Show More
@@ -1,866 +1,871 | |||
|
1 | 1 | # Subversion 1.4/1.5 Python API backend |
|
2 | 2 | # |
|
3 | 3 | # Copyright(C) 2007 Daniel Holth et al |
|
4 | 4 | # |
|
5 | 5 | # Configuration options: |
|
6 | 6 | # |
|
7 | 7 | # convert.svn.trunk |
|
8 | 8 | # Relative path to the trunk (default: "trunk") |
|
9 | 9 | # convert.svn.branches |
|
10 | 10 | # Relative path to tree of branches (default: "branches") |
|
11 | 11 | # convert.svn.tags |
|
12 | 12 | # Relative path to tree of tags (default: "tags") |
|
13 | 13 | # |
|
14 | 14 | # Set these in a hgrc, or on the command line as follows: |
|
15 | 15 | # |
|
16 | 16 | # hg convert --config convert.svn.trunk=wackoname [...] |
|
17 | 17 | |
|
18 | 18 | import locale |
|
19 | 19 | import os |
|
20 | 20 | import re |
|
21 | 21 | import sys |
|
22 | 22 | import cPickle as pickle |
|
23 | 23 | import tempfile |
|
24 | 24 | |
|
25 | 25 | from mercurial import strutil, util |
|
26 | 26 | from mercurial.i18n import _ |
|
27 | 27 | |
|
28 | 28 | # Subversion stuff. Works best with very recent Python SVN bindings |
|
29 | 29 | # e.g. SVN 1.5 or backports. Thanks to the bzr folks for enhancing |
|
30 | 30 | # these bindings. |
|
31 | 31 | |
|
32 | 32 | from cStringIO import StringIO |
|
33 | 33 | |
|
34 | 34 | from common import NoRepo, commit, converter_source, encodeargs, decodeargs |
|
35 | 35 | from common import commandline, converter_sink, mapfile |
|
36 | 36 | |
|
37 | 37 | try: |
|
38 | 38 | from svn.core import SubversionException, Pool |
|
39 | 39 | import svn |
|
40 | 40 | import svn.client |
|
41 | 41 | import svn.core |
|
42 | 42 | import svn.ra |
|
43 | 43 | import svn.delta |
|
44 | 44 | import transport |
|
45 | 45 | except ImportError: |
|
46 | 46 | pass |
|
47 | 47 | |
|
48 | 48 | def geturl(path): |
|
49 | 49 | try: |
|
50 | 50 | return svn.client.url_from_path(svn.core.svn_path_canonicalize(path)) |
|
51 | 51 | except SubversionException: |
|
52 | 52 | pass |
|
53 | 53 | if os.path.isdir(path): |
|
54 | 54 | return 'file://%s' % os.path.normpath(os.path.abspath(path)) |
|
55 | 55 | return path |
|
56 | 56 | |
|
57 | 57 | def optrev(number): |
|
58 | 58 | optrev = svn.core.svn_opt_revision_t() |
|
59 | 59 | optrev.kind = svn.core.svn_opt_revision_number |
|
60 | 60 | optrev.value.number = number |
|
61 | 61 | return optrev |
|
62 | 62 | |
|
63 | 63 | class changedpath(object): |
|
64 | 64 | def __init__(self, p): |
|
65 | 65 | self.copyfrom_path = p.copyfrom_path |
|
66 | 66 | self.copyfrom_rev = p.copyfrom_rev |
|
67 | 67 | self.action = p.action |
|
68 | 68 | |
|
69 | 69 | def get_log_child(fp, url, paths, start, end, limit=0, discover_changed_paths=True, |
|
70 | 70 | strict_node_history=False): |
|
71 | 71 | protocol = -1 |
|
72 | 72 | def receiver(orig_paths, revnum, author, date, message, pool): |
|
73 | 73 | if orig_paths is not None: |
|
74 | 74 | for k, v in orig_paths.iteritems(): |
|
75 | 75 | orig_paths[k] = changedpath(v) |
|
76 | 76 | pickle.dump((orig_paths, revnum, author, date, message), |
|
77 | 77 | fp, protocol) |
|
78 | 78 | |
|
79 | 79 | try: |
|
80 | 80 | # Use an ra of our own so that our parent can consume |
|
81 | 81 | # our results without confusing the server. |
|
82 | 82 | t = transport.SvnRaTransport(url=url) |
|
83 | 83 | svn.ra.get_log(t.ra, paths, start, end, limit, |
|
84 | 84 | discover_changed_paths, |
|
85 | 85 | strict_node_history, |
|
86 | 86 | receiver) |
|
87 | 87 | except SubversionException, (inst, num): |
|
88 | 88 | pickle.dump(num, fp, protocol) |
|
89 | 89 | else: |
|
90 | 90 | pickle.dump(None, fp, protocol) |
|
91 | 91 | fp.close() |
|
92 | 92 | |
|
93 | 93 | def debugsvnlog(ui, **opts): |
|
94 | 94 | """Fetch SVN log in a subprocess and channel them back to parent to |
|
95 | 95 | avoid memory collection issues. |
|
96 | 96 | """ |
|
97 | 97 | util.set_binary(sys.stdin) |
|
98 | 98 | util.set_binary(sys.stdout) |
|
99 | 99 | args = decodeargs(sys.stdin.read()) |
|
100 | 100 | get_log_child(sys.stdout, *args) |
|
101 | 101 | |
|
102 | 102 | # SVN conversion code stolen from bzr-svn and tailor |
|
103 | 103 | class svn_source(converter_source): |
|
104 | 104 | def __init__(self, ui, url, rev=None): |
|
105 | 105 | super(svn_source, self).__init__(ui, url, rev=rev) |
|
106 | 106 | |
|
107 | 107 | try: |
|
108 | 108 | SubversionException |
|
109 | 109 | except NameError: |
|
110 | 110 | raise NoRepo('Subversion python bindings could not be loaded') |
|
111 | 111 | |
|
112 | 112 | self.encoding = locale.getpreferredencoding() |
|
113 | 113 | self.lastrevs = {} |
|
114 | 114 | |
|
115 | 115 | latest = None |
|
116 | 116 | try: |
|
117 | 117 | # Support file://path@rev syntax. Useful e.g. to convert |
|
118 | 118 | # deleted branches. |
|
119 | 119 | at = url.rfind('@') |
|
120 | 120 | if at >= 0: |
|
121 | 121 | latest = int(url[at+1:]) |
|
