
diff changeset
795 return c
515
03f27b1381f9 Whitespace cleanups
mpm@selenic.com
parents: 484
diff changeset
796
3494
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
797 def _match(self, id):
3210
7240f9e47144 correctly find the type of 'id' in revlog.lookup
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3153
diff changeset
798 if isinstance(id, (long, int)):
3152
d01e4cb2f5f2 cleanups in revlog.lookup
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3139
diff changeset
799 # rev
2650
156fb1feab62 lookup should allow -1 to represent nullid (if passed an int as arg)
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 2600
diff changeset
800 return self.node(id)
3438
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
801 if len(id) == 20:
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
802 # possibly a binary node
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
803 # odds of a binary node being all hex in ASCII are 1 in 10**25
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
804 try:
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
805 node = id
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
806 r = self.rev(node) # quick search the index
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
807 return node
3930
01d98d68d697 Add revlog.LookupError exception, and use it instead of RevlogError.
Brendan Cully <brendan@kublai.com>
parents: 3928
diff changeset
808 except LookupError:
3438
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
809 pass # may be partial hex id
36
da28286bf6b7 Add smart node lookup by substring or by rev number
mpm@selenic.com
parents: 26
diff changeset
810 try:
3152
d01e4cb2f5f2 cleanups in revlog.lookup
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3139
diff changeset
811 # str(rev)
36
da28286bf6b7 Add smart node lookup by substring or by rev number
mpm@selenic.com
parents: 26
diff changeset
812 rev = int(id)
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
813 if str(rev) != id:
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
814 raise ValueError
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
815 if rev < 0:
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
816 rev = self.count() + rev
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
817 if rev < 0 or rev >= self.count():
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
818 raise ValueError
36
da28286bf6b7 Add smart node lookup by substring or by rev number
mpm@selenic.com
parents: 26
diff changeset
819 return self.node(rev)
469
e205194ca7ef Various node id lookup tweaks
mpm@selenic.com
parents: 451
diff changeset
820 except (ValueError, OverflowError):
3152
d01e4cb2f5f2 cleanups in revlog.lookup
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3139
diff changeset
821 pass
3494
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
822 if len(id) == 40:
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
823 try:
3438
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
824 # a full hex nodeid?
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
825 node = bin(id)
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
826 r = self.rev(node)
3153
4fe41a9e4591 optimize revlog.lookup when passed hex(node)[:...]
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 3152
diff changeset
827 return node
3494
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
828 except TypeError:
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
829 pass
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
830
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
831 def _partialmatch(self, id):
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
832 if len(id) < 40:
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
833 try:
3438
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
834 # hex(node)[:...]
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
835 bin_id = bin(id[:len(id) & ~1]) # grab an even number of digits
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
836 node = None
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
837 for n in self.nodemap:
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
838 if n.startswith(bin_id) and hex(n).startswith(id):
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
839 if node is not None:
3930
01d98d68d697 Add revlog.LookupError exception, and use it instead of RevlogError.
Brendan Cully <brendan@kublai.com>
parents: 3928
diff changeset
840 raise LookupError(_("Ambiguous identifier"))
3438
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
841 node = n
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
842 if node is not None:
b17f9d3eda74 revlog.lookup tweaks
Matt Mackall <mpm@selenic.com>
parents: 3390
diff changeset
843 return node
3494
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
844 except TypeError:
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
845 pass
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
846
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
847 def lookup(self, id):
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
848 """locate a node based on:
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
849 - revision number or str(revision number)
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
850 - nodeid or subset of hex nodeid
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
851 """
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
852 n = self._match(id)
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
853 if n is not None:
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
854 return n
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
855 n = self._partialmatch(id)
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
856 if n:
dba3cadef789 Only look up tags and branches as a last resort
Matt Mackall <mpm@selenic.com>
parents: 3438
diff changeset
857 return n
515
03f27b1381f9 Whitespace cleanups
mpm@selenic.com
parents: 484
diff changeset
858
3930
01d98d68d697 Add revlog.LookupError exception, and use it instead of RevlogError.
Brendan Cully <brendan@kublai.com>
parents: 3928
diff changeset
859 raise LookupError(_("No match found"))
36
da28286bf6b7 Add smart node lookup by substring or by rev number
mpm@selenic.com
parents: 26
diff changeset
860
2910
5df3e5cf16bc Move cmp bits from filelog to revlog
Matt Mackall <mpm@selenic.com>
parents: 2858
diff changeset
861 def cmp(self, node, text):
5df3e5cf16bc Move cmp bits from filelog to revlog
Matt Mackall <mpm@selenic.com>
parents: 2858
diff changeset
862 """compare text with a given file revision"""
5df3e5cf16bc Move cmp bits from filelog to revlog
Matt Mackall <mpm@selenic.com>
parents: 2858
diff changeset
863 p1, p2 = self.parents(node)
5df3e5cf16bc Move cmp bits from filelog to revlog
Matt Mackall <mpm@selenic.com>
parents: 2858
diff changeset
864 return hash(text, p1, p2) != node
5df3e5cf16bc Move cmp bits from filelog to revlog
Matt Mackall <mpm@selenic.com>
parents: 2858
diff changeset
865
4988
14486eea8e7a revlog: speed up chunkcache
Matt Mackall <mpm@selenic.com>
parents: 4987
diff changeset
866 def chunk(self, rev, df=None):
2072
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
867 def loadcache(df):
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
868 if not df:
4988
14486eea8e7a revlog: speed up chunkcache
Matt Mackall <mpm@selenic.com>
parents: 4987
diff changeset
869 if self._inline:
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
870 df = self.opener(self.indexfile)
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
871 else:
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
872 df = self.opener(self.datafile)
1598
14d1f1868bf6 cleanup of revlog.group when repository is local
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1559
diff changeset
