|
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 |
|
23 import sys, os, zlib, sha |
|
24 from mercurial import hg, ui |
|
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 |
|
34 def getfile(self, name, rev): |
|
35 a = file(self.path + ("/.git/objects/%s/%s" |
|
36 % (rev[:2], rev[2:]))).read() |
|
37 b = zlib.decompress(a) |
|
38 if sha.sha(b).hexdigest() != rev: raise "bad hash" |
|
39 head, text = b.split('\0', 1) |
|
40 return text |
|
41 |
|
42 def getchanges(self, version): |
|
43 path = os.getcwd() |
|
44 os.chdir(self.path) |
|
45 fh = os.popen("git-diff-tree -m -r %s" % (version)) |
|
46 os.chdir(path) |
|
47 |
|
48 changes = [] |
|
49 for l in fh: |
|
50 if "\t" not in l: continue |
|
51 m, f = l[:-1].split("\t") |
|
52 m = m.split() |
|
53 h = m[3] |
|
54 p = (m[1] == "100755") |
|
55 changes.append((f, h, p)) |
|
56 return changes |
|
57 |
|
58 def getcommit(self, version): |
|
59 c = self.getfile("", version) # read the commit hash |
|
60 end = c.find("\n\n") |
|
61 message = c[end+2:] |
|
62 l = c[:end].splitlines() |
|
63 manifest = l[0].split()[1] |
|
64 parents = [] |
|
65 for e in l[1:]: |
|
66 n,v = e.split(" ", 1) |
|
67 if n == "author": |
|
68 p = v.split() |
|
69 date = " ".join(p[-2:]) |
|
70 author = " ".join(p[:-2]) |
|
71 if author[0] == "<": author = author[1:-1] |
|
72 if n == "parent": parents.append(v) |
|
73 return (parents, author, date, message) |
|
74 |
|
75 class convert_mercurial: |
|
76 def __init__(self, path): |
|
77 self.path = path |
|
78 u = ui.ui() |
|
79 self.repo = hg.repository(u, path) |
|
80 |
|
81 def getheads(self): |
|
82 h = self.repo.changelog.heads() |
|
83 h = [ hg.hex(x) for x in h ] |
|
84 return h |
|
85 |
|
86 def putfile(self, f, e, data): |
|
87 self.repo.wfile(f, "w").write(data) |
|
88 hg.set_exec(self.repo.wjoin(f), e) |
|
89 |
|
90 def delfile(self, f): |
|
91 try: |
|
92 os.unlink(self.repo.wjoin(f)) |
|
93 self.repo.remove([f]) |
|
94 except: |
|
95 pass |
|
96 |
|
97 def putcommit(self, files, parents, author, dest, text): |
|
98 p1, p2 = "0"*40, "0"*40 |
|
99 if len(parents) > 0: p1 = parents[0] |
|
100 if len(parents) > 1: p2 = parents[1] |
|
101 if len(parents) > 2: raise "the dreaded octopus merge!" |
|
102 self.repo.rawcommit(files, text, author, dest, |
|
103 hg.bin(p1), hg.bin(p2)) |
|
104 |
|
105 return hg.hex(self.repo.changelog.tip()) |
|
106 |
|
107 class convert: |
|
108 def __init__(self, source, dest, mapfile): |
|
109 self.source = source |
|
110 self.dest = dest |
|
111 self.mapfile = mapfile |
|
112 self.commitcache = {} |
|
113 |
|
114 self.map = {} |
|
115 for l in file(self.mapfile): |
|
116 sv, dv = l[:-1].split() |
|
117 self.map[sv] = dv |
|
118 |
|
119 def walktree(self, heads): |
|
120 visit = heads |
|
121 known = {} |
|
122 parents = {} |
|
123 while visit: |
|
124 n = visit.pop(0) |
|
125 if n in known or n in self.map: continue |
|
126 known[n] = 1 |
|
127 self.commitcache[n] = self.source.getcommit(n) |
|
128 cp = self.commitcache[n][0] |
|
129 for p in cp: |
|
130 parents.setdefault(n, []).append(p) |
|
131 visit.append(p) |
|
132 |
|
133 return parents |
|
134 |
|
135 def toposort(self, parents): |
|
136 visit = parents.keys() |
|
137 seen = {} |
|
138 children = {} |
|
139 while visit: |
|
140 n = visit.pop(0) |
|
141 if n in seen: continue |
|
142 seen[n] = 1 |
|
143 pc = 0 |
|
144 if n in parents: |
|
145 for p in parents[n]: |
|
146 if p not in self.map: pc += 1 |
|
147 visit.append(p) |
|
148 children.setdefault(p, []).append(n) |
|
149 if not pc: root = n |
|
150 |
|
151 s = [] |
|
152 removed = {} |
|
153 visit = parents.keys() |
|
154 while visit: |
|
155 n = visit.pop(0) |
|
156 if n in removed: continue |
|
157 dep = 0 |
|
158 if n in parents: |
|
159 for p in parents[n]: |
|
160 if p in self.map: continue |
|
161 if p not in removed: |
|
162 # we're still dependent |
|
163 visit.append(n) |
|
164 dep = 1 |
|
165 break |
|
166 |
|
167 if not dep: |
|
168 # all n's parents are in the list |
|
169 removed[n] = 1 |
|
170 s.append(n) |
|
171 if n in children: |
|
172 for c in children[n]: |
|
173 visit.insert(0, c) |
|
174 |
|
175 return s |
|
176 |
|
177 def copy(self, rev): |
|
178 p, a, d, t = self.commitcache[rev] |
|
179 files = self.source.getchanges(rev) |
|
180 |
|
181 for f,v,e in files: |
|
182 try: |
|
183 data = self.source.getfile(f, v) |
|
184 except IOError, inst: |
|
185 self.dest.delfile(f) |
|
186 else: |
|
187 self.dest.putfile(f, e, data) |
|
188 |
|
189 r = [self.map[v] for v in p] |
|
190 f = [f for f,v,e in files] |
|
191 self.map[rev] = self.dest.putcommit(f, r, a, d, t) |
|
192 file(self.mapfile, "a").write("%s %s\n" % (rev, self.map[rev])) |
|
193 |
|
194 def convert(self): |
|
195 heads = self.source.getheads() |
|
196 parents = self.walktree(heads) |
|
197 t = self.toposort(parents) |
|
198 num = len(t) |
|
199 |
|
200 for c in t: |
|
201 num -= 1 |
|
202 if c in self.map: continue |
|
203 desc = self.commitcache[c][3].splitlines()[0] |
|
204 print num, desc |
|
205 self.copy(c) |
|
206 |
|
207 gitpath, hgpath, mapfile = sys.argv[1:] |
|
208 |
|
209 c = convert(convert_git(gitpath), convert_mercurial(hgpath), mapfile) |
|
210 c.convert() |