Show More
@@ -1,289 +1,289 b'' | |||
|
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, time |
|
24 | 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 | return [file(self.path + "/HEAD").read()[:-1]] |
|
32 | 32 | |
|
33 | 33 | def catfile(self, rev, type): |
|
34 | 34 | if rev == "0" * 40: raise IOError() |
|
35 | 35 | fh = os.popen("GIT_DIR=%s git-cat-file %s %s 2>/dev/null" % (self.path, type, rev)) |
|
36 | 36 | return fh.read() |
|
37 | 37 | |
|
38 | 38 | def getfile(self, name, rev): |
|
39 | 39 | return self.catfile(rev, "blob") |
|
40 | 40 | |
|
41 | 41 | def getchanges(self, version): |
|
42 | 42 | fh = os.popen("GIT_DIR=%s git-diff-tree --root -m -r %s" % (self.path, version)) |
|
43 | 43 | changes = [] |
|
44 | 44 | for l in fh: |
|
45 | 45 | if "\t" not in l: continue |
|
46 | 46 | m, f = l[:-1].split("\t") |
|
47 | 47 | m = m.split() |
|
48 | 48 | h = m[3] |
|
49 | 49 | p = (m[1] == "100755") |
|
50 | 50 | changes.append((f, h, p)) |
|
51 | 51 | return changes |
|
52 | 52 | |
|
53 | 53 | def getcommit(self, version): |
|
54 | 54 | c = self.catfile(version, "commit") # read the commit hash |
|
55 | 55 | end = c.find("\n\n") |
|
56 | 56 | message = c[end+2:] |
|
57 | 57 | l = c[:end].splitlines() |
|
58 | 58 | manifest = l[0].split()[1] |
|
59 | 59 | parents = [] |
|
60 | 60 | for e in l[1:]: |
|
61 | 61 | n,v = e.split(" ", 1) |
|
62 | 62 | if n == "author": |
|
63 | 63 | p = v.split() |
|
64 | 64 | tm, tz = p[-2:] |
|
65 | 65 | author = " ".join(p[:-2]) |
|
66 | 66 | if author[0] == "<": author = author[1:-1] |
|
67 | 67 | if n == "committer": |
|
68 | 68 | p = v.split() |
|
69 | 69 | tm, tz = p[-2:] |
|
70 | 70 | committer = " ".join(p[:-2]) |
|
71 | 71 | if committer[0] == "<": committer = committer[1:-1] |
|
72 | 72 | message += "\ncommitter: %s\n" % v |
|
73 | 73 | if n == "parent": parents.append(v) |
|
74 | 74 | |
|
75 | 75 | tzs, tzh, tzm = tz[-5:-4] + "1", tz[-4:-2], tz[-2:] |
|
76 | tz = int(tzs) * (int(tzh) * 3600 + int(tzm)) | |
|
76 | tz = -int(tzs) * (int(tzh) * 3600 + int(tzm)) | |
|
77 | 77 | date = tm + " " + str(tz) |
|
78 | 78 | return (parents, author, date, message) |
|
79 | 79 | |
|
80 | 80 | def gettags(self): |
|
81 | 81 | tags = {} |
|
82 | 82 | for f in os.listdir(self.path + "/refs/tags"): |
|
83 | 83 | try: |
|
84 | 84 | h = file(self.path + "/refs/tags/" + f).read().strip() |
|
85 | 85 | c = self.catfile(h, "tag") # read the commit hash |
|
86 | 86 | h = c.splitlines()[0].split()[1] |
|
87 | 87 | tags[f] = h |
|
88 | 88 | except: |
|
89 | 89 | pass |
|
90 | 90 | return tags |
|
91 | 91 | |
|
92 | 92 | class convert_mercurial: |
|
93 | 93 | def __init__(self, path): |
|
94 | 94 | self.path = path |
|
95 | 95 | u = ui.ui() |
|
96 | 96 | self.repo = hg.repository(u, path) |
|
97 | 97 | |
|
98 | 98 | def getheads(self): |
|
99 | 99 | h = self.repo.changelog.heads() |
|
100 | 100 | return [ hg.hex(x) for x in h ] |
|
101 | 101 | |
|
102 | 102 | def putfile(self, f, e, data): |
|
103 | 103 | self.repo.wfile(f, "w").write(data) |
|
104 | 104 | if self.repo.dirstate.state(f) == '?': |
|
105 | 105 | self.repo.dirstate.update([f], "a") |
|
106 | 106 | |
|
107 | 107 | util.set_exec(self.repo.wjoin(f), e) |
|
108 | 108 | |
|
109 | 109 | def delfile(self, f): |