122 | 122 | url = url[:at] |
|
123 | 123 | except ValueError, e: |
|
124 | 124 | pass |
|
125 | 125 | self.url = geturl(url) |
|
126 | 126 | self.encoding = 'UTF-8' # Subversion is always nominal UTF-8 |
|
127 | 127 | try: |
|
128 | 128 | self.transport = transport.SvnRaTransport(url=self.url) |
|
129 | 129 | self.ra = self.transport.ra |
|
130 | 130 | self.ctx = self.transport.client |
|
131 | 131 | self.base = svn.ra.get_repos_root(self.ra) |
|
132 | 132 | self.module = self.url[len(self.base):] |
|
133 | 133 | self.modulemap = {} # revision, module |
|
134 | 134 | self.commits = {} |
|
135 | 135 | self.paths = {} |
|
136 | 136 | self.uuid = svn.ra.get_uuid(self.ra).decode(self.encoding) |
|
137 | 137 | except SubversionException, e: |
|
138 | 138 | ui.print_exc() |
|
139 | 139 | raise NoRepo("%s does not look like a Subversion repo" % self.url) |
|
140 | 140 | |
|
141 | 141 | if rev: |
|
142 | 142 | try: |
|
143 | 143 | latest = int(rev) |
|
144 | 144 | except ValueError: |
|
145 | 145 | raise util.Abort('svn: revision %s is not an integer' % rev) |
|
146 | 146 | |
|
147 | 147 | try: |
|
148 | 148 | self.get_blacklist() |
|
149 | 149 | except IOError, e: |
|
150 | 150 | pass |
|
151 | 151 | |
|
152 | 152 | self.last_changed = self.latest(self.module, latest) |
|
153 | 153 | |
|
154 | 154 | self.head = self.revid(self.last_changed) |
|
155 | 155 | self._changescache = None |
|
156 | 156 | |
|
157 | 157 | def setrevmap(self, revmap): |
|
158 | 158 | lastrevs = {} |
|
159 | 159 | for revid in revmap.iterkeys(): |
|
160 | 160 | uuid, module, revnum = self.revsplit(revid) |
|
161 | 161 | lastrevnum = lastrevs.setdefault(module, revnum) |
|
162 | 162 | if revnum > lastrevnum: |
|
163 | 163 | lastrevs[module] = revnum |
|
164 | 164 | self.lastrevs = lastrevs |
|
165 | 165 | |
|
166 | 166 | def exists(self, path, optrev): |
|
167 | 167 | try: |
|
168 | 168 | svn.client.ls(self.url.rstrip('/') + '/' + path, |
|
169 | 169 | optrev, False, self.ctx) |
|
170 | 170 | return True |
|
171 | 171 | except SubversionException, err: |
|
172 | 172 | return False |
|
173 | 173 | |
|
174 | 174 | def getheads(self): |
|
175 | 175 | # detect standard /branches, /tags, /trunk layout |
|
176 | 176 | rev = optrev(self.last_changed) |
|
177 | 177 | rpath = self.url.strip('/') |
|
178 | 178 | cfgtrunk = self.ui.config('convert', 'svn.trunk') |
|
179 | 179 | cfgbranches = self.ui.config('convert', 'svn.branches') |
|
180 | 180 | cfgtags = self.ui.config('convert', 'svn.tags') |
|
181 | 181 | trunk = (cfgtrunk or 'trunk').strip('/') |
|
182 | 182 | branches = (cfgbranches or 'branches').strip('/') |
|
183 | 183 | tags = (cfgtags or 'tags').strip('/') |
|
184 | 184 | if self.exists(trunk, rev) and self.exists(branches, rev) and self.exists(tags, rev): |
|
185 | 185 | self.ui.note('found trunk at %r, branches at %r and tags at %r\n' % |
|
186 | 186 | (trunk, branches, tags)) |
|
187 | 187 | oldmodule = self.module |
|
188 | 188 | self.module += '/' + trunk |
|
189 | 189 | lt = self.latest(self.module, self.last_changed) |
|
190 | 190 | self.head = self.revid(lt) |
|
191 | 191 | self.heads = [self.head] |
|
192 | 192 | branchnames = svn.client.ls(rpath + '/' + branches, rev, False, |
|
193 | 193 | self.ctx) |
|
194 | 194 | for branch in branchnames.keys(): |
|
195 | 195 | if oldmodule: |
|
196 | 196 | module = oldmodule + '/' + branches + '/' + branch |
|
197 | 197 | else: |
|
198 | 198 | module = '/' + branches + '/' + branch |
|
199 | 199 | brevnum = self.latest(module, self.last_changed) |
|
200 | 200 | brev = self.revid(brevnum, module) |
|
201 | 201 | self.ui.note('found branch %s at %d\n' % (branch, brevnum)) |
|
202 | 202 | self.heads.append(brev) |
|
203 | 203 | |
|
204 | 204 | if oldmodule: |
|
205 | 205 | self.tags = '%s/%s' % (oldmodule, tags) |
|
206 | 206 | else: |
|
207 | 207 | self.tags = '/%s' % tags |
|
208 | 208 | |
|
209 | 209 | elif cfgtrunk or cfgbranches or cfgtags: |
|
210 | 210 | raise util.Abort('trunk/branch/tags layout expected, but not found') |
|
211 | 211 | else: |
|
212 | 212 | self.ui.note('working with one branch\n') |
|
213 | 213 | self.heads = [self.head] |
|
214 | 214 | self.tags = tags |
|
215 | 215 | return self.heads |
|
216 | 216 | |
|
217 | 217 | def getfile(self, file, rev): |
|
218 | 218 | data, mode = self._getfile(file, rev) |
|
219 | 219 | self.modecache[(file, rev)] = mode |
|
220 | 220 | return data |
|
221 | 221 | |
|
222 | 222 | def getmode(self, file, rev): |
|
223 | 223 | return self.modecache[(file, rev)] |
|
224 | 224 | |
|
225 | 225 | def getchanges(self, rev): |
|
226 | 226 | if self._changescache and self._changescache[0] == rev: |
|
227 | 227 | return self._changescache[1] |
|
228 | 228 | self._changescache = None |
|
229 | 229 | self.modecache = {} |
|
230 | 230 | (paths, parents) = self.paths[rev] |
|
