Show More
@@ -1,229 +1,229 | |||
|
1 | 1 | #!/usr/bin/env python |
|
2 | 2 | # |
|
3 | 3 | # This is a generalized framework for converting between SCM |
|
4 | 4 | # repository formats. |
|
5 | 5 | # |
|
6 | 6 | # In its current form, it's hardcoded to convert incrementally between |
|
7 | 7 | # git and Mercurial. |
|
8 | 8 | # |
|
9 | 9 | # To use, you must first import the first git version into Mercurial, |
|
10 | 10 | # and establish a mapping between the git commit hash and the hash in |
|
11 | 11 | # Mercurial for that version. This mapping is kept in a simple text |
|
12 | 12 | # file with lines like so: |
|
13 | 13 | # |
|
14 | 14 | # <git hash> <mercurial hash> |
|
15 | 15 | # |
|
16 | 16 | # To convert the rest of the repo, run: |
|
17 | 17 | # |
|
18 | 18 | # convert-repo <git-dir> <hg-dir> <mapfile> |
|
19 | 19 | # |
|
20 | 20 | # This updates the mapfile on each commit copied, so it can be |
|
21 | 21 | # interrupted and can be run repeatedly to copy new commits. |
|
22 | 22 | |
|
23 | 23 | import sys, os, zlib, sha |
|
24 | from mercurial import hg, ui | |
|
24 | from mercurial import hg, ui, util | |
|
25 | 25 | |
|
26 | 26 | class convert_git: |
|
27 | 27 | def __init__(self, path): |
|
28 | 28 | self.path = path |
|
29 | 29 | |
|
30 | 30 | def getheads(self): |
|
31 | 31 | h = file(self.path + "/.git/HEAD").read()[:-1] |
|
32 | 32 | return [h] |
|
33 | 33 | |
|
34 | 34 | def getfile(self, name, rev): |
|
35 | 35 | a = file(self.path + ("/.git/objects/%s/%s" |
|
36 | 36 | % (rev[:2], rev[2:]))).read() |
|
37 | 37 | b = zlib.decompress(a) |
|
38 | 38 | if sha.sha(b).hexdigest() != rev: raise "bad hash" |
|
39 | 39 | head, text = b.split('\0', 1) |
|
40 | 40 | return text |
|
41 | 41 | |
|
42 | 42 | def getchanges(self, version): |
|
43 | 43 | path = os.getcwd() |
|
44 | 44 | os.chdir(self.path) |
|
45 | 45 | fh = os.popen("git-diff-tree -m -r %s" % (version)) |
|
46 | 46 | os.chdir(path) |
|
47 | 47 | |
|
48 | 48 | changes = [] |
|
49 | 49 | for l in fh: |
|
50 | 50 | if "\t" not in l: continue |
|
51 | 51 | m, f = l[:-1].split("\t") |
|
52 | 52 | m = m.split() |
|
53 | 53 | h = m[3] |
|
54 | 54 | p = (m[1] == "100755") |
|
55 | 55 | changes.append((f, h, p)) |
|
56 | 56 | return changes |
|
57 | 57 | |
|
58 | 58 | def getcommit(self, version): |
|
59 | 59 | c = self.getfile("", version) # read the commit hash |
|
60 | 60 | end = c.find("\n\n") |
|
61 | 61 | message = c[end+2:] |
|
62 | 62 | l = c[:end].splitlines() |
|
63 | 63 | manifest = l[0].split()[1] |
|
64 | 64 | parents = [] |
|
65 | 65 | for e in l[1:]: |
|
66 | 66 | n,v = e.split(" ", 1) |
|
67 | 67 | if n == "author": |
|
68 | 68 | p = v.split() |
|
69 | 69 | date = " ".join(p[-2:]) |
|
70 | 70 | author = " ".join(p[:-2]) |
|
71 | 71 | if author[0] == "<": author = author[1:-1] |
|
72 | 72 | if n == "committer": |
