Show More
@@ -1,952 +1,952 b'' | |||
|
1 | 1 | # Mercurial built-in replacement for cvsps. |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2008, Frank Kingswood <frank@kingswood-consulting.co.uk> |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | from __future__ import absolute_import |
|
8 | 8 | |
|
9 | 9 | import os |
|
10 | 10 | import re |
|
11 | 11 | |
|
12 | 12 | from mercurial.i18n import _ |
|
13 | 13 | from mercurial import ( |
|
14 | 14 | encoding, |
|
15 | 15 | error, |
|
16 | 16 | hook, |
|
17 | 17 | pycompat, |
|
18 | 18 | util, |
|
19 | 19 | ) |
|
20 | 20 | |
|
21 | 21 | pickle = util.pickle |
|
22 | 22 | |
|
23 | 23 | class logentry(object): |
|
24 | 24 | '''Class logentry has the following attributes: |
|
25 | 25 | .author - author name as CVS knows it |
|
26 | 26 | .branch - name of branch this revision is on |
|
27 | 27 | .branches - revision tuple of branches starting at this revision |
|
28 | 28 | .comment - commit message |
|
29 | 29 | .commitid - CVS commitid or None |
|
30 | 30 | .date - the commit date as a (time, tz) tuple |
|
31 | 31 | .dead - true if file revision is dead |
|
32 | 32 | .file - Name of file |
|
33 | 33 | .lines - a tuple (+lines, -lines) or None |
|
34 | 34 | .parent - Previous revision of this entry |
|
35 | 35 | .rcs - name of file as returned from CVS |
|
36 | 36 | .revision - revision number as tuple |
|
37 | 37 | .tags - list of tags on the file |
|
38 | 38 | .synthetic - is this a synthetic "file ... added on ..." revision? |
|
39 | 39 | .mergepoint - the branch that has been merged from (if present in |
|
40 | 40 | rlog output) or None |
|
41 | 41 | .branchpoints - the branches that start at the current entry or empty |
|
42 | 42 | ''' |
|
43 | 43 | def __init__(self, **entries): |
|
44 | 44 | self.synthetic = False |
|
45 | 45 | self.__dict__.update(entries) |
|
46 | 46 | |
|
47 | 47 | def __repr__(self): |
|
48 | 48 | items = ("%s=%r"%(k, self.__dict__[k]) for k in sorted(self.__dict__)) |
|
49 | 49 | return "%s(%s)"%(type(self).__name__, ", ".join(items)) |
|
50 | 50 | |
|
51 | 51 | class logerror(Exception): |
|
52 | 52 | pass |
|
53 | 53 | |
|
54 | 54 | def getrepopath(cvspath): |
|
55 | 55 | """Return the repository path from a CVS path. |
|
56 | 56 | |
|
57 | 57 | >>> getrepopath(b'/foo/bar') |
|
58 | 58 | '/foo/bar' |
|
59 | 59 | >>> getrepopath(b'c:/foo/bar') |
|
60 | 60 | '/foo/bar' |
|
61 | 61 | >>> getrepopath(b':pserver:10/foo/bar') |
|
62 | 62 | '/foo/bar' |
|
63 | 63 | >>> getrepopath(b':pserver:10c:/foo/bar') |
|
64 | 64 | '/foo/bar' |
|
65 | 65 | >>> getrepopath(b':pserver:/foo/bar') |
|
66 | 66 | '/foo/bar' |
|
67 | 67 | >>> getrepopath(b':pserver:c:/foo/bar') |
|
68 | 68 | '/foo/bar' |
|
69 | 69 | >>> getrepopath(b':pserver:truc@foo.bar:/foo/bar') |
|
70 | 70 | '/foo/bar' |
|
71 | 71 | >>> getrepopath(b':pserver:truc@foo.bar:c:/foo/bar') |
|
72 | 72 | '/foo/bar' |
|
73 | 73 | >>> getrepopath(b'user@server/path/to/repository') |
|
74 | 74 | '/path/to/repository' |
|
75 | 75 | """ |
|
76 | 76 | # According to CVS manual, CVS paths are expressed like: |
|
77 | 77 | # [:method:][[user][:password]@]hostname[:[port]]/path/to/repository |
|
78 | 78 | # |
|
79 | 79 | # CVSpath is splitted into parts and then position of the first occurrence |
|
80 | 80 | # of the '/' char after the '@' is located. The solution is the rest of the |
|
81 | 81 | # string after that '/' sign including it |
|
82 | 82 | |
|
83 | 83 | parts = cvspath.split(':') |
|
84 | 84 | atposition = parts[-1].find('@') |
|
85 | 85 | start = 0 |
|
86 | 86 | |
|
87 | 87 | if atposition != -1: |
|
88 | 88 | start = atposition |
|
89 | 89 | |
|
90 | 90 | repopath = parts[-1][parts[-1].find('/', start):] |
|
91 | 91 | return repopath |
|
92 | 92 | |
|
93 | 93 | def createlog(ui, directory=None, root="", rlog=True, cache=None): |
|
94 | 94 | '''Collect the CVS rlog''' |
|
95 | 95 | |
|
96 | 96 | # Because we store many duplicate commit log messages, reusing strings |
|
97 | 97 | # saves a lot of memory and pickle storage space. |
|
98 | 98 | _scache = {} |
|
99 | 99 | def scache(s): |
|
100 | 100 | "return a shared version of a string" |
|
101 | 101 | return _scache.setdefault(s, s) |
|
102 | 102 | |
|
103 | 103 | ui.status(_('collecting CVS rlog\n')) |
|
104 | 104 | |
|
105 | 105 | log = [] # list of logentry objects containing the CVS state |
|
106 | 106 | |
|
107 | 107 | # patterns to match in CVS (r)log output, by state of use |
|
108 | 108 | re_00 = re.compile('RCS file: (.+)$') |
|
109 | 109 | re_01 = re.compile('cvs \\[r?log aborted\\]: (.+)$') |
|
110 | 110 | re_02 = re.compile('cvs (r?log|server): (.+)\n$') |
|
111 | 111 | re_03 = re.compile("(Cannot access.+CVSROOT)|" |
|
112 | 112 | "(can't create temporary directory.+)$") |
|
113 | 113 | re_10 = re.compile('Working file: (.+)$') |
|
114 | 114 | re_20 = re.compile('symbolic names:') |
|
115 | 115 | re_30 = re.compile('\t(.+): ([\\d.]+)$') |
|
116 | 116 | re_31 = re.compile('----------------------------$') |
|
117 | 117 | re_32 = re.compile('=======================================' |
|
118 | 118 | '======================================$') |
|
119 | 119 | re_50 = re.compile('revision ([\\d.]+)(\s+locked by:\s+.+;)?$') |
|
120 | 120 | re_60 = re.compile(r'date:\s+(.+);\s+author:\s+(.+);\s+state:\s+(.+?);' |
|
121 | 121 | r'(\s+lines:\s+(\+\d+)?\s+(-\d+)?;)?' |
|
122 | 122 | r'(\s+commitid:\s+([^;]+);)?' |
|
123 | 123 | r'(.*mergepoint:\s+([^;]+);)?') |
|
124 | 124 | re_70 = re.compile('branches: (.+);$') |
|
125 | 125 | |
|
126 | 126 | file_added_re = re.compile(r'file [^/]+ was (initially )?added on branch') |
|
127 | 127 | |
|
128 | 128 | prefix = '' # leading path to strip of what we get from CVS |
|
129 | 129 | |
|
130 | 130 | if directory is None: |
|
131 | 131 | # Current working directory |
|
132 | 132 | |
|
133 | 133 | # Get the real directory in the repository |
|
134 | 134 | try: |
|
135 | 135 | prefix = open(os.path.join('CVS','Repository'), 'rb').read().strip() |
|
136 | 136 | directory = prefix |
|
137 | 137 | if prefix == ".": |
|
138 | 138 | prefix = "" |
|
139 | 139 | except IOError: |
|
140 | 140 | raise logerror(_('not a CVS sandbox')) |
|
141 | 141 | |
|
142 | 142 | if prefix and not prefix.endswith(pycompat.ossep): |
|
143 | 143 | prefix += pycompat.ossep |
|
144 | 144 | |
|
145 | 145 | # Use the Root file in the sandbox, if it exists |
|
146 | 146 | try: |
|
147 | 147 | root = open(os.path.join('CVS','Root'), 'rb').read().strip() |
|
148 | 148 | except IOError: |
|
149 | 149 | pass |
|
150 | 150 | |
|
151 | 151 | if not root: |
|
152 | 152 | root = encoding.environ.get('CVSROOT', '') |
|
153 | 153 | |
|
154 | 154 | # read log cache if one exists |
|
155 | 155 | oldlog = [] |
|
156 | 156 | date = None |
|
157 | 157 | |
|
158 | 158 | if cache: |
|
159 | 159 | cachedir = os.path.expanduser('~/.hg.cvsps') |
|
160 | 160 | if not os.path.exists(cachedir): |
|
161 | 161 | os.mkdir(cachedir) |
|
162 | 162 | |
|
163 | 163 | # The cvsps cache pickle needs a uniquified name, based on the |
|
164 | 164 | # repository location. The address may have all sort of nasties |
|
165 | 165 | # in it, slashes, colons and such. So here we take just the |
|
166 | 166 | # alphanumeric characters, concatenated in a way that does not |
|
167 | 167 | # mix up the various components, so that |
|
168 | 168 | # :pserver:user@server:/path |
|
169 | 169 | # and |
|
170 | 170 | # /pserver/user/server/path |
|
171 | 171 | # are mapped to different cache file names. |
|
172 | 172 | cachefile = root.split(":") + [directory, "cache"] |
|
173 | cachefile = ['-'.join(re.findall(r'\w+', s)) for s in cachefile if s] | |
|
173 | cachefile = ['-'.join(re.findall(br'\w+', s)) for s in cachefile if s] | |
|
174 | 174 | cachefile = os.path.join(cachedir, |
|
175 | 175 | '.'.join([s for s in cachefile if s])) |
|
176 | 176 | |
|
177 | 177 | if cache == 'update': |
|
178 | 178 | try: |
|
179 | 179 | ui.note(_('reading cvs log cache %s\n') % cachefile) |
|
180 | 180 | oldlog = pickle.load(open(cachefile, 'rb')) |
|
181 | 181 | for e in oldlog: |
|
182 | 182 | if not (util.safehasattr(e, 'branchpoints') and |
|
183 | 183 | util.safehasattr(e, 'commitid') and |
|
184 | 184 | util.safehasattr(e, 'mergepoint')): |
|
185 | 185 | ui.status(_('ignoring old cache\n')) |
|
186 | 186 | oldlog = [] |
|
187 | 187 | break |
|
188 | 188 | |
|
189 | 189 | ui.note(_('cache has %d log entries\n') % len(oldlog)) |
|
190 | 190 | except Exception as e: |
|
191 | 191 | ui.note(_('error reading cache: %r\n') % e) |
|
192 | 192 | |
|
193 | 193 | if oldlog: |
|
194 | 194 | date = oldlog[-1].date # last commit date as a (time,tz) tuple |
|
195 | 195 | date = util.datestr(date, '%Y/%m/%d %H:%M:%S %1%2') |
|
196 | 196 | |
|
197 | 197 | # build the CVS commandline |
|
198 | 198 | cmd = ['cvs', '-q'] |
|
199 | 199 | if root: |
|
200 | 200 | cmd.append('-d%s' % root) |
|
201 | 201 | p = util.normpath(getrepopath(root)) |
|
202 | 202 | if not p.endswith('/'): |
|
203 | 203 | p += '/' |
|
204 | 204 | if prefix: |
|
205 | 205 | # looks like normpath replaces "" by "." |
|
206 | 206 | prefix = p + util.normpath(prefix) |
|
207 | 207 | else: |
|
208 | 208 | prefix = p |
|
209 | 209 | cmd.append(['log', 'rlog'][rlog]) |
|
210 | 210 | if date: |
|
211 | 211 | # no space between option and date string |
|
212 | 212 | cmd.append('-d>%s' % date) |
|
213 | 213 | cmd.append(directory) |
|
214 | 214 | |
|
215 | 215 | # state machine begins here |
|
216 | 216 | tags = {} # dictionary of revisions on current file with their tags |
|
217 | 217 | branchmap = {} # mapping between branch names and revision numbers |
|
218 | 218 | rcsmap = {} |
|
219 | 219 | state = 0 |
|
220 | 220 | store = False # set when a new record can be appended |
|
221 | 221 | |
|
222 | 222 | cmd = [util.shellquote(arg) for arg in cmd] |
|
223 | 223 | ui.note(_("running %s\n") % (' '.join(cmd))) |
|
224 | 224 | ui.debug("prefix=%r directory=%r root=%r\n" % (prefix, directory, root)) |
|
225 | 225 | |
|
226 | 226 | pfp = util.popen(' '.join(cmd)) |
|
227 | 227 | peek = pfp.readline() |
|
228 | 228 | while True: |
|
229 | 229 | line = peek |
|
230 | 230 | if line == '': |
|
231 | 231 | break |
|
232 | 232 | peek = pfp.readline() |
|
233 | 233 | if line.endswith('\n'): |
|
234 | 234 | line = line[:-1] |
|
235 | 235 | #ui.debug('state=%d line=%r\n' % (state, line)) |
|
236 | 236 | |
|
237 | 237 | if state == 0: |
|
238 | 238 | # initial state, consume input until we see 'RCS file' |
|
239 | 239 | match = re_00.match(line) |
|
240 | 240 | if match: |
|
241 | 241 | rcs = match.group(1) |
|
242 | 242 | tags = {} |
|
243 | 243 | if rlog: |
|
244 | 244 | filename = util.normpath(rcs[:-2]) |
|
245 | 245 | if filename.startswith(prefix): |
|
246 | 246 | filename = filename[len(prefix):] |
|
247 | 247 | if filename.startswith('/'): |
|
248 | 248 | filename = filename[1:] |
|
249 | 249 | if filename.startswith('Attic/'): |
|
250 | 250 | filename = filename[6:] |
|
251 | 251 | else: |
|
252 | 252 | filename = filename.replace('/Attic/', '/') |
|
253 | 253 | state = 2 |
|
254 | 254 | continue |
|
255 | 255 | state = 1 |
|
256 | 256 | continue |
|
257 | 257 | match = re_01.match(line) |
|
258 | 258 | if match: |
|
259 | 259 | raise logerror(match.group(1)) |
|
260 | 260 | match = re_02.match(line) |
|
261 | 261 | if match: |
|
262 | 262 | raise logerror(match.group(2)) |
|
263 | 263 | if re_03.match(line): |
|
264 | 264 | raise logerror(line) |
|
265 | 265 | |
|
266 | 266 | elif state == 1: |
|
267 | 267 | # expect 'Working file' (only when using log instead of rlog) |
|
268 | 268 | match = re_10.match(line) |
|
269 | 269 | assert match, _('RCS file must be followed by working file') |
|
270 | 270 | filename = util.normpath(match.group(1)) |
|
271 | 271 | state = 2 |
|
272 | 272 | |
|
273 | 273 | elif state == 2: |
|
274 | 274 | # expect 'symbolic names' |
|
275 | 275 | if re_20.match(line): |
|
276 | 276 | branchmap = {} |
|
277 | 277 | state = 3 |
|
278 | 278 | |
|
279 | 279 | elif state == 3: |
|
280 | 280 | # read the symbolic names and store as tags |
|
281 | 281 | match = re_30.match(line) |
|
282 | 282 | if match: |
|
283 | 283 | rev = [int(x) for x in match.group(2).split('.')] |
|
284 | 284 | |
|
285 | 285 | # Convert magic branch number to an odd-numbered one |
|
286 | 286 | revn = len(rev) |
|
287 | 287 | if revn > 3 and (revn % 2) == 0 and rev[-2] == 0: |
|
288 | 288 | rev = rev[:-2] + rev[-1:] |
|
289 | 289 | rev = tuple(rev) |
|
290 | 290 | |
|
291 | 291 | if rev not in tags: |
|
292 | 292 | tags[rev] = [] |
|
293 | 293 | tags[rev].append(match.group(1)) |
|
294 | 294 | branchmap[match.group(1)] = match.group(2) |
|
295 | 295 | |
|
296 | 296 | elif re_31.match(line): |
|
297 | 297 | state = 5 |
|
298 | 298 | elif re_32.match(line): |
|
299 | 299 | state = 0 |
|
300 | 300 | |
|
301 | 301 | elif state == 4: |
|
302 | 302 | # expecting '------' separator before first revision |
|
303 | 303 | if re_31.match(line): |
|
304 | 304 | state = 5 |
|
305 | 305 | else: |
|
306 | 306 | assert not re_32.match(line), _('must have at least ' |
|
307 | 307 | 'some revisions') |
|
308 | 308 | |
|
309 | 309 | elif state == 5: |
|
310 | 310 | # expecting revision number and possibly (ignored) lock indication |
|
311 | 311 | # we create the logentry here from values stored in states 0 to 4, |
|
312 | 312 | # as this state is re-entered for subsequent revisions of a file. |
|
313 | 313 | match = re_50.match(line) |
|
314 | 314 | assert match, _('expected revision number') |
|
315 | 315 | e = logentry(rcs=scache(rcs), |
|
316 | 316 | file=scache(filename), |
|
317 | 317 | revision=tuple([int(x) for x in |
|
318 | 318 | match.group(1).split('.')]), |
|
319 | 319 | branches=[], |
|
320 | 320 | parent=None, |
|
321 | 321 | commitid=None, |
|
322 | 322 | mergepoint=None, |
|
323 | 323 | branchpoints=set()) |
|
324 | 324 | |
|
325 | 325 | state = 6 |
|
326 | 326 | |
|
327 | 327 | elif state == 6: |
|