231 | 231 | files, copies = self.expandpaths(rev, paths, parents) |
|
232 | 232 | files.sort() |
|
233 | 233 | files = zip(files, [rev] * len(files)) |
|
234 | 234 | |
|
235 | 235 | # caller caches the result, so free it here to release memory |
|
236 | 236 | del self.paths[rev] |
|
237 | 237 | return (files, copies) |
|
238 | 238 | |
|
239 | 239 | def getchangedfiles(self, rev, i): |
|
240 | 240 | changes = self.getchanges(rev) |
|
241 | 241 | self._changescache = (rev, changes) |
|
242 | 242 | return [f[0] for f in changes[0]] |
|
243 | 243 | |
|
244 | 244 | def getcommit(self, rev): |
|
245 | 245 | if rev not in self.commits: |
|
246 | 246 | uuid, module, revnum = self.revsplit(rev) |
|
247 | 247 | self.module = module |
|
248 | 248 | self.reparent(module) |
|
249 | 249 | stop = self.lastrevs.get(module, 0) |
|
250 | 250 | self._fetch_revisions(from_revnum=revnum, to_revnum=stop) |
|
251 | 251 | commit = self.commits[rev] |
|
252 | 252 | # caller caches the result, so free it here to release memory |
|
253 | 253 | del self.commits[rev] |
|
254 | 254 | return commit |
|
255 | 255 | |
|
256 | 256 | def get_log(self, paths, start, end, limit=0, discover_changed_paths=True, |
|
257 | 257 | strict_node_history=False): |
|
258 | 258 | |
|
259 | 259 | def parent(fp): |
|
260 | 260 | while True: |
|
261 | 261 | entry = pickle.load(fp) |
|
262 | 262 | try: |
|
263 | 263 | orig_paths, revnum, author, date, message = entry |
|
264 | 264 | except: |
|
265 | 265 | if entry is None: |
|
266 | 266 | break |
|
267 | 267 | raise SubversionException("child raised exception", entry) |
|
268 | 268 | yield entry |
|
269 | 269 | |
|
270 | 270 | args = [self.url, paths, start, end, limit, discover_changed_paths, |
|
271 | 271 | strict_node_history] |
|
272 | 272 | arg = encodeargs(args) |
|
273 | 273 | hgexe = util.hgexecutable() |
|
274 | 274 | cmd = '%s debugsvnlog' % util.shellquote(hgexe) |
|
275 | 275 | stdin, stdout = os.popen2(cmd, 'b') |
|
276 | 276 | |
|
277 | 277 | stdin.write(arg) |
|
278 | 278 | stdin.close() |
|
279 | 279 | |
|
280 | 280 | for p in parent(stdout): |
|
281 | 281 | yield p |
|
282 | 282 | |
|
283 | 283 | def gettags(self): |
|
284 | 284 | tags = {} |
|
285 | 285 | start = self.revnum(self.head) |
|
286 | 286 | try: |
|
287 | 287 | for entry in self.get_log([self.tags], 0, start): |
|
288 | 288 | orig_paths, revnum, author, date, message = entry |
|
289 | 289 | for path in orig_paths: |
|
290 | 290 | if not path.startswith(self.tags+'/'): |
|
291 | 291 | continue |
|
292 | 292 | ent = orig_paths[path] |
|
293 | 293 | source = ent.copyfrom_path |
|
294 | 294 | rev = ent.copyfrom_rev |
|
295 | 295 | tag = path.split('/')[-1] |
|
296 | 296 | tags[tag] = self.revid(rev, module=source) |
|
297 | 297 | except SubversionException, (inst, num): |
|
298 | 298 | self.ui.note('no tags found at revision %d\n' % start) |
|
299 | 299 | return tags |
|
300 | 300 | |
|
301 | 301 | # -- helper functions -- |
|
302 | 302 | |
|
303 | 303 | def revid(self, revnum, module=None): |
|
304 | 304 | if not module: |
|
305 | 305 | module = self.module |
|
306 | 306 | return u"svn:%s%s@%s" % (self.uuid, module.decode(self.encoding), |
|
307 | 307 | revnum) |
|
308 | 308 | |
|
309 | 309 | def revnum(self, rev): |
|
310 | 310 | return int(rev.split('@')[-1]) |
|
311 | 311 | |
|
312 | 312 | def revsplit(self, rev): |
|
313 | 313 | url, revnum = rev.encode(self.encoding).split('@', 1) |
|
314 | 314 | revnum = int(revnum) |
|
315 | 315 | parts = url.split('/', 1) |
|
316 | 316 | uuid = parts.pop(0)[4:] |
|
317 | 317 | mod = '' |
|
318 | 318 | if parts: |
|
319 | 319 | mod = '/' + parts[0] |
|
320 | 320 | return uuid, mod, revnum |
|
321 | 321 | |
|
322 | 322 | def latest(self, path, stop=0): |
|
323 | 323 | 'find the latest revision affecting path, up to stop' |
|
324 | 324 | if not stop: |
|
325 | 325 | stop = svn.ra.get_latest_revnum(self.ra) |
|
326 | 326 | try: |
|
327 | 327 | self.reparent('') |
|
328 | 328 | dirent = svn.ra.stat(self.ra, path.strip('/'), stop) |
|
329 | 329 | self.reparent(self.module) |
|
330 | 330 | except SubversionException: |
|
331 | 331 | dirent = None |
|
332 | 332 | if not dirent: |
|
333 | 333 | raise util.Abort('%s not found up to revision %d' % (path, stop)) |
|
334 | 334 | |
|
335 | 335 | return dirent.created_rev |
|
336 | 336 | |
|
337 | 337 | def get_blacklist(self): |
|
338 | 338 | """Avoid certain revision numbers. |
|
339 | 339 | It is not uncommon for two nearby revisions to cancel each other |
|
340 | 340 | out, e.g. 'I copied trunk into a subdirectory of itself instead |
|
341 | 341 | of making a branch'. The converted repository is significantly |
|
342 | 342 | smaller if we ignore such revisions.""" |
|
343 | 343 | self.blacklist = util.set() |
|
344 | 344 | blacklist = self.blacklist |
|
345 | 345 | for line in file("blacklist.txt", "r"): |
|
