Mercurial > hg > mercurial-crew-with-dirclash
annotate contrib/convert-repo @ 694:51eb248d3348
Teach convert-repo about tags
Git tags are bad, very bad. More importantly, they're horribly
inconsistent. This drops tags which don't appear to work like most of
the others.
manifest hash: f2dda9e9a3ae8a0d84b19e496059b8a795b8e603
| author | mpm@selenic.com |
|---|---|
| date | Thu, 14 Jul 2005 22:37:46 -0800 |
| parents | 695dd9a491da |
| children | 574869103985 |
| 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): | |
| 31 h = file(self.path + "/.git/HEAD").read()[:-1] | |
| 32 return [h] | |
| 33 | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
34 def catfile(self, rev, type): |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
35 if rev == "0" * 40: raise IOError() |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
36 path = os.getcwd() |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
37 os.chdir(self.path) |
| 694 | 38 fh = os.popen("git-cat-file %s %s 2>/dev/null" % (type, rev)) |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
39 os.chdir(path) |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
40 return fh.read() |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
41 |
| 316 | 42 def getfile(self, name, rev): |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
43 return self.catfile(rev, "blob") |
| 316 | 44 |
| 45 def getchanges(self, version): | |
| 46 path = os.getcwd() | |
| 47 os.chdir(self.path) | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
48 fh = os.popen("git-diff-tree --root -m -r %s" % (version)) |
| 316 | 49 os.chdir(path) |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
50 |
| 316 | 51 changes = [] |
| 52 for l in fh: | |
| 53 if "\t" not in l: continue | |
| 54 m, f = l[:-1].split("\t") | |
| 55 m = m.split() | |
| 56 h = m[3] | |
| 57 p = (m[1] == "100755") | |
| 58 changes.append((f, h, p)) | |
| 59 return changes | |
| 60 | |
| 61 def getcommit(self, version): | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
62 c = self.catfile(version, "commit") # read the commit hash |
| 316 | 63 end = c.find("\n\n") |
| 64 message = c[end+2:] | |
| 65 l = c[:end].splitlines() | |
| 66 manifest = l[0].split()[1] | |
| 67 parents = [] | |
| 68 for e in l[1:]: | |
| 69 n,v = e.split(" ", 1) | |
| 70 if n == "author": | |
| 71 p = v.split() | |
| 72 date = " ".join(p[-2:]) | |
| 73 author = " ".join(p[:-2]) | |
| 74 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
|
75 if n == "committer": |
| 431 | 76 p = v.split() |
| 77 date = " ".join(p[-2:]) | |
| 78 committer = " ".join(p[:-2]) | |
| 79 if committer[0] == "<": committer = committer[1:-1] | |
| 80 message += "\ncommitter: %s %s\n" % (committer, date) | |
| 316 | 81 if n == "parent": parents.append(v) |
| 82 return (parents, author, date, message) | |
| 83 | |
| 694 | 84 def gettags(self): |
| 85 tags = {} | |
| 86 for f in os.listdir(self.path + "/.git/refs/tags"): | |
| 87 try: | |
| 88 h = file(self.path + "/.git/refs/tags/" + f).read().strip() | |
| 89 p, a, d, m = self.getcommit(h) | |
| 90 if not p: p = [h] # git is ugly, don't blame me | |
| 91 tags[f] = p[0] | |
| 92 except: | |
| 93 pass | |
| 94 return tags | |
| 95 | |
| 316 | 96 class convert_mercurial: |
| 97 def __init__(self, path): | |
| 98 self.path = path | |
| 99 u = ui.ui() | |
| 100 self.repo = hg.repository(u, path) | |
| 101 | |
| 102 def getheads(self): | |
| 103 h = self.repo.changelog.heads() | |
| 104 h = [ hg.hex(x) for x in h ] | |
| 105 return h | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
106 |
| 316 | 107 def putfile(self, f, e, data): |
| 108 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
|
109 if self.repo.dirstate.state(f) == '?': |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
110 self.repo.dirstate.update([f], "a") |
|
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
111 |
| 450 | 112 util.set_exec(self.repo.wjoin(f), e) |
| 316 | 113 |
| 114 def delfile(self, f): | |
| 115 try: | |
| 116 os.unlink(self.repo.wjoin(f)) | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
117 #self.repo.remove([f]) |
| 316 | 118 except: |
| 119 pass | |
| 120 | |
| 121 def putcommit(self, files, parents, author, dest, text): | |
| 431 | 122 seen = {} |
| 123 pl = [] | |
| 124 for p in parents: | |
| 125 if p not in seen: | |
| 126 pl.append(p) | |
| 127 seen[p] = 1 | |
| 128 parents = pl | |
| 316 | 129 |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
130 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
|
131 if len(parents) < 2: parents.append("0" * 40) |
| 431 | 132 p2 = parents.pop(0) |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