873 df.seek(start)
4983
4dbcfc6e359e revlog: pull chunkcache back into revlog
Matt Mackall <mpm@selenic.com>
parents: 4982
diff changeset
874 self._chunkcache = (start, df.read(cache_length))
1598
14d1f1868bf6 cleanup of revlog.group when repository is local
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1559
diff changeset
875
5007
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
876 start, length = self.start(rev), self.length(rev)
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
877 if self._inline:
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
878 start += (rev + 1) * self._io.size
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
879 end = start + length
1598
14d1f1868bf6 cleanup of revlog.group when repository is local
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1559
diff changeset
880
5007
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
881 offset = 0
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
882 if not self._chunkcache:
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
883 cache_length = max(65536, length)
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
884 loadcache(df)
1598
14d1f1868bf6 cleanup of revlog.group when repository is local
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1559
diff changeset
885 else:
5007
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
886 cache_start = self._chunkcache[0]
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
887 cache_length = len(self._chunkcache[1])
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
888 cache_end = cache_start + cache_length
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
889 if start >= cache_start and end <= cache_end:
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
890 # it is cached
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
891 offset = start - cache_start
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
892 else:
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
893 cache_length = max(65536, length)
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
894 loadcache(df)
1598
14d1f1868bf6 cleanup of revlog.group when repository is local
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1559
diff changeset
895
4988
14486eea8e7a revlog: speed up chunkcache
Matt Mackall <mpm@selenic.com>
parents: 4987
diff changeset
896 # avoid copying large chunks
14486eea8e7a revlog: speed up chunkcache
Matt Mackall <mpm@selenic.com>
parents: 4987
diff changeset
897 c = self._chunkcache[1]
5007
c2febf5420e9 revlog: minor chunk speed-up
Matt Mackall <mpm@selenic.com>
parents: 5006
diff changeset
898 if cache_length != length:
4988
14486eea8e7a revlog: speed up chunkcache
Matt Mackall <mpm@selenic.com>
parents: 4987
diff changeset
899 c = c[offset:offset + length]
14486eea8e7a revlog: speed up chunkcache
Matt Mackall <mpm@selenic.com>
parents: 4987
diff changeset
900
14486eea8e7a revlog: speed up chunkcache
Matt Mackall <mpm@selenic.com>
parents: 4987
diff changeset
901 return decompress(c)
1598
14d1f1868bf6 cleanup of revlog.group when repository is local
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1559
diff changeset
902
119
c7a66f9752a4 Add code to retrieve or construct a revlog delta
mpm@selenic.com
parents: 117
diff changeset
903 def delta(self, node):
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
904 """return or calculate a delta between a node and its predecessor"""
119
c7a66f9752a4 Add code to retrieve or construct a revlog delta
mpm@selenic.com
parents: 117
diff changeset
905 r = self.rev(node)
1941
7518823709a2 revlog.py: factorization and fixes for rev < 0 (nullid)
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1853
diff changeset
906 return self.revdiff(r - 1, r)
7518823709a2 revlog.py: factorization and fixes for rev < 0 (nullid)
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1853
diff changeset
907
7518823709a2 revlog.py: factorization and fixes for rev < 0 (nullid)
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1853
diff changeset
908 def revdiff(self, rev1, rev2):
7518823709a2 revlog.py: factorization and fixes for rev < 0 (nullid)
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1853
diff changeset
909 """return or calculate a delta between two revisions"""
5006
72082bfced9a revlog: minor revdiff reorganization
Matt Mackall <mpm@selenic.com>
parents: 5005
diff changeset
910 if rev1 + 1 == rev2 and self.base(rev1) == self.base(rev2):
1941
7518823709a2 revlog.py: factorization and fixes for rev < 0 (nullid)
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1853
diff changeset
911 return self.chunk(rev2)
5006
72082bfced9a revlog: minor revdiff reorganization
Matt Mackall <mpm@selenic.com>
parents: 5005
diff changeset
912
72082bfced9a revlog: minor revdiff reorganization
Matt Mackall <mpm@selenic.com>
parents: 5005
diff changeset
913 return mdiff.textdiff(self.revision(self.node(rev1)),
72082bfced9a revlog: minor revdiff reorganization
Matt Mackall <mpm@selenic.com>
parents: 5005
diff changeset
914 self.revision(self.node(rev2)))
119
c7a66f9752a4 Add code to retrieve or construct a revlog delta
mpm@selenic.com
parents: 117
diff changeset
915
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
916 def revision(self, node):
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
917 """return an uncompressed revision of a given"""
4980
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
918 if node == nullid:
fc44c8df9d99 revlog: some codingstyle cleanups
Matt Mackall <mpm@selenic.com>
parents: 4979
diff changeset
919 return ""
4984
b4066fcbd6ba revlog: mark cache private
Matt Mackall <mpm@selenic.com>
parents: 4983
diff changeset
920 if self._cache and self._cache[0] == node:
5450
c728424d44c6 revlog: fix caching of buffer objects
Matt Mackall <mpm@selenic.com>
parents: 5448
diff changeset
921 return str(self._cache[2])
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
922
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
923 # look up what we need to read
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
924 text = None
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
925 rev = self.rev(node)
1598
14d1f1868bf6 cleanup of revlog.group when repository is local
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1559
diff changeset
926 base = self.base(rev)
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
927
5005
825516d16b25 revlog: move flag checking out of the offset fastpath
Matt Mackall <mpm@selenic.com>
parents: 4996
diff changeset
928 # check rev flags
825516d16b25 revlog: move flag checking out of the offset fastpath
Matt Mackall <mpm@selenic.com>
parents: 4996
diff changeset
929 if self.index[rev][0] & 0xFFFF:
5322
fb070713ff36 revlog: more robust for damaged indexes
Matt Mackall <mpm@selenic.com>
parents: 5008
diff changeset
930 raise RevlogError(_('incompatible revision flag %x') %
fb070713ff36 revlog: more robust for damaged indexes
Matt Mackall <mpm@selenic.com>
parents: 5008
diff changeset
931 (self.index[rev][0] & 0xFFFF))
5005
825516d16b25 revlog: move flag checking out of the offset fastpath
Matt Mackall <mpm@selenic.com>
parents: 4996
diff changeset
932
4982
9672e3c42b0c revlog: change _inline from a function to a variable
Matt Mackall <mpm@selenic.com>
parents: 4981
diff changeset
933 if self._inline:
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
934 # we probably have the whole chunk cached
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
935 df = None
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
936 else:
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
937 df = self.opener(self.datafile)
2072
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
938
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
939 # do we have useful data cached?