346 | 346 | if not line.startswith("#"): |
|
347 | 347 | try: |
|
348 | 348 | svn_rev = int(line.strip()) |
|
349 | 349 | blacklist.add(svn_rev) |
|
350 | 350 | except ValueError, e: |
|
351 | 351 | pass # not an integer or a comment |
|
352 | 352 | |
|
353 | 353 | def is_blacklisted(self, svn_rev): |
|
354 | 354 | return svn_rev in self.blacklist |
|
355 | 355 | |
|
356 | 356 | def reparent(self, module): |
|
357 | 357 | svn_url = self.base + module |
|
358 | 358 | self.ui.debug("reparent to %s\n" % svn_url.encode(self.encoding)) |
|
359 | 359 | svn.ra.reparent(self.ra, svn_url.encode(self.encoding)) |
|
360 | 360 | |
|
361 | 361 | def expandpaths(self, rev, paths, parents): |
|
362 | 362 | def get_entry_from_path(path, module=self.module): |
|
363 | 363 | # Given the repository url of this wc, say |
|
364 | 364 | # "http://server/plone/CMFPlone/branches/Plone-2_0-branch" |
|
365 | 365 | # extract the "entry" portion (a relative path) from what |
|
366 | 366 | # svn log --xml says, ie |
|
367 | 367 | # "/CMFPlone/branches/Plone-2_0-branch/tests/PloneTestCase.py" |
|
368 | 368 | # that is to say "tests/PloneTestCase.py" |
|
369 | 369 | if path.startswith(module): |
|
370 | 370 | relative = path[len(module):] |
|
371 | 371 | if relative.startswith('/'): |
|
372 | 372 | return relative[1:] |
|
373 | 373 | else: |
|
374 | 374 | return relative |
|
375 | 375 | |
|
376 | 376 | # The path is outside our tracked tree... |
|
377 | 377 | self.ui.debug('%r is not under %r, ignoring\n' % (path, module)) |
|
378 | 378 | return None |
|
379 | 379 | |
|
380 | 380 | entries = [] |
|
381 | 381 | copyfrom = {} # Map of entrypath, revision for finding source of deleted revisions. |
|
382 | 382 | copies = {} |
|
383 | 383 | revnum = self.revnum(rev) |
|
384 | 384 | |
|
385 | 385 | if revnum in self.modulemap: |
|
386 | 386 | new_module = self.modulemap[revnum] |
|
387 | 387 | if new_module != self.module: |
|
388 | 388 | self.module = new_module |
|
389 | 389 | self.reparent(self.module) |
|
390 | 390 | |
|
391 | 391 | for path, ent in paths: |
|
392 | 392 | entrypath = get_entry_from_path(path, module=self.module) |
|
393 | 393 | entry = entrypath.decode(self.encoding) |
|
394 | 394 | |
|
395 | 395 | kind = svn.ra.check_path(self.ra, entrypath, revnum) |
|
396 | 396 | if kind == svn.core.svn_node_file: |
|
397 | 397 | if ent.copyfrom_path: |
|
398 | 398 | copyfrom_path = get_entry_from_path(ent.copyfrom_path) |
|
399 | 399 | if copyfrom_path: |
|
400 | 400 | self.ui.debug("Copied to %s from %s@%s\n" % (entry, copyfrom_path, ent.copyfrom_rev)) |
|
401 | 401 | # It's probably important for hg that the source |
|
402 | 402 | # exists in the revision's parent, not just the |
|
403 | 403 | # ent.copyfrom_rev |
|
404 | 404 | fromkind = svn.ra.check_path(self.ra, copyfrom_path, ent.copyfrom_rev) |
|
405 | 405 | if fromkind != 0: |
|
406 | 406 | copies[self.recode(entry)] = self.recode(copyfrom_path) |
|
407 | 407 | entries.append(self.recode(entry)) |
|
408 | 408 | elif kind == 0: # gone, but had better be a deleted *file* |
|
409 | 409 | self.ui.debug("gone from %s\n" % ent.copyfrom_rev) |
|
410 | 410 | |
|
411 | 411 | # if a branch is created but entries are removed in the same |
|
412 | 412 | # changeset, get the right fromrev |
|
413 | 413 | if parents: |
|
414 | 414 | uuid, old_module, fromrev = self.revsplit(parents[0]) |
|
415 | 415 | else: |
|
416 | 416 | fromrev = revnum - 1 |
|
417 | 417 | # might always need to be revnum - 1 in these 3 lines? |
|
418 | 418 | old_module = self.modulemap.get(fromrev, self.module) |
|
419 | 419 | |
|
420 | 420 | basepath = old_module + "/" + get_entry_from_path(path, module=self.module) |
|
421 | 421 | entrypath = old_module + "/" + get_entry_from_path(path, module=self.module) |
|
422 | 422 | |
|
423 | 423 | def lookup_parts(p): |
|
424 | 424 | rc = None |
|
425 | 425 | parts = p.split("/") |
|
426 | 426 | for i in range(len(parts)): |
|
427 | 427 | part = "/".join(parts[:i]) |
|
428 | 428 | info = part, copyfrom.get(part, None) |
|
429 | 429 | if info[1] is not None: |
|
430 | 430 | self.ui.debug("Found parent directory %s\n" % info[1]) |
|
431 | 431 | rc = info |
|
432 | 432 | return rc |
|
433 | 433 | |
|
434 | 434 | self.ui.debug("base, entry %s %s\n" % (basepath, entrypath)) |
|
435 | 435 | |
|
436 | 436 | frompath, froment = lookup_parts(entrypath) or (None, revnum - 1) |
|
437 | 437 | |
|
438 | 438 | # need to remove fragment from lookup_parts and replace with copyfrom_path |
|
439 | 439 | if frompath is not None: |
|
440 | 440 | self.ui.debug("munge-o-matic\n") |
|
441 | 441 | self.ui.debug(entrypath + '\n') |
|
442 | 442 | self.ui.debug(entrypath[len(frompath):] + '\n') |
|
443 | 443 | entrypath = froment.copyfrom_path + entrypath[len(frompath):] |
|
444 | 444 | fromrev = froment.copyfrom_rev |
|
445 | 445 | self.ui.debug("Info: %s %s %s %s\n" % (frompath, froment, ent, entrypath)) |