328 | 328 | # expecting date, author, state, lines changed |
|
329 | 329 | match = re_60.match(line) |
|
330 | 330 | assert match, _('revision must be followed by date line') |
|
331 | 331 | d = match.group(1) |
|
332 | 332 | if d[2] == '/': |
|
333 | 333 | # Y2K |
|
334 | 334 | d = '19' + d |
|
335 | 335 | |
|
336 | 336 | if len(d.split()) != 3: |
|
337 | 337 | # cvs log dates always in GMT |
|
338 | 338 | d = d + ' UTC' |
|
339 | 339 | e.date = util.parsedate(d, ['%y/%m/%d %H:%M:%S', |
|
340 | 340 | '%Y/%m/%d %H:%M:%S', |
|
341 | 341 | '%Y-%m-%d %H:%M:%S']) |
|
342 | 342 | e.author = scache(match.group(2)) |
|
343 | 343 | e.dead = match.group(3).lower() == 'dead' |
|
344 | 344 | |
|
345 | 345 | if match.group(5): |
|
346 | 346 | if match.group(6): |
|
347 | 347 | e.lines = (int(match.group(5)), int(match.group(6))) |
|
348 | 348 | else: |
|
349 | 349 | e.lines = (int(match.group(5)), 0) |
|
350 | 350 | elif match.group(6): |
|
351 | 351 | e.lines = (0, int(match.group(6))) |
|
352 | 352 | else: |
|
353 | 353 | e.lines = None |
|
354 | 354 | |
|
355 | 355 | if match.group(7): # cvs 1.12 commitid |
|
356 | 356 | e.commitid = match.group(8) |
|
357 | 357 | |
|
358 | 358 | if match.group(9): # cvsnt mergepoint |
|
359 | 359 | myrev = match.group(10).split('.') |
|
360 | 360 | if len(myrev) == 2: # head |
|
361 | 361 | e.mergepoint = 'HEAD' |
|
362 | 362 | else: |
|
363 | 363 | myrev = '.'.join(myrev[:-2] + ['0', myrev[-2]]) |
|
364 | 364 | branches = [b for b in branchmap if branchmap[b] == myrev] |
|
365 | 365 | assert len(branches) == 1, ('unknown branch: %s' |
|
366 | 366 | % e.mergepoint) |
|
367 | 367 | e.mergepoint = branches[0] |
|
368 | 368 | |
|
369 | 369 | e.comment = [] |
|
370 | 370 | state = 7 |
|
371 | 371 | |
|
372 | 372 | elif state == 7: |
|
373 | 373 | # read the revision numbers of branches that start at this revision |
|
374 | 374 | # or store the commit log message otherwise |
|
375 | 375 | m = re_70.match(line) |
|
376 | 376 | if m: |
|
377 | 377 | e.branches = [tuple([int(y) for y in x.strip().split('.')]) |
|
378 | 378 | for x in m.group(1).split(';')] |
|
379 | 379 | state = 8 |
|
380 | 380 | elif re_31.match(line) and re_50.match(peek): |
|
381 | 381 | state = 5 |
|
382 | 382 | store = True |
|
383 | 383 | elif re_32.match(line): |
|
384 | 384 | state = 0 |
|
385 | 385 | store = True |
|
386 | 386 | else: |
|
387 | 387 | e.comment.append(line) |
|
388 | 388 | |
|
389 | 389 | elif state == 8: |
|
390 | 390 | # store commit log message |
|
391 | 391 | if re_31.match(line): |
|
392 | 392 | cpeek = peek |
|
393 | 393 | if cpeek.endswith('\n'): |
|
394 | 394 | cpeek = cpeek[:-1] |
|
395 | 395 | if re_50.match(cpeek): |
|
396 | 396 | state = 5 |
|
397 | 397 | store = True |
|
398 | 398 | else: |
|
399 | 399 | e.comment.append(line) |
|
400 | 400 | elif re_32.match(line): |
|
401 | 401 | state = 0 |
|
402 | 402 | store = True |
|
403 | 403 | else: |
|
404 | 404 | e.comment.append(line) |
|
405 | 405 | |
|
406 | 406 | # When a file is added on a branch B1, CVS creates a synthetic |
|
407 | 407 | # dead trunk revision 1.1 so that the branch has a root. |
|
408 | 408 | # Likewise, if you merge such a file to a later branch B2 (one |
|
409 | 409 | # that already existed when the file was added on B1), CVS |
|
410 | 410 | # creates a synthetic dead revision 1.1.x.1 on B2. Don't drop |
|
411 | 411 | # these revisions now, but mark them synthetic so |
|
412 | 412 | # createchangeset() can take care of them. |
|
413 | 413 | if (store and |
|
414 | 414 | e.dead and |
|
415 | 415 | e.revision[-1] == 1 and # 1.1 or 1.1.x.1 |
|
416 | 416 | len(e.comment) == 1 and |
|
417 | 417 | file_added_re.match(e.comment[0])): |
|
418 | 418 | ui.debug('found synthetic revision in %s: %r\n' |
|
419 | 419 | % (e.rcs, e.comment[0])) |
|
420 | 420 | e.synthetic = True |
|
421 | 421 | |
|
422 | 422 | if store: |
|
423 | 423 | # clean up the results and save in the log. |
|
424 | 424 | store = False |
|
425 | 425 | e.tags = sorted([scache(x) for x in tags.get(e.revision, [])]) |
|
426 | 426 | e.comment = scache('\n'.join(e.comment)) |
|
427 | 427 | |
|
428 | 428 | revn = len(e.revision) |
|
429 | 429 | if revn > 3 and (revn % 2) == 0: |
|
430 | 430 | e.branch = tags.get(e.revision[:-1], [None])[0] |
|
431 | 431 | else: |
|
432 | 432 | e.branch = None |
|
433 | 433 | |
|
434 | 434 | # find the branches starting from this revision |
|
435 | 435 | branchpoints = set() |
|
436 | 436 | for branch, revision in branchmap.iteritems(): |
|
437 | 437 | revparts = tuple([int(i) for i in revision.split('.')]) |
|
438 | 438 | if len(revparts) < 2: # bad tags |
|
439 | 439 | continue |
|
440 | 440 | if revparts[-2] == 0 and revparts[-1] % 2 == 0: |
|
441 | 441 | # normal branch |
|
442 | 442 | if revparts[:-2] == e.revision: |
|
443 | 443 | branchpoints.add(branch) |
|
444 | 444 | elif revparts == (1, 1, 1): # vendor branch |
|
445 | 445 | if revparts in e.branches: |
|
446 | 446 | branchpoints.add(branch) |
|
447 | 447 | e.branchpoints = branchpoints |
|
448 | 448 | |
|
449 | 449 | log.append(e) |
|
450 | 450 | |
|
451 | 451 | rcsmap[e.rcs.replace('/Attic/', '/')] = e.rcs |
|
452 | 452 | |
|
453 | 453 | if len(log) % 100 == 0: |
|
454 | 454 | ui.status(util.ellipsis('%d %s' % (len(log), e.file), 80)+'\n') |
|
455 | 455 | |
|
456 | 456 | log.sort(key=lambda x: (x.rcs, x.revision)) |
|
457 | 457 | |
|
458 | 458 | # find parent revisions of individual files |
|
459 | 459 | versions = {} |
|
460 | 460 | for e in sorted(oldlog, key=lambda x: (x.rcs, x.revision)): |
|
461 | 461 | rcs = e.rcs.replace('/Attic/', '/') |
|
462 | 462 | if rcs in rcsmap: |
|
463 | 463 | e.rcs = rcsmap[rcs] |
|
464 | 464 | branch = e.revision[:-1] |
|
465 | 465 | versions[(e.rcs, branch)] = e.revision |
|
466 | 466 | |
|
467 | 467 | for e in log: |
|
468 | 468 | branch = e.revision[:-1] |
|
469 | 469 | p = versions.get((e.rcs, branch), None) |
|
470 | 470 | if p is None: |
|
471 | 471 | p = e.revision[:-2] |
|
472 | 472 | e.parent = p |
|
473 | 473 | versions[(e.rcs, branch)] = e.revision |
|
474 | 474 | |
|
475 | 475 | # update the log cache |
|
476 | 476 | if cache: |
|
477 | 477 | if log: |
|
478 | 478 | # join up the old and new logs |
|
479 | 479 | log.sort(key=lambda x: x.date) |
|
480 | 480 | |
|
481 | 481 | if oldlog and oldlog[-1].date >= log[0].date: |
|
482 | 482 | raise logerror(_('log cache overlaps with new log entries,' |
|
483 | 483 | ' re-run without cache.')) |
|
484 | 484 | |
|
485 | 485 | log = oldlog + log |
|
486 | 486 | |
|
487 | 487 | # write the new cachefile |
|
488 | 488 | ui.note(_('writing cvs log cache %s\n') % cachefile) |
|
489 | 489 | pickle.dump(log, open(cachefile, 'wb')) |
|
490 | 490 | else: |
|
491 | 491 | log = oldlog |
|
492 | 492 | |
|
493 | 493 | ui.status(_('%d log entries\n') % len(log)) |
|
494 | 494 | |
|
495 | 495 | encodings = ui.configlist('convert', 'cvsps.logencoding') |
|
496 | 496 | if encodings: |
|
497 | 497 | def revstr(r): |
|
498 | 498 | # this is needed, because logentry.revision is a tuple of "int" |
|
499 | 499 | # (e.g. (1, 2) for "1.2") |
|
500 | 500 | return '.'.join(pycompat.maplist(pycompat.bytestr, r)) |
|
501 | 501 | |
|
502 | 502 | for entry in log: |
|
503 | 503 | comment = entry.comment |
|
504 | 504 | for e in encodings: |
|
505 | 505 | try: |
|
506 | 506 | entry.comment = comment.decode(e).encode('utf-8') |
|
507 | 507 | if ui.debugflag: |
|
508 | 508 | ui.debug("transcoding by %s: %s of %s\n" % |
|
509 | 509 | (e, revstr(entry.revision), entry.file)) |
|
510 | 510 | break |
|
511 | 511 | except UnicodeDecodeError: |
|
512 | 512 | pass # try next encoding |
|
513 | 513 | except LookupError as inst: # unknown encoding, maybe |
|
514 | 514 | raise error.Abort(inst, |
|
515 | 515 | hint=_('check convert.cvsps.logencoding' |
|
516 | 516 | ' configuration')) |
|
517 | 517 | else: |
|
518 | 518 | raise error.Abort(_("no encoding can transcode" |
|
519 | 519 | " CVS log message for %s of %s") |
|
520 | 520 | % (revstr(entry.revision), entry.file), |
|
521 | 521 | hint=_('check convert.cvsps.logencoding' |
|
522 | 522 | ' configuration')) |
|
523 | 523 | |
|
524 | 524 | hook.hook(ui, None, "cvslog", True, log=log) |
|
525 | 525 | |
|
526 | 526 | return log |
|
527 | 527 | |
|
528 | 528 | |
|
529 | 529 | class changeset(object): |
|
530 | 530 | '''Class changeset has the following attributes: |
|
531 | 531 | .id - integer identifying this changeset (list index) |
|
532 | 532 | .author - author name as CVS knows it |
|
533 | 533 | .branch - name of branch this changeset is on, or None |
|
534 | 534 | .comment - commit message |
|
535 | 535 | .commitid - CVS commitid or None |
|
536 | 536 | .date - the commit date as a (time,tz) tuple |
|
537 | 537 | .entries - list of logentry objects in this changeset |
|
538 | 538 | .parents - list of one or two parent changesets |
|
539 | 539 | .tags - list of tags on this changeset |
|
540 | 540 | .synthetic - from synthetic revision "file ... added on branch ..." |
|
541 | 541 | .mergepoint- the branch that has been merged from or None |
|
542 | 542 | .branchpoints- the branches that start at the current entry or empty |
|
543 | 543 | ''' |
|
544 | 544 | def __init__(self, **entries): |
|
545 | 545 | self.id = None |
|
546 | 546 | self.synthetic = False |
|
547 | 547 | self.__dict__.update(entries) |
|
548 | 548 | |
|
549 | 549 | def __repr__(self): |
|
550 | 550 | items = ("%s=%r"%(k, self.__dict__[k]) for k in sorted(self.__dict__)) |
|
551 | 551 | return "%s(%s)"%(type(self).__name__, ", ".join(items)) |
|
552 | 552 | |
|
553 | 553 | def createchangeset(ui, log, fuzz=60, mergefrom=None, mergeto=None): |
|
554 | 554 | '''Convert log into changesets.''' |
|
555 | 555 | |
|
556 | 556 | ui.status(_('creating changesets\n')) |
|
557 | 557 | |
|
558 | 558 | # try to order commitids by date |
|
559 | 559 | mindate = {} |
|
560 | 560 | for e in log: |
|
561 | 561 | if e.commitid: |
|
562 | 562 | mindate[e.commitid] = min(e.date, mindate.get(e.commitid)) |
|
563 | 563 | |
|
564 | 564 | # Merge changesets |
|
565 | 565 | log.sort(key=lambda x: (mindate.get(x.commitid), x.commitid, x.comment, |
|
566 | 566 | x.author, x.branch, x.date, x.branchpoints)) |
|
567 | 567 | |
|
568 | 568 | changesets = [] |
|
569 | 569 | files = set() |
|
570 | 570 | c = None |
|
571 | 571 | for i, e in enumerate(log): |
|
572 | 572 | |
|
573 | 573 | # Check if log entry belongs to the current changeset or not. |
|
574 | 574 | |
|
575 | 575 | # Since CVS is file-centric, two different file revisions with |
|
576 | 576 | # different branchpoints should be treated as belonging to two |
|
577 | 577 | # different changesets (and the ordering is important and not |
|
578 | 578 | # honoured by cvsps at this point). |
|
579 | 579 | # |
|
580 | 580 | # Consider the following case: |
|
581 | 581 | # foo 1.1 branchpoints: [MYBRANCH] |
|
582 | 582 | # bar 1.1 branchpoints: [MYBRANCH, MYBRANCH2] |
|
583 | 583 | # |
|
584 | 584 | # Here foo is part only of MYBRANCH, but not MYBRANCH2, e.g. a |
|
585 | 585 | # later version of foo may be in MYBRANCH2, so foo should be the |
|
586 | 586 | # first changeset and bar the next and MYBRANCH and MYBRANCH2 |
|
587 | 587 | # should both start off of the bar changeset. No provisions are |
|
588 | 588 | # made to ensure that this is, in fact, what happens. |
|
589 | 589 | if not (c and e.branchpoints == c.branchpoints and |
|
590 | 590 | (# cvs commitids |
|
591 | 591 | (e.commitid is not None and e.commitid == c.commitid) or |
|
592 | 592 | (# no commitids, use fuzzy commit detection |
|
593 | 593 | (e.commitid is None or c.commitid is None) and |
|
594 | 594 | e.comment == c.comment and |
|
595 | 595 | e.author == c.author and |
|
596 | 596 | e.branch == c.branch and |
|
597 | 597 | ((c.date[0] + c.date[1]) <= |
|
598 | 598 | (e.date[0] + e.date[1]) <= |
|
599 | 599 | (c.date[0] + c.date[1]) + fuzz) and |
|
600 | 600 | e.file not in files))): |
|
601 | 601 | c = changeset(comment=e.comment, author=e.author, |
|
602 | 602 | branch=e.branch, date=e.date, |
|
603 | 603 | entries=[], mergepoint=e.mergepoint, |
|
604 | 604 | branchpoints=e.branchpoints, commitid=e.commitid) |
|
605 | 605 | changesets.append(c) |
|
606 | 606 | |
|
607 | 607 | files = set() |
|
608 | 608 | if len(changesets) % 100 == 0: |
|
609 | 609 | t = '%d %s' % (len(changesets), repr(e.comment)[1:-1]) |
|
610 | 610 | ui.status(util.ellipsis(t, 80) + '\n') |
|
611 | 611 | |
|
612 | 612 | c.entries.append(e) |
|
613 | 613 | files.add(e.file) |
|
614 | 614 | c.date = e.date # changeset date is date of latest commit in it |
|
615 | 615 | |
|
616 | 616 | # Mark synthetic changesets |
|
617 | 617 | |
|
618 | 618 | for c in changesets: |
|
619 | 619 | # Synthetic revisions always get their own changeset, because |
|
620 | 620 | # the log message includes the filename. E.g. if you add file3 |
|
621 | 621 | # and file4 on a branch, you get four log entries and three |
|
622 | 622 | # changesets: |
|
623 | 623 | # "File file3 was added on branch ..." (synthetic, 1 entry) |
|
624 | 624 | # "File file4 was added on branch ..." (synthetic, 1 entry) |
|
625 | 625 | # "Add file3 and file4 to fix ..." (real, 2 entries) |
|
626 | 626 | # Hence the check for 1 entry here. |
|
627 | 627 | c.synthetic = len(c.entries) == 1 and c.entries[0].synthetic |
|
628 | 628 | |
|
629 | 629 | # Sort files in each changeset |
|
630 | 630 | |
|
631 | 631 | def entitycompare(l, r): |
|
632 | 632 | 'Mimic cvsps sorting order' |
|
633 | 633 | l = l.file.split('/') |
|
634 | 634 | r = r.file.split('/') |
|
635 | 635 | nl = len(l) |
|
636 | 636 | nr = len(r) |
|
637 | 637 | n = min(nl, nr) |
|
638 | 638 | for i in range(n): |
|
639 | 639 | if i + 1 == nl and nl < nr: |
|
640 | 640 | return -1 |
|
641 | 641 | elif i + 1 == nr and nl > nr: |
|
642 | 642 | return +1 |
|
643 | 643 | elif l[i] < r[i]: |
|
644 | 644 | return -1 |
|
645 | 645 | elif l[i] > r[i]: |
|
646 | 646 | return +1 |
|
647 | 647 | return 0 |
