Mercurial > hg > mercurial-crew-with-dirclash
annotate contrib/convert-repo @ 3861:db36a4f490f6
Indicate the purpose of the dummy changelog file in itself.
Otherwise people getting an error message with an old installation might
wonder why this file is corrupted.
| author | Thomas Arendsen Hein <thomas@intevation.de> |
|---|---|
| date | Sun, 10 Dec 2006 23:03:53 +0100 |
| parents | 158fce02dc40 |
| children | 4bc5a2405b12 |
| rev | line source |
|---|---|
| 316 | 1 #!/usr/bin/env python |
| 2 # | |
| 3 # This is a generalized framework for converting between SCM | |
| 4 # repository formats. | |
| 5 # | |
| 6 # In its current form, it's hardcoded to convert incrementally between | |
| 7 # git and Mercurial. | |
| 8 # | |
| 9 # To use, you must first import the first git version into Mercurial, | |
| 10 # and establish a mapping between the git commit hash and the hash in | |
| 11 # Mercurial for that version. This mapping is kept in a simple text | |
| 12 # file with lines like so: | |
| 13 # | |
| 14 # <git hash> <mercurial hash> | |
| 15 # | |
| 16 # To convert the rest of the repo, run: | |
| 17 # | |
| 18 # convert-repo <git-dir> <hg-dir> <mapfile> | |
| 19 # | |
| 20 # This updates the mapfile on each commit copied, so it can be | |
| 21 # interrupted and can be run repeatedly to copy new commits. | |
| 22 | |
| 694 | 23 import sys, os, zlib, sha, time |
|
3825
158fce02dc40
Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents:
2649
diff
changeset
|
24 |
|
158fce02dc40
Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents:
2649
diff
changeset
|
25 os.environ["HGENCODING"] = "utf-8" |
|
158fce02dc40
Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents:
2649
diff
changeset
|
26 |
|
1715
40346aa66b0f
Revert convert-repo changes
Matt Mackall <mpm@selenic.com>
parents:
1656
diff
changeset
|
27 from mercurial import hg, ui, util |
| 316 | 28 |
|
3825
158fce02dc40
Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents:
2649
diff
changeset
|
29 def recode(s): |
|
158fce02dc40
Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents:
2649
diff
changeset
|
30 try: |
|
158fce02dc40
Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents:
2649
diff
changeset
|
31 return s.decode("utf-8").encode("utf-8") |
|
158fce02dc40
Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents:
2649
diff
changeset
|
32 except: |
|
158fce02dc40
Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents:
2649
diff
changeset
|
33 try: |
|
158fce02dc40
Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents:
2649
diff
changeset
|
34 return s.decode("latin-1").encode("utf-8") |
|
158fce02dc40
Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents:
2649
diff
changeset
|
35 except: |
|
158fce02dc40
Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents:
2649
diff
changeset
|
36 return s.decode("utf-8", "replace").encode("utf-8") |
|
158fce02dc40
Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents:
2649
diff
changeset
|
37 |
| 316 | 38 class convert_git: |
| 39 def __init__(self, path): | |
| 40 self.path = path | |
| 41 | |
| 42 def getheads(self): | |
|
2649
e6a7a6a33a62
make convert-repo deal with git symbolic refs.
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents:
2093
diff
changeset
|
43 fh = os.popen("GIT_DIR=%s git-rev-parse --verify HEAD" % self.path) |
|
e6a7a6a33a62
make convert-repo deal with git symbolic refs.