|
446 | 446 | |
|
447 | 447 | fromkind = svn.ra.check_path(self.ra, entrypath, fromrev) |
|
448 | 448 | if fromkind == svn.core.svn_node_file: # a deleted file |
|
449 | 449 | entries.append(self.recode(entry)) |
|
450 | 450 | elif fromkind == svn.core.svn_node_dir: |
|
451 | 451 | # print "Deleted/moved non-file:", revnum, path, ent |
|
452 | 452 | # children = self._find_children(path, revnum - 1) |
|
453 | 453 | # print "find children %s@%d from %d action %s" % (path, revnum, ent.copyfrom_rev, ent.action) |
|
454 | 454 | # Sometimes this is tricky. For example: in |
|
455 | 455 | # The Subversion Repository revision 6940 a dir |
|
456 | 456 | # was copied and one of its files was deleted |
|
457 | 457 | # from the new location in the same commit. This |
|
458 | 458 | # code can't deal with that yet. |
|
459 | 459 | if ent.action == 'C': |
|
460 | 460 | children = self._find_children(path, fromrev) |
|
461 | 461 | else: |
|
462 | 462 | oroot = entrypath.strip('/') |
|
463 | 463 | nroot = path.strip('/') |
|
464 | 464 | children = self._find_children(oroot, fromrev) |
|
465 | 465 | children = [s.replace(oroot,nroot) for s in children] |
|
466 | 466 | # Mark all [files, not directories] as deleted. |
|
467 | 467 | for child in children: |
|
468 | 468 | # Can we move a child directory and its |
|
469 | 469 | # parent in the same commit? (probably can). Could |
|
470 | 470 | # cause problems if instead of revnum -1, |
|
471 | 471 | # we have to look in (copyfrom_path, revnum - 1) |
|
472 | 472 | entrypath = get_entry_from_path("/" + child, module=old_module) |
|
473 | 473 | if entrypath: |
|
474 | 474 | entry = self.recode(entrypath.decode(self.encoding)) |
|
475 | 475 | if entry in copies: |
|
476 | 476 | # deleted file within a copy |
|
477 | 477 | del copies[entry] |
|
478 | 478 | else: |
|
479 | 479 | entries.append(entry) |
|
480 | 480 | else: |
|
481 | 481 | self.ui.debug('unknown path in revision %d: %s\n' % \ |
|
482 | 482 | (revnum, path)) |
|
483 | 483 | elif kind == svn.core.svn_node_dir: |
|
484 | 484 | # Should probably synthesize normal file entries |
|
485 | 485 | # and handle as above to clean up copy/rename handling. |
|
486 | 486 | |
|
487 | 487 | # If the directory just had a prop change, |
|
488 | 488 | # then we shouldn't need to look for its children. |
|
489 | 489 | # Also this could create duplicate entries. Not sure |
|
490 | 490 | # whether this will matter. Maybe should make entries a set. |
|
491 | 491 | # print "Changed directory", revnum, path, ent.action, ent.copyfrom_path, ent.copyfrom_rev |
|
492 | 492 | # This will fail if a directory was copied |
|
493 | 493 | # from another branch and then some of its files |
|
494 | 494 | # were deleted in the same transaction. |
|
495 | 495 | children = self._find_children(path, revnum) |
|
496 | 496 | children.sort() |
|
497 | 497 | for child in children: |
|
498 | 498 | # Can we move a child directory and its |
|
499 | 499 | # parent in the same commit? (probably can). Could |
|
500 | 500 | # cause problems if instead of revnum -1, |
|
501 | 501 | # we have to look in (copyfrom_path, revnum - 1) |
|
502 | 502 | entrypath = get_entry_from_path("/" + child, module=self.module) |
|
503 | 503 | # print child, self.module, entrypath |
|
504 | 504 | if entrypath: |
|
505 | 505 | # Need to filter out directories here... |
|
506 | 506 | kind = svn.ra.check_path(self.ra, entrypath, revnum) |
|
507 | 507 | if kind != svn.core.svn_node_dir: |
|
508 | 508 | entries.append(self.recode(entrypath)) |
|
509 | 509 | |
|
510 | 510 | # Copies here (must copy all from source) |
|
511 | 511 | # Probably not a real problem for us if |
|
512 | 512 | # source does not exist |
|
513 | 513 | |
|
514 | 514 | # Can do this with the copy command "hg copy" |
|
515 | 515 | # if ent.copyfrom_path: |
|
516 | 516 | # copyfrom_entry = get_entry_from_path(ent.copyfrom_path.decode(self.encoding), |
|
517 | 517 | # module=self.module) |
|
518 | 518 | # copyto_entry = entrypath |
|
519 | 519 | # |
|
520 | 520 | # print "copy directory", copyfrom_entry, 'to', copyto_entry |
|
521 | 521 | # |
|
522 | 522 | # copies.append((copyfrom_entry, copyto_entry)) |
|
523 | 523 | |
|
524 | 524 | if ent.copyfrom_path: |
|
525 | 525 | copyfrom_path = ent.copyfrom_path.decode(self.encoding) |
|
526 | 526 | copyfrom_entry = get_entry_from_path(copyfrom_path, module=self.module) |
|
527 | 527 | if copyfrom_entry: |
|
528 | 528 | copyfrom[path] = ent |
|
529 | 529 | self.ui.debug("mark %s came from %s\n" % (path, copyfrom[path])) |
|
530 | 530 | |
|
531 | 531 | # Good, /probably/ a regular copy. Really should check |
|
532 | 532 | # to see whether the parent revision actually contains |
|
533 | 533 | # the directory in question. |
|
534 | 534 | children = self._find_children(self.recode(copyfrom_path), ent.copyfrom_rev) |