133 |
| 431 | 134 while parents: |
| 135 p1 = p2 | |
| 136 p2 = parents.pop(0) | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
137 self.repo.rawcommit(files, text, author, dest, |
| 431 | 138 hg.bin(p1), hg.bin(p2)) |
| 139 text = "(octopus merge fixup)\n" | |
| 140 | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
141 return hg.hex(self.repo.changelog.tip()) |
| 316 | 142 |
| 694 | 143 def puttags(self, tags): |
| 144 try: | |
| 145 old = self.repo.wfile(".hgtags").read() | |
| 146 oldlines = old.splitlines(1) | |
| 147 oldlines.sort() | |
| 148 except: | |
| 149 oldlines = [] | |
| 150 | |
| 151 k = tags.keys() | |
| 152 k.sort() | |
| 153 newlines = [] | |
| 154 for tag in k: | |
| 155 newlines.append("%s %s\n" % (tags[tag], tag)) | |
| 156 | |
| 157 newlines.sort() | |
| 158 | |
| 159 if newlines != oldlines: | |
| 160 print "updating tags" | |
| 161 f = self.repo.wfile(".hgtags", "w") | |
| 162 f.write("".join(newlines)) | |
| 163 f.close() | |
| 164 if not oldlines: self.repo.add([".hgtags"]) | |
| 165 date = "%s 0" % time.mktime(time.gmtime()) | |
| 166 self.repo.rawcommit([".hgtags"], "update tags", "convert-repo", | |
| 167 date, self.repo.changelog.tip(), hg.nullid) | |
| 168 | |
| 316 | 169 class convert: |
| 170 def __init__(self, source, dest, mapfile): | |
| 171 self.source = source | |
| 172 self.dest = dest | |
| 173 self.mapfile = mapfile | |
| 174 self.commitcache = {} | |
| 175 | |
| 176 self.map = {} | |
| 177 for l in file(self.mapfile): | |
| 178 sv, dv = l[:-1].split() | |
| 179 self.map[sv] = dv | |
| 180 | |
| 181 def walktree(self, heads): | |
| 182 visit = heads | |
| 183 known = {} | |
| 184 parents = {} | |
| 185 while visit: | |
| 186 n = visit.pop(0) | |
| 187 if n in known or n in self.map: continue | |
| 188 known[n] = 1 | |
| 189 self.commitcache[n] = self.source.getcommit(n) | |
| 190 cp = self.commitcache[n][0] | |
| 191 for p in cp: | |
| 192 parents.setdefault(n, []).append(p) | |
| 193 visit.append(p) | |
| 194 | |
| 195 return parents | |
| 196 | |
| 197 def toposort(self, parents): | |
| 198 visit = parents.keys() | |
| 199 seen = {} | |
| 200 children = {} | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
201 |
| 316 | 202 while visit: |
| 203 n = visit.pop(0) | |
| 204 if n in seen: continue | |
| 205 seen[n] = 1 | |
| 206 pc = 0 | |
| 207 if n in parents: | |
| 208 for p in parents[n]: | |
| 209 if p not in self.map: pc += 1 | |
| 210 visit.append(p) | |
| 211 children.setdefault(p, []).append(n) | |
| 212 if not pc: root = n | |
| 213 | |
| 214 s = [] | |
| 215 removed = {} | |
|
692
695dd9a491da
convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents:
450
diff
changeset
|
216 visit = children.keys() |
| 316 | 217 while visit: |
| 218 n = visit.pop(0) | |
| 219 if n in removed: continue | |
| 220 dep = 0 | |
| 221 if n in parents: | |
| 222 for p in parents[n]: | |
| 223 if p in self.map: continue | |
| 224 if p not in removed: | |
| 225 # we're still dependent | |
| 226 visit.append(n) | |
| 227 dep = 1 | |
| 228 break | |
| 229 | |
| 230 if not dep: | |
| 231 # all n's parents are in the list | |
| 232 removed[n] = 1 | |
| 233 s.append(n) | |
| 234 if n in children: | |
| 235 for c in children[n]: | |
| 236 visit.insert(0, c) | |
| 237 | |
| 238 return s | |
| 239 | |
| 240 def copy(self, rev): | |
| 241 p, a, d, t = self.commitcache[rev] | |
| 242 files = self.source.getchanges(rev) | |
| 243 | |
| 244 for f,v,e in files: | |
| 245 try: | |
| 246 data = self.source.getfile(f, v) | |
| 247 except IOError, inst: | |
| 248 self.dest.delfile(f) | |
| 249 else: | |
| 250 self.dest.putfile(f, e, data) | |
| 251 | |
| 252 r = [self.map[v] for v in p] | |
| 253 f = [f for f,v,e in files] | |
| 254 self.map[rev] = self.dest.putcommit(f, r, a, d, t) | |
| 255 file(self.mapfile, "a").write("%s %s\n" % (rev, self.map[rev])) | |
| 256 | |
| 257 def convert(self): | |
| 258 heads = self.source.getheads() | |
| 259 parents = self.walktree(heads) | |
| 260 t = self.toposort(parents) | |
| 261 num = len(t) | |
| 262 | |
| 263 for c in t: | |
| 264 num -= 1 | |
| 265 if c in self.map: continue | |
| 266 desc = self.commitcache[c][3].splitlines()[0] | |
| 267 print num, desc | |
| 268 self.copy(c) | |
| 269 | |
| 694 | 270 tags = self.source.gettags() |
| 271 ctags = {} | |
| 272 for k in tags: | |
| 273 v = tags[k] | |
| 274 if v in self.map: | |
| 275 ctags[k] = self.map[v] | |
| 276 | |
| 277 self.dest.puttags(ctags) | |
| 278 | |
| 316 | 279 gitpath, hgpath, mapfile = sys.argv[1:] |
| 280 | |
| 281 c = convert(convert_git(gitpath), convert_mercurial(hgpath), mapfile) | |
| 282 c.convert() |