|
110 | 110 | try: |
|
111 | 111 | os.unlink(self.repo.wjoin(f)) |
|
112 | 112 | #self.repo.remove([f]) |
|
113 | 113 | except: |
|
114 | 114 | pass |
|
115 | 115 | |
|
116 | 116 | def putcommit(self, files, parents, author, dest, text): |
|
117 | 117 | seen = {} |
|
118 | 118 | pl = [] |
|
119 | 119 | for p in parents: |
|
120 | 120 | if p not in seen: |
|
121 | 121 | pl.append(p) |
|
122 | 122 | seen[p] = 1 |
|
123 | 123 | parents = pl |
|
124 | 124 | |
|
125 | 125 | if len(parents) < 2: parents.append("0" * 40) |
|
126 | 126 | if len(parents) < 2: parents.append("0" * 40) |
|
127 | 127 | p2 = parents.pop(0) |
|
128 | 128 | |
|
129 | 129 | while parents: |
|
130 | 130 | p1 = p2 |
|
131 | 131 | p2 = parents.pop(0) |
|
132 | 132 | self.repo.rawcommit(files, text, author, dest, |
|
133 | 133 | hg.bin(p1), hg.bin(p2)) |
|
134 | 134 | text = "(octopus merge fixup)\n" |
|
135 | 135 | p2 = hg.hex(self.repo.changelog.tip()) |
|
136 | 136 | |
|
137 | 137 | return p2 |
|
138 | 138 | |
|
139 | 139 | def puttags(self, tags): |
|
140 | 140 | try: |
|
141 | 141 | old = self.repo.wfile(".hgtags").read() |
|
142 | 142 | oldlines = old.splitlines(1) |
|
143 | 143 | oldlines.sort() |
|
144 | 144 | except: |
|
145 | 145 | oldlines = [] |
|
146 | 146 | |
|
147 | 147 | k = tags.keys() |
|
148 | 148 | k.sort() |
|
149 | 149 | newlines = [] |
|
150 | 150 | for tag in k: |
|
151 | 151 | newlines.append("%s %s\n" % (tags[tag], tag)) |
|
152 | 152 | |
|
153 | 153 | newlines.sort() |
|
154 | 154 | |
|
155 | 155 | if newlines != oldlines: |
|
156 | 156 | #print "updating tags" |
|
157 | 157 | f = self.repo.wfile(".hgtags", "w") |
|
158 | 158 | f.write("".join(newlines)) |
|
159 | 159 | f.close() |
|
160 | 160 | if not oldlines: self.repo.add([".hgtags"]) |
|
161 | 161 | date = "%s 0" % int(time.mktime(time.gmtime())) |
|
162 | 162 | self.repo.rawcommit([".hgtags"], "update tags", "convert-repo", |
|
163 | 163 | date, self.repo.changelog.tip(), hg.nullid) |
|
164 | 164 | return hg.hex(self.repo.changelog.tip()) |
|
165 | 165 | |
|
166 | 166 | class convert: |
|
167 | 167 | def __init__(self, source, dest, mapfile): |
|
168 | 168 | self.source = source |
|
169 | 169 | self.dest = dest |
|
170 | 170 | self.mapfile = mapfile |
|
171 | 171 | self.commitcache = {} |
|
172 | 172 | |
|
173 | 173 | self.map = {} |
|
174 | 174 | try: |
|
175 | 175 | for l in file(self.mapfile): |
|
176 | 176 | sv, dv = l[:-1].split() |
|
177 | 177 | self.map[sv] = dv |
|
178 | 178 | except IOError: |
|
179 | 179 | pass |
|
180 | 180 | |
|
181 | 181 | def walktree(self, heads): |
|
182 | 182 | visit = heads |
|
183 | 183 | known = {} |
|
184 | 184 | parents = {} |
|
185 | 185 | while visit: |
|
186 | 186 | n = visit.pop(0) |
|
187 | 187 | if n in known or n in self.map: continue |
|
188 | 188 | known[n] = 1 |
|
189 | 189 | self.commitcache[n] = self.source.getcommit(n) |
|
190 | 190 | cp = self.commitcache[n][0] |
|
191 | 191 | for p in cp: |
|
192 | 192 | parents.setdefault(n, []).append(p) |
|
193 | 193 | visit.append(p) |
|
194 | 194 | |
|
195 | 195 | return parents |
|
196 | 196 | |
|
197 | 197 | def toposort(self, parents): |
|
198 | 198 | visit = parents.keys() |
|
199 | 199 | seen = {} |
|
200 | 200 | children = {} |
|
201 | 201 | |