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents:
2093
diff
changeset
|
44 return [fh.read()[:-1]] |
| 316 | 45 |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
46 def catfile(self, rev, type): |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
47 if rev == "0" * 40: raise IOError() |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
48 fh = os.popen("GIT_DIR=%s git-cat-file %s %s 2>/dev/null" % (self.path, type, rev)) |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
49 return fh.read() |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
50 |
| 316 | 51 def getfile(self, name, rev): |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
52 return self.catfile(rev, "blob") |
| 316 | 53 |
| 54 def getchanges(self, version): | |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
55 fh = os.popen("GIT_DIR=%s git-diff-tree --root -m -r %s" % (self.path, version)) |
| 316 | 56 changes = [] |
| 57 for l in fh: | |
| 58 if "\t" not in l: continue | |
| 59 m, f = l[:-1].split("\t") | |
| 60 m = m.split() | |
| 61 h = m[3] | |
| 62 p = (m[1] == "100755") | |
| 63 changes.append((f, h, p)) | |
| 64 return changes | |
| 65 | |
| 66 def getcommit(self, version): | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
67 c = self.catfile(version, "commit") # read the commit hash |
| 316 | 68 end = c.find("\n\n") |
| 69 message = c[end+2:] | |
|
3825
158fce02dc40
Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents:
2649
diff
changeset
|
70 message = recode(message) |
| 316 | 71 l = c[:end].splitlines() |
| 72 manifest = l[0].split()[1] | |
| 73 parents = [] | |
| 74 for e in l[1:]: | |
| 75 n,v = e.split(" ", 1) | |
| 76 if n == "author": | |
| 77 p = v.split() | |
|
1385
adb3de56635b
convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents:
1335
diff
changeset
|
78 tm, tz = p[-2:] |
| 316 | 79 author = " ".join(p[:-2]) |
| 80 if author[0] == "<": author = author[1:-1] | |
|
3825
158fce02dc40
Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents:
2649
diff
changeset
|
81 author = recode(author) |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
82 if n == "committer": |
| 431 | 83 p = v.split() |
|
1385
adb3de56635b
convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents:
1335
diff
changeset
|
84 tm, tz = p[-2:] |
| 431 | 85 committer = " ".join(p[:-2]) |
| 86 if committer[0] == "<": committer = committer[1:-1] | |
|
3825
158fce02dc40
Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents:
2649
diff
changeset
|
87 committer = recode(committer) |
|
1385
adb3de56635b
convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents:
1335
diff
changeset
|
88 message += "\ncommitter: %s\n" % v |
| 316 | 89 if n == "parent": parents.append(v) |
|
1385
adb3de56635b
convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents:
1335
diff
changeset
|
90 |
|
adb3de56635b
convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents:
1335
diff
changeset
|
91 tzs, tzh, tzm = tz[-5:-4] + "1", tz[-4:-2], tz[-2:] |
|
2093
5cc414722587
convert-repo: fix reversed time zone offset
Vadim Gelfer <vadim.gelfer@gmail.com>
parents:
1715
diff
changeset
|
92 tz = -int(tzs) * (int(tzh) * 3600 + int(tzm)) |
|
1385
adb3de56635b
convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents:
1335
diff
changeset
|
93 date = tm + " " + str(tz) |
| 316 | 94 return (parents, author, date, message) |
| 95 | |
| 694 | 96 def gettags(self): |
| 97 tags = {} | |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
98 for f in os.listdir(self.path + "/refs/tags"): |
| 694 | 99 try: |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
100 h = file(self.path + "/refs/tags/" + f).read().strip() |
|
1386
a1040345fdda
convert-repo: retrieve the commit hash from the tag object for tag import
Matt Mackall <mpm@selenic.com>
parents:
1385
diff
changeset
|
101 c = self.catfile(h, "tag") # read the commit hash |
|
a1040345fdda
convert-repo: retrieve the commit hash from the tag object for tag import
Matt Mackall <mpm@selenic.com>
parents:
1385
diff
changeset
|
102 h = c.splitlines()[0].split()[1] |
| 1237 | 103 tags[f] = h |
| 694 | 104 except: |
| 105 pass | |
| 106 return tags | |
| 107 | |
| 316 | 108 class convert_mercurial: |
| 109 def __init__(self, path): | |
| 110 self.path = path | |
| 111 u = ui.ui() | |
| 112 self.repo = hg.repository(u, path) | |
| 113 | |
| 114 def getheads(self): | |