|
648 | 648 | |
|
649 | 649 | for c in changesets: |
|
650 | 650 | c.entries.sort(entitycompare) |
|
651 | 651 | |
|
652 | 652 | # Sort changesets by date |
|
653 | 653 | |
|
654 | 654 | odd = set() |
|
655 | 655 | def cscmp(l, r): |
|
656 | 656 | d = sum(l.date) - sum(r.date) |
|
657 | 657 | if d: |
|
658 | 658 | return d |
|
659 | 659 | |
|
660 | 660 | # detect vendor branches and initial commits on a branch |
|
661 | 661 | le = {} |
|
662 | 662 | for e in l.entries: |
|
663 | 663 | le[e.rcs] = e.revision |
|
664 | 664 | re = {} |
|
665 | 665 | for e in r.entries: |
|
666 | 666 | re[e.rcs] = e.revision |
|
667 | 667 | |
|
668 | 668 | d = 0 |
|
669 | 669 | for e in l.entries: |
|
670 | 670 | if re.get(e.rcs, None) == e.parent: |
|
671 | 671 | assert not d |
|
672 | 672 | d = 1 |
|
673 | 673 | break |
|
674 | 674 | |
|
675 | 675 | for e in r.entries: |
|
676 | 676 | if le.get(e.rcs, None) == e.parent: |
|
677 | 677 | if d: |
|
678 | 678 | odd.add((l, r)) |
|
679 | 679 | d = -1 |
|
680 | 680 | break |
|
681 | 681 | # By this point, the changesets are sufficiently compared that |
|
682 | 682 | # we don't really care about ordering. However, this leaves |
|
683 | 683 | # some race conditions in the tests, so we compare on the |
|
684 | 684 | # number of files modified, the files contained in each |
|
685 | 685 | # changeset, and the branchpoints in the change to ensure test |
|
686 | 686 | # output remains stable. |
|
687 | 687 | |
|
688 | 688 | # recommended replacement for cmp from |
|
689 | 689 | # https://docs.python.org/3.0/whatsnew/3.0.html |
|
690 | 690 | c = lambda x, y: (x > y) - (x < y) |
|
691 | 691 | # Sort bigger changes first. |
|
692 | 692 | if not d: |
|
693 | 693 | d = c(len(l.entries), len(r.entries)) |
|
694 | 694 | # Try sorting by filename in the change. |
|
695 | 695 | if not d: |
|
696 | 696 | d = c([e.file for e in l.entries], [e.file for e in r.entries]) |
|
697 | 697 | # Try and put changes without a branch point before ones with |
|
698 | 698 | # a branch point. |
|
699 | 699 | if not d: |
|
700 | 700 | d = c(len(l.branchpoints), len(r.branchpoints)) |
|
701 | 701 | return d |
|
702 | 702 | |
|
703 | 703 | changesets.sort(cscmp) |
|
704 | 704 | |
|
705 | 705 | # Collect tags |
|
706 | 706 | |
|
707 | 707 | globaltags = {} |
|
708 | 708 | for c in changesets: |
|
709 | 709 | for e in c.entries: |
|
710 | 710 | for tag in e.tags: |
|
711 | 711 | # remember which is the latest changeset to have this tag |
|
712 | 712 | globaltags[tag] = c |
|
713 | 713 | |
|
714 | 714 | for c in changesets: |
|
715 | 715 | tags = set() |
|
716 | 716 | for e in c.entries: |
|
717 | 717 | tags.update(e.tags) |
|
718 | 718 | # remember tags only if this is the latest changeset to have it |
|
719 | 719 | c.tags = sorted(tag for tag in tags if globaltags[tag] is c) |
|
720 | 720 | |
|
721 | 721 | # Find parent changesets, handle {{mergetobranch BRANCHNAME}} |
|
722 | 722 | # by inserting dummy changesets with two parents, and handle |
|
723 | 723 | # {{mergefrombranch BRANCHNAME}} by setting two parents. |
|
724 | 724 | |
|
725 | 725 | if mergeto is None: |
|
726 | 726 | mergeto = r'{{mergetobranch ([-\w]+)}}' |
|
727 | 727 | if mergeto: |
|
728 | 728 | mergeto = re.compile(mergeto) |
|
729 | 729 | |
|
730 | 730 | if mergefrom is None: |
|
731 | 731 | mergefrom = r'{{mergefrombranch ([-\w]+)}}' |
|
732 | 732 | if mergefrom: |
|
733 | 733 | mergefrom = re.compile(mergefrom) |
|
734 | 734 | |
|
735 | 735 | versions = {} # changeset index where we saw any particular file version |
|
736 | 736 | branches = {} # changeset index where we saw a branch |
|
737 | 737 | n = len(changesets) |
|
738 | 738 | i = 0 |
|
739 | 739 | while i < n: |
|
740 | 740 | c = changesets[i] |
|
741 | 741 | |
|
742 | 742 | for f in c.entries: |
|
743 | 743 | versions[(f.rcs, f.revision)] = i |
|
744 | 744 | |
|
745 | 745 | p = None |
|
746 | 746 | if c.branch in branches: |
|
747 | 747 | p = branches[c.branch] |
|
748 | 748 | else: |
|
749 | 749 | # first changeset on a new branch |
|
750 | 750 | # the parent is a changeset with the branch in its |
|
751 | 751 | # branchpoints such that it is the latest possible |
|
752 | 752 | # commit without any intervening, unrelated commits. |
|
753 | 753 | |
|
754 | 754 | for candidate in xrange(i): |
|
755 | 755 | if c.branch not in changesets[candidate].branchpoints: |
|
756 | 756 | if p is not None: |
|
757 | 757 | break |
|
758 | 758 | continue |
|
759 | 759 | p = candidate |
|
760 | 760 | |
|
761 | 761 | c.parents = [] |
|
762 | 762 | if p is not None: |
|
763 | 763 | p = changesets[p] |
|
764 | 764 | |
|
765 | 765 | # Ensure no changeset has a synthetic changeset as a parent. |
|
766 | 766 | while p.synthetic: |
|
767 | 767 | assert len(p.parents) <= 1, \ |
|
768 | 768 | _('synthetic changeset cannot have multiple parents') |
|
769 | 769 | if p.parents: |
|
770 | 770 | p = p.parents[0] |
|
771 | 771 | else: |
|
772 | 772 | p = None |
|
773 | 773 | break |
|
774 | 774 | |
|
775 | 775 | if p is not None: |
|
776 | 776 | c.parents.append(p) |
|
777 | 777 | |
|
778 | 778 | if c.mergepoint: |
|
779 | 779 | if c.mergepoint == 'HEAD': |
|
780 | 780 | c.mergepoint = None |
|
781 | 781 | c.parents.append(changesets[branches[c.mergepoint]]) |
|
782 | 782 | |
|
783 | 783 | if mergefrom: |
|
784 | 784 | m = mergefrom.search(c.comment) |
|
785 | 785 | if m: |
|
786 | 786 | m = m.group(1) |
|
787 | 787 | if m == 'HEAD': |
|
788 | 788 | m = None |
|
789 | 789 | try: |
|
790 | 790 | candidate = changesets[branches[m]] |
|
791 | 791 | except KeyError: |
|
792 | 792 | ui.warn(_("warning: CVS commit message references " |
|
793 | 793 | "non-existent branch %r:\n%s\n") |
|
794 | 794 | % (m, c.comment)) |
|
795 | 795 | if m in branches and c.branch != m and not candidate.synthetic: |
|
796 | 796 | c.parents.append(candidate) |
|
797 | 797 | |
|
798 | 798 | if mergeto: |
|
799 | 799 | m = mergeto.search(c.comment) |
|
800 | 800 | if m: |
|
801 | 801 | if m.groups(): |
|
802 | 802 | m = m.group(1) |
|
803 | 803 | if m == 'HEAD': |
|
804 | 804 | m = None |
|
805 | 805 | else: |
|
806 | 806 | m = None # if no group found then merge to HEAD |
|
807 | 807 | if m in branches and c.branch != m: |
|
808 | 808 | # insert empty changeset for merge |
|
809 | 809 | cc = changeset( |
|
810 | 810 | author=c.author, branch=m, date=c.date, |
|
811 | 811 | comment='convert-repo: CVS merge from branch %s' |
|
812 | 812 | % c.branch, |
|
813 | 813 | entries=[], tags=[], |
|
814 | 814 | parents=[changesets[branches[m]], c]) |
|
815 | 815 | changesets.insert(i + 1, cc) |
|
816 | 816 | branches[m] = i + 1 |
|
817 | 817 | |
|
818 | 818 | # adjust our loop counters now we have inserted a new entry |
|
819 | 819 | n += 1 |
|
820 | 820 | i += 2 |
|
821 | 821 | continue |
|
822 | 822 | |
|
823 | 823 | branches[c.branch] = i |
|
824 | 824 | i += 1 |
|
825 | 825 | |
|
826 | 826 | # Drop synthetic changesets (safe now that we have ensured no other |
|
827 | 827 | # changesets can have them as parents). |
|
828 | 828 | i = 0 |
|
829 | 829 | while i < len(changesets): |
|
830 | 830 | if changesets[i].synthetic: |
|
831 | 831 | del changesets[i] |
|
832 | 832 | else: |
|
833 | 833 | i += 1 |
|
834 | 834 | |
|
835 | 835 | # Number changesets |
|
836 | 836 | |
|
837 | 837 | for i, c in enumerate(changesets): |
|
838 | 838 | c.id = i + 1 |
|
839 | 839 | |
|
840 | 840 | if odd: |
|
841 | 841 | for l, r in odd: |
|
842 | 842 | if l.id is not None and r.id is not None: |
|
843 | 843 | ui.warn(_('changeset %d is both before and after %d\n') |
|
844 | 844 | % (l.id, r.id)) |
|
845 | 845 | |
|
846 | 846 | ui.status(_('%d changeset entries\n') % len(changesets)) |
|
847 | 847 | |
|
848 | 848 | hook.hook(ui, None, "cvschangesets", True, changesets=changesets) |
|
849 | 849 | |
|
850 | 850 | return changesets |
|
851 | 851 | |
|
852 | 852 | |
|
853 | 853 | def debugcvsps(ui, *args, **opts): |
|
854 | 854 | '''Read CVS rlog for current directory or named path in |
|
855 | 855 | repository, and convert the log to changesets based on matching |
|
856 | 856 | commit log entries and dates. |
|
857 | 857 | ''' |
|
858 | 858 | opts = pycompat.byteskwargs(opts) |
|
859 | 859 | if opts["new_cache"]: |
|
860 | 860 | cache = "write" |
|
861 | 861 | elif opts["update_cache"]: |
|
862 | 862 | cache = "update" |
|
863 | 863 | else: |
|
864 | 864 | cache = None |
|
865 | 865 | |
|
866 | 866 | revisions = opts["revisions"] |
|
867 | 867 | |
|
868 | 868 | try: |
|
869 | 869 | if args: |
|
870 | 870 | log = [] |
|
871 | 871 | for d in args: |
|
872 | 872 | log += createlog(ui, d, root=opts["root"], cache=cache) |
|
873 | 873 | else: |
|
874 | 874 | log = createlog(ui, root=opts["root"], cache=cache) |
|
875 | 875 | except logerror as e: |
|
876 | 876 | ui.write("%r\n"%e) |
|
877 | 877 | return |
|
878 | 878 | |
|
879 | 879 | changesets = createchangeset(ui, log, opts["fuzz"]) |
|
880 | 880 | del log |
|
881 | 881 | |
|
882 | 882 | # Print changesets (optionally filtered) |
|
883 | 883 | |
|
884 | 884 | off = len(revisions) |
|
885 | 885 | branches = {} # latest version number in each branch |
|
886 | 886 | ancestors = {} # parent branch |
|
887 | 887 | for cs in changesets: |
|
888 | 888 | |
|
889 | 889 | if opts["ancestors"]: |
|
890 | 890 | if cs.branch not in branches and cs.parents and cs.parents[0].id: |
|
891 | 891 | ancestors[cs.branch] = (changesets[cs.parents[0].id - 1].branch, |
|
892 | 892 | cs.parents[0].id) |
|
893 | 893 | branches[cs.branch] = cs.id |
|
894 | 894 | |
|
895 | 895 | # limit by branches |
|
896 | 896 | if opts["branches"] and (cs.branch or 'HEAD') not in opts["branches"]: |
|
897 | 897 | continue |
|
898 | 898 | |
|
899 | 899 | if not off: |
|
900 | 900 | # Note: trailing spaces on several lines here are needed to have |
|
901 | 901 | # bug-for-bug compatibility with cvsps. |
|
902 | 902 | ui.write('---------------------\n') |
|
903 | 903 | ui.write(('PatchSet %d \n' % cs.id)) |
|
904 | 904 | ui.write(('Date: %s\n' % util.datestr(cs.date, |
|
905 | 905 | '%Y/%m/%d %H:%M:%S %1%2'))) |
|
906 | 906 | ui.write(('Author: %s\n' % cs.author)) |
|
907 | 907 | ui.write(('Branch: %s\n' % (cs.branch or 'HEAD'))) |
|
908 | 908 | ui.write(('Tag%s: %s \n' % (['', 's'][len(cs.tags) > 1], |
|
909 | 909 | ','.join(cs.tags) or '(none)'))) |
|
910 | 910 | if cs.branchpoints: |
|
911 | 911 | ui.write(('Branchpoints: %s \n') % |
|
912 | 912 | ', '.join(sorted(cs.branchpoints))) |
|
913 | 913 | if opts["parents"] and cs.parents: |
|
914 | 914 | if len(cs.parents) > 1: |
|
915 | 915 | ui.write(('Parents: %s\n' % |
|
916 | 916 | (','.join([str(p.id) for p in cs.parents])))) |
|
917 | 917 | else: |
|
918 | 918 | ui.write(('Parent: %d\n' % cs.parents[0].id)) |
|
919 | 919 | |
|
920 | 920 | if opts["ancestors"]: |
|
921 | 921 | b = cs.branch |
|
922 | 922 | r = [] |
|
923 | 923 | while b: |
|
924 | 924 | b, c = ancestors[b] |
|
925 | 925 | r.append('%s:%d:%d' % (b or "HEAD", c, branches[b])) |
|
926 | 926 | if r: |
|
927 | 927 | ui.write(('Ancestors: %s\n' % (','.join(r)))) |
|
928 | 928 | |
|
929 | 929 | ui.write(('Log:\n')) |
|
930 | 930 | ui.write('%s\n\n' % cs.comment) |
|
931 | 931 | ui.write(('Members: \n')) |
|
932 | 932 | for f in cs.entries: |
|
933 | 933 | fn = f.file |
|
934 | 934 | if fn.startswith(opts["prefix"]): |
|
935 | 935 | fn = fn[len(opts["prefix"]):] |
|
936 | 936 | ui.write('\t%s:%s->%s%s \n' % ( |
|
937 | 937 | fn, '.'.join([str(x) for x in f.parent]) or 'INITIAL', |
|
938 | 938 | '.'.join([str(x) for x in f.revision]), |
|
939 | 939 | ['', '(DEAD)'][f.dead])) |
|
940 | 940 | ui.write('\n') |
|
941 | 941 | |
|
942 | 942 | # have we seen the start tag? |
|
943 | 943 | if revisions and off: |
|
944 | 944 | if revisions[0] == str(cs.id) or \ |
|
945 | 945 | revisions[0] in cs.tags: |
|
946 | 946 | off = False |
|
947 | 947 | |
|
948 | 948 | # see if we reached the end tag |
|
949 | 949 | if len(revisions) > 1 and not off: |
|
950 | 950 | if revisions[1] == str(cs.id) or \ |
|
951 | 951 | revisions[1] in cs.tags: |
|
952 | 952 | break |
@@ -1,1356 +1,1356 b'' | |||
|
1 | 1 | # Subversion 1.4/1.5 Python API backend |
|
2 | 2 | # |
|
3 | 3 | # Copyright(C) 2007 Daniel Holth et al |
|
4 | 4 | from __future__ import absolute_import |
|
5 | 5 | |
|
6 | 6 | import os |
|
7 | 7 | import re |
|
8 | 8 | import tempfile |
|
9 | 9 | import xml.dom.minidom |
|
10 | 10 | |
|
11 | 11 | from mercurial.i18n import _ |
|
12 | 12 | from mercurial import ( |
|
13 | 13 | encoding, |
|
14 | 14 | error, |
|
15 | 15 | pycompat, |
|
16 | 16 | util, |
|
17 | 17 | vfs as vfsmod, |
|
18 | 18 | ) |
|
19 | 19 | |
|
20 | 20 | from . import common |
|
21 | 21 | |
|
22 | 22 | pickle = util.pickle |
|
23 | 23 | stringio = util.stringio |
|
24 | 24 | propertycache = util.propertycache |
|
25 | 25 | urlerr = util.urlerr |
|
26 | 26 | urlreq = util.urlreq |
|
27 | 27 | |
|
28 | 28 | commandline = common.commandline |
|
29 | 29 | commit = common.commit |
|
30 | 30 | converter_sink = common.converter_sink |
|
31 | 31 | converter_source = common.converter_source |
|
32 | 32 | decodeargs = common.decodeargs |
|
33 | 33 | encodeargs = common.encodeargs |
|
34 | 34 | makedatetimestamp = common.makedatetimestamp |
|
35 | 35 | mapfile = common.mapfile |
|
36 | 36 | MissingTool = common.MissingTool |
|
37 | 37 | NoRepo = common.NoRepo |
|
38 | 38 | |
|
39 | 39 | # Subversion stuff. Works best with very recent Python SVN bindings |
|
40 | 40 | # e.g. SVN 1.5 or backports. Thanks to the bzr folks for enhancing |
|
41 | 41 | # these bindings. |
|
42 | 42 | |
|
43 | 43 | try: |
|
44 | 44 | import svn |
|
45 | 45 | import svn.client |
|
46 | 46 | import svn.core |
|
47 | 47 | import svn.ra |
|
48 | 48 | import svn.delta |
|
49 | 49 | from . import transport |
|
50 | 50 | import warnings |
|
51 | 51 | warnings.filterwarnings('ignore', |
|
52 | 52 | module='svn.core', |
|
53 | 53 | category=DeprecationWarning) |
|