|
73 | 73 | p = v.split() |
|
74 | 74 | date = " ".join(p[-2:]) |
|
75 | 75 | committer = " ".join(p[:-2]) |
|
76 | 76 | if committer[0] == "<": committer = committer[1:-1] |
|
77 | 77 | message += "\ncommitter: %s %s\n" % (committer, date) |
|
78 | 78 | if n == "parent": parents.append(v) |
|
79 | 79 | return (parents, author, date, message) |
|
80 | 80 | |
|
81 | 81 | class convert_mercurial: |
|
82 | 82 | def __init__(self, path): |
|
83 | 83 | self.path = path |
|
84 | 84 | u = ui.ui() |
|
85 | 85 | self.repo = hg.repository(u, path) |
|
86 | 86 | |
|
87 | 87 | def getheads(self): |
|
88 | 88 | h = self.repo.changelog.heads() |
|
89 | 89 | h = [ hg.hex(x) for x in h ] |
|
90 | 90 | return h |
|
91 | 91 | |
|
92 | 92 | def putfile(self, f, e, data): |
|
93 | 93 | self.repo.wfile(f, "w").write(data) |
|
94 |
|
|
|
94 | util.set_exec(self.repo.wjoin(f), e) | |
|
95 | 95 | |
|
96 | 96 | def delfile(self, f): |
|
97 | 97 | try: |
|
98 | 98 | os.unlink(self.repo.wjoin(f)) |
|
99 | 99 | self.repo.remove([f]) |
|
100 | 100 | except: |
|
101 | 101 | pass |
|
102 | 102 | |
|
103 | 103 | def putcommit(self, files, parents, author, dest, text): |
|
104 | 104 | if not parents: parents = ["0" * 40] |
|
105 | 105 | if len(parents) < 2: parents.append("0" * 40) |
|
106 | 106 | |
|
107 | 107 | seen = {} |
|
108 | 108 | pl = [] |
|
109 | 109 | for p in parents: |
|
110 | 110 | if p not in seen: |
|
111 | 111 | pl.append(p) |
|
112 | 112 | seen[p] = 1 |
|
113 | 113 | parents = pl |
|
114 | 114 | |
|
115 | 115 | p2 = parents.pop(0) |
|
116 | 116 | c = self.repo.changelog.count() |
|
117 | 117 | while parents: |
|
118 | 118 | p1 = p2 |
|
119 | 119 | p2 = parents.pop(0) |
|
120 | 120 | self.repo.rawcommit(files, text, author, dest, |
|
121 | 121 | hg.bin(p1), hg.bin(p2)) |
|
122 | 122 | text = "(octopus merge fixup)\n" |
|
123 | 123 | |
|
124 | 124 | return hg.hex(self.repo.changelog.node(c)) |
|
125 | 125 | |
|
126 | 126 | class convert: |
|
127 | 127 | def __init__(self, source, dest, mapfile): |
|
128 | 128 | self.source = source |
|
129 | 129 | self.dest = dest |
|
130 | 130 | self.mapfile = mapfile |
|
131 | 131 | self.commitcache = {} |
|
132 | 132 | |
|
133 | 133 | self.map = {} |
|
134 | 134 | for l in file(self.mapfile): |
|
135 | 135 | sv, dv = l[:-1].split() |
|
136 | 136 | self.map[sv] = dv |
|
137 | 137 | |
|
138 | 138 | def walktree(self, heads): |
|
139 | 139 | visit = heads |
|
140 | 140 | known = {} |
|
141 | 141 | parents = {} |
|
142 | 142 | while visit: |
|
143 | 143 | n = visit.pop(0) |
|
144 | 144 | if n in known or n in self.map: continue |
|
145 | 145 | known[n] = 1 |
|
146 | 146 | self.commitcache[n] = self.source.getcommit(n) |
|
147 | 147 | cp = self.commitcache[n][0] |
|
148 | 148 | for p in cp: |
|
149 | 149 | parents.setdefault(n, []).append(p) |