4984
b4066fcbd6ba revlog: mark cache private
Matt Mackall <mpm@selenic.com>
parents: 4983
diff changeset
940 if self._cache and self._cache[1] >= base and self._cache[1] < rev:
b4066fcbd6ba revlog: mark cache private
Matt Mackall <mpm@selenic.com>
parents: 4983
diff changeset
941 base = self._cache[1]
5450
c728424d44c6 revlog: fix caching of buffer objects
Matt Mackall <mpm@selenic.com>
parents: 5448
diff changeset
942 text = str(self._cache[2])
4964
ee983d0dbea8 revlog: privatize some methods
Matt Mackall <mpm@selenic.com>
parents: 4963
diff changeset
943 self._loadindex(base, rev + 1)
1598
14d1f1868bf6 cleanup of revlog.group when repository is local
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1559
diff changeset
944 else:
4964
ee983d0dbea8 revlog: privatize some methods
Matt Mackall <mpm@selenic.com>
parents: 4963
diff changeset
945 self._loadindex(base, rev + 1)
2072
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
946 text = self.chunk(base, df=df)
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
947
4989
1aaed3d69772 revlog: eliminate diff and patches functions
Matt Mackall <mpm@selenic.com>
parents: 4988
diff changeset
948 bins = [self.chunk(r, df) for r in xrange(base + 1, rev + 1)]
1aaed3d69772 revlog: eliminate diff and patches functions
Matt Mackall <mpm@selenic.com>
parents: 4988
diff changeset
949 text = mdiff.patches(text, bins)
1598
14d1f1868bf6 cleanup of revlog.group when repository is local
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1559
diff changeset
950 p1, p2 = self.parents(node)
26
9cf83bf9ad38 Simplify integrity checking
mpm@selenic.com
parents: 14
diff changeset
951 if node != hash(text, p1, p2):
1402
9d2c2e6b32b5 i18n part2: use '_' for all strings who are part of the user interface
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1400
diff changeset
952 raise RevlogError(_("integrity check failed on %s:%d")
3680
69cf255a55a1 Indentation cleanups for 2956948b81f3.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3679
diff changeset
953 % (self.datafile, rev))
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
954
4984
b4066fcbd6ba revlog: mark cache private
Matt Mackall <mpm@selenic.com>
parents: 4983
diff changeset
955 self._cache = (node, rev, text)
515
03f27b1381f9 Whitespace cleanups
mpm@selenic.com
parents: 484
diff changeset
956 return text
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
957
2075
343aeefb553b Make the appendfile class inline-data index friendly
mason@suse.com
parents: 2073
diff changeset
958 def checkinlinesize(self, tr, fp=None):
4982
9672e3c42b0c revlog: change _inline from a function to a variable
Matt Mackall <mpm@selenic.com>
parents: 4981
diff changeset
959 if not self._inline:
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
960 return
2075
343aeefb553b Make the appendfile class inline-data index friendly
mason@suse.com
parents: 2073
diff changeset
961 if not fp:
343aeefb553b Make the appendfile class inline-data index friendly
mason@suse.com
parents: 2073
diff changeset
962 fp = self.opener(self.indexfile, 'r')
2082
856f0ba200bc Additional appendfile fixes for interleaved data/index files
mason@suse.com
parents: 2081
diff changeset
963 fp.seek(0, 2)
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
964 size = fp.tell()
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
965 if size < 131072:
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
966 return
2084
Chris Mason <mason@suse.com>
parents: 2082
diff changeset
967 trinfo = tr.find(self.indexfile)
Chris Mason <mason@suse.com>
parents: 2082
diff changeset
968 if trinfo == None:
3680
69cf255a55a1 Indentation cleanups for 2956948b81f3.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3679
diff changeset
969 raise RevlogError(_("%s not found in the transaction")
69cf255a55a1 Indentation cleanups for 2956948b81f3.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3679
diff changeset
970 % self.indexfile)
2084
Chris Mason <mason@suse.com>
parents: 2082
diff changeset
971
Chris Mason <mason@suse.com>
parents: 2082
diff changeset
972 trindex = trinfo[2]
Chris Mason <mason@suse.com>
parents: 2082
diff changeset
973 dataoff = self.start(trindex)
Chris Mason <mason@suse.com>
parents: 2082
diff changeset
974
Chris Mason <mason@suse.com>
parents: 2082
diff changeset
975 tr.add(self.datafile, dataoff)
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
976 df = self.opener(self.datafile, 'w')
4977
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
977 calc = self._io.size
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
978 for r in xrange(self.count()):
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
979 start = self.start(r) + (r + 1) * calc
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
980 length = self.length(r)
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
981 fp.seek(start)
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
982 d = fp.read(length)
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
983 df.write(d)
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
984 fp.close()
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
985 df.close()
2076
d007df6daf8e Create an atomic opener that does not automatically rename on close
mason@suse.com
parents: 2075
diff changeset
986 fp = self.opener(self.indexfile, 'w', atomictemp=True)
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
987 self.version &= ~(REVLOGNGINLINEDATA)
4982
9672e3c42b0c revlog: change _inline from a function to a variable
Matt Mackall <mpm@selenic.com>
parents: 4981
diff changeset
988 self._inline = False
4986
58cc017ec7e0 revlog: abstract out index entry packing
Matt Mackall <mpm@selenic.com>
parents: 4985
diff changeset
989 for i in xrange(self.count()):
5338
f87685355c9c revlog: fix revlogio.packentry corner case
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 5325
diff changeset
990 e = self._io.packentry(self.index[i], self.node, self.version, i)
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
991 fp.write(e)
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
992
2076
d007df6daf8e Create an atomic opener that does not automatically rename on close
mason@suse.com
parents: 2075
diff changeset
993 # if we don't call rename, the temp file will never replace the
d007df6daf8e Create an atomic opener that does not automatically rename on close
mason@suse.com
parents: 2075
diff changeset
994 # real index
d007df6daf8e Create an atomic opener that does not automatically rename on close
mason@suse.com
parents: 2075
diff changeset
995 fp.rename()
2084
Chris Mason <mason@suse.com>
parents: 2082
diff changeset
996
Chris Mason <mason@suse.com>
parents: 2082
diff changeset
997 tr.replace(self.indexfile, trindex * calc)
