Show More
@@ -1,199 +1,201 | |||
|
1 | 1 | # revlog.py - storage back-end for mercurial |
|
2 | 2 | # |
|
3 | 3 | # This provides efficient delta storage with O(1) retrieve and append |
|
4 | 4 | # and O(changes) merge between branches |
|
5 | 5 | # |
|
6 | 6 | # Copyright 2005 Matt Mackall <mpm@selenic.com> |
|
7 | 7 | # |
|
8 | 8 | # This software may be used and distributed according to the terms |
|
9 | 9 | # of the GNU General Public License, incorporated herein by reference. |
|
10 | 10 | |
|
11 |
import zlib, struct, sha, |
|
|
11 | import zlib, struct, sha, os, tempfile | |
|
12 | 12 | from mercurial import mdiff |
|
13 | 13 | |
|
14 | 14 | def compress(text): |
|
15 | 15 | return zlib.compress(text) |
|
16 | 16 | |
|
17 | 17 | def decompress(bin): |
|
18 | 18 | return zlib.decompress(bin) |
|
19 | 19 | |
|
20 | 20 | def hash(text, p1, p2): |
|
21 | 21 | l = [p1, p2] |
|
22 | 22 | l.sort() |
|
23 | 23 | return sha.sha(l[0] + l[1] + text).digest() |
|
24 | 24 | |
|
25 | 25 | nullid = "\0" * 20 |
|
26 | 26 | indexformat = ">4l20s20s20s" |
|
27 | 27 | |
|
28 | 28 | class revlog: |
|
29 | 29 | def __init__(self, opener, indexfile, datafile): |
|
30 | 30 | self.indexfile = indexfile |
|
31 | 31 | self.datafile = datafile |
|
32 | 32 | self.index = [] |
|
33 | 33 | self.opener = opener |
|
34 | 34 | self.cache = None |
|
35 | 35 | self.nodemap = { -1: nullid, nullid: -1 } |
|
36 | 36 | # read the whole index for now, handle on-demand later |
|
37 | 37 | try: |
|
38 | 38 | n = 0 |
|
39 | 39 | i = self.opener(self.indexfile).read() |
|
40 | 40 | s = struct.calcsize(indexformat) |
|
41 | 41 | for f in range(0, len(i), s): |
|
42 | 42 | # offset, size, base, linkrev, p1, p2, nodeid, changeset |
|
43 | 43 | e = struct.unpack(indexformat, i[f:f + s]) |
|
44 | 44 | self.nodemap[e[6]] = n |
|
45 | 45 | self.index.append(e) |
|
46 | 46 | n += 1 |
|
47 | 47 | except IOError: pass |
|
48 | 48 | |
|
49 | 49 | def tip(self): return self.node(len(self.index) - 1) |
|
50 | 50 | def count(self): return len(self.index) |
|
51 | 51 | def node(self, rev): return rev < 0 and nullid or self.index[rev][6] |
|
52 | 52 | def rev(self, node): return self.nodemap[node] |
|
53 | 53 | def linkrev(self, node): return self.index[self.nodemap[node]][3] |
|
54 | def parents(self, node): return self.index[self.nodemap[node]][4:6] | |
|
54 | def parents(self, node): | |
|
55 | if node == nullid: return (nullid, nullid) | |
|
56 | return self.index[self.nodemap[node]][4:6] | |
|
55 | 57 | |
|
56 | 58 | def start(self, rev): return self.index[rev][0] |
|
57 | 59 | def length(self, rev): return self.index[rev][1] |
|
58 | 60 | def end(self, rev): return self.start(rev) + self.length(rev) |
|
59 | 61 | def base(self, rev): return self.index[rev][2] |
|
60 | 62 | |
|
61 | 63 | def revisions(self, list): |
|
62 | 64 | # this can be optimized to do spans, etc |
|
63 | 65 | # be stupid for now |
|
64 | 66 | for r in list: |
|
65 | 67 | yield self.revision(r) |
|
66 | 68 | |
|
67 | 69 | def diff(self, a, b): |
|
68 | 70 | return mdiff.textdiff(a, b) |
|
69 | 71 | |
|
70 | 72 | def patch(self, text, patch): |
|
71 | 73 | return mdiff.patch(text, patch) |
|
72 | 74 | |
|
73 | 75 | def revision(self, node): |
|
74 | 76 | if node is nullid: return "" |
|
75 | 77 | if self.cache and self.cache[0] == node: return self.cache[2] |
|
76 | 78 | |
|
77 | 79 | text = None |
|
78 | 80 | rev = self.rev(node) |
|
79 | 81 | base = self.base(rev) |
|
80 | 82 | start = self.start(base) |
|
81 | 83 | end = self.end(rev) |
|
82 | 84 | |
|
83 | 85 | if self.cache and self.cache[1] >= base and self.cache[1] < rev: |
|
84 | 86 | base = self.cache[1] |
|
85 | 87 | start = self.start(base + 1) |
|
86 | 88 | text = self.cache[2] |
|
87 | 89 | last = 0 |
|
88 | 90 | |
|
89 | 91 | f = self.opener(self.datafile) |
|
90 | 92 | f.seek(start) |
|
91 | 93 | data = f.read(end - start) |
|
92 | 94 | |
|
93 | 95 | if not text: |
|
94 | 96 | last = self.length(base) |
|
95 | 97 | text = decompress(data[:last]) |
|
96 | 98 | |
|