54 | 54 | svn.core.SubversionException # trigger import to catch error |
|
55 | 55 | |
|
56 | 56 | except ImportError: |
|
57 | 57 | svn = None |
|
58 | 58 | |
|
59 | 59 | class SvnPathNotFound(Exception): |
|
60 | 60 | pass |
|
61 | 61 | |
|
62 | 62 | def revsplit(rev): |
|
63 | 63 | """Parse a revision string and return (uuid, path, revnum). |
|
64 | 64 | >>> revsplit(b'svn:a2147622-4a9f-4db4-a8d3-13562ff547b2' |
|
65 | 65 | ... b'/proj%20B/mytrunk/mytrunk@1') |
|
66 | 66 | ('a2147622-4a9f-4db4-a8d3-13562ff547b2', '/proj%20B/mytrunk/mytrunk', 1) |
|
67 | 67 | >>> revsplit(b'svn:8af66a51-67f5-4354-b62c-98d67cc7be1d@1') |
|
68 | 68 | ('', '', 1) |
|
69 | 69 | >>> revsplit(b'@7') |
|
70 | 70 | ('', '', 7) |
|
71 | 71 | >>> revsplit(b'7') |
|
72 | 72 | ('', '', 0) |
|
73 | 73 | >>> revsplit(b'bad') |
|
74 | 74 | ('', '', 0) |
|
75 | 75 | """ |
|
76 | 76 | parts = rev.rsplit('@', 1) |
|
77 | 77 | revnum = 0 |
|
78 | 78 | if len(parts) > 1: |
|
79 | 79 | revnum = int(parts[1]) |
|
80 | 80 | parts = parts[0].split('/', 1) |
|
81 | 81 | uuid = '' |
|
82 | 82 | mod = '' |
|
83 | 83 | if len(parts) > 1 and parts[0].startswith('svn:'): |
|
84 | 84 | uuid = parts[0][4:] |
|
85 | 85 | mod = '/' + parts[1] |
|
86 | 86 | return uuid, mod, revnum |
|
87 | 87 | |
|
88 | 88 | def quote(s): |
|
89 | 89 | # As of svn 1.7, many svn calls expect "canonical" paths. In |
|
90 | 90 | # theory, we should call svn.core.*canonicalize() on all paths |
|
91 | 91 | # before passing them to the API. Instead, we assume the base url |
|
92 | 92 | # is canonical and copy the behaviour of svn URL encoding function |
|
93 | 93 | # so we can extend it safely with new components. The "safe" |
|
94 | 94 | # characters were taken from the "svn_uri__char_validity" table in |
|
95 | 95 | # libsvn_subr/path.c. |
|
96 | 96 | return urlreq.quote(s, "!$&'()*+,-./:=@_~") |
|
97 | 97 | |
|
98 | 98 | def geturl(path): |
|
99 | 99 | try: |
|
100 | 100 | return svn.client.url_from_path(svn.core.svn_path_canonicalize(path)) |
|
101 | 101 | except svn.core.SubversionException: |
|
102 | 102 | # svn.client.url_from_path() fails with local repositories |
|
103 | 103 | pass |
|
104 | 104 | if os.path.isdir(path): |
|
105 | 105 | path = os.path.normpath(os.path.abspath(path)) |
|
106 | 106 | if pycompat.iswindows: |
|
107 | 107 | path = '/' + util.normpath(path) |
|
108 | 108 | # Module URL is later compared with the repository URL returned |
|
109 | 109 | # by svn API, which is UTF-8. |
|
110 | 110 | path = encoding.tolocal(path) |
|
111 | 111 | path = 'file://%s' % quote(path) |
|
112 | 112 | return svn.core.svn_path_canonicalize(path) |
|
113 | 113 | |
|
114 | 114 | def optrev(number): |
|
115 | 115 | optrev = svn.core.svn_opt_revision_t() |
|
116 | 116 | optrev.kind = svn.core.svn_opt_revision_number |
|
117 | 117 | optrev.value.number = number |
|
118 | 118 | return optrev |
|
119 | 119 | |
|
120 | 120 | class changedpath(object): |
|
121 | 121 | def __init__(self, p): |
|
122 | 122 | self.copyfrom_path = p.copyfrom_path |
|
123 | 123 | self.copyfrom_rev = p.copyfrom_rev |
|
124 | 124 | self.action = p.action |
|
125 | 125 | |
|
126 | 126 | def get_log_child(fp, url, paths, start, end, limit=0, |
|
127 | 127 | discover_changed_paths=True, strict_node_history=False): |
|
128 | 128 | protocol = -1 |
|
129 | 129 | def receiver(orig_paths, revnum, author, date, message, pool): |
|
130 | 130 | paths = {} |
|
131 | 131 | if orig_paths is not None: |
|
132 | 132 | for k, v in orig_paths.iteritems(): |
|
133 | 133 | paths[k] = changedpath(v) |
|
134 | 134 | pickle.dump((paths, revnum, author, date, message), |
|
135 | 135 | fp, protocol) |
|
136 | 136 | |
|
137 | 137 | try: |
|
138 | 138 | # Use an ra of our own so that our parent can consume |
|
139 | 139 | # our results without confusing the server. |
|
140 | 140 | t = transport.SvnRaTransport(url=url) |
|
141 | 141 | svn.ra.get_log(t.ra, paths, start, end, limit, |
|
142 | 142 | discover_changed_paths, |
|
143 | 143 | strict_node_history, |
|
144 | 144 | receiver) |
|
145 | 145 | except IOError: |
|
146 | 146 | # Caller may interrupt the iteration |
|
147 | 147 | pickle.dump(None, fp, protocol) |
|
148 | 148 | except Exception as inst: |
|
149 | 149 | pickle.dump(str(inst), fp, protocol) |
|
150 | 150 | else: |
|
151 | 151 | pickle.dump(None, fp, protocol) |
|
152 | 152 | fp.close() |
|
153 | 153 | # With large history, cleanup process goes crazy and suddenly |
|
154 | 154 | # consumes *huge* amount of memory. The output file being closed, |
|
155 | 155 | # there is no need for clean termination. |
|
156 | 156 | os._exit(0) |
|
157 | 157 | |
|
158 | 158 | def debugsvnlog(ui, **opts): |
|
159 | 159 | """Fetch SVN log in a subprocess and channel them back to parent to |
|
160 | 160 | avoid memory collection issues. |
|
161 | 161 | """ |
|
162 | 162 | if svn is None: |
|
163 | 163 | raise error.Abort(_('debugsvnlog could not load Subversion python ' |
|
164 | 164 | 'bindings')) |
|
165 | 165 | |
|
166 | 166 | args = decodeargs(ui.fin.read()) |
|
167 | 167 | get_log_child(ui.fout, *args) |
|
168 | 168 | |
|
169 | 169 | class logstream(object): |
|
170 | 170 | """Interruptible revision log iterator.""" |
|
171 | 171 | def __init__(self, stdout): |
|
172 | 172 | self._stdout = stdout |
|
173 | 173 | |
|
174 | 174 | def __iter__(self): |
|
175 | 175 | while True: |
|
176 | 176 | try: |
|
177 | 177 | entry = pickle.load(self._stdout) |
|
178 | 178 | except EOFError: |
|
179 | 179 | raise error.Abort(_('Mercurial failed to run itself, check' |
|
180 | 180 | ' hg executable is in PATH')) |
|
181 | 181 | try: |
|
182 | 182 | orig_paths, revnum, author, date, message = entry |
|
183 | 183 | except (TypeError, ValueError): |
|
184 | 184 | if entry is None: |
|
185 | 185 | break |
|
186 | 186 | raise error.Abort(_("log stream exception '%s'") % entry) |
|
187 | 187 | yield entry |
|
188 | 188 | |
|
189 | 189 | def close(self): |
|
190 | 190 | if self._stdout: |
|
191 | 191 | self._stdout.close() |
|
192 | 192 | self._stdout = None |
|
193 | 193 | |
|
194 | 194 | class directlogstream(list): |
|
195 | 195 | """Direct revision log iterator. |
|
196 | 196 | This can be used for debugging and development but it will probably leak |
|
197 | 197 | memory and is not suitable for real conversions.""" |
|
198 | 198 | def __init__(self, url, paths, start, end, limit=0, |
|
199 | 199 | discover_changed_paths=True, strict_node_history=False): |
|
200 | 200 | |
|
201 | 201 | def receiver(orig_paths, revnum, author, date, message, pool): |
|
202 | 202 | paths = {} |
|
203 | 203 | if orig_paths is not None: |
|
204 | 204 | for k, v in orig_paths.iteritems(): |
|
205 | 205 | paths[k] = changedpath(v) |
|
206 | 206 | self.append((paths, revnum, author, date, message)) |
|
207 | 207 | |
|
208 | 208 | # Use an ra of our own so that our parent can consume |
|
209 | 209 | # our results without confusing the server. |
|
210 | 210 | t = transport.SvnRaTransport(url=url) |
|
211 | 211 | svn.ra.get_log(t.ra, paths, start, end, limit, |
|
212 | 212 | discover_changed_paths, |
|
213 | 213 | strict_node_history, |
|
214 | 214 | receiver) |
|
215 | 215 | |
|
216 | 216 | def close(self): |
|
217 | 217 | pass |
|
218 | 218 | |
|
219 | 219 | # Check to see if the given path is a local Subversion repo. Verify this by |
|
220 | 220 | # looking for several svn-specific files and directories in the given |
|
221 | 221 | # directory. |
|
222 | 222 | def filecheck(ui, path, proto): |
|
223 | 223 | for x in ('locks', 'hooks', 'format', 'db'): |
|
224 | 224 | if not os.path.exists(os.path.join(path, x)): |
|
225 | 225 | return False |
|
226 | 226 | return True |
|
227 | 227 | |
|
228 | 228 | # Check to see if a given path is the root of an svn repo over http. We verify |
|
229 | 229 | # this by requesting a version-controlled URL we know can't exist and looking |
|
230 | 230 | # for the svn-specific "not found" XML. |
|
231 | 231 | def httpcheck(ui, path, proto): |
|
232 | 232 | try: |
|
233 | 233 | opener = urlreq.buildopener() |
|
234 | 234 | rsp = opener.open('%s://%s/!svn/ver/0/.svn' % (proto, path), 'rb') |
|
235 | 235 | data = rsp.read() |
|
236 | 236 | except urlerr.httperror as inst: |
|
237 | 237 | if inst.code != 404: |
|
238 | 238 | # Except for 404 we cannot know for sure this is not an svn repo |
|
239 | 239 | ui.warn(_('svn: cannot probe remote repository, assume it could ' |
|
240 | 240 | 'be a subversion repository. Use --source-type if you ' |
|
241 | 241 | 'know better.\n')) |
|
242 | 242 | return True |
|
243 | 243 | data = inst.fp.read() |
|
244 | 244 | except Exception: |
|
245 | 245 | # Could be urlerr.urlerror if the URL is invalid or anything else. |
|
246 | 246 | return False |
|
247 | 247 | return '<m:human-readable errcode="160013">' in data |
|
248 | 248 | |
|
249 | 249 | protomap = {'http': httpcheck, |
|
250 | 250 | 'https': httpcheck, |
|
251 | 251 | 'file': filecheck, |
|
252 | 252 | } |
|
253 | 253 | def issvnurl(ui, url): |
|
254 | 254 | try: |
|
255 | 255 | proto, path = url.split('://', 1) |
|
256 | 256 | if proto == 'file': |
|
257 | 257 | if (pycompat.iswindows and path[:1] == '/' |
|
258 | 258 | and path[1:2].isalpha() and path[2:6].lower() == '%3a/'): |
|
259 | 259 | path = path[:2] + ':/' + path[6:] |
|
260 | 260 | path = urlreq.url2pathname(path) |
|
261 | 261 | except ValueError: |
|
262 | 262 | proto = 'file' |
|
263 | 263 | path = os.path.abspath(url) |
|
264 | 264 | if proto == 'file': |
|
265 | 265 | path = util.pconvert(path) |
|
266 | 266 | check = protomap.get(proto, lambda *args: False) |
|
267 | 267 | while '/' in path: |
|
268 | 268 | if check(ui, path, proto): |
|
269 | 269 | return True |
|
270 | 270 | path = path.rsplit('/', 1)[0] |
|
271 | 271 | return False |
|
272 | 272 | |
|
273 | 273 | # SVN conversion code stolen from bzr-svn and tailor |
|
274 | 274 | # |
|
275 | 275 | # Subversion looks like a versioned filesystem, branches structures |
|
276 | 276 | # are defined by conventions and not enforced by the tool. First, |
|
277 | 277 | # we define the potential branches (modules) as "trunk" and "branches" |
|
278 | 278 | # children directories. Revisions are then identified by their |
|
279 | 279 | # module and revision number (and a repository identifier). |
|
280 | 280 | # |
|
281 | 281 | # The revision graph is really a tree (or a forest). By default, a |
|
282 | 282 | # revision parent is the previous revision in the same module. If the |
|
283 | 283 | # module directory is copied/moved from another module then the |
|
284 | 284 | # revision is the module root and its parent the source revision in |
|
285 | 285 | # the parent module. A revision has at most one parent. |
|
286 | 286 | # |
|
287 | 287 | class svn_source(converter_source): |
|
288 | 288 | def __init__(self, ui, repotype, url, revs=None): |
|
289 | 289 | super(svn_source, self).__init__(ui, repotype, url, revs=revs) |
|
290 | 290 | |
|
291 | 291 | if not (url.startswith('svn://') or url.startswith('svn+ssh://') or |
|
292 | 292 | (os.path.exists(url) and |
|
293 | 293 | os.path.exists(os.path.join(url, '.svn'))) or |
|
294 | 294 | issvnurl(ui, url)): |
|
295 | 295 | raise NoRepo(_("%s does not look like a Subversion repository") |
|
296 | 296 | % url) |
|
297 | 297 | if svn is None: |
|
298 | 298 | raise MissingTool(_('could not load Subversion python bindings')) |
|
299 | 299 | |
|
300 | 300 | try: |
|
301 | 301 | version = svn.core.SVN_VER_MAJOR, svn.core.SVN_VER_MINOR |
|
302 | 302 | if version < (1, 4): |
|
303 | 303 | raise MissingTool(_('Subversion python bindings %d.%d found, ' |
|
304 | 304 | '1.4 or later required') % version) |
|
305 | 305 | except AttributeError: |
|
306 | 306 | raise MissingTool(_('Subversion python bindings are too old, 1.4 ' |
|
307 | 307 | 'or later required')) |
|
308 | 308 | |
|
309 | 309 | self.lastrevs = {} |
|
310 | 310 | |
|
311 | 311 | latest = None |
|
312 | 312 | try: |
|
313 | 313 | # Support file://path@rev syntax. Useful e.g. to convert |
|
314 | 314 | # deleted branches. |
|
315 | 315 | at = url.rfind('@') |
|
316 | 316 | if at >= 0: |
|
317 | 317 | latest = int(url[at + 1:]) |
|
318 | 318 | url = url[:at] |
|
319 | 319 | except ValueError: |
|
320 | 320 | pass |
|
321 | 321 | self.url = geturl(url) |
|
322 | 322 | self.encoding = 'UTF-8' # Subversion is always nominal UTF-8 |
|
323 | 323 | try: |
|
324 | 324 | self.transport = transport.SvnRaTransport(url=self.url) |
|
325 | 325 | self.ra = self.transport.ra |
|
326 | 326 | self.ctx = self.transport.client |
|
327 | 327 | self.baseurl = svn.ra.get_repos_root(self.ra) |
|
328 | 328 | # Module is either empty or a repository path starting with |
|
329 | 329 | # a slash and not ending with a slash. |
|
330 | 330 | self.module = urlreq.unquote(self.url[len(self.baseurl):]) |
|
331 | 331 | self.prevmodule = None |
|
332 | 332 | self.rootmodule = self.module |
|
333 | 333 | self.commits = {} |
|
334 | 334 | self.paths = {} |
|
335 | 335 | self.uuid = svn.ra.get_uuid(self.ra) |
|
336 | 336 | except svn.core.SubversionException: |
|
337 | 337 | ui.traceback() |
|
338 | 338 | svnversion = '%d.%d.%d' % (svn.core.SVN_VER_MAJOR, |
|
339 | 339 | svn.core.SVN_VER_MINOR, |
|
340 | 340 | svn.core.SVN_VER_MICRO) |
|
341 | 341 | raise NoRepo(_("%s does not look like a Subversion repository " |
|
342 | 342 | "to libsvn version %s") |
|
343 | 343 | % (self.url, svnversion)) |
|
344 | 344 | |
|
345 | 345 | if revs: |
|
346 | 346 | if len(revs) > 1: |
|
347 | 347 | raise error.Abort(_('subversion source does not support ' |
|
348 | 348 | 'specifying multiple revisions')) |
|
349 | 349 | try: |
|
350 | 350 | latest = int(revs[0]) |
|
351 | 351 | except ValueError: |
|
352 | 352 | raise error.Abort(_('svn: revision %s is not an integer') % |
|
353 | 353 | revs[0]) |
|
354 | 354 | |
|
355 | 355 | trunkcfg = self.ui.config('convert', 'svn.trunk') |
|
356 | 356 | if trunkcfg is None: |
|
357 | 357 | trunkcfg = 'trunk' |
|
358 | 358 | self.trunkname = trunkcfg.strip('/') |