4983
4dbcfc6e359e revlog: pull chunkcache back into revlog
Matt Mackall <mpm@selenic.com>
parents: 4982
diff changeset
998 self._chunkcache = None
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
999
4981
e7131935fbb3 revlog: simplify addrevision
Matt Mackall <mpm@selenic.com>
parents: 4980
diff changeset
1000 def addrevision(self, text, transaction, link, p1, p2, d=None):
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1001 """add a revision to the log
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1002
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1003 text - the revision data to add
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1004 transaction - the transaction object used for rollback
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1005 link - the linkrev data to add
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1006 p1, p2 - the parent nodeids of the revision
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1007 d - an optional precomputed delta
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1008 """
4981
e7131935fbb3 revlog: simplify addrevision
Matt Mackall <mpm@selenic.com>
parents: 4980
diff changeset
1009 dfh = None
4982
9672e3c42b0c revlog: change _inline from a function to a variable
Matt Mackall <mpm@selenic.com>
parents: 4981
diff changeset
1010 if not self._inline:
3390
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1011 dfh = self.opener(self.datafile, "a")
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1012 ifh = self.opener(self.indexfile, "a+")
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1013 return self._addrevision(text, transaction, link, p1, p2, d, ifh, dfh)
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1014
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1015 def _addrevision(self, text, transaction, link, p1, p2, d, ifh, dfh):
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
1016 node = hash(text, p1, p2)
301
5add718d92db revlog: allow duplicates
mpm@selenic.com
parents: 243
diff changeset
1017 if node in self.nodemap:
5add718d92db revlog: allow duplicates
mpm@selenic.com
parents: 243
diff changeset
1018 return node
5add718d92db revlog: allow duplicates
mpm@selenic.com
parents: 243
diff changeset
1019
4981
e7131935fbb3 revlog: simplify addrevision
Matt Mackall <mpm@selenic.com>
parents: 4980
diff changeset
1020 curr = self.count()
e7131935fbb3 revlog: simplify addrevision
Matt Mackall <mpm@selenic.com>
parents: 4980
diff changeset
1021 prev = curr - 1
e7131935fbb3 revlog: simplify addrevision
Matt Mackall <mpm@selenic.com>
parents: 4980
diff changeset
1022 base = self.base(prev)
e7131935fbb3 revlog: simplify addrevision
Matt Mackall <mpm@selenic.com>
parents: 4980
diff changeset
1023 offset = self.end(prev)
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
1024
4981
e7131935fbb3 revlog: simplify addrevision
Matt Mackall <mpm@selenic.com>
parents: 4980
diff changeset
1025 if curr:
644
6ebe118280bd Performance enhancements for manifest.add()
mason@suse.com
parents: 547
diff changeset
1026 if not d:
4981
e7131935fbb3 revlog: simplify addrevision
Matt Mackall <mpm@selenic.com>
parents: 4980
diff changeset
1027 ptext = self.revision(self.node(prev))
4989
1aaed3d69772 revlog: eliminate diff and patches functions
Matt Mackall <mpm@selenic.com>
parents: 4988
diff changeset
1028 d = mdiff.textdiff(ptext, text)
98
3dde7c87e36d Add paranoia to diff code
mpm@selenic.com
parents: 97
diff changeset
1029 data = compress(d)
1533
3d11f81c9145 Reduce string duplication in compression code
mason@suse.com
parents: 1509
diff changeset
1030 l = len(data[1]) + len(data[0])
4981
e7131935fbb3 revlog: simplify addrevision
Matt Mackall <mpm@selenic.com>
parents: 4980
diff changeset
1031 dist = l + offset - self.start(base)
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
1032
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
1033 # full versions are inserted when the needed deltas
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
1034 # become comparable to the uncompressed text
4981
e7131935fbb3 revlog: simplify addrevision
Matt Mackall <mpm@selenic.com>
parents: 4980
diff changeset
1035 if not curr or dist > len(text) * 2:
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
1036 data = compress(text)
1533
3d11f81c9145 Reduce string duplication in compression code
mason@suse.com
parents: 1509
diff changeset
1037 l = len(data[1]) + len(data[0])
4981
e7131935fbb3 revlog: simplify addrevision
Matt Mackall <mpm@selenic.com>
parents: 4980
diff changeset
1038 base = curr
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
1039
4977
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
1040 e = (offset_type(offset, 0), l, len(text),
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
1041 base, link, self.rev(p1), self.rev(p2), node)
4979
06abdaf78788 revlog: add a magic null revision to our index
Matt Mackall <mpm@selenic.com>
parents: 4978
diff changeset
1042 self.index.insert(-1, e)
4981
e7131935fbb3 revlog: simplify addrevision
Matt Mackall <mpm@selenic.com>
parents: 4980
diff changeset
1043 self.nodemap[node] = curr
4977
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
1044
5338
f87685355c9c revlog: fix revlogio.packentry corner case
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 5325
diff changeset
1045 entry = self._io.packentry(e, self.node, self.version, curr)
4982
9672e3c42b0c revlog: change _inline from a function to a variable
Matt Mackall <mpm@selenic.com>
parents: 4981
diff changeset
1046 if not self._inline:
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1047 transaction.add(self.datafile, offset)
4981
e7131935fbb3 revlog: simplify addrevision
Matt Mackall <mpm@selenic.com>
parents: 4980
diff changeset
1048 transaction.add(self.indexfile, curr * len(entry))
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1049 if data[0]:
3390
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1050 dfh.write(data[0])
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1051 dfh.write(data[1])
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1052 dfh.flush()
4981
e7131935fbb3 revlog: simplify addrevision
Matt Mackall <mpm@selenic.com>
parents: 4980
diff changeset
1053 ifh.write(entry)
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1054 else:
4996
a0d37976cd5b revlog: avoid some unnecessary seek/tell syscalls
Matt Mackall <mpm@selenic.com>
parents: 4994
diff changeset
1055 offset += curr * self._io.size
5320
8409a2e3a78d revlog: fix inlined revision transaction extra data (issue 749)
Patrick Mezard <pmezard@gmail.com>
parents: 5008
diff changeset
1056 transaction.add(self.indexfile, offset, curr)
4981
e7131935fbb3 revlog: simplify addrevision
