Mercurial > hg > mercurial-crew-with-dirclash
annotate contrib/convert-repo @ 1388:5eb2d3c54165
convert-repo: change duplicate elimination
| author | Matt Mackall <mpm@selenic.com> |
|---|---|
| date | Wed, 05 Oct 2005 19:26:41 -0700 |
| parents | 0c7e8d345564 |
| children | 9b3ef6f3cef5 |
| 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 |
| 450 | 24 from mercurial import hg, ui, util |
| 316 | 25 |
| 26 class convert_git: | |
| 27 def __init__(self, path): | |
| 28 self.path = path | |
| 29 | |
| 30 def getheads(self): | |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
31 return [file(self.path + "/HEAD").read()[:-1]] |
| 316 | 32 |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
33 def catfile(self, rev, type): |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
34 if rev == "0" * 40: raise IOError() |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
35 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
|
36 return fh.read() |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
37 |
| 316 | 38 def getfile(self, name, rev): |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
39 return self.catfile(rev, "blob") |
| 316 | 40 |
| 41 def getchanges(self, version): | |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
42 fh = os.popen("GIT_DIR=%s git-diff-tree --root -m -r %s" % (self.path, version)) |
| 316 | 43 changes = [] |
| 44 for l in fh: | |
| 45 if "\t" not in l: continue | |
| 46 m, f = l[:-1].split("\t") | |
| 47 m = m.split() | |
| 48 h = m[3] | |
| 49 p = (m[1] == "100755") | |
| 50 changes.append((f, h, p)) | |
| 51 return changes | |
| 52 | |
| 53 def getcommit(self, version): | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
54 c = self.catfile(version, "commit") # read the commit hash |
| 316 | 55 end = c.find("\n\n") |
| 56 message = c[end+2:] | |
| 57 l = c[:end].splitlines() | |
| 58 manifest = l[0].split()[1] | |
| 59 parents = [] | |
| 60 for e in l[1:]: | |
| 61 n,v = e.split(" ", 1) | |
| 62 if n == "author": | |
| 63 p = v.split() | |
|
1385
adb3de56635b
convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents:
1335
diff
changeset
|
64 tm, tz = p[-2:] |
| 316 | 65 author = " ".join(p[:-2]) |
| 66 if author[0] == "<": author = author[1:-1] | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
67 if n == "committer": |
| 431 | 68 p = v.split() |
|
1385
adb3de56635b
convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents:
1335
diff
changeset
|
69 tm, tz = p[-2:] |
| 431 | 70 committer = " ".join(p[:-2]) |
| 71 if committer[0] == "<": committer = committer[1:-1] | |
|
1385
adb3de56635b
convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents:
1335
diff
changeset
|
72 message += "\ncommitter: %s\n" % v |
| 316 | 73 if n == "parent": parents.append(v) |
|
1385
adb3de56635b
convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents:
1335
diff
changeset
|
74 |
|
adb3de56635b
convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents:
1335
diff
changeset
|
75 tzs, tzh, tzm = tz[-5:-4] + "1", tz[-4:-2], tz[-2:] |
|
adb3de56635b
convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents:
1335
diff
changeset
|
76 tz = int(tzs) * (int(tzh) * 3600 + int(tzm)) |
|
adb3de56635b
convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents:
1335
diff
changeset
|
77 date = tm + " " + str(tz) |
| 316 | 78 return (parents, author, date, message) |
| 79 | |
| 694 | 80 def gettags(self): |
| 81 tags = {} | |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
82 for f in os.listdir(self.path + "/refs/tags"): |
| 694 | 83 try: |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
84 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
|
85 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
|
86 h = c.splitlines()[0].split()[1] |