|
359 | 359 | self.startrev = self.ui.config('convert', 'svn.startrev') |
|
360 | 360 | try: |
|
361 | 361 | self.startrev = int(self.startrev) |
|
362 | 362 | if self.startrev < 0: |
|
363 | 363 | self.startrev = 0 |
|
364 | 364 | except ValueError: |
|
365 | 365 | raise error.Abort(_('svn: start revision %s is not an integer') |
|
366 | 366 | % self.startrev) |
|
367 | 367 | |
|
368 | 368 | try: |
|
369 | 369 | self.head = self.latest(self.module, latest) |
|
370 | 370 | except SvnPathNotFound: |
|
371 | 371 | self.head = None |
|
372 | 372 | if not self.head: |
|
373 | 373 | raise error.Abort(_('no revision found in module %s') |
|
374 | 374 | % self.module) |
|
375 | 375 | self.last_changed = self.revnum(self.head) |
|
376 | 376 | |
|
377 | 377 | self._changescache = (None, None) |
|
378 | 378 | |
|
379 | 379 | if os.path.exists(os.path.join(url, '.svn/entries')): |
|
380 | 380 | self.wc = url |
|
381 | 381 | else: |
|
382 | 382 | self.wc = None |
|
383 | 383 | self.convertfp = None |
|
384 | 384 | |
|
385 | 385 | def setrevmap(self, revmap): |
|
386 | 386 | lastrevs = {} |
|
387 | 387 | for revid in revmap: |
|
388 | 388 | uuid, module, revnum = revsplit(revid) |
|
389 | 389 | lastrevnum = lastrevs.setdefault(module, revnum) |
|
390 | 390 | if revnum > lastrevnum: |
|
391 | 391 | lastrevs[module] = revnum |
|
392 | 392 | self.lastrevs = lastrevs |
|
393 | 393 | |
|
394 | 394 | def exists(self, path, optrev): |
|
395 | 395 | try: |
|
396 | 396 | svn.client.ls(self.url.rstrip('/') + '/' + quote(path), |
|
397 | 397 | optrev, False, self.ctx) |
|
398 | 398 | return True |
|
399 | 399 | except svn.core.SubversionException: |
|
400 | 400 | return False |
|
401 | 401 | |
|
402 | 402 | def getheads(self): |
|
403 | 403 | |
|
404 | 404 | def isdir(path, revnum): |
|
405 | 405 | kind = self._checkpath(path, revnum) |
|
406 | 406 | return kind == svn.core.svn_node_dir |
|
407 | 407 | |
|
408 | 408 | def getcfgpath(name, rev): |
|
409 | 409 | cfgpath = self.ui.config('convert', 'svn.' + name) |
|
410 | 410 | if cfgpath is not None and cfgpath.strip() == '': |
|
411 | 411 | return None |
|
412 | 412 | path = (cfgpath or name).strip('/') |
|
413 | 413 | if not self.exists(path, rev): |
|
414 | 414 | if self.module.endswith(path) and name == 'trunk': |
|
415 | 415 | # we are converting from inside this directory |
|
416 | 416 | return None |
|
417 | 417 | if cfgpath: |
|
418 | 418 | raise error.Abort(_('expected %s to be at %r, but not found' |
|
419 | 419 | ) % (name, path)) |
|
420 | 420 | return None |
|
421 | 421 | self.ui.note(_('found %s at %r\n') % (name, path)) |
|
422 | 422 | return path |
|
423 | 423 | |
|
424 | 424 | rev = optrev(self.last_changed) |
|
425 | 425 | oldmodule = '' |
|
426 | 426 | trunk = getcfgpath('trunk', rev) |
|
427 | 427 | self.tags = getcfgpath('tags', rev) |
|
428 | 428 | branches = getcfgpath('branches', rev) |
|
429 | 429 | |
|
430 | 430 | # If the project has a trunk or branches, we will extract heads |
|
431 | 431 | # from them. We keep the project root otherwise. |
|
432 | 432 | if trunk: |
|
433 | 433 | oldmodule = self.module or '' |
|
434 | 434 | self.module += '/' + trunk |
|
435 | 435 | self.head = self.latest(self.module, self.last_changed) |
|
436 | 436 | if not self.head: |
|
437 | 437 | raise error.Abort(_('no revision found in module %s') |
|
438 | 438 | % self.module) |
|
439 | 439 | |
|
440 | 440 | # First head in the list is the module's head |
|
441 | 441 | self.heads = [self.head] |
|
442 | 442 | if self.tags is not None: |
|
443 | 443 | self.tags = '%s/%s' % (oldmodule , (self.tags or 'tags')) |
|
444 | 444 | |
|
445 | 445 | # Check if branches bring a few more heads to the list |
|
446 | 446 | if branches: |
|
447 | 447 | rpath = self.url.strip('/') |
|
448 | 448 | branchnames = svn.client.ls(rpath + '/' + quote(branches), |
|
449 | 449 | rev, False, self.ctx) |
|
450 | 450 | for branch in sorted(branchnames): |
|
451 | 451 | module = '%s/%s/%s' % (oldmodule, branches, branch) |
|
452 | 452 | if not isdir(module, self.last_changed): |
|
453 | 453 | continue |
|
454 | 454 | brevid = self.latest(module, self.last_changed) |
|
455 | 455 | if not brevid: |
|
456 | 456 | self.ui.note(_('ignoring empty branch %s\n') % branch) |
|
457 | 457 | continue |
|
458 | 458 | self.ui.note(_('found branch %s at %d\n') % |
|
459 | 459 | (branch, self.revnum(brevid))) |
|
460 | 460 | self.heads.append(brevid) |
|
461 | 461 | |
|
462 | 462 | if self.startrev and self.heads: |
|
463 | 463 | if len(self.heads) > 1: |
|
464 | 464 | raise error.Abort(_('svn: start revision is not supported ' |
|
465 | 465 | 'with more than one branch')) |
|
466 | 466 | revnum = self.revnum(self.heads[0]) |
|
467 | 467 | if revnum < self.startrev: |
|
468 | 468 | raise error.Abort( |
|
469 | 469 | _('svn: no revision found after start revision %d') |
|
470 | 470 | % self.startrev) |
|
471 | 471 | |
|
472 | 472 | return self.heads |
|
473 | 473 | |
|
474 | 474 | def _getchanges(self, rev, full): |
|
475 | 475 | (paths, parents) = self.paths[rev] |
|
476 | 476 | copies = {} |
|
477 | 477 | if parents: |
|
478 | 478 | files, self.removed, copies = self.expandpaths(rev, paths, parents) |
|
479 | 479 | if full or not parents: |
|
480 | 480 | # Perform a full checkout on roots |
|
481 | 481 | uuid, module, revnum = revsplit(rev) |
|
482 | 482 | entries = svn.client.ls(self.baseurl + quote(module), |
|
483 | 483 | optrev(revnum), True, self.ctx) |
|
484 | 484 | files = [n for n, e in entries.iteritems() |
|
485 | 485 | if e.kind == svn.core.svn_node_file] |
|
486 | 486 | self.removed = set() |
|
487 | 487 | |
|
488 | 488 | files.sort() |
|
489 | 489 | files = zip(files, [rev] * len(files)) |
|
490 | 490 | return (files, copies) |
|
491 | 491 | |
|
492 | 492 | def getchanges(self, rev, full): |
|
493 | 493 | # reuse cache from getchangedfiles |
|
494 | 494 | if self._changescache[0] == rev and not full: |
|
495 | 495 | (files, copies) = self._changescache[1] |
|
496 | 496 | else: |
|
497 | 497 | (files, copies) = self._getchanges(rev, full) |
|
498 | 498 | # caller caches the result, so free it here to release memory |
|
499 | 499 | del self.paths[rev] |
|
500 | 500 | return (files, copies, set()) |
|
501 | 501 | |
|
502 | 502 | def getchangedfiles(self, rev, i): |
|
503 | 503 | # called from filemap - cache computed values for reuse in getchanges |
|
504 | 504 | (files, copies) = self._getchanges(rev, False) |
|
505 | 505 | self._changescache = (rev, (files, copies)) |
|
506 | 506 | return [f[0] for f in files] |
|
507 | 507 | |
|
508 | 508 | def getcommit(self, rev): |
|
509 | 509 | if rev not in self.commits: |
|
510 | 510 | uuid, module, revnum = revsplit(rev) |
|
511 | 511 | self.module = module |
|
512 | 512 | self.reparent(module) |
|
513 | 513 | # We assume that: |
|
514 | 514 | # - requests for revisions after "stop" come from the |
|
515 | 515 | # revision graph backward traversal. Cache all of them |
|
516 | 516 | # down to stop, they will be used eventually. |
|
517 | 517 | # - requests for revisions before "stop" come to get |
|
518 | 518 | # isolated branches parents. Just fetch what is needed. |
|
519 | 519 | stop = self.lastrevs.get(module, 0) |
|
520 | 520 | if revnum < stop: |
|
521 | 521 | stop = revnum + 1 |
|
522 | 522 | self._fetch_revisions(revnum, stop) |
|
523 | 523 | if rev not in self.commits: |
|
524 | 524 | raise error.Abort(_('svn: revision %s not found') % revnum) |
|
525 | 525 | revcommit = self.commits[rev] |
|
526 | 526 | # caller caches the result, so free it here to release memory |
|
527 | 527 | del self.commits[rev] |
|
528 | 528 | return revcommit |
|
529 | 529 | |
|
530 | 530 | def checkrevformat(self, revstr, mapname='splicemap'): |
|
531 | 531 | """ fails if revision format does not match the correct format""" |
|
532 | 532 | if not re.match(r'svn:[0-9a-f]{8,8}-[0-9a-f]{4,4}-' |
|
533 | 533 | r'[0-9a-f]{4,4}-[0-9a-f]{4,4}-[0-9a-f]' |
|
534 | 534 | r'{12,12}(.*)\@[0-9]+$',revstr): |
|
535 | 535 | raise error.Abort(_('%s entry %s is not a valid revision' |
|
536 | 536 | ' identifier') % (mapname, revstr)) |
|
537 | 537 | |
|
538 | 538 | def numcommits(self): |
|
539 | 539 | return int(self.head.rsplit('@', 1)[1]) - self.startrev |
|
540 | 540 | |
|
541 | 541 | def gettags(self): |
|
542 | 542 | tags = {} |
|
543 | 543 | if self.tags is None: |
|
544 | 544 | return tags |
|
545 | 545 | |
|
546 | 546 | # svn tags are just a convention, project branches left in a |
|
547 | 547 | # 'tags' directory. There is no other relationship than |
|
548 | 548 | # ancestry, which is expensive to discover and makes them hard |
|
549 | 549 | # to update incrementally. Worse, past revisions may be |
|
550 | 550 | # referenced by tags far away in the future, requiring a deep |
|
551 | 551 | # history traversal on every calculation. Current code |
|
552 | 552 | # performs a single backward traversal, tracking moves within |
|
553 | 553 | # the tags directory (tag renaming) and recording a new tag |
|
554 | 554 | # everytime a project is copied from outside the tags |
|
555 | 555 | # directory. It also lists deleted tags, this behaviour may |
|
556 | 556 | # change in the future. |
|
557 | 557 | pendings = [] |
|
558 | 558 | tagspath = self.tags |
|
559 | 559 | start = svn.ra.get_latest_revnum(self.ra) |
|
560 | 560 | stream = self._getlog([self.tags], start, self.startrev) |
|
561 | 561 | try: |
|
562 | 562 | for entry in stream: |
|
563 | 563 | origpaths, revnum, author, date, message = entry |
|
564 | 564 | if not origpaths: |
|
565 | 565 | origpaths = [] |
|
566 | 566 | copies = [(e.copyfrom_path, e.copyfrom_rev, p) for p, e |
|
567 | 567 | in origpaths.iteritems() if e.copyfrom_path] |
|
568 | 568 | # Apply moves/copies from more specific to general |
|
569 | 569 | copies.sort(reverse=True) |
|
570 | 570 | |
|
571 | 571 | srctagspath = tagspath |
|
572 | 572 | if copies and copies[-1][2] == tagspath: |
|
573 | 573 | # Track tags directory moves |
|
574 | 574 | srctagspath = copies.pop()[0] |
|
575 | 575 | |
|
576 | 576 | for source, sourcerev, dest in copies: |
|
577 | 577 | if not dest.startswith(tagspath + '/'): |
|
578 | 578 | continue |
|
579 | 579 | for tag in pendings: |
|
580 | 580 | if tag[0].startswith(dest): |
|
581 | 581 | tagpath = source + tag[0][len(dest):] |
|
582 | 582 | tag[:2] = [tagpath, sourcerev] |
|
583 | 583 | break |
|
584 | 584 | else: |
|
585 | 585 | pendings.append([source, sourcerev, dest]) |
|
586 | 586 | |
|
587 | 587 | # Filter out tags with children coming from different |
|
588 | 588 | # parts of the repository like: |
|
589 | 589 | # /tags/tag.1 (from /trunk:10) |
|
590 | 590 | # /tags/tag.1/foo (from /branches/foo:12) |
|
591 | 591 | # Here/tags/tag.1 discarded as well as its children. |
|
592 | 592 | # It happens with tools like cvs2svn. Such tags cannot |
|
593 | 593 | # be represented in mercurial. |
|
594 | 594 | addeds = dict((p, e.copyfrom_path) for p, e |
|
595 | 595 | in origpaths.iteritems() |
|
596 | 596 | if e.action == 'A' and e.copyfrom_path) |
|
597 | 597 | badroots = set() |
|
598 | 598 | for destroot in addeds: |
|
599 | 599 | for source, sourcerev, dest in pendings: |
|
600 | 600 | if (not dest.startswith(destroot + '/') |
|
601 | 601 | or source.startswith(addeds[destroot] + '/')): |
|
602 | 602 | continue |
|
603 | 603 | badroots.add(destroot) |
|
604 | 604 | break |
|
605 | 605 | |
|
606 | 606 | for badroot in badroots: |
|
607 | 607 | pendings = [p for p in pendings if p[2] != badroot |
|
608 | 608 | and not p[2].startswith(badroot + '/')] |
|
609 | 609 | |
|
610 | 610 | # Tell tag renamings from tag creations |
|
611 | 611 | renamings = [] |
|
612 | 612 | for source, sourcerev, dest in pendings: |
|
613 | 613 | tagname = dest.split('/')[-1] |
|
614 | 614 | if source.startswith(srctagspath): |
|
615 | 615 | renamings.append([source, sourcerev, tagname]) |
|
616 | 616 | continue |
|
617 | 617 | if tagname in tags: |
|
618 | 618 | # Keep the latest tag value |
|
619 | 619 | continue |
|
620 | 620 | # From revision may be fake, get one with changes |
|
621 | 621 | try: |
|
622 | 622 | tagid = self.latest(source, sourcerev) |
|
623 | 623 | if tagid and tagname not in tags: |
|
624 | 624 | tags[tagname] = tagid |
|
625 | 625 | except SvnPathNotFound: |
|
626 | 626 | # It happens when we are following directories |
|
627 | 627 | # we assumed were copied with their parents |
|
628 | 628 | # but were really created in the tag |
|
629 | 629 | # directory. |
|
630 | 630 | pass |
|
631 | 631 | pendings = renamings |
|
632 | 632 | tagspath = srctagspath |
|
633 | 633 | finally: |
|
634 | 634 | stream.close() |
|
635 | 635 | return tags |
|
636 | 636 | |
|
637 | 637 | def converted(self, rev, destrev): |
|
638 | 638 | if not self.wc: |
|
639 | 639 | return |
|
640 | 640 | if self.convertfp is None: |
|
641 | 641 | self.convertfp = open(os.path.join(self.wc, '.svn', 'hg-shamap'), |
|
642 | 642 | 'ab') |
|
643 | 643 | self.convertfp.write(util.tonativeeol('%s %d\n' |
|
644 | 644 | % (destrev, self.revnum(rev)))) |
|
645 | 645 | self.convertfp.flush() |
|
646 | 646 | |
|
647 | 647 | def revid(self, revnum, module=None): |
|
648 | 648 | return 'svn:%s%s@%s' % (self.uuid, module or self.module, revnum) |
|
649 | 649 | |
|
650 | 650 | def revnum(self, rev): |
|
651 | 651 | return int(rev.split('@')[-1]) |
|
652 | 652 | |
|
653 | 653 | def latest(self, path, stop=None): |
|
654 | 654 | """Find the latest revid affecting path, up to stop revision |
|
655 | 655 | number. If stop is None, default to repository latest |
|
656 | 656 | revision. It may return a revision in a different module, |
|
657 | 657 | since a branch may be moved without a change being |
|
658 | 658 | reported. Return None if computed module does not belong to |
|
659 | 659 | rootmodule subtree. |
|
660 | 660 | """ |
|
661 | 661 | def findchanges(path, start, stop=None): |
|
662 | 662 | stream = self._getlog([path], start, stop or 1) |