|
535 | 535 | children.sort() |
|
536 | 536 | for child in children: |
|
537 | 537 | entrypath = get_entry_from_path("/" + child, module=self.module) |
|
538 | 538 | if entrypath: |
|
539 | 539 | entry = entrypath.decode(self.encoding) |
|
540 | 540 | # print "COPY COPY From", copyfrom_entry, entry |
|
541 | 541 | copyto_path = path + entry[len(copyfrom_entry):] |
|
542 | 542 | copyto_entry = get_entry_from_path(copyto_path, module=self.module) |
|
543 | 543 | # print "COPY", entry, "COPY To", copyto_entry |
|
544 | 544 | copies[self.recode(copyto_entry)] = self.recode(entry) |
|
545 | 545 | # copy from quux splort/quuxfile |
|
546 | 546 | |
|
547 | 547 | return (entries, copies) |
|
548 | 548 | |
|
549 | 549 | def _fetch_revisions(self, from_revnum = 0, to_revnum = 347): |
|
550 | 550 | self.child_cset = None |
|
551 | 551 | def parselogentry(orig_paths, revnum, author, date, message): |
|
552 | 552 | self.ui.debug("parsing revision %d (%d changes)\n" % |
|
553 | 553 | (revnum, len(orig_paths))) |
|
554 | 554 | |
|
555 | 555 | if revnum in self.modulemap: |
|
556 | 556 | new_module = self.modulemap[revnum] |
|
557 | 557 | if new_module != self.module: |
|
558 | 558 | self.module = new_module |
|
559 | 559 | self.reparent(self.module) |
|
560 | 560 | |
|
561 | 561 | rev = self.revid(revnum) |
|
562 | 562 | # branch log might return entries for a parent we already have |
|
563 | 563 | if (rev in self.commits or |
|
564 | 564 | (revnum < self.lastrevs.get(self.module, 0))): |
|
565 | 565 | return |
|
566 | 566 | |
|
567 | 567 | parents = [] |
|
568 | 568 | # check whether this revision is the start of a branch |
|
569 | 569 | if self.module in orig_paths: |
|
570 | 570 | ent = orig_paths[self.module] |
|
571 | 571 | if ent.copyfrom_path: |
|
572 | 572 | # ent.copyfrom_rev may not be the actual last revision |
|
573 | 573 | prev = self.latest(ent.copyfrom_path, ent.copyfrom_rev) |
|
574 | 574 | self.modulemap[prev] = ent.copyfrom_path |
|
575 | 575 | parents = [self.revid(prev, ent.copyfrom_path)] |
|
576 | 576 | self.ui.note('found parent of branch %s at %d: %s\n' % \ |
|
577 | 577 | (self.module, prev, ent.copyfrom_path)) |
|
578 | 578 | else: |
|
579 | 579 | self.ui.debug("No copyfrom path, don't know what to do.\n") |
|
580 | 580 | |
|
581 | 581 | self.modulemap[revnum] = self.module # track backwards in time |
|
582 | 582 | |
|
583 | 583 | orig_paths = orig_paths.items() |
|
584 | 584 | orig_paths.sort() |
|
585 | 585 | paths = [] |
|
586 | 586 | # filter out unrelated paths |
|
587 | 587 | for path, ent in orig_paths: |
|
588 | 588 | if not path.startswith(self.module): |
|
589 | 589 | self.ui.debug("boring@%s: %s\n" % (revnum, path)) |
|
590 | 590 | continue |
|
591 | 591 | paths.append((path, ent)) |
|
592 | 592 | |
|
593 | 593 | self.paths[rev] = (paths, parents) |
|
594 | 594 | |
|
595 | 595 | # Example SVN datetime. Includes microseconds. |
|
596 | 596 | # ISO-8601 conformant |
|
597 | 597 | # '2007-01-04T17:35:00.902377Z' |
|
598 | 598 | date = util.parsedate(date[:18] + " UTC", ["%Y-%m-%dT%H:%M:%S"]) |
|
599 | 599 | |
|
600 | 600 | log = message and self.recode(message) |
|
601 | 601 | author = author and self.recode(author) or '' |
|
602 | 602 | try: |
|
603 | 603 | branch = self.module.split("/")[-1] |
|
604 | 604 | if branch == 'trunk': |
|
605 | 605 | branch = '' |
|
606 | 606 | except IndexError: |
|
607 | 607 | branch = None |
|
608 | 608 | |
|
609 | 609 | cset = commit(author=author, |
|
610 | 610 | date=util.datestr(date), |
|
611 | 611 | desc=log, |
|
612 | 612 | parents=parents, |
|
613 | 613 | branch=branch, |
|
614 | 614 | rev=rev.encode('utf-8')) |
|
615 | 615 | |
|
616 | 616 | self.commits[rev] = cset |
|
617 | 617 | if self.child_cset and not self.child_cset.parents: |
|
618 | 618 | self.child_cset.parents = [rev] |
|
619 | 619 | self.child_cset = cset |
|
620 | 620 | |
|
621 | 621 | self.ui.note('fetching revision log for "%s" from %d to %d\n' % |
|
622 | 622 | (self.module, from_revnum, to_revnum)) |
|
623 | 623 | |
|
624 | 624 | try: |
|
625 | 625 | for entry in self.get_log([self.module], from_revnum, to_revnum): |
|
626 | 626 | orig_paths, revnum, author, date, message = entry |
|
627 | 627 | if self.is_blacklisted(revnum): |
|
628 | 628 | self.ui.note('skipping blacklisted revision %d\n' % revnum) |
|
629 | 629 | continue |
|
630 | 630 | if orig_paths is None: |
|
631 | 631 | self.ui.debug('revision %d has no entries\n' % revnum) |
|
632 | 632 | continue |
|
633 | 633 | parselogentry(orig_paths, revnum, author, date, message) |
|
634 | 634 | except SubversionException, (inst, num): |
|
635 | 635 | if num == svn.core.SVN_ERR_FS_NO_SUCH_REVISION: |
|
636 | 636 | raise NoSuchRevision(branch=self, |
|
637 | 637 | revision="Revision number %d" % to_revnum) |
|
638 | 638 | raise |
|
639 | 639 | |
|