Matt Mackall <mpm@selenic.com>
parents: 4980
diff changeset
1057 ifh.write(entry)
3390
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1058 ifh.write(data[0])
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1059 ifh.write(data[1])
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1060 self.checkinlinesize(transaction, ifh)
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1061
4984
b4066fcbd6ba revlog: mark cache private
Matt Mackall <mpm@selenic.com>
parents: 4983
diff changeset
1062 self._cache = (node, curr, text)
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
1063 return node
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
1064
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
1065 def ancestor(self, a, b):
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1066 """calculate the least common ancestor of nodes a and b"""
2081
416d8b2a75b8 Speedup revlog.ancestors for the linear case
Chris Mason <mason@suse.com>
parents: 2080
diff changeset
1067
3139
1fd1cdcc4200 Switch revlog.ancestor to use revisions rather than nodeids
Matt Mackall <mpm@selenic.com>
parents: 3136
diff changeset
1068 def parents(rev):
3578
3b4e00cba57a Define and use nullrev (revision of nullid) instead of -1.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3505
diff changeset
1069 return [p for p in self.parentrevs(rev) if p != nullrev]
515
03f27b1381f9 Whitespace cleanups
mpm@selenic.com
parents: 484
diff changeset
1070
3139
1fd1cdcc4200 Switch revlog.ancestor to use revisions rather than nodeids
Matt Mackall <mpm@selenic.com>
parents: 3136
diff changeset
1071 c = ancestor.ancestor(self.rev(a), self.rev(b), parents)
1fd1cdcc4200 Switch revlog.ancestor to use revisions rather than nodeids
Matt Mackall <mpm@selenic.com>
parents: 3136
diff changeset
1072 if c is None:
1fd1cdcc4200 Switch revlog.ancestor to use revisions rather than nodeids
Matt Mackall <mpm@selenic.com>
parents: 3136
diff changeset
1073 return nullid
1fd1cdcc4200 Switch revlog.ancestor to use revisions rather than nodeids
Matt Mackall <mpm@selenic.com>
parents: 3136
diff changeset
1074
1fd1cdcc4200 Switch revlog.ancestor to use revisions rather than nodeids
Matt Mackall <mpm@selenic.com>
parents: 3136
diff changeset
1075 return self.node(c)
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
1076
1598
14d1f1868bf6 cleanup of revlog.group when repository is local
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1559
diff changeset
1077 def group(self, nodelist, lookup, infocollect=None):
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1078 """calculate a delta group
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1079
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1080 Given a list of changeset revs, return a set of deltas and
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1081 metadata corresponding to nodes. the first delta is
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1082 parent(nodes[0]) -> nodes[0] the receiver is guaranteed to
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1083 have this parent as it has all history before these
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1084 changesets. parent is parent[0]
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1085 """
1458
1033892bbb87 This changes the revlog.group and re-implements the localrepo.changeroup
Eric Hopper <hopper@omnifarious.org>
parents: 1457
diff changeset
1086 revs = [self.rev(n) for n in nodelist]
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1087
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1088 # if we don't have any revisions touched by these changesets, bail
192
5d8553352d2e Changes to network protocol
mpm@selenic.com
parents: 155
diff changeset
1089 if not revs:
1981
736b6c96bbbc make incoming work via ssh (issue139); move chunk code into separate module.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1941
diff changeset
1090 yield changegroup.closechunk()
192
5d8553352d2e Changes to network protocol
mpm@selenic.com
parents: 155
diff changeset
1091 return
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1092
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1093 # add the parent of the first rev
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1094 p = self.parents(self.node(revs[0]))[0]
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1095 revs.insert(0, self.rev(p))
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1096
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1097 # build deltas
71
47c9a869adee Add mdiff.patches to speed up applying thousands of patches to the manifest
mpm@selenic.com
parents: 67
diff changeset
1098 for d in xrange(0, len(revs) - 1):
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1099 a, b = revs[d], revs[d + 1]
1598
14d1f1868bf6 cleanup of revlog.group when repository is local
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1559
diff changeset
1100 nb = self.node(b)
192
5d8553352d2e Changes to network protocol
mpm@selenic.com
parents: 155
diff changeset
1101
1458
1033892bbb87 This changes the revlog.group and re-implements the localrepo.changeroup
Eric Hopper <hopper@omnifarious.org>
parents: 1457
diff changeset
1102 if infocollect is not None:
1598
14d1f1868bf6 cleanup of revlog.group when repository is local
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1559
diff changeset
1103 infocollect(nb)
1458
1033892bbb87 This changes the revlog.group and re-implements the localrepo.changeroup
Eric Hopper <hopper@omnifarious.org>
parents: 1457
diff changeset
1104
1598
14d1f1868bf6 cleanup of revlog.group when repository is local
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1559
diff changeset
1105 p = self.parents(nb)
14d1f1868bf6 cleanup of revlog.group when repository is local
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1559
diff changeset
1106 meta = nb + p[0] + p[1] + lookup(nb)
5369
7530334bf301 revlog: generate trivial deltas against null revision
Matt Mackall <mpm@selenic.com>
parents: 5338
diff changeset
1107 if a == -1:
7530334bf301 revlog: generate trivial deltas against null revision
Matt Mackall <mpm@selenic.com>
parents: 5338
diff changeset
1108 d = self.revision(nb)
7530334bf301 revlog: generate trivial deltas against null revision
Matt Mackall <mpm@selenic.com>
parents: 5338
diff changeset
1109 meta += mdiff.trivialdiffheader(len(d))
7530334bf301 revlog: generate trivial deltas against null revision
Matt Mackall <mpm@selenic.com>
parents: 5338
diff changeset
1110 else:
7530334bf301 revlog: generate trivial deltas against null revision
Matt Mackall <mpm@selenic.com>
parents: 5338
diff changeset
1111 d = self.revdiff(a, b)
5370
61462e7d62ed changegroup: avoid large copies