|
150 | 150 | visit.append(p) |
|
151 | 151 | |
|
152 | 152 | return parents |
|
153 | 153 | |
|
154 | 154 | def toposort(self, parents): |
|
155 | 155 | visit = parents.keys() |
|
156 | 156 | seen = {} |
|
157 | 157 | children = {} |
|
158 | 158 | while visit: |
|
159 | 159 | n = visit.pop(0) |
|
160 | 160 | if n in seen: continue |
|
161 | 161 | seen[n] = 1 |
|
162 | 162 | pc = 0 |
|
163 | 163 | if n in parents: |
|
164 | 164 | for p in parents[n]: |
|
165 | 165 | if p not in self.map: pc += 1 |
|
166 | 166 | visit.append(p) |
|
167 | 167 | children.setdefault(p, []).append(n) |
|
168 | 168 | if not pc: root = n |
|
169 | 169 | |
|
170 | 170 | s = [] |
|
171 | 171 | removed = {} |
|
172 | 172 | visit = parents.keys() |
|
173 | 173 | while visit: |
|
174 | 174 | n = visit.pop(0) |
|
175 | 175 | if n in removed: continue |
|
176 | 176 | dep = 0 |
|
177 | 177 | if n in parents: |
|
178 | 178 | for p in parents[n]: |
|
179 | 179 | if p in self.map: continue |
|
180 | 180 | if p not in removed: |
|
181 | 181 | # we're still dependent |
|
182 | 182 | visit.append(n) |
|
183 | 183 | dep = 1 |
|
184 | 184 | break |
|
185 | 185 | |
|
186 | 186 | if not dep: |
|
187 | 187 | # all n's parents are in the list |
|
188 | 188 | removed[n] = 1 |
|
189 | 189 | s.append(n) |
|
190 | 190 | if n in children: |
|
191 | 191 | for c in children[n]: |
|
192 | 192 | visit.insert(0, c) |
|
193 | 193 | |
|
194 | 194 | return s |
|
195 | 195 | |
|
196 | 196 | def copy(self, rev): |
|
197 | 197 | p, a, d, t = self.commitcache[rev] |
|
198 | 198 | files = self.source.getchanges(rev) |
|
199 | 199 | |
|
200 | 200 | for f,v,e in files: |
|
201 | 201 | try: |
|
202 | 202 | data = self.source.getfile(f, v) |
|
203 | 203 | except IOError, inst: |
|
204 | 204 | self.dest.delfile(f) |
|
205 | 205 | else: |
|
206 | 206 | self.dest.putfile(f, e, data) |
|
207 | 207 | |
|
208 | 208 | r = [self.map[v] for v in p] |
|
209 | 209 | f = [f for f,v,e in files] |
|
210 | 210 | self.map[rev] = self.dest.putcommit(f, r, a, d, t) |
|
211 | 211 | file(self.mapfile, "a").write("%s %s\n" % (rev, self.map[rev])) |
|
212 | 212 | |
|
213 | 213 | def convert(self): |
|
214 | 214 | heads = self.source.getheads() |
|
215 | 215 | parents = self.walktree(heads) |
|
216 | 216 | t = self.toposort(parents) |
|
217 | 217 | num = len(t) |
|
218 | 218 | |
|
219 | 219 | for c in t: |
|
220 | 220 | num -= 1 |
|
221 | 221 | if c in self.map: continue |
|
222 | 222 | desc = self.commitcache[c][3].splitlines()[0] |
|
223 | 223 | print num, desc |
|
224 | 224 | self.copy(c) |
|
225 | 225 | |
|
226 | 226 | gitpath, hgpath, mapfile = sys.argv[1:] |
|
227 | 227 | |
|
228 | 228 | c = convert(convert_git(gitpath), convert_mercurial(hgpath), mapfile) |
|
229 | 229 | c.convert() |
General Comments 0
You need to be logged in to leave comments.
Login now