| 1237 | 87 tags[f] = h |
| 694 | 88 except: |
| 89 pass | |
| 90 return tags | |
| 91 | |
| 316 | 92 class convert_mercurial: |
| 93 def __init__(self, path): | |
| 94 self.path = path | |
| 95 u = ui.ui() | |
| 96 self.repo = hg.repository(u, path) | |
| 97 | |
| 98 def getheads(self): | |
| 99 h = self.repo.changelog.heads() | |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
100 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
|
101 |
| 316 | 102 def putfile(self, f, e, data): |
| 103 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
|
104 if self.repo.dirstate.state(f) == '?': |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
105 self.repo.dirstate.update([f], "a") |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
106 |
| 450 | 107 util.set_exec(self.repo.wjoin(f), e) |
| 316 | 108 |
| 109 def delfile(self, f): | |
| 110 try: | |
| 111 os.unlink(self.repo.wjoin(f)) | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
112 #self.repo.remove([f]) |
| 316 | 113 except: |
| 114 pass | |
| 115 | |
| 116 def putcommit(self, files, parents, author, dest, text): | |
| 431 | 117 seen = {} |
| 118 pl = [] | |
| 119 for p in parents: | |
| 120 if p not in seen: | |
| 121 pl.append(p) | |
| 122 seen[p] = 1 | |
| 123 parents = pl | |
| 316 | 124 |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
125 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
|
126 if len(parents) < 2: parents.append("0" * 40) |
| 431 | 127 p2 = parents.pop(0) |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
128 |
| 431 | 129 while parents: |
| 130 p1 = p2 | |
| 131 p2 = parents.pop(0) | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
132 self.repo.rawcommit(files, text, author, dest, |
| 431 | 133 hg.bin(p1), hg.bin(p2)) |
| 134 text = "(octopus merge fixup)\n" | |
| 135 | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
136 return hg.hex(self.repo.changelog.tip()) |
| 316 | 137 |
| 694 | 138 def puttags(self, tags): |
| 139 try: | |
| 140 old = self.repo.wfile(".hgtags").read() | |
| 141 oldlines = old.splitlines(1) | |
| 142 oldlines.sort() | |
| 143 except: | |
| 144 oldlines = [] | |
| 145 | |
| 146 k = tags.keys() | |
| 147 k.sort() | |
| 148 newlines = [] | |
| 149 for tag in k: | |
| 150 newlines.append("%s %s\n" % (tags[tag], tag)) | |
| 151 | |
| 152 newlines.sort() | |
| 153 | |
| 154 if newlines != oldlines: | |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
155 #print "updating tags" |
| 694 | 156 f = self.repo.wfile(".hgtags", "w") |
| 157 f.write("".join(newlines)) | |
| 158 f.close() | |
| 159 if not oldlines: self.repo.add([".hgtags"]) | |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
160 date = "%s 0" % int(time.mktime(time.gmtime())) |
| 694 | 161 self.repo.rawcommit([".hgtags"], "update tags", "convert-repo", |
| 162 date, self.repo.changelog.tip(), hg.nullid) | |
|
1387
0c7e8d345564
convert-repo: linearize the tag commit
Matt Mackall <mpm@selenic.com>
parents:
1386
diff
changeset
|
163 return hg.hex(self.repo.changelog.tip()) |
| 694 | 164 |
| 316 | 165 class convert: |
| 166 def __init__(self, source, dest, mapfile): | |
| 167 self.source = source | |
| 168 self.dest = dest | |
| 169 self.mapfile = mapfile | |
| 170 self.commitcache = {} | |
| 171 | |
| 172 self.map = {} | |
| 173 for l in file(self.mapfile): | |
| 174 sv, dv = l[:-1].split() | |
| 175 self.map[sv] = dv | |
| 176 | |
| 177 def walktree(self, heads): | |
| 178 visit = heads | |
| 179 known = {} | |
| 180 parents = {} | |
| 181 while visit: | |
| 182 n = visit.pop(0) | |
| 183 if n in known or n in self.map: continue | |
| 184 known[n] = 1 | |
| 185 self.commitcache[n] = self.source.getcommit(n) | |
| 186 cp = self.commitcache[n][0] | |
| 187 for p in cp: | |