640 | 640 | def _getfile(self, file, rev): |
|
641 | 641 | io = StringIO() |
|
642 | 642 | # TODO: ra.get_file transmits the whole file instead of diffs. |
|
643 | 643 | mode = '' |
|
644 | 644 | try: |
|
645 | 645 | revnum = self.revnum(rev) |
|
646 | 646 | if self.module != self.modulemap[revnum]: |
|
647 | 647 | self.module = self.modulemap[revnum] |
|
648 | 648 | self.reparent(self.module) |
|
649 | 649 | info = svn.ra.get_file(self.ra, file, revnum, io) |
|
650 | 650 | if isinstance(info, list): |
|
651 | 651 | info = info[-1] |
|
652 | 652 | mode = ("svn:executable" in info) and 'x' or '' |
|
653 | 653 | mode = ("svn:special" in info) and 'l' or mode |
|
654 | 654 | except SubversionException, e: |
|
655 | 655 | notfound = (svn.core.SVN_ERR_FS_NOT_FOUND, |
|
656 | 656 | svn.core.SVN_ERR_RA_DAV_PATH_NOT_FOUND) |
|
657 | 657 | if e.apr_err in notfound: # File not found |
|
658 | 658 | raise IOError() |
|
659 | 659 | raise |
|
660 | 660 | data = io.getvalue() |
|
661 | 661 | if mode == 'l': |
|
662 | 662 | link_prefix = "link " |
|
663 | 663 | if data.startswith(link_prefix): |
|
664 | 664 | data = data[len(link_prefix):] |
|
665 | 665 | return data, mode |
|
666 | 666 | |
|
667 | 667 | def _find_children(self, path, revnum): |
|
668 | 668 | path = path.strip('/') |
|
669 | 669 | pool = Pool() |
|
670 | 670 | rpath = '/'.join([self.base, path]).strip('/') |
|
671 | 671 | return ['%s/%s' % (path, x) for x in svn.client.ls(rpath, optrev(revnum), True, self.ctx, pool).keys()] |
|
672 | 672 | |
|
673 | 673 | pre_revprop_change = '''#!/bin/sh |
|
674 | 674 | |
|
675 | 675 | REPOS="$1" |
|
676 | 676 | REV="$2" |
|
677 | 677 | USER="$3" |
|
678 | 678 | PROPNAME="$4" |
|
679 | 679 | ACTION="$5" |
|
680 | 680 | |
|
681 | 681 | if [ "$ACTION" = "M" -a "$PROPNAME" = "svn:log" ]; then exit 0; fi |
|
682 | 682 | if [ "$ACTION" = "A" -a "$PROPNAME" = "hg:convert-branch" ]; then exit 0; fi |
|
683 | 683 | if [ "$ACTION" = "A" -a "$PROPNAME" = "hg:convert-rev" ]; then exit 0; fi |
|
684 | 684 | |
|
685 | 685 | echo "Changing prohibited revision property" >&2 |
|
686 | 686 | exit 1 |
|
687 | 687 | ''' |
|
688 | 688 | |
|
689 | 689 | class svn_sink(converter_sink, commandline): |
|
690 | 690 | commit_re = re.compile(r'Committed revision (\d+).', re.M) |
|
691 | 691 | |
|
692 | 692 | def prerun(self): |
|
693 | 693 | if self.wc: |
|
694 | 694 | os.chdir(self.wc) |
|
695 | 695 | |
|
696 | 696 | def postrun(self): |
|
697 | 697 | if self.wc: |
|
698 | 698 | os.chdir(self.cwd) |
|
699 | 699 | |
|
700 | 700 | def join(self, name): |
|
701 | 701 | return os.path.join(self.wc, '.svn', name) |
|
702 | 702 | |
|
703 | 703 | def revmapfile(self): |
|
704 | 704 | return self.join('hg-shamap') |
|
705 | 705 | |
|
706 | 706 | def authorfile(self): |
|
707 | 707 | return self.join('hg-authormap') |
|
708 | 708 | |
|
709 | 709 | def __init__(self, ui, path): |
|
710 | 710 | converter_sink.__init__(self, ui, path) |
|
711 | 711 | commandline.__init__(self, ui, 'svn') |
|
712 | 712 | self.delete = [] |
|
713 | 713 | self.wc = None |
|
714 | 714 | self.cwd = os.getcwd() |
|
715 | 715 | |
|
716 | 716 | path = os.path.realpath(path) |
|
717 | 717 | |
|
718 | 718 | created = False |
|
719 | 719 | if os.path.isfile(os.path.join(path, '.svn', 'entries')): |
|
720 | 720 | self.wc = path |
|
721 | 721 | self.run0('update') |
|
722 | 722 | else: |
|
723 | wcpath = os.path.join(os.getcwd(), os.path.basename(path) + '-wc') | |
|
724 | ||
|
723 | 725 | if os.path.isdir(os.path.dirname(path)): |
|
724 | 726 | if not os.path.exists(os.path.join(path, 'db', 'fs-type')): |
|
725 | 727 | ui.status(_('initializing svn repo %r\n') % |
|
726 | 728 | os.path.basename(path)) |
|
727 | 729 | commandline(ui, 'svnadmin').run0('create', path) |
|
728 | 730 | created = path |
|
731 | path = path.replace('\\', '/') | |
|
732 | if not path.startswith('/'): | |
|
733 | path = '/' + path | |
|
729 | 734 | path = 'file://' + path |
|
730 | wcpath = os.path.join(os.getcwd(), os.path.basename(path) + '-wc') | |
|
735 | ||
|
731 | 736 | ui.status(_('initializing svn wc %r\n') % os.path.basename(wcpath)) |
|
732 | 737 | self.run0('checkout', path, wcpath) |
|
733 | 738 | |
|
734 | 739 | self.wc = wcpath |
|
735 | 740 | self.opener = util.opener(self.wc) |
|
736 | 741 | self.wopener = util.opener(self.wc) |
|
737 | 742 | self.childmap = mapfile(ui, self.join('hg-childmap')) |
|
738 | 743 | |
|
739 | 744 | if created: |
|
740 | 745 | hook = os.path.join(created, 'hooks', 'pre-revprop-change') |
|
741 | 746 | fp = open(hook, 'w') |
|
742 | 747 | fp.write(pre_revprop_change) |
|
743 | 748 | fp.close() |
|
744 | 749 | util.set_exec(hook, True) |
|
745 | 750 | |
|
746 | 751 | def wjoin(self, *names): |
|
747 | 752 | return os.path.join(self.wc, *names) |
|
748 | 753 | |
|
749 | 754 | def putfile(self, filename, flags, data): |