Matt Mackall <mpm@selenic.com>
parents: 5369
diff changeset
1112 yield changegroup.chunkheader(len(meta) + len(d))
61462e7d62ed changegroup: avoid large copies
Matt Mackall <mpm@selenic.com>
parents: 5369
diff changeset
1113 yield meta
5448
e038738714fd revlog: avoid large yields in group()
Matt Mackall <mpm@selenic.com>
parents: 5445
diff changeset
1114 if len(d) > 2**20:
e038738714fd revlog: avoid large yields in group()
Matt Mackall <mpm@selenic.com>
parents: 5445
diff changeset
1115 pos = 0
e038738714fd revlog: avoid large yields in group()
Matt Mackall <mpm@selenic.com>
parents: 5445
diff changeset
1116 while pos < len(d):
e038738714fd revlog: avoid large yields in group()
Matt Mackall <mpm@selenic.com>
parents: 5445
diff changeset
1117 pos2 = pos + 2 ** 18
e038738714fd revlog: avoid large yields in group()
Matt Mackall <mpm@selenic.com>
parents: 5445
diff changeset
1118 yield d[pos:pos2]
e038738714fd revlog: avoid large yields in group()
Matt Mackall <mpm@selenic.com>
parents: 5445
diff changeset
1119 pos = pos2
e038738714fd revlog: avoid large yields in group()
Matt Mackall <mpm@selenic.com>
parents: 5445
diff changeset
1120 else:
e038738714fd revlog: avoid large yields in group()
Matt Mackall <mpm@selenic.com>
parents: 5445
diff changeset
1121 yield d
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1122
1981
736b6c96bbbc make incoming work via ssh (issue139); move chunk code into separate module.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1941
diff changeset
1123 yield changegroup.closechunk()
192
5d8553352d2e Changes to network protocol
mpm@selenic.com
parents: 155
diff changeset
1124
1062
6d5a62a549fa pep-0008 cleanup
benoit.boissinot@ens-lyon.fr
parents: 902
diff changeset
1125 def addgroup(self, revs, linkmapper, transaction, unique=0):
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1126 """
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1127 add a delta group
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1128
1083
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1129 given a set of deltas, add them to the revision log. the
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1130 first delta is against its parent, which should be in our
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1131 log, the rest are against the previous delta.
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1132 """
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1133
30974cf73435 Add some docstrings to revlog.py
mpm@selenic.com
parents: 1074
diff changeset
1134 #track the base of the current delta log
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1135 r = self.count()
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1136 t = r - 1
2002
4aab906517c6 Calling revlog.addgroup with an empty changegroup now raises RevlogError.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1981
diff changeset
1137 node = None
515
03f27b1381f9 Whitespace cleanups
mpm@selenic.com
parents: 484
diff changeset
1138
3578
3b4e00cba57a Define and use nullrev (revision of nullid) instead of -1.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3505
diff changeset
1139 base = prev = nullrev
2078
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
1140 start = end = textlen = 0
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1141 if r:
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1142 end = self.end(t)
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1143
2072
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
1144 ifh = self.opener(self.indexfile, "a+")
4996
a0d37976cd5b revlog: avoid some unnecessary seek/tell syscalls
Matt Mackall <mpm@selenic.com>
parents: 4994
diff changeset
1145 isize = r * self._io.size
4982
9672e3c42b0c revlog: change _inline from a function to a variable
Matt Mackall <mpm@selenic.com>
parents: 4981
diff changeset
1146 if self._inline:
4996
a0d37976cd5b revlog: avoid some unnecessary seek/tell syscalls
Matt Mackall <mpm@selenic.com>
parents: 4994
diff changeset
1147 transaction.add(self.indexfile, end + isize, r)
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1148 dfh = None
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1149 else:
4996
a0d37976cd5b revlog: avoid some unnecessary seek/tell syscalls
Matt Mackall <mpm@selenic.com>
parents: 4994
diff changeset
1150 transaction.add(self.indexfile, isize, r)
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1151 transaction.add(self.datafile, end)
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1152 dfh = self.opener(self.datafile, "a")
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1153
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1154 # loop through our set of deltas
192
5d8553352d2e Changes to network protocol
mpm@selenic.com
parents: 155
diff changeset
1155 chain = None
5d8553352d2e Changes to network protocol
mpm@selenic.com
parents: 155
diff changeset
1156 for chunk in revs:
5d8553352d2e Changes to network protocol
mpm@selenic.com
parents: 155
diff changeset
1157 node, p1, p2, cs = struct.unpack("20s20s20s20s", chunk[:80])
94
7daef883134f Refactor merge code
mpm@selenic.com
parents: 86
diff changeset
1158 link = linkmapper(cs)
77
bed15e766511 Fix bug in lazymap code
mpm@selenic.com
parents: 76
diff changeset
1159 if node in self.nodemap:
224
ccbcc4d76f81 fix bad assumption about uniqueness of file versions
mpm@selenic.com
parents: 221
diff changeset
1160 # this can happen if two branches make the same change
1218
cde6818e082a Add preliminary support for the bundle and unbundle commands
mpm@selenic.com
parents: 1214
diff changeset
1161 # if unique:
1402
9d2c2e6b32b5 i18n part2: use '_' for all strings who are part of the user interface
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1400
diff changeset
1162 # raise RevlogError(_("already have %s") % hex(node[:4]))
653
94cdd02792b5 Fix corruption resulting from skipping parts of a revision group
Matt Mackall <mpm@selenic.com>
parents: 651
diff changeset
1163 chain = node
224
ccbcc4d76f81 fix bad assumption about uniqueness of file versions
mpm@selenic.com
parents: 221
diff changeset
1164 continue
5445
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1165 delta = buffer(chunk, 80)
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1166 del chunk
192
5d8553352d2e Changes to network protocol
mpm@selenic.com
parents: 155
diff changeset
1167
1509
46a07392cf28 Add safety check for addgroup
Matt Mackall <mpm@selenic.com>
parents: 1494
diff changeset
1168 for p in (p1, p2):
46a07392cf28 Add safety check for addgroup
Matt Mackall <mpm@selenic.com>
parents: 1494
diff changeset
1169 if not p in self.nodemap:
3930
01d98d68d697 Add revlog.LookupError exception, and use it instead of RevlogError.