|
663 | 663 | try: |
|
664 | 664 | for entry in stream: |
|
665 | 665 | paths, revnum, author, date, message = entry |
|
666 | 666 | if stop is None and paths: |
|
667 | 667 | # We do not know the latest changed revision, |
|
668 | 668 | # keep the first one with changed paths. |
|
669 | 669 | break |
|
670 | 670 | if revnum <= stop: |
|
671 | 671 | break |
|
672 | 672 | |
|
673 | 673 | for p in paths: |
|
674 | 674 | if (not path.startswith(p) or |
|
675 | 675 | not paths[p].copyfrom_path): |
|
676 | 676 | continue |
|
677 | 677 | newpath = paths[p].copyfrom_path + path[len(p):] |
|
678 | 678 | self.ui.debug("branch renamed from %s to %s at %d\n" % |
|
679 | 679 | (path, newpath, revnum)) |
|
680 | 680 | path = newpath |
|
681 | 681 | break |
|
682 | 682 | if not paths: |
|
683 | 683 | revnum = None |
|
684 | 684 | return revnum, path |
|
685 | 685 | finally: |
|
686 | 686 | stream.close() |
|
687 | 687 | |
|
688 | 688 | if not path.startswith(self.rootmodule): |
|
689 | 689 | # Requests on foreign branches may be forbidden at server level |
|
690 | 690 | self.ui.debug('ignoring foreign branch %r\n' % path) |
|
691 | 691 | return None |
|
692 | 692 | |
|
693 | 693 | if stop is None: |
|
694 | 694 | stop = svn.ra.get_latest_revnum(self.ra) |
|
695 | 695 | try: |
|
696 | 696 | prevmodule = self.reparent('') |
|
697 | 697 | dirent = svn.ra.stat(self.ra, path.strip('/'), stop) |
|
698 | 698 | self.reparent(prevmodule) |
|
699 | 699 | except svn.core.SubversionException: |
|
700 | 700 | dirent = None |
|
701 | 701 | if not dirent: |
|
702 | 702 | raise SvnPathNotFound(_('%s not found up to revision %d') |
|
703 | 703 | % (path, stop)) |
|
704 | 704 | |
|
705 | 705 | # stat() gives us the previous revision on this line of |
|
706 | 706 | # development, but it might be in *another module*. Fetch the |
|
707 | 707 | # log and detect renames down to the latest revision. |
|
708 | 708 | revnum, realpath = findchanges(path, stop, dirent.created_rev) |
|
709 | 709 | if revnum is None: |
|
710 | 710 | # Tools like svnsync can create empty revision, when |
|
711 | 711 | # synchronizing only a subtree for instance. These empty |
|
712 | 712 | # revisions created_rev still have their original values |
|
713 | 713 | # despite all changes having disappeared and can be |
|
714 | 714 | # returned by ra.stat(), at least when stating the root |
|
715 | 715 | # module. In that case, do not trust created_rev and scan |
|
716 | 716 | # the whole history. |
|
717 | 717 | revnum, realpath = findchanges(path, stop) |
|
718 | 718 | if revnum is None: |
|
719 | 719 | self.ui.debug('ignoring empty branch %r\n' % realpath) |
|
720 | 720 | return None |
|
721 | 721 | |
|
722 | 722 | if not realpath.startswith(self.rootmodule): |
|
723 | 723 | self.ui.debug('ignoring foreign branch %r\n' % realpath) |
|
724 | 724 | return None |
|
725 | 725 | return self.revid(revnum, realpath) |
|
726 | 726 | |
|
727 | 727 | def reparent(self, module): |
|
728 | 728 | """Reparent the svn transport and return the previous parent.""" |
|
729 | 729 | if self.prevmodule == module: |
|
730 | 730 | return module |
|
731 | 731 | svnurl = self.baseurl + quote(module) |
|
732 | 732 | prevmodule = self.prevmodule |
|
733 | 733 | if prevmodule is None: |
|
734 | 734 | prevmodule = '' |
|
735 | 735 | self.ui.debug("reparent to %s\n" % svnurl) |
|
736 | 736 | svn.ra.reparent(self.ra, svnurl) |
|
737 | 737 | self.prevmodule = module |
|
738 | 738 | return prevmodule |
|
739 | 739 | |
|
740 | 740 | def expandpaths(self, rev, paths, parents): |
|
741 | 741 | changed, removed = set(), set() |
|
742 | 742 | copies = {} |
|
743 | 743 | |
|
744 | 744 | new_module, revnum = revsplit(rev)[1:] |
|
745 | 745 | if new_module != self.module: |
|
746 | 746 | self.module = new_module |
|
747 | 747 | self.reparent(self.module) |
|
748 | 748 | |
|
749 | 749 | for i, (path, ent) in enumerate(paths): |
|
750 | 750 | self.ui.progress(_('scanning paths'), i, item=path, |
|
751 | 751 | total=len(paths), unit=_('paths')) |
|
752 | 752 | entrypath = self.getrelpath(path) |
|
753 | 753 | |
|
754 | 754 | kind = self._checkpath(entrypath, revnum) |
|
755 | 755 | if kind == svn.core.svn_node_file: |
|
756 | 756 | changed.add(self.recode(entrypath)) |
|
757 | 757 | if not ent.copyfrom_path or not parents: |
|
758 | 758 | continue |
|
759 | 759 | # Copy sources not in parent revisions cannot be |
|
760 | 760 | # represented, ignore their origin for now |
|
761 | 761 | pmodule, prevnum = revsplit(parents[0])[1:] |
|
762 | 762 | if ent.copyfrom_rev < prevnum: |
|
763 | 763 | continue |
|
764 | 764 | copyfrom_path = self.getrelpath(ent.copyfrom_path, pmodule) |
|
765 | 765 | if not copyfrom_path: |
|
766 | 766 | continue |
|
767 | 767 | self.ui.debug("copied to %s from %s@%s\n" % |
|
768 | 768 | (entrypath, copyfrom_path, ent.copyfrom_rev)) |
|
769 | 769 | copies[self.recode(entrypath)] = self.recode(copyfrom_path) |
|
770 | 770 | elif kind == 0: # gone, but had better be a deleted *file* |
|
771 | 771 | self.ui.debug("gone from %s\n" % ent.copyfrom_rev) |
|
772 | 772 | pmodule, prevnum = revsplit(parents[0])[1:] |
|
773 | 773 | parentpath = pmodule + "/" + entrypath |
|
774 | 774 | fromkind = self._checkpath(entrypath, prevnum, pmodule) |
|
775 | 775 | |
|
776 | 776 | if fromkind == svn.core.svn_node_file: |
|
777 | 777 | removed.add(self.recode(entrypath)) |
|
778 | 778 | elif fromkind == svn.core.svn_node_dir: |
|
779 | 779 | oroot = parentpath.strip('/') |
|
780 | 780 | nroot = path.strip('/') |
|
781 | 781 | children = self._iterfiles(oroot, prevnum) |
|
782 | 782 | for childpath in children: |
|
783 | 783 | childpath = childpath.replace(oroot, nroot) |
|
784 | 784 | childpath = self.getrelpath("/" + childpath, pmodule) |
|
785 | 785 | if childpath: |
|
786 | 786 | removed.add(self.recode(childpath)) |
|
787 | 787 | else: |
|
788 | 788 | self.ui.debug('unknown path in revision %d: %s\n' % \ |
|
789 | 789 | (revnum, path)) |
|
790 | 790 | elif kind == svn.core.svn_node_dir: |
|
791 | 791 | if ent.action == 'M': |
|
792 | 792 | # If the directory just had a prop change, |
|
793 | 793 | # then we shouldn't need to look for its children. |
|
794 | 794 | continue |
|
795 | 795 | if ent.action == 'R' and parents: |
|
796 | 796 | # If a directory is replacing a file, mark the previous |
|
797 | 797 | # file as deleted |
|
798 | 798 | pmodule, prevnum = revsplit(parents[0])[1:] |
|
799 | 799 | pkind = self._checkpath(entrypath, prevnum, pmodule) |
|
800 | 800 | if pkind == svn.core.svn_node_file: |
|
801 | 801 | removed.add(self.recode(entrypath)) |
|
802 | 802 | elif pkind == svn.core.svn_node_dir: |
|
803 | 803 | # We do not know what files were kept or removed, |
|
804 | 804 | # mark them all as changed. |
|
805 | 805 | for childpath in self._iterfiles(pmodule, prevnum): |
|
806 | 806 | childpath = self.getrelpath("/" + childpath) |
|
807 | 807 | if childpath: |
|
808 | 808 | changed.add(self.recode(childpath)) |
|
809 | 809 | |
|
810 | 810 | for childpath in self._iterfiles(path, revnum): |
|
811 | 811 | childpath = self.getrelpath("/" + childpath) |
|
812 | 812 | if childpath: |
|
813 | 813 | changed.add(self.recode(childpath)) |
|
814 | 814 | |
|
815 | 815 | # Handle directory copies |
|
816 | 816 | if not ent.copyfrom_path or not parents: |
|
817 | 817 | continue |
|
818 | 818 | # Copy sources not in parent revisions cannot be |
|
819 | 819 | # represented, ignore their origin for now |
|
820 | 820 | pmodule, prevnum = revsplit(parents[0])[1:] |
|
821 | 821 | if ent.copyfrom_rev < prevnum: |
|
822 | 822 | continue |
|
823 | 823 | copyfrompath = self.getrelpath(ent.copyfrom_path, pmodule) |
|
824 | 824 | if not copyfrompath: |
|
825 | 825 | continue |
|
826 | 826 | self.ui.debug("mark %s came from %s:%d\n" |
|
827 | 827 | % (path, copyfrompath, ent.copyfrom_rev)) |
|
828 | 828 | children = self._iterfiles(ent.copyfrom_path, ent.copyfrom_rev) |
|
829 | 829 | for childpath in children: |
|
830 | 830 | childpath = self.getrelpath("/" + childpath, pmodule) |
|
831 | 831 | if not childpath: |
|
832 | 832 | continue |
|
833 | 833 | copytopath = path + childpath[len(copyfrompath):] |
|
834 | 834 | copytopath = self.getrelpath(copytopath) |
|
835 | 835 | copies[self.recode(copytopath)] = self.recode(childpath) |
|
836 | 836 | |
|
837 | 837 | self.ui.progress(_('scanning paths'), None) |
|
838 | 838 | changed.update(removed) |
|
839 | 839 | return (list(changed), removed, copies) |
|
840 | 840 | |
|
841 | 841 | def _fetch_revisions(self, from_revnum, to_revnum): |
|
842 | 842 | if from_revnum < to_revnum: |
|
843 | 843 | from_revnum, to_revnum = to_revnum, from_revnum |
|
844 | 844 | |
|
845 | 845 | self.child_cset = None |
|
846 | 846 | |
|
847 | 847 | def parselogentry(orig_paths, revnum, author, date, message): |
|
848 | 848 | """Return the parsed commit object or None, and True if |
|
849 | 849 | the revision is a branch root. |
|
850 | 850 | """ |
|
851 | 851 | self.ui.debug("parsing revision %d (%d changes)\n" % |
|
852 | 852 | (revnum, len(orig_paths))) |
|
853 | 853 | |
|
854 | 854 | branched = False |
|
855 | 855 | rev = self.revid(revnum) |
|
856 | 856 | # branch log might return entries for a parent we already have |
|
857 | 857 | |
|
858 | 858 | if rev in self.commits or revnum < to_revnum: |
|
859 | 859 | return None, branched |
|
860 | 860 | |
|
861 | 861 | parents = [] |
|
862 | 862 | # check whether this revision is the start of a branch or part |
|
863 | 863 | # of a branch renaming |
|
864 | 864 | orig_paths = sorted(orig_paths.iteritems()) |
|
865 | 865 | root_paths = [(p, e) for p, e in orig_paths |
|
866 | 866 | if self.module.startswith(p)] |
|
867 | 867 | if root_paths: |
|
868 | 868 | path, ent = root_paths[-1] |
|
869 | 869 | if ent.copyfrom_path: |
|
870 | 870 | branched = True |
|
871 | 871 | newpath = ent.copyfrom_path + self.module[len(path):] |
|
872 | 872 | # ent.copyfrom_rev may not be the actual last revision |
|
873 | 873 | previd = self.latest(newpath, ent.copyfrom_rev) |
|
874 | 874 | if previd is not None: |
|
875 | 875 | prevmodule, prevnum = revsplit(previd)[1:] |
|
876 | 876 | if prevnum >= self.startrev: |
|
877 | 877 | parents = [previd] |
|
878 | 878 | self.ui.note( |
|
879 | 879 | _('found parent of branch %s at %d: %s\n') % |
|
880 | 880 | (self.module, prevnum, prevmodule)) |
|
881 | 881 | else: |
|
882 | 882 | self.ui.debug("no copyfrom path, don't know what to do.\n") |
|
883 | 883 | |
|
884 | 884 | paths = [] |
|
885 | 885 | # filter out unrelated paths |
|
886 | 886 | for path, ent in orig_paths: |
|
887 | 887 | if self.getrelpath(path) is None: |
|
888 | 888 | continue |
|
889 | 889 | paths.append((path, ent)) |
|
890 | 890 | |
|
891 | 891 | # Example SVN datetime. Includes microseconds. |
|
892 | 892 | # ISO-8601 conformant |
|
893 | 893 | # '2007-01-04T17:35:00.902377Z' |
|
894 | 894 | date = util.parsedate(date[:19] + " UTC", ["%Y-%m-%dT%H:%M:%S"]) |
|
895 | 895 | if self.ui.configbool('convert', 'localtimezone'): |
|
896 | 896 | date = makedatetimestamp(date[0]) |
|
897 | 897 | |
|
898 | 898 | if message: |
|
899 | 899 | log = self.recode(message) |
|
900 | 900 | else: |
|
901 | 901 | log = '' |
|
902 | 902 | |
|
903 | 903 | if author: |
|
904 | 904 | author = self.recode(author) |
|
905 | 905 | else: |
|
906 | 906 | author = '' |
|
907 | 907 | |
|
908 | 908 | try: |
|
909 | 909 | branch = self.module.split("/")[-1] |
|
910 | 910 | if branch == self.trunkname: |
|
911 | 911 | branch = None |
|
912 | 912 | except IndexError: |
|
913 | 913 | branch = None |
|
914 | 914 | |
|
915 | 915 | cset = commit(author=author, |
|
916 | 916 | date=util.datestr(date, '%Y-%m-%d %H:%M:%S %1%2'), |
|
917 | 917 | desc=log, |
|
918 | 918 | parents=parents, |
|
919 | 919 | branch=branch, |
|
920 | 920 | rev=rev) |
|
921 | 921 | |
|
922 | 922 | self.commits[rev] = cset |
|
923 | 923 | # The parents list is *shared* among self.paths and the |
|
924 | 924 | # commit object. Both will be updated below. |
|
925 | 925 | self.paths[rev] = (paths, cset.parents) |
|
926 | 926 | if self.child_cset and not self.child_cset.parents: |
|
927 | 927 | self.child_cset.parents[:] = [rev] |
|
928 | 928 | self.child_cset = cset |
|
929 | 929 | return cset, branched |
|
930 | 930 | |
|
931 | 931 | self.ui.note(_('fetching revision log for "%s" from %d to %d\n') % |
|
932 | 932 | (self.module, from_revnum, to_revnum)) |
|
933 | 933 | |
|
934 | 934 | try: |
|
935 | 935 | firstcset = None |
|
936 | 936 | lastonbranch = False |
|
937 | 937 | stream = self._getlog([self.module], from_revnum, to_revnum) |
|
938 | 938 | try: |
|
939 | 939 | for entry in stream: |
|
940 | 940 | paths, revnum, author, date, message = entry |
|
941 | 941 | if revnum < self.startrev: |
|
942 | 942 | lastonbranch = True |
|
943 | 943 | break |
|
944 | 944 | if not paths: |
|
945 | 945 | self.ui.debug('revision %d has no entries\n' % revnum) |
|
946 | 946 | # If we ever leave the loop on an empty |
|
947 | 947 | # revision, do not try to get a parent branch |
|
948 | 948 | lastonbranch = lastonbranch or revnum == 0 |
|
949 | 949 | continue |
|
950 | 950 | cset, lastonbranch = parselogentry(paths, revnum, author, |
|
951 | 951 | date, message) |
|
952 | 952 | if cset: |
|
953 | 953 | firstcset = cset |
|
954 | 954 | if lastonbranch: |
|
955 | 955 | break |
|
956 | 956 | finally: |
|
957 | 957 | stream.close() |
|
958 | 958 | |
|
959 | 959 | if not lastonbranch and firstcset and not firstcset.parents: |
|
960 | 960 | # The first revision of the sequence (the last fetched one) |
|
961 | 961 | # has invalid parents if not a branch root. Find the parent |
|
962 | 962 | # revision now, if any. |
|
963 | 963 | try: |
|
964 | 964 | firstrevnum = self.revnum(firstcset.rev) |
|
965 | 965 | if firstrevnum > 1: |
|
966 | 966 | latest = self.latest(self.module, firstrevnum - 1) |
|
967 | 967 | if latest: |
|
968 | 968 | firstcset.parents.append(latest) |
|
969 | 969 | except SvnPathNotFound: |