|
750 | 755 | if 'l' in flags: |
|
751 | 756 | self.wopener.symlink(data, filename) |
|
752 | 757 | else: |
|
753 | 758 | try: |
|
754 | 759 | if os.path.islink(self.wjoin(filename)): |
|
755 | 760 | os.unlink(filename) |
|
756 | 761 | except OSError: |
|
757 | 762 | pass |
|
758 | 763 | self.wopener(filename, 'w').write(data) |
|
759 | 764 | was_exec = util.is_exec(self.wjoin(filename)) |
|
760 | 765 | util.set_exec(self.wjoin(filename), 'x' in flags) |
|
761 | 766 | if was_exec: |
|
762 | 767 | if 'x' not in flags: |
|
763 | 768 | self.run0('propdel', 'svn:executable', filename) |
|
764 | 769 | else: |
|
765 | 770 | if 'x' in flags: |
|
766 | 771 | self.run0('propset', 'svn:executable', '*', filename) |
|
767 | 772 | |
|
768 | 773 | def delfile(self, name): |
|
769 | 774 | self.delete.append(name) |
|
770 | 775 | |
|
771 | 776 | def copyfile(self, source, dest): |
|
772 | 777 | # SVN's copy command pukes if the destination file exists, but |
|
773 | 778 | # our copyfile method expects to record a copy that has |
|
774 | 779 | # already occurred. Cross the semantic gap. |
|
775 | 780 | wdest = self.wjoin(dest) |
|
776 | 781 | exists = os.path.exists(wdest) |
|
777 | 782 | if exists: |
|
778 | 783 | fd, tempname = tempfile.mkstemp( |
|
779 | 784 | prefix='hg-copy-', dir=os.path.dirname(wdest)) |
|
780 | 785 | os.close(fd) |
|
781 | 786 | os.unlink(tempname) |
|
782 | 787 | os.rename(wdest, tempname) |
|
783 | 788 | try: |
|
784 | 789 | self.run0('copy', source, dest) |
|
785 | 790 | finally: |
|
786 | 791 | if exists: |
|
787 | 792 | try: |
|
788 | 793 | os.unlink(wdest) |
|
789 | 794 | except OSError: |
|
790 | 795 | pass |
|
791 | 796 | os.rename(tempname, wdest) |
|
792 | 797 | |
|
793 | 798 | def dirs_of(self, files): |
|
794 | 799 | dirs = set() |
|
795 | 800 | for f in files: |
|
796 | 801 | if os.path.isdir(self.wjoin(f)): |
|
797 | 802 | dirs.add(f) |
|
798 | 803 | for i in strutil.rfindall(f, '/'): |
|
799 | 804 | dirs.add(f[:i]) |
|
800 | 805 | return dirs |
|
801 | 806 | |
|
802 | 807 | def add_files(self, files): |
|
803 | 808 | add_dirs = [d for d in self.dirs_of(files) |
|
804 | 809 | if not os.path.exists(self.wjoin(d, '.svn', 'entries'))] |
|
805 | 810 | if add_dirs: |
|
806 | 811 | self.run('add', non_recursive=True, quiet=True, *add) |
|
807 | 812 | if files: |
|
808 | 813 | self.run('add', quiet=True, *files) |
|
809 | 814 | return files.union(add_dirs) |
|
810 | 815 | |
|
811 | 816 | def tidy_dirs(self, names): |
|
812 | 817 | dirs = list(self.dirs_of(names)) |
|
813 | 818 | dirs.sort(reverse=True) |
|
814 | 819 | deleted = [] |
|
815 | 820 | for d in dirs: |
|
816 | 821 | wd = self.wjoin(d) |
|
817 | 822 | if os.listdir(wd) == '.svn': |
|
818 | 823 | self.run0('delete', d) |
|
819 | 824 | deleted.append(d) |
|
820 | 825 | return deleted |
|
821 | 826 | |
|
822 | 827 | def addchild(self, parent, child): |
|
823 | 828 | self.childmap[parent] = child |
|
824 | 829 | |
|
825 | 830 | def putcommit(self, files, parents, commit): |
|
826 | 831 | for parent in parents: |
|
827 | 832 | try: |
|
828 | 833 | return self.childmap[parent] |
|
829 | 834 | except KeyError: |
|
830 | 835 | pass |
|
831 | 836 | entries = set(self.delete) |
|
832 | 837 | if self.delete: |
|
833 | 838 | self.run0('delete', *self.delete) |
|
834 | 839 | self.delete = [] |
|
835 | 840 | files = util.frozenset(files) |
|
836 | 841 | entries.update(self.add_files(files.difference(entries))) |
|
837 | 842 | entries.update(self.tidy_dirs(entries)) |
|
838 | 843 | fd, messagefile = tempfile.mkstemp(prefix='hg-convert-') |
|
839 | 844 | fp = os.fdopen(fd, 'w') |
|
840 | 845 | fp.write(commit.desc) |
|
841 | 846 | fp.close() |
|
842 | 847 | try: |
|
843 | 848 | output = self.run0('commit', |
|
844 | 849 | username=util.shortuser(commit.author), |
|
845 | 850 | file=messagefile, |
|
846 | 851 | *list(entries)) |
|
847 | 852 | try: |
|
848 | 853 | rev = self.commit_re.search(output).group(1) |
|
849 | 854 | except AttributeError: |
|
850 | 855 | self.ui.warn(_('unexpected svn output:\n')) |
|
851 | 856 | self.ui.warn(output) |
|
852 | 857 | raise util.Abort(_('unable to cope with svn output')) |
|
853 | 858 | if commit.rev: |
|
854 | 859 | self.run('propset', 'hg:convert-rev', commit.rev, |
|
855 | 860 | revprop=True, revision=rev) |
|
856 | 861 | if commit.branch and commit.branch != 'default': |
|
857 | 862 | self.run('propset', 'hg:convert-branch', commit.branch, |
|
858 | 863 | revprop=True, revision=rev) |
|
859 | 864 | for parent in parents: |
|
860 | 865 | self.addchild(parent, rev) |
|
861 | 866 | return rev |
|
862 | 867 | finally: |
|
863 | 868 | os.unlink(messagefile) |
|
864 | 869 | |
|
865 | 870 | def puttags(self, tags): |
|
866 | 871 | self.ui.warn(_('XXX TAGS NOT IMPLEMENTED YET\n')) |
General Comments 0
You need to be logged in to leave comments.
Login now