97 | 99 | for r in range(base + 1, rev + 1): |
|
98 | 100 | s = self.length(r) |
|
99 | 101 | b = decompress(data[last:last + s]) |
|
100 | 102 | text = self.patch(text, b) |
|
101 | 103 | last = last + s |
|
102 | 104 | |
|
103 | 105 | (p1, p2) = self.parents(node) |
|
104 | 106 | if self.node(rev) != hash(text, p1, p2): |
|
105 | 107 | raise "Consistency check failed on %s:%d" % (self.datafile, rev) |
|
106 | 108 | |
|
107 | 109 | self.cache = (node, rev, text) |
|
108 | 110 | return text |
|
109 | 111 | |
|
110 | 112 | def addrevision(self, text, transaction, link, p1=None, p2=None): |
|
111 | 113 | if text is None: text = "" |
|
112 | 114 | if p1 is None: p1 = self.tip() |
|
113 | 115 | if p2 is None: p2 = nullid |
|
114 | 116 | |
|
115 | 117 | node = hash(text, p1, p2) |
|
116 | 118 | |
|
117 | 119 | n = self.count() |
|
118 | 120 | t = n - 1 |
|
119 | 121 | |
|
120 | 122 | if n: |
|
121 | 123 | start = self.start(self.base(t)) |
|
122 | 124 | end = self.end(t) |
|
123 | 125 | prev = self.revision(self.tip()) |
|
124 | 126 | if 0: |
|
125 | 127 | dd = self.diff(prev, text) |
|
126 | 128 | tt = self.patch(prev, dd) |
|
127 | 129 | if tt != text: |
|
128 | 130 | print prev |
|
129 | 131 | print text |
|
130 | 132 | print tt |
|
131 | 133 | raise "diff+patch failed" |
|
132 | 134 | data = compress(self.diff(prev, text)) |
|
133 | 135 | |
|
134 | 136 | # full versions are inserted when the needed deltas |
|
135 | 137 | # become comparable to the uncompressed text |
|
136 | 138 | if not n or (end + len(data) - start) > len(text) * 2: |
|
137 | 139 | data = compress(text) |
|
138 | 140 | base = n |
|
139 | 141 | else: |
|
140 | 142 | base = self.base(t) |
|
141 | 143 | |
|
142 | 144 | offset = 0 |
|
143 | 145 | if t >= 0: |
|
144 | 146 | offset = self.end(t) |
|
145 | 147 | |
|
146 | 148 | e = (offset, len(data), base, link, p1, p2, node) |
|
147 | 149 | |
|
148 | 150 | self.index.append(e) |
|
149 | 151 | self.nodemap[node] = n |
|
150 | 152 | entry = struct.pack(indexformat, *e) |
|
151 | 153 | |
|
152 | 154 | transaction.add(self.datafile, e[0]) |
|
153 | 155 | self.opener(self.datafile, "a").write(data) |
|
154 | 156 | transaction.add(self.indexfile, n * len(entry)) |
|
155 | 157 | self.opener(self.indexfile, "a").write(entry) |
|
156 | 158 | |
|
157 | 159 | self.cache = (node, n, text) |
|
158 | 160 | return node |
|
159 | 161 | |
|
160 | 162 | def ancestor(self, a, b): |
|
161 | 163 | def expand(e1, e2, a1, a2): |
|
162 | 164 | ne = [] |
|
163 | 165 | for n in e1: |
|
164 | 166 | (p1, p2) = self.parents(n) |
|
165 | 167 | if p1 in a2: return p1 |
|
166 | 168 | if p2 in a2: return p2 |
|
167 | 169 | if p1 != nullid and p1 not in a1: |
|
168 | 170 | a1[p1] = 1 |
|
169 | 171 | ne.append(p1) |
|
170 | 172 | if p2 != nullid and p2 not in a1: |
|
171 | 173 | a1[p2] = 1 |
|
172 | 174 | ne.append(p2) |
|
173 | 175 | return expand(e2, ne, a2, a1) |
|
174 | 176 | return expand([a], [b], {a:1}, {b:1}) |
|
175 | 177 | |
|
176 | 178 | def mergedag(self, other, transaction, linkseq, accumulate = None): |
|
177 | 179 | """combine the nodes from other's DAG into ours""" |
|
178 | 180 | old = self.tip() |
|
179 | 181 | i = self.count() |
|
180 | 182 | l = [] |
|
181 | 183 | |
|
182 | 184 | # merge the other revision log into our DAG |
|
183 | 185 | for r in range(other.count()): |
|
184 | 186 | id = other.node(r) |
|
185 | 187 | if id not in self.nodemap: |
|
186 | 188 | (xn, yn) = other.parents(id) |
|
187 | 189 | l.append((id, xn, yn)) |
|
188 | 190 | self.nodemap[id] = i |
|
189 | 191 | i += 1 |
|
190 | 192 | |
|
191 | 193 | # merge node date for new nodes |
|
192 | 194 | r = other.revisions([e[0] for e in l]) |
|
193 | 195 | for e in l: |
|
194 | 196 | t = r.next() |
|
195 | 197 | if accumulate: accumulate(t) |
|
196 | 198 | self.addrevision(t, transaction, linkseq.next(), e[1], e[2]) |
|
197 | 199 | |
|
198 | 200 | # return the unmerged heads for later resolving |
|
199 | 201 | return (old, self.tip()) |
General Comments 0
You need to be logged in to leave comments.
Login now