|
970 | 970 | pass |
|
971 | 971 | except svn.core.SubversionException as xxx_todo_changeme: |
|
972 | 972 | (inst, num) = xxx_todo_changeme.args |
|
973 | 973 | if num == svn.core.SVN_ERR_FS_NO_SUCH_REVISION: |
|
974 | 974 | raise error.Abort(_('svn: branch has no revision %s') |
|
975 | 975 | % to_revnum) |
|
976 | 976 | raise |
|
977 | 977 | |
|
978 | 978 | def getfile(self, file, rev): |
|
979 | 979 | # TODO: ra.get_file transmits the whole file instead of diffs. |
|
980 | 980 | if file in self.removed: |
|
981 | 981 | return None, None |
|
982 | 982 | mode = '' |
|
983 | 983 | try: |
|
984 | 984 | new_module, revnum = revsplit(rev)[1:] |
|
985 | 985 | if self.module != new_module: |
|
986 | 986 | self.module = new_module |
|
987 | 987 | self.reparent(self.module) |
|
988 | 988 | io = stringio() |
|
989 | 989 | info = svn.ra.get_file(self.ra, file, revnum, io) |
|
990 | 990 | data = io.getvalue() |
|
991 | 991 | # ra.get_file() seems to keep a reference on the input buffer |
|
992 | 992 | # preventing collection. Release it explicitly. |
|
993 | 993 | io.close() |
|
994 | 994 | if isinstance(info, list): |
|
995 | 995 | info = info[-1] |
|
996 | 996 | mode = ("svn:executable" in info) and 'x' or '' |
|
997 | 997 | mode = ("svn:special" in info) and 'l' or mode |
|
998 | 998 | except svn.core.SubversionException as e: |
|
999 | 999 | notfound = (svn.core.SVN_ERR_FS_NOT_FOUND, |
|
1000 | 1000 | svn.core.SVN_ERR_RA_DAV_PATH_NOT_FOUND) |
|
1001 | 1001 | if e.apr_err in notfound: # File not found |
|
1002 | 1002 | return None, None |
|
1003 | 1003 | raise |
|
1004 | 1004 | if mode == 'l': |
|
1005 | 1005 | link_prefix = "link " |
|
1006 | 1006 | if data.startswith(link_prefix): |
|
1007 | 1007 | data = data[len(link_prefix):] |
|
1008 | 1008 | return data, mode |
|
1009 | 1009 | |
|
1010 | 1010 | def _iterfiles(self, path, revnum): |
|
1011 | 1011 | """Enumerate all files in path at revnum, recursively.""" |
|
1012 | 1012 | path = path.strip('/') |
|
1013 | 1013 | pool = svn.core.Pool() |
|
1014 | 1014 | rpath = '/'.join([self.baseurl, quote(path)]).strip('/') |
|
1015 | 1015 | entries = svn.client.ls(rpath, optrev(revnum), True, self.ctx, pool) |
|
1016 | 1016 | if path: |
|
1017 | 1017 | path += '/' |
|
1018 | 1018 | return ((path + p) for p, e in entries.iteritems() |
|
1019 | 1019 | if e.kind == svn.core.svn_node_file) |
|
1020 | 1020 | |
|
1021 | 1021 | def getrelpath(self, path, module=None): |
|
1022 | 1022 | if module is None: |
|
1023 | 1023 | module = self.module |
|
1024 | 1024 | # Given the repository url of this wc, say |
|
1025 | 1025 | # "http://server/plone/CMFPlone/branches/Plone-2_0-branch" |
|
1026 | 1026 | # extract the "entry" portion (a relative path) from what |
|
1027 | 1027 | # svn log --xml says, i.e. |
|
1028 | 1028 | # "/CMFPlone/branches/Plone-2_0-branch/tests/PloneTestCase.py" |
|
1029 | 1029 | # that is to say "tests/PloneTestCase.py" |
|
1030 | 1030 | if path.startswith(module): |
|
1031 | 1031 | relative = path.rstrip('/')[len(module):] |
|
1032 | 1032 | if relative.startswith('/'): |
|
1033 | 1033 | return relative[1:] |
|
1034 | 1034 | elif relative == '': |
|
1035 | 1035 | return relative |
|
1036 | 1036 | |
|
1037 | 1037 | # The path is outside our tracked tree... |
|
1038 | 1038 | self.ui.debug('%r is not under %r, ignoring\n' % (path, module)) |
|
1039 | 1039 | return None |
|
1040 | 1040 | |
|
1041 | 1041 | def _checkpath(self, path, revnum, module=None): |
|
1042 | 1042 | if module is not None: |
|
1043 | 1043 | prevmodule = self.reparent('') |
|
1044 | 1044 | path = module + '/' + path |
|
1045 | 1045 | try: |
|
1046 | 1046 | # ra.check_path does not like leading slashes very much, it leads |
|
1047 | 1047 | # to PROPFIND subversion errors |
|
1048 | 1048 | return svn.ra.check_path(self.ra, path.strip('/'), revnum) |
|
1049 | 1049 | finally: |
|
1050 | 1050 | if module is not None: |
|
1051 | 1051 | self.reparent(prevmodule) |
|
1052 | 1052 | |
|
1053 | 1053 | def _getlog(self, paths, start, end, limit=0, discover_changed_paths=True, |
|
1054 | 1054 | strict_node_history=False): |
|
1055 | 1055 | # Normalize path names, svn >= 1.5 only wants paths relative to |
|
1056 | 1056 | # supplied URL |
|
1057 | 1057 | relpaths = [] |
|
1058 | 1058 | for p in paths: |
|
1059 | 1059 | if not p.startswith('/'): |
|
1060 | 1060 | p = self.module + '/' + p |
|
1061 | 1061 | relpaths.append(p.strip('/')) |
|
1062 | 1062 | args = [self.baseurl, relpaths, start, end, limit, |
|
1063 | 1063 | discover_changed_paths, strict_node_history] |
|
1064 | 1064 | # developer config: convert.svn.debugsvnlog |
|
1065 | 1065 | if not self.ui.configbool('convert', 'svn.debugsvnlog'): |
|
1066 | 1066 | return directlogstream(*args) |
|
1067 | 1067 | arg = encodeargs(args) |
|
1068 | 1068 | hgexe = util.hgexecutable() |
|
1069 | 1069 | cmd = '%s debugsvnlog' % util.shellquote(hgexe) |
|
1070 | 1070 | stdin, stdout = util.popen2(util.quotecommand(cmd)) |
|
1071 | 1071 | stdin.write(arg) |
|
1072 | 1072 | try: |
|
1073 | 1073 | stdin.close() |
|
1074 | 1074 | except IOError: |
|
1075 | 1075 | raise error.Abort(_('Mercurial failed to run itself, check' |
|
1076 | 1076 | ' hg executable is in PATH')) |
|
1077 | 1077 | return logstream(stdout) |
|
1078 | 1078 | |
|
1079 | 1079 | pre_revprop_change = '''#!/bin/sh |
|
1080 | 1080 | |
|
1081 | 1081 | REPOS="$1" |
|
1082 | 1082 | REV="$2" |
|
1083 | 1083 | USER="$3" |
|
1084 | 1084 | PROPNAME="$4" |
|
1085 | 1085 | ACTION="$5" |
|
1086 | 1086 | |
|
1087 | 1087 | if [ "$ACTION" = "M" -a "$PROPNAME" = "svn:log" ]; then exit 0; fi |
|
1088 | 1088 | if [ "$ACTION" = "A" -a "$PROPNAME" = "hg:convert-branch" ]; then exit 0; fi |
|
1089 | 1089 | if [ "$ACTION" = "A" -a "$PROPNAME" = "hg:convert-rev" ]; then exit 0; fi |
|
1090 | 1090 | |
|
1091 | 1091 | echo "Changing prohibited revision property" >&2 |
|
1092 | 1092 | exit 1 |
|
1093 | 1093 | ''' |
|
1094 | 1094 | |
|
1095 | 1095 | class svn_sink(converter_sink, commandline): |
|
1096 | 1096 | commit_re = re.compile(r'Committed revision (\d+).', re.M) |
|
1097 | 1097 | uuid_re = re.compile(r'Repository UUID:\s*(\S+)', re.M) |
|
1098 | 1098 | |
|
1099 | 1099 | def prerun(self): |
|
1100 | 1100 | if self.wc: |
|
1101 | 1101 | os.chdir(self.wc) |
|
1102 | 1102 | |
|
1103 | 1103 | def postrun(self): |
|
1104 | 1104 | if self.wc: |
|
1105 | 1105 | os.chdir(self.cwd) |
|
1106 | 1106 | |
|
1107 | 1107 | def join(self, name): |
|
1108 | 1108 | return os.path.join(self.wc, '.svn', name) |
|
1109 | 1109 | |
|
1110 | 1110 | def revmapfile(self): |
|
1111 | 1111 | return self.join('hg-shamap') |
|
1112 | 1112 | |
|
1113 | 1113 | def authorfile(self): |
|
1114 | 1114 | return self.join('hg-authormap') |
|
1115 | 1115 | |
|
1116 | 1116 | def __init__(self, ui, repotype, path): |
|
1117 | 1117 | |
|
1118 | 1118 | converter_sink.__init__(self, ui, repotype, path) |
|
1119 | 1119 | commandline.__init__(self, ui, 'svn') |
|
1120 | 1120 | self.delete = [] |
|
1121 | 1121 | self.setexec = [] |
|
1122 | 1122 | self.delexec = [] |
|
1123 | 1123 | self.copies = [] |
|
1124 | 1124 | self.wc = None |
|
1125 | 1125 | self.cwd = pycompat.getcwd() |
|
1126 | 1126 | |
|
1127 | 1127 | created = False |
|
1128 | 1128 | if os.path.isfile(os.path.join(path, '.svn', 'entries')): |
|
1129 | 1129 | self.wc = os.path.realpath(path) |
|
1130 | 1130 | self.run0('update') |
|
1131 | 1131 | else: |
|
1132 | if not re.search(r'^(file|http|https|svn|svn\+ssh)\://', path): | |
|
1132 | if not re.search(br'^(file|http|https|svn|svn\+ssh)\://', path): | |
|
1133 | 1133 | path = os.path.realpath(path) |
|
1134 | 1134 | if os.path.isdir(os.path.dirname(path)): |
|
1135 | 1135 | if not os.path.exists(os.path.join(path, 'db', 'fs-type')): |
|
1136 | 1136 | ui.status(_('initializing svn repository %r\n') % |
|
1137 | 1137 | os.path.basename(path)) |
|
1138 | 1138 | commandline(ui, 'svnadmin').run0('create', path) |
|
1139 | 1139 | created = path |
|
1140 | 1140 | path = util.normpath(path) |
|
1141 | 1141 | if not path.startswith('/'): |
|
1142 | 1142 | path = '/' + path |
|
1143 | 1143 | path = 'file://' + path |
|
1144 | 1144 | |
|
1145 | 1145 | wcpath = os.path.join(pycompat.getcwd(), os.path.basename(path) + |
|
1146 | 1146 | '-wc') |
|
1147 | 1147 | ui.status(_('initializing svn working copy %r\n') |
|
1148 | 1148 | % os.path.basename(wcpath)) |
|
1149 | 1149 | self.run0('checkout', path, wcpath) |
|
1150 | 1150 | |
|
1151 | 1151 | self.wc = wcpath |
|
1152 | 1152 | self.opener = vfsmod.vfs(self.wc) |
|
1153 | 1153 | self.wopener = vfsmod.vfs(self.wc) |
|
1154 | 1154 | self.childmap = mapfile(ui, self.join('hg-childmap')) |
|
1155 | 1155 | if util.checkexec(self.wc): |
|
1156 | 1156 | self.is_exec = util.isexec |
|
1157 | 1157 | else: |
|
1158 | 1158 | self.is_exec = None |
|
1159 | 1159 | |
|
1160 | 1160 | if created: |
|
1161 | 1161 | hook = os.path.join(created, 'hooks', 'pre-revprop-change') |
|
1162 | 1162 | fp = open(hook, 'wb') |
|
1163 | 1163 | fp.write(pre_revprop_change) |
|
1164 | 1164 | fp.close() |
|
1165 | 1165 | util.setflags(hook, False, True) |
|
1166 | 1166 | |
|
1167 | 1167 | output = self.run0('info') |
|
1168 | 1168 | self.uuid = self.uuid_re.search(output).group(1).strip() |
|
1169 | 1169 | |
|
1170 | 1170 | def wjoin(self, *names): |
|
1171 | 1171 | return os.path.join(self.wc, *names) |
|
1172 | 1172 | |
|
1173 | 1173 | @propertycache |
|
1174 | 1174 | def manifest(self): |
|
1175 | 1175 | # As of svn 1.7, the "add" command fails when receiving |
|
1176 | 1176 | # already tracked entries, so we have to track and filter them |
|
1177 | 1177 | # ourselves. |
|
1178 | 1178 | m = set() |
|
1179 | 1179 | output = self.run0('ls', recursive=True, xml=True) |
|
1180 | 1180 | doc = xml.dom.minidom.parseString(output) |
|
1181 | 1181 | for e in doc.getElementsByTagName('entry'): |
|
1182 | 1182 | for n in e.childNodes: |
|
1183 | 1183 | if n.nodeType != n.ELEMENT_NODE or n.tagName != 'name': |
|
1184 | 1184 | continue |
|
1185 | 1185 | name = ''.join(c.data for c in n.childNodes |
|
1186 | 1186 | if c.nodeType == c.TEXT_NODE) |
|
1187 | 1187 | # Entries are compared with names coming from |
|
1188 | 1188 | # mercurial, so bytes with undefined encoding. Our |
|
1189 | 1189 | # best bet is to assume they are in local |
|
1190 | 1190 | # encoding. They will be passed to command line calls |
|
1191 | 1191 | # later anyway, so they better be. |
|
1192 | 1192 | m.add(encoding.unitolocal(name)) |
|
1193 | 1193 | break |
|
1194 | 1194 | return m |
|
1195 | 1195 | |
|
1196 | 1196 | def putfile(self, filename, flags, data): |
|
1197 | 1197 | if 'l' in flags: |
|
1198 | 1198 | self.wopener.symlink(data, filename) |
|
1199 | 1199 | else: |
|
1200 | 1200 | try: |
|
1201 | 1201 | if os.path.islink(self.wjoin(filename)): |
|
1202 | 1202 | os.unlink(filename) |
|
1203 | 1203 | except OSError: |
|
1204 | 1204 | pass |
|
1205 | 1205 | self.wopener.write(filename, data) |
|
1206 | 1206 | |
|
1207 | 1207 | if self.is_exec: |
|
1208 | 1208 | if self.is_exec(self.wjoin(filename)): |
|
1209 | 1209 | if 'x' not in flags: |
|
1210 | 1210 | self.delexec.append(filename) |
|
1211 | 1211 | else: |
|
1212 | 1212 | if 'x' in flags: |
|
1213 | 1213 | self.setexec.append(filename) |
|
1214 | 1214 | util.setflags(self.wjoin(filename), False, 'x' in flags) |
|
1215 | 1215 | |
|
1216 | 1216 | def _copyfile(self, source, dest): |
|
1217 | 1217 | # SVN's copy command pukes if the destination file exists, but |
|
1218 | 1218 | # our copyfile method expects to record a copy that has |
|
1219 | 1219 | # already occurred. Cross the semantic gap. |
|
1220 | 1220 | wdest = self.wjoin(dest) |
|
1221 | 1221 | exists = os.path.lexists(wdest) |
|
1222 | 1222 | if exists: |
|
1223 | 1223 | fd, tempname = tempfile.mkstemp( |
|
1224 | 1224 | prefix='hg-copy-', dir=os.path.dirname(wdest)) |
|
1225 | 1225 | os.close(fd) |
|
1226 | 1226 | os.unlink(tempname) |
|
1227 | 1227 | os.rename(wdest, tempname) |
|
1228 | 1228 | try: |
|
1229 | 1229 | self.run0('copy', source, dest) |
|
1230 | 1230 | finally: |
|
1231 | 1231 | self.manifest.add(dest) |
|
1232 | 1232 | if exists: |
|
1233 | 1233 | try: |
|
1234 | 1234 | os.unlink(wdest) |
|
1235 | 1235 | except OSError: |
|
1236 | 1236 | pass |
|
1237 | 1237 | os.rename(tempname, wdest) |
|
1238 | 1238 | |
|
1239 | 1239 | def dirs_of(self, files): |
|
1240 | 1240 | dirs = set() |
|
1241 | 1241 | for f in files: |
|
1242 | 1242 | if os.path.isdir(self.wjoin(f)): |
|
1243 | 1243 | dirs.add(f) |
|
1244 | 1244 | i = len(f) |
|
1245 | 1245 | for i in iter(lambda: f.rfind('/', 0, i), -1): |
|
1246 | 1246 | dirs.add(f[:i]) |
|
1247 | 1247 | return dirs |
|
1248 | 1248 | |
|
1249 | 1249 | def add_dirs(self, files): |
|
1250 | 1250 | add_dirs = [d for d in sorted(self.dirs_of(files)) |
|
1251 | 1251 | if d not in self.manifest] |
|
1252 | 1252 | if add_dirs: |
|
1253 | 1253 | self.manifest.update(add_dirs) |
|
1254 | 1254 | self.xargs(add_dirs, 'add', non_recursive=True, quiet=True) |
|
1255 | 1255 | return add_dirs |
|
1256 | 1256 | |
|
1257 | 1257 | def add_files(self, files): |
|
1258 | 1258 | files = [f for f in files if f not in self.manifest] |
|
1259 | 1259 | if files: |
|
1260 | 1260 | self.manifest.update(files) |
|
1261 | 1261 | self.xargs(files, 'add', quiet=True) |
|
1262 | 1262 | return files |
|
1263 | 1263 | |
|
1264 | 1264 | def addchild(self, parent, child): |
|
1265 | 1265 | self.childmap[parent] = child |
|
1266 | 1266 | |
|
1267 | 1267 | def revid(self, rev): |
|
1268 | 1268 | return u"svn:%s@%s" % (self.uuid, rev) |
|
1269 | 1269 | |
|
1270 | 1270 | def putcommit(self, files, copies, parents, commit, source, revmap, full, |
|
1271 | 1271 | cleanp2): |
|
1272 | 1272 | for parent in parents: |
|
1273 | 1273 | try: |
|
1274 | 1274 | return self.revid(self.childmap[parent]) |
|
1275 | 1275 | except KeyError: |
|
1276 | 1276 | pass |
|
1277 | 1277 | |
|
1278 | 1278 | # Apply changes to working copy |