| 115 h = self.repo.changelog.heads() | |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
116 return [ hg.hex(x) for x in h ] |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
117 |
| 316 | 118 def putfile(self, f, e, data): |
| 119 self.repo.wfile(f, "w").write(data) | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
120 if self.repo.dirstate.state(f) == '?': |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
121 self.repo.dirstate.update([f], "a") |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
122 |
| 450 | 123 util.set_exec(self.repo.wjoin(f), e) |
| 316 | 124 |
| 125 def delfile(self, f): | |
| 126 try: | |
| 127 os.unlink(self.repo.wjoin(f)) | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
128 #self.repo.remove([f]) |
| 316 | 129 except: |
| 130 pass | |
| 131 | |
|
1715
40346aa66b0f
Revert convert-repo changes
Matt Mackall <mpm@selenic.com>
parents:
1656
diff
changeset
|
132 def putcommit(self, files, parents, author, dest, text): |
| 431 | 133 seen = {} |
| 134 pl = [] | |
| 135 for p in parents: | |
| 136 if p not in seen: | |
| 137 pl.append(p) | |
| 138 seen[p] = 1 | |
| 139 parents = pl | |
| 316 | 140 |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
141 if len(parents) < 2: parents.append("0" * 40) |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
142 if len(parents) < 2: parents.append("0" * 40) |
| 431 | 143 p2 = parents.pop(0) |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
144 |
| 431 | 145 while parents: |
| 146 p1 = p2 | |
| 147 p2 = parents.pop(0) | |
|
1715
40346aa66b0f
Revert convert-repo changes
Matt Mackall <mpm@selenic.com>
parents:
1656
diff
changeset
|
148 self.repo.rawcommit(files, text, author, dest, |
|
40346aa66b0f
Revert convert-repo changes
Matt Mackall <mpm@selenic.com>
parents:
1656
diff
changeset
|
149 hg.bin(p1), hg.bin(p2)) |
| 431 | 150 text = "(octopus merge fixup)\n" |
|
1389
9b3ef6f3cef5
convert-repo: fix up octopus merge conversion
Matt Mackall <mpm@selenic.com>
parents:
1388
diff
changeset
|
151 p2 = hg.hex(self.repo.changelog.tip()) |
| 431 | 152 |
|
1389
9b3ef6f3cef5
convert-repo: fix up octopus merge conversion
Matt Mackall <mpm@selenic.com>
parents:
1388
diff
changeset
|
153 return p2 |
| 316 | 154 |
| 694 | 155 def puttags(self, tags): |
| 156 try: | |
| 157 old = self.repo.wfile(".hgtags").read() | |
| 158 oldlines = old.splitlines(1) | |
| 159 oldlines.sort() | |
| 160 except: | |
| 161 oldlines = [] | |
| 162 | |
| 163 k = tags.keys() | |
| 164 k.sort() | |
| 165 newlines = [] | |
| 166 for tag in k: | |
| 167 newlines.append("%s %s\n" % (tags[tag], tag)) | |
| 168 | |
| 169 newlines.sort() | |
| 170 | |
| 171 if newlines != oldlines: | |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
172 #print "updating tags" |
| 694 | 173 f = self.repo.wfile(".hgtags", "w") |
| 174 f.write("".join(newlines)) | |
| 175 f.close() | |
| 176 if not oldlines: self.repo.add([".hgtags"]) | |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
177 date = "%s 0" % int(time.mktime(time.gmtime())) |
| 694 | 178 self.repo.rawcommit([".hgtags"], "update tags", "convert-repo", |
| 179 date, self.repo.changelog.tip(), hg.nullid) | |
|
1387
0c7e8d345564
convert-repo: linearize the tag commit
Matt Mackall <mpm@selenic.com>
parents:
1386
diff
changeset
|
180 return hg.hex(self.repo.changelog.tip()) |
| 694 | 181 |
| 316 | 182 class convert: |
| 183 def __init__(self, source, dest, mapfile): | |
| 184 self.source = source | |
| 185 self.dest = dest | |
| 186 self.mapfile = mapfile | |
| 187 self.commitcache = {} | |
| 188 | |
| 189 self.map = {} | |
|
1655
7bfd4724932a
convert-repo: automatically create empty map file
Matt Mackall <mpm@selenic.com>
parents:
1389
diff
changeset
|
190 try: |
|
7bfd4724932a
convert-repo: automatically create empty map file
Matt Mackall <mpm@selenic.com>
parents:
1389
diff
changeset
|
191 for l in file(self.mapfile): |
|
7bfd4724932a
convert-repo: automatically create empty map file
Matt Mackall <mpm@selenic.com>
parents:
1389
diff
changeset
|
192 sv, dv = l[:-1].split() |
|
7bfd4724932a
convert-repo: automatically create empty map file
Matt Mackall <mpm@selenic.com>
parents:
1389
diff
changeset
|
193 self.map[sv] = dv |
|
7bfd4724932a
convert-repo: automatically create empty map file
Matt Mackall <mpm@selenic.com>
parents:
1389
diff
changeset
|
194 except IOError: |
|
7bfd4724932a
convert-repo: automatically create empty map file