Brendan Cully <brendan@kublai.com>
parents: 3928
diff changeset
1170 raise LookupError(_("unknown parent %s") % short(p))
1509
46a07392cf28 Add safety check for addgroup
Matt Mackall <mpm@selenic.com>
parents: 1494
diff changeset
1171
192
5d8553352d2e Changes to network protocol
mpm@selenic.com
parents: 155
diff changeset
1172 if not chain:
5d8553352d2e Changes to network protocol
mpm@selenic.com
parents: 155
diff changeset
1173 # retrieve the parent revision of the delta chain
5d8553352d2e Changes to network protocol
mpm@selenic.com
parents: 155
diff changeset
1174 chain = p1
5d8553352d2e Changes to network protocol
mpm@selenic.com
parents: 155
diff changeset
1175 if not chain in self.nodemap:
3930
01d98d68d697 Add revlog.LookupError exception, and use it instead of RevlogError.
Brendan Cully <brendan@kublai.com>
parents: 3928
diff changeset
1176 raise LookupError(_("unknown base %s") % short(chain[:4]))
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1177
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1178 # full versions are inserted when the needed deltas become
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1179 # comparable to the uncompressed text or when the previous
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1180 # version is not the one we have a delta against. We use
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1181 # the size of the previous full rev as a proxy for the
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1182 # current size.
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1183
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1184 if chain == prev:
5445
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1185 cdelta = compress(delta)
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1186 cdeltalen = len(cdelta[0]) + len(cdelta[1])
2078
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
1187 textlen = mdiff.patchedsize(textlen, delta)
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1188
5445
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1189 if chain != prev or (end - start + cdeltalen) > textlen * 2:
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1190 # flush our writes here so we can read it in revision
2072
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
1191 if dfh:
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
1192 dfh.flush()
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1193 ifh.flush()
65
d40cc5aacc31 Fix up a bunch of bugs in the new merge code
mpm@selenic.com
parents: 64
diff changeset
1194 text = self.revision(chain)
5445
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1195 if len(text) == 0:
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1196 # skip over trivial delta header
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1197 text = buffer(delta, 12)
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1198 else:
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1199 text = mdiff.patches(text, [delta])
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1200 del delta
3390
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1201 chk = self._addrevision(text, transaction, link, p1, p2, None,
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1202 ifh, dfh)
4982
9672e3c42b0c revlog: change _inline from a function to a variable
Matt Mackall <mpm@selenic.com>
parents: 4981
diff changeset
1203 if not dfh and not self._inline:
3390
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1204 # addrevision switched from inline to conventional
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1205 # reopen the index
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1206 dfh = self.opener(self.datafile, "a")
a74addddd092 make revlog.addgroup pass its file handles to addrevision
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 3360
diff changeset
1207 ifh = self.opener(self.indexfile, "a")
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1208 if chk != node:
1402
9d2c2e6b32b5 i18n part2: use '_' for all strings who are part of the user interface
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1400
diff changeset
1209 raise RevlogError(_("consistency error adding group"))
2078
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 2077
diff changeset
1210 textlen = len(text)
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1211 else:
5445
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1212 e = (offset_type(end, 0), cdeltalen, textlen, base,
4977
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
1213 link, self.rev(p1), self.rev(p2), node)
4979
06abdaf78788 revlog: add a magic null revision to our index
Matt Mackall <mpm@selenic.com>
parents: 4978
diff changeset
1214 self.index.insert(-1, e)
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1215 self.nodemap[node] = r
5338
f87685355c9c revlog: fix revlogio.packentry corner case
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 5325
diff changeset
1216 entry = self._io.packentry(e, self.node, self.version, r)
4982
9672e3c42b0c revlog: change _inline from a function to a variable
Matt Mackall <mpm@selenic.com>
parents: 4981
diff changeset
1217 if self._inline:
4986
58cc017ec7e0 revlog: abstract out index entry packing
Matt Mackall <mpm@selenic.com>
parents: 4985
diff changeset
1218 ifh.write(entry)
5445
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1219 ifh.write(cdelta[0])
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1220 ifh.write(cdelta[1])
2075
343aeefb553b Make the appendfile class inline-data index friendly
mason@suse.com
parents: 2073
diff changeset
1221 self.checkinlinesize(transaction, ifh)
4982
9672e3c42b0c revlog: change _inline from a function to a variable
Matt Mackall <mpm@selenic.com>
parents: 4981
diff changeset
1222 if not self._inline:
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1223 dfh = self.opener(self.datafile, "a")
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1224 ifh = self.opener(self.indexfile, "a")
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1225 else:
5445
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1226 dfh.write(cdelta[0])
64cf1c853674 revlog: reduce memory usage in addgroup
Matt Mackall <mpm@selenic.com>
parents: 5370
diff changeset
1227 dfh.write(cdelta[1])
4977
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
1228 ifh.write(entry)
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1229
65
d40cc5aacc31 Fix up a bunch of bugs in the new merge code
mpm@selenic.com
parents: 64
diff changeset
1230 t, r, chain, prev = r, r + 1, node, node
1749
d457fec76ab0 fix warnings from pychecker (unused variables and shadowing)
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1711
diff changeset
1231 base = self.base(t)
d457fec76ab0 fix warnings from pychecker (unused variables and shadowing)
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1711
diff changeset
1232 start = self.start(base)
46
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1233 end = self.end(t)
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1234
93e868fa0db8 Add changegroup support
mpm@selenic.com
parents: 45
diff changeset
1235 return node
1493
1a216cb4ee64 verify: add check for mismatch of index and data length
Matt Mackall <mpm@selenic.com>
parents: 1469
diff changeset
1236
1535
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1237 def strip(self, rev, minlink):
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1238 if self.count() == 0 or rev >= self.count():
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1239 return
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1240
2072
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
1241 if isinstance(self.index, lazyindex):
4964
ee983d0dbea8 revlog: privatize some methods
Matt Mackall <mpm@selenic.com>
parents: 4963
diff changeset
1242 self._loadindexmap()
2072
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
1243
1535
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1244 # When stripping away a revision, we need to make sure it
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1245 # does not actually belong to an older changeset.