| 188 parents.setdefault(n, []).append(p) | |
| 189 visit.append(p) | |
| 190 | |
| 191 return parents | |
| 192 | |
| 193 def toposort(self, parents): | |
| 194 visit = parents.keys() | |
| 195 seen = {} | |
| 196 children = {} | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
197 |
| 316 | 198 while visit: |
| 199 n = visit.pop(0) | |
| 200 if n in seen: continue | |
| 201 seen[n] = 1 | |
| 202 pc = 0 | |
| 203 if n in parents: | |
| 204 for p in parents[n]: | |
| 205 if p not in self.map: pc += 1 | |
| 206 visit.append(p) | |
| 207 children.setdefault(p, []).append(n) | |
| 208 if not pc: root = n | |
| 209 | |
| 210 s = [] | |
| 211 removed = {} | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
212 visit = children.keys() |
| 316 | 213 while visit: |
| 214 n = visit.pop(0) | |
| 215 if n in removed: continue | |
| 216 dep = 0 | |
| 217 if n in parents: | |
| 218 for p in parents[n]: | |
| 219 if p in self.map: continue | |
| 220 if p not in removed: | |
| 221 # we're still dependent | |
| 222 visit.append(n) | |
| 223 dep = 1 | |
| 224 break | |
| 225 | |
| 226 if not dep: | |
| 227 # all n's parents are in the list | |
| 228 removed[n] = 1 | |
| 229 s.append(n) | |
| 230 if n in children: | |
| 231 for c in children[n]: | |
| 232 visit.insert(0, c) | |
| 233 | |
| 234 return s | |
| 235 | |
| 236 def copy(self, rev): | |
| 237 p, a, d, t = self.commitcache[rev] | |
| 238 files = self.source.getchanges(rev) | |
| 239 | |
| 240 for f,v,e in files: | |
| 241 try: | |
| 242 data = self.source.getfile(f, v) | |
| 243 except IOError, inst: | |
| 244 self.dest.delfile(f) | |
| 245 else: | |
| 246 self.dest.putfile(f, e, data) | |
| 247 | |
| 248 r = [self.map[v] for v in p] | |
| 249 f = [f for f,v,e in files] | |
| 250 self.map[rev] = self.dest.putcommit(f, r, a, d, t) | |
| 251 file(self.mapfile, "a").write("%s %s\n" % (rev, self.map[rev])) | |
| 252 | |
| 253 def convert(self): | |
| 254 heads = self.source.getheads() | |
| 255 parents = self.walktree(heads) | |
| 256 t = self.toposort(parents) | |
|
1388
5eb2d3c54165
convert-repo: change duplicate elimination
Matt Mackall <mpm@selenic.com>
parents:
1387
diff
changeset
|
257 t = [n for n in t if n not in self.map] |
| 316 | 258 num = len(t) |
| 259 | |
| 260 for c in t: | |
| 261 num -= 1 | |
| 262 desc = self.commitcache[c][3].splitlines()[0] | |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
263 #print num, desc |
| 316 | 264 self.copy(c) |
| 265 | |
| 694 | 266 tags = self.source.gettags() |
| 267 ctags = {} | |
| 268 for k in tags: | |
| 269 v = tags[k] | |
| 270 if v in self.map: | |
| 271 ctags[k] = self.map[v] | |
| 272 | |
|
1387
0c7e8d345564
convert-repo: linearize the tag commit
Matt Mackall <mpm@selenic.com>
parents:
1386
diff
changeset
|
273 if ctags: |
|
0c7e8d345564
convert-repo: linearize the tag commit
Matt Mackall <mpm@selenic.com>
parents:
1386
diff
changeset
|
274 nrev = self.dest.puttags(ctags) |
|
0c7e8d345564
convert-repo: linearize the tag commit
Matt Mackall <mpm@selenic.com>
parents:
1386
diff
changeset
|
275 # write another hash correspondence to override the previous |
|
0c7e8d345564
convert-repo: linearize the tag commit
Matt Mackall <mpm@selenic.com>
parents:
1386
diff
changeset
|
276 # 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
|
277 file(self.mapfile, "a").write("%s %s\n" % (c, nrev)) |
| 694 | 278 |
| 316 | 279 gitpath, hgpath, mapfile = sys.argv[1:] |
|
1335
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
280 if os.path.isdir(gitpath + "/.git"): |
|
bea6356b8bca
git -> hg conversion script
Florian La Roche <laroche@redhat.com>
parents:
1237
diff
changeset
|
281 gitpath += "/.git" |
| 316 | 282 |
| 283 c = convert(convert_git(gitpath), convert_mercurial(hgpath), mapfile) | |
| 284 c.convert() |