Matt Mackall <mpm@selenic.com>
parents:
1389
diff
changeset
|
195 pass |
| 316 | 196 |
| 197 def walktree(self, heads): | |
| 198 visit = heads | |
| 199 known = {} | |
| 200 parents = {} | |
| 201 while visit: | |
| 202 n = visit.pop(0) | |
| 203 if n in known or n in self.map: continue | |
| 204 known[n] = 1 | |
| 205 self.commitcache[n] = self.source.getcommit(n) | |
| 206 cp = self.commitcache[n][0] | |
| 207 for p in cp: | |
| 208 parents.setdefault(n, []).append(p) | |
| 209 visit.append(p) | |
| 210 | |
| 211 return parents | |
| 212 | |
| 213 def toposort(self, parents): | |
| 214 visit = parents.keys() | |
| 215 seen = {} | |
| 216 children = {} | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
217 |
| 316 | 218 while visit: |
| 219 n = visit.pop(0) | |
| 220 if n in seen: continue | |
| 221 seen[n] = 1 | |
| 222 pc = 0 | |
| 223 if n in parents: | |
| 224 for p in parents[n]: | |
| 225 if p not in self.map: pc += 1 | |
| 226 visit.append(p) | |
| 227 children.setdefault(p, []).append(n) | |
| 228 if not pc: root = n | |
| 229 | |
| 230 s = [] | |
| 231 removed = {} | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
232 visit = children.keys() |
| 316 | 233 while visit: |
| 234 n = visit.pop(0) | |
| 235 if n in removed: continue | |
| 236 dep = 0 | |
| 237 if n in parents: | |
| 238 for p in parents[n]: | |
| 239 if p in self.map: continue | |
| 240 if p not in removed: | |
| 241 # we're still dependent | |
| 242 visit.append(n) | |
| 243 dep = 1 | |
| 244 break | |
| 245 | |
| 246 if not dep: | |
| 247 # all n's parents are in the list | |
| 248 removed[n] = 1 | |
| 249 s.append(n) | |
| 250 if n in children: | |
| 251 for c in children[n]: | |
| 252 visit.insert(0, c) | |
| 253 | |
| 254 return s | |
| 255 | |
| 256 def copy(self, rev): | |
| 257 p, a, d, t = self.commitcache[rev] | |
| 258 files = self.source.getchanges(rev) | |
| 259 | |
| 260 for f,v,e in files: | |
| 261 try: | |
| 262 data = self.source.getfile(f, v) | |
| 263 except IOError, inst: | |
| 264 self.dest.delfile(f) | |
| 265 else: | |
| 266 self.dest.putfile(f, e, data) | |
| 267 | |
| 268 r = [self.map[v] for v in p] | |
| 269 f = [f for f,v,e in files] | |
| 270 self.map[rev] = self.dest.putcommit(f, r, a, d, t) | |
| 271 file(self.mapfile, "a").write("%s %s\n" % (rev, self.map[rev])) | |
| 272 | |
| 273 def convert(self): | |
| 274 heads = self.source.getheads() | |
| 275 parents = self.walktree(heads) | |
| 276 t = self.toposort(parents) | |
|
1388
5eb2d3c54165
convert-repo: change duplicate elimination
Matt Mackall <mpm@selenic.com>
parents:
1387
diff
changeset
|
277 t = [n for n in t if n not in self.map] |
| 316 | 278 num = len(t) |
|
1715
40346aa66b0f
Revert convert-repo changes
Matt Mackall <mpm@selenic.com>
parents:
1656
diff
changeset
|
279 c = None |
| 316 | 280 |
| 281 for c in t: | |
| 282 num -= 1 | |
| 283 desc = self.commitcache[c][3].splitlines()[0] | |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
284 #print num, desc |
| 316 | 285 self.copy(c) |
| 286 | |
| 694 | 287 tags = self.source.gettags() |
| 288 ctags = {} | |
| 289 for k in tags: | |
| 290 v = tags[k] | |
| 291 if v in self.map: | |
| 292 ctags[k] = self.map[v] | |
| 293 | |
|
1715
40346aa66b0f
Revert convert-repo changes
Matt Mackall <mpm@selenic.com>
parents:
1656
diff
changeset
|
294 if c and ctags: |
|
1387
0c7e8d345564
convert-repo: linearize the tag commit
Matt Mackall <mpm@selenic.com>
parents:
1386
diff
changeset
|
295 nrev = self.dest.puttags(ctags) |
|
0c7e8d345564
convert-repo: linearize the tag commit
Matt Mackall <mpm@selenic.com>
parents:
1386
diff
changeset
|
296 # write another hash correspondence to override the previous |
|
0c7e8d345564
convert-repo: linearize the tag commit
Matt Mackall <mpm@selenic.com>
parents:
1386
diff
changeset
|
297 # one so we don't end up with extra tag heads |
|
0c7e8d345564
convert-repo: linearize the tag commit
Matt Mackall <mpm@selenic.com>
parents:
1386
diff
changeset
|
298 file(self.mapfile, "a").write("%s %s\n" % (c, nrev)) |
| 694 | 299 |
| 316 | 300 gitpath, hgpath, mapfile = sys.argv[1:] |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
301 if os.path.isdir(gitpath + "/.git"): |
|
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
302 gitpath += "/.git" |
| 316 | 303 |
| 304 c = convert(convert_git(gitpath), convert_mercurial(hgpath), mapfile) | |
| 305 c.convert() |