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1246 # The minlink parameter defines the oldest revision
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1247 # we're allowed to strip away.
4978
93d48a8fa496 revlog: change accesses to index entry elements to use positive offsets
Matt Mackall <mpm@selenic.com>
parents: 4977
diff changeset
1248 while minlink > self.index[rev][4]:
1535
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1249 rev += 1
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1250 if rev >= self.count():
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1251 return
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1252
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1253 # first truncate the files on disk
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1254 end = self.start(rev)
4982
9672e3c42b0c revlog: change _inline from a function to a variable
Matt Mackall <mpm@selenic.com>
parents: 4981
diff changeset
1255 if not self._inline:
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1256 df = self.opener(self.datafile, "a")
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1257 df.truncate(end)
4977
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
1258 end = rev * self._io.size
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1259 else:
4977
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
1260 end += rev * self._io.size
2072
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
1261
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
1262 indexf = self.opener(self.indexfile, "a")
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
1263 indexf.truncate(end)
1535
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1264
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1265 # then reset internal state in memory to forget those revisions
4984
b4066fcbd6ba revlog: mark cache private
Matt Mackall <mpm@selenic.com>
parents: 4983
diff changeset
1266 self._cache = None
4983
4dbcfc6e359e revlog: pull chunkcache back into revlog
Matt Mackall <mpm@selenic.com>
parents: 4982
diff changeset
1267 self._chunkcache = None
2072
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
1268 for x in xrange(rev, self.count()):
74d3f5336b66 Implement revlogng.
mason@suse.com
parents: 2002
diff changeset
1269 del self.nodemap[self.node(x)]
1535
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1270
4979
06abdaf78788 revlog: add a magic null revision to our index
Matt Mackall <mpm@selenic.com>
parents: 4978
diff changeset
1271 del self.index[rev:-1]
1535
7ae0ce7a3dc4 Add revlog.strip to truncate away revisions.
mason@suse.com
parents: 1533
diff changeset
1272
1493
1a216cb4ee64 verify: add check for mismatch of index and data length
Matt Mackall <mpm@selenic.com>
parents: 1469
diff changeset
1273 def checksize(self):
1a216cb4ee64 verify: add check for mismatch of index and data length
Matt Mackall <mpm@selenic.com>
parents: 1469
diff changeset
1274 expected = 0
1a216cb4ee64 verify: add check for mismatch of index and data length
Matt Mackall <mpm@selenic.com>
parents: 1469
diff changeset
1275 if self.count():
5322
fb070713ff36 revlog: more robust for damaged indexes
Matt Mackall <mpm@selenic.com>
parents: 5008
diff changeset
1276 expected = max(0, self.end(self.count() - 1))
1667
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1277
1494
249ca10d37f4 Handle empty logs in repo.checksize
Matt Mackall <mpm@selenic.com>
parents: 1493
diff changeset
1278 try:
249ca10d37f4 Handle empty logs in repo.checksize
Matt Mackall <mpm@selenic.com>
parents: 1493
diff changeset
1279 f = self.opener(self.datafile)
249ca10d37f4 Handle empty logs in repo.checksize
Matt Mackall <mpm@selenic.com>
parents: 1493
diff changeset
1280 f.seek(0, 2)
249ca10d37f4 Handle empty logs in repo.checksize
Matt Mackall <mpm@selenic.com>
parents: 1493
diff changeset
1281 actual = f.tell()
1667
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1282 dd = actual - expected
1494
249ca10d37f4 Handle empty logs in repo.checksize
Matt Mackall <mpm@selenic.com>
parents: 1493
diff changeset
1283 except IOError, inst:
1667
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1284 if inst.errno != errno.ENOENT:
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1285 raise
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1286 dd = 0
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1287
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1288 try:
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1289 f = self.opener(self.indexfile)
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1290 f.seek(0, 2)
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1291 actual = f.tell()
4977
6cb30bc4ca32 revlog: parse revlogv0 indexes into v1 internally
Matt Mackall <mpm@selenic.com>
parents: 4976
diff changeset
1292 s = self._io.size
5322
fb070713ff36 revlog: more robust for damaged indexes
Matt Mackall <mpm@selenic.com>
parents: 5008
diff changeset
1293 i = max(0, actual / s)
1667
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1294 di = actual - (i * s)
4982
9672e3c42b0c revlog: change _inline from a function to a variable
Matt Mackall <mpm@selenic.com>
parents: 4981
diff changeset
1295 if self._inline:
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1296 databytes = 0
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1297 for r in xrange(self.count()):
5322
fb070713ff36 revlog: more robust for damaged indexes
Matt Mackall <mpm@selenic.com>
parents: 5008
diff changeset
1298 databytes += max(0, self.length(r))
2073
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1299 dd = 0
1e6745f78989 Implement data inlined with the index file
mason@suse.com
parents: 2072
diff changeset
1300 di = actual - self.count() * s - databytes
1667
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1301 except IOError, inst:
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1302 if inst.errno != errno.ENOENT:
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1303 raise
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1304 di = 0
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1305
daff3ef0de8d verify: notice extra data in indices
Matt Mackall <mpm@selenic.com>
parents: 1660
diff changeset
1306 return (dd, di)