|
202 | 202 | while visit: |
|
203 | 203 | n = visit.pop(0) |
|
204 | 204 | if n in seen: continue |
|
205 | 205 | seen[n] = 1 |
|
206 | 206 | pc = 0 |
|
207 | 207 | if n in parents: |
|
208 | 208 | for p in parents[n]: |
|
209 | 209 | if p not in self.map: pc += 1 |
|
210 | 210 | visit.append(p) |
|
211 | 211 | children.setdefault(p, []).append(n) |
|
212 | 212 | if not pc: root = n |
|
213 | 213 | |
|
214 | 214 | s = [] |
|
215 | 215 | removed = {} |
|
216 | 216 | visit = children.keys() |
|
217 | 217 | while visit: |
|
218 | 218 | n = visit.pop(0) |
|
219 | 219 | if n in removed: continue |
|
220 | 220 | dep = 0 |
|
221 | 221 | if n in parents: |
|
222 | 222 | for p in parents[n]: |
|
223 | 223 | if p in self.map: continue |
|
224 | 224 | if p not in removed: |
|
225 | 225 | # we're still dependent |
|
226 | 226 | visit.append(n) |
|
227 | 227 | dep = 1 |
|
228 | 228 | break |
|
229 | 229 | |
|
230 | 230 | if not dep: |
|
231 | 231 | # all n's parents are in the list |
|
232 | 232 | removed[n] = 1 |
|
233 | 233 | s.append(n) |
|
234 | 234 | if n in children: |
|
235 | 235 | for c in children[n]: |
|
236 | 236 | visit.insert(0, c) |
|
237 | 237 | |
|
238 | 238 | return s |
|
239 | 239 | |
|
240 | 240 | def copy(self, rev): |
|
241 | 241 | p, a, d, t = self.commitcache[rev] |
|
242 | 242 | files = self.source.getchanges(rev) |
|
243 | 243 | |
|
244 | 244 | for f,v,e in files: |
|
245 | 245 | try: |
|
246 | 246 | data = self.source.getfile(f, v) |
|
247 | 247 | except IOError, inst: |
|
248 | 248 | self.dest.delfile(f) |
|
249 | 249 | else: |
|
250 | 250 | self.dest.putfile(f, e, data) |
|
251 | 251 | |
|
252 | 252 | r = [self.map[v] for v in p] |
|
253 | 253 | f = [f for f,v,e in files] |
|
254 | 254 | self.map[rev] = self.dest.putcommit(f, r, a, d, t) |
|
255 | 255 | file(self.mapfile, "a").write("%s %s\n" % (rev, self.map[rev])) |
|
256 | 256 | |
|
257 | 257 | def convert(self): |
|
258 | 258 | heads = self.source.getheads() |
|
259 | 259 | parents = self.walktree(heads) |
|
260 | 260 | t = self.toposort(parents) |
|
261 | 261 | t = [n for n in t if n not in self.map] |
|
262 | 262 | num = len(t) |
|
263 | 263 | c = None |
|
264 | 264 | |
|
265 | 265 | for c in t: |
|
266 | 266 | num -= 1 |
|
267 | 267 | desc = self.commitcache[c][3].splitlines()[0] |
|
268 | 268 | #print num, desc |
|
269 | 269 | self.copy(c) |
|
270 | 270 | |
|
271 | 271 | tags = self.source.gettags() |
|
272 | 272 | ctags = {} |
|
273 | 273 | for k in tags: |
|
274 | 274 | v = tags[k] |
|
275 | 275 | if v in self.map: |
|
276 | 276 | ctags[k] = self.map[v] |
|
277 | 277 | |
|
278 | 278 | if c and ctags: |
|
279 | 279 | nrev = self.dest.puttags(ctags) |
|
280 | 280 | # write another hash correspondence to override the previous |
|
281 | 281 | # one so we don't end up with extra tag heads |
|
282 | 282 | file(self.mapfile, "a").write("%s %s\n" % (c, nrev)) |
|
283 | 283 | |
|
284 | 284 | gitpath, hgpath, mapfile = sys.argv[1:] |
|
285 | 285 | if os.path.isdir(gitpath + "/.git"): |
|
286 | 286 | gitpath += "/.git" |
|
287 | 287 | |
|
288 | 288 | c = convert(convert_git(gitpath), convert_mercurial(hgpath), mapfile) |
|
289 | 289 | c.convert() |
General Comments 0
You need to be logged in to leave comments.
Login now