|
1279 | 1279 | for f, v in files: |
|
1280 | 1280 | data, mode = source.getfile(f, v) |
|
1281 | 1281 | if data is None: |
|
1282 | 1282 | self.delete.append(f) |
|
1283 | 1283 | else: |
|
1284 | 1284 | self.putfile(f, mode, data) |
|
1285 | 1285 | if f in copies: |
|
1286 | 1286 | self.copies.append([copies[f], f]) |
|
1287 | 1287 | if full: |
|
1288 | 1288 | self.delete.extend(sorted(self.manifest.difference(files))) |
|
1289 | 1289 | files = [f[0] for f in files] |
|
1290 | 1290 | |
|
1291 | 1291 | entries = set(self.delete) |
|
1292 | 1292 | files = frozenset(files) |
|
1293 | 1293 | entries.update(self.add_dirs(files.difference(entries))) |
|
1294 | 1294 | if self.copies: |
|
1295 | 1295 | for s, d in self.copies: |
|
1296 | 1296 | self._copyfile(s, d) |
|
1297 | 1297 | self.copies = [] |
|
1298 | 1298 | if self.delete: |
|
1299 | 1299 | self.xargs(self.delete, 'delete') |
|
1300 | 1300 | for f in self.delete: |
|
1301 | 1301 | self.manifest.remove(f) |
|
1302 | 1302 | self.delete = [] |
|
1303 | 1303 | entries.update(self.add_files(files.difference(entries))) |
|
1304 | 1304 | if self.delexec: |
|
1305 | 1305 | self.xargs(self.delexec, 'propdel', 'svn:executable') |
|
1306 | 1306 | self.delexec = [] |
|
1307 | 1307 | if self.setexec: |
|
1308 | 1308 | self.xargs(self.setexec, 'propset', 'svn:executable', '*') |
|
1309 | 1309 | self.setexec = [] |
|
1310 | 1310 | |
|
1311 | 1311 | fd, messagefile = tempfile.mkstemp(prefix='hg-convert-') |
|
1312 | 1312 | fp = os.fdopen(fd, pycompat.sysstr('wb')) |
|
1313 | 1313 | fp.write(util.tonativeeol(commit.desc)) |
|
1314 | 1314 | fp.close() |
|
1315 | 1315 | try: |
|
1316 | 1316 | output = self.run0('commit', |
|
1317 | 1317 | username=util.shortuser(commit.author), |
|
1318 | 1318 | file=messagefile, |
|
1319 | 1319 | encoding='utf-8') |
|
1320 | 1320 | try: |
|
1321 | 1321 | rev = self.commit_re.search(output).group(1) |
|
1322 | 1322 | except AttributeError: |
|
1323 | 1323 | if parents and not files: |
|
1324 | 1324 | return parents[0] |
|
1325 | 1325 | self.ui.warn(_('unexpected svn output:\n')) |
|
1326 | 1326 | self.ui.warn(output) |
|
1327 | 1327 | raise error.Abort(_('unable to cope with svn output')) |
|
1328 | 1328 | if commit.rev: |
|
1329 | 1329 | self.run('propset', 'hg:convert-rev', commit.rev, |
|
1330 | 1330 | revprop=True, revision=rev) |
|
1331 | 1331 | if commit.branch and commit.branch != 'default': |
|
1332 | 1332 | self.run('propset', 'hg:convert-branch', commit.branch, |
|
1333 | 1333 | revprop=True, revision=rev) |
|
1334 | 1334 | for parent in parents: |
|
1335 | 1335 | self.addchild(parent, rev) |
|
1336 | 1336 | return self.revid(rev) |
|
1337 | 1337 | finally: |
|
1338 | 1338 | os.unlink(messagefile) |
|
1339 | 1339 | |
|
1340 | 1340 | def puttags(self, tags): |
|
1341 | 1341 | self.ui.warn(_('writing Subversion tags is not yet implemented\n')) |
|
1342 | 1342 | return None, None |
|
1343 | 1343 | |
|
1344 | 1344 | def hascommitfrommap(self, rev): |
|
1345 | 1345 | # We trust that revisions referenced in a map still is present |
|
1346 | 1346 | # TODO: implement something better if necessary and feasible |
|
1347 | 1347 | return True |
|
1348 | 1348 | |
|
1349 | 1349 | def hascommitforsplicemap(self, rev): |
|
1350 | 1350 | # This is not correct as one can convert to an existing subversion |
|
1351 | 1351 | # repository and childmap would not list all revisions. Too bad. |
|
1352 | 1352 | if rev in self.childmap: |
|
1353 | 1353 | return True |
|
1354 | 1354 | raise error.Abort(_('splice map revision %s not found in subversion ' |
|
1355 | 1355 | 'child map (revision lookups are not implemented)') |
|
1356 | 1356 | % rev) |
@@ -1,481 +1,481 b'' | |||
|
1 | 1 | # blobstore.py - local and remote (speaking Git-LFS protocol) blob storages |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2017 Facebook, Inc. |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | import hashlib |
|
11 | 11 | import json |
|
12 | 12 | import os |
|
13 | 13 | import re |
|
14 | 14 | import socket |
|
15 | 15 | |
|
16 | 16 | from mercurial.i18n import _ |
|
17 | 17 | |
|
18 | 18 | from mercurial import ( |
|
19 | 19 | error, |
|
20 | 20 | pathutil, |
|
21 | 21 | url as urlmod, |
|
22 | 22 | util, |
|
23 | 23 | vfs as vfsmod, |
|
24 | 24 | worker, |
|
25 | 25 | ) |
|
26 | 26 | |
|
27 | 27 | from ..largefiles import lfutil |
|
28 | 28 | |
|
29 | 29 | # 64 bytes for SHA256 |
|
30 | _lfsre = re.compile(r'\A[a-f0-9]{64}\Z') | |
|
30 | _lfsre = re.compile(br'\A[a-f0-9]{64}\Z') | |
|
31 | 31 | |
|
32 | 32 | class lfsvfs(vfsmod.vfs): |
|
33 | 33 | def join(self, path): |
|
34 | 34 | """split the path at first two characters, like: XX/XXXXX...""" |
|
35 | 35 | if not _lfsre.match(path): |
|
36 | 36 | raise error.ProgrammingError('unexpected lfs path: %s' % path) |
|
37 | 37 | return super(lfsvfs, self).join(path[0:2], path[2:]) |
|
38 | 38 | |
|
39 | 39 | def walk(self, path=None, onerror=None): |
|
40 | 40 | """Yield (dirpath, [], oids) tuple for blobs under path |
|
41 | 41 | |
|
42 | 42 | Oids only exist in the root of this vfs, so dirpath is always ''. |
|
43 | 43 | """ |
|
44 | 44 | root = os.path.normpath(self.base) |
|
45 | 45 | # when dirpath == root, dirpath[prefixlen:] becomes empty |
|
46 | 46 | # because len(dirpath) < prefixlen. |
|
47 | 47 | prefixlen = len(pathutil.normasprefix(root)) |
|
48 | 48 | oids = [] |
|
49 | 49 | |
|
50 | 50 | for dirpath, dirs, files in os.walk(self.reljoin(self.base, path or ''), |
|
51 | 51 | onerror=onerror): |
|
52 | 52 | dirpath = dirpath[prefixlen:] |
|
53 | 53 | |
|
54 | 54 | # Silently skip unexpected files and directories |
|
55 | 55 | if len(dirpath) == 2: |
|
56 | 56 | oids.extend([dirpath + f for f in files |
|
57 | 57 | if _lfsre.match(dirpath + f)]) |
|
58 | 58 | |
|
59 | 59 | yield ('', [], oids) |
|
60 | 60 | |
|
61 | 61 | class filewithprogress(object): |
|
62 | 62 | """a file-like object that supports __len__ and read. |
|
63 | 63 | |
|
64 | 64 | Useful to provide progress information for how many bytes are read. |
|
65 | 65 | """ |
|
66 | 66 | |
|
67 | 67 | def __init__(self, fp, callback): |
|
68 | 68 | self._fp = fp |
|
69 | 69 | self._callback = callback # func(readsize) |
|
70 | 70 | fp.seek(0, os.SEEK_END) |
|
71 | 71 | self._len = fp.tell() |
|
72 | 72 | fp.seek(0) |
|
73 | 73 | |
|
74 | 74 | def __len__(self): |
|
75 | 75 | return self._len |
|
76 | 76 | |
|
77 | 77 | def read(self, size): |
|
78 | 78 | if self._fp is None: |
|
79 | 79 | return b'' |
|
80 | 80 | data = self._fp.read(size) |
|
81 | 81 | if data: |
|
82 | 82 | if self._callback: |
|
83 | 83 | self._callback(len(data)) |
|
84 | 84 | else: |
|
85 | 85 | self._fp.close() |
|
86 | 86 | self._fp = None |
|
87 | 87 | return data |
|
88 | 88 | |
|
89 | 89 | class local(object): |
|
90 | 90 | """Local blobstore for large file contents. |
|
91 | 91 | |
|
92 | 92 | This blobstore is used both as a cache and as a staging area for large blobs |
|
93 | 93 | to be uploaded to the remote blobstore. |
|
94 | 94 | """ |
|
95 | 95 | |
|
96 | 96 | def __init__(self, repo): |
|
97 | 97 | fullpath = repo.svfs.join('lfs/objects') |
|
98 | 98 | self.vfs = lfsvfs(fullpath) |
|
99 | 99 | usercache = lfutil._usercachedir(repo.ui, 'lfs') |
|
100 | 100 | self.cachevfs = lfsvfs(usercache) |
|
101 | 101 | self.ui = repo.ui |
|
102 | 102 | |
|
103 | 103 | def open(self, oid): |
|
104 | 104 | """Open a read-only file descriptor to the named blob, in either the |
|
105 | 105 | usercache or the local store.""" |
|
106 | 106 | # The usercache is the most likely place to hold the file. Commit will |
|
107 | 107 | # write to both it and the local store, as will anything that downloads |
|
108 | 108 | # the blobs. However, things like clone without an update won't |
|
109 | 109 | # populate the local store. For an init + push of a local clone, |
|
110 | 110 | # the usercache is the only place it _could_ be. If not present, the |
|
111 | 111 | # missing file msg here will indicate the local repo, not the usercache. |
|
112 | 112 | if self.cachevfs.exists(oid): |
|
113 | 113 | return self.cachevfs(oid, 'rb') |
|
114 | 114 | |
|
115 | 115 | return self.vfs(oid, 'rb') |
|
116 | 116 | |
|
117 | 117 | def download(self, oid, src): |
|
118 | 118 | """Read the blob from the remote source in chunks, verify the content, |
|
119 | 119 | and write to this local blobstore.""" |
|
120 | 120 | sha256 = hashlib.sha256() |
|
121 | 121 | |
|
122 | 122 | with self.vfs(oid, 'wb', atomictemp=True) as fp: |
|
123 | 123 | for chunk in util.filechunkiter(src, size=1048576): |
|
124 | 124 | fp.write(chunk) |
|
125 | 125 | sha256.update(chunk) |
|
126 | 126 | |
|
127 | 127 | realoid = sha256.hexdigest() |
|
128 | 128 | if realoid != oid: |
|
129 | 129 | raise error.Abort(_('corrupt remote lfs object: %s') % oid) |
|
130 | 130 | |
|
131 | 131 | # XXX: should we verify the content of the cache, and hardlink back to |
|
132 | 132 | # the local store on success, but truncate, write and link on failure? |
|
133 | 133 | if not self.cachevfs.exists(oid): |
|
134 | 134 | self.ui.note(_('lfs: adding %s to the usercache\n') % oid) |
|
135 | 135 | lfutil.link(self.vfs.join(oid), self.cachevfs.join(oid)) |
|
136 | 136 | |
|
137 | 137 | def write(self, oid, data): |
|
138 | 138 | """Write blob to local blobstore. |
|
139 | 139 | |
|
140 | 140 | This should only be called from the filelog during a commit or similar. |
|
141 | 141 | As such, there is no need to verify the data. Imports from a remote |
|
142 | 142 | store must use ``download()`` instead.""" |
|
143 | 143 | with self.vfs(oid, 'wb', atomictemp=True) as fp: |
|
144 | 144 | fp.write(data) |
|
145 | 145 | |
|
146 | 146 | # XXX: should we verify the content of the cache, and hardlink back to |
|
147 | 147 | # the local store on success, but truncate, write and link on failure? |
|
148 | 148 | if not self.cachevfs.exists(oid): |
|
149 | 149 | self.ui.note(_('lfs: adding %s to the usercache\n') % oid) |
|
150 | 150 | lfutil.link(self.vfs.join(oid), self.cachevfs.join(oid)) |
|
151 | 151 | |
|
152 | 152 | def read(self, oid, verify=True): |
|
153 | 153 | """Read blob from local blobstore.""" |
|
154 | 154 | if not self.vfs.exists(oid): |
|
155 | 155 | blob = self._read(self.cachevfs, oid, verify) |
|
156 | 156 | |
|
157 | 157 | # Even if revlog will verify the content, it needs to be verified |
|
158 | 158 | # now before making the hardlink to avoid propagating corrupt blobs. |
|
159 | 159 | # Don't abort if corruption is detected, because `hg verify` will |
|
160 | 160 | # give more useful info about the corruption- simply don't add the |
|
161 | 161 | # hardlink. |
|
162 | 162 | if verify or hashlib.sha256(blob).hexdigest() == oid: |
|
163 | 163 | self.ui.note(_('lfs: found %s in the usercache\n') % oid) |
|
164 | 164 | lfutil.link(self.cachevfs.join(oid), self.vfs.join(oid)) |
|
165 | 165 | else: |
|
166 | 166 | self.ui.note(_('lfs: found %s in the local lfs store\n') % oid) |
|
167 | 167 | blob = self._read(self.vfs, oid, verify) |
|
168 | 168 | return blob |
|
169 | 169 | |
|
170 | 170 | def _read(self, vfs, oid, verify): |
|
171 | 171 | """Read blob (after verifying) from the given store""" |
|
172 | 172 | blob = vfs.read(oid) |
|
173 | 173 | if verify: |
|
174 | 174 | _verify(oid, blob) |
|
175 | 175 | return blob |
|
176 | 176 | |
|
177 | 177 | def has(self, oid): |
|
178 | 178 | """Returns True if the local blobstore contains the requested blob, |
|
179 | 179 | False otherwise.""" |
|
180 | 180 | return self.cachevfs.exists(oid) or self.vfs.exists(oid) |
|
181 | 181 | |
|
182 | 182 | class _gitlfsremote(object): |
|
183 | 183 | |
|
184 | 184 | def __init__(self, repo, url): |
|
185 | 185 | ui = repo.ui |
|
186 | 186 | self.ui = ui |
|
187 | 187 | baseurl, authinfo = url.authinfo() |
|
188 | 188 | self.baseurl = baseurl.rstrip('/') |
|
189 | 189 | useragent = repo.ui.config('experimental', 'lfs.user-agent') |
|
190 | 190 | if not useragent: |
|
191 | 191 | useragent = 'git-lfs/2.3.4 (Mercurial %s)' % util.version() |
|
192 | 192 | self.urlopener = urlmod.opener(ui, authinfo, useragent) |
|
193 | 193 | self.retry = ui.configint('lfs', 'retry') |
|
194 | 194 | |
|
195 | 195 | def writebatch(self, pointers, fromstore): |
|
196 | 196 | """Batch upload from local to remote blobstore.""" |
|
197 | 197 | self._batch(_deduplicate(pointers), fromstore, 'upload') |
|
198 | 198 | |
|
199 | 199 | def readbatch(self, pointers, tostore): |
|
200 | 200 | """Batch download from remote to local blostore.""" |
|
201 | 201 | self._batch(_deduplicate(pointers), tostore, 'download') |
|
202 | 202 | |
|
203 | 203 | def _batchrequest(self, pointers, action): |
|
204 | 204 | """Get metadata about objects pointed by pointers for given action |
|
205 | 205 | |
|
206 | 206 | Return decoded JSON object like {'objects': [{'oid': '', 'size': 1}]} |
|
207 | 207 | See https://github.com/git-lfs/git-lfs/blob/master/docs/api/batch.md |
|
208 | 208 | """ |
|
209 | 209 | objects = [{'oid': p.oid(), 'size': p.size()} for p in pointers] |
|
210 | 210 | requestdata = json.dumps({ |
|
211 | 211 | 'objects': objects, |
|
212 | 212 | 'operation': action, |
|
213 | 213 | }) |
|
214 | 214 | batchreq = util.urlreq.request('%s/objects/batch' % self.baseurl, |
|
215 | 215 | data=requestdata) |
|
216 | 216 | batchreq.add_header('Accept', 'application/vnd.git-lfs+json') |
|
217 | 217 | batchreq.add_header('Content-Type', 'application/vnd.git-lfs+json') |
|
218 | 218 | try: |
|
219 | 219 | rawjson = self.urlopener.open(batchreq).read() |
|
220 | 220 | except util.urlerr.httperror as ex: |
|
221 | 221 | raise LfsRemoteError(_('LFS HTTP error: %s (action=%s)') |
|
222 | 222 | % (ex, action)) |
|
223 | 223 | try: |
|
224 | 224 | response = json.loads(rawjson) |
|
225 | 225 | except ValueError: |
|
226 | 226 | raise LfsRemoteError(_('LFS server returns invalid JSON: %s') |
|
227 | 227 | % rawjson) |
|
228 | 228 | return response |
|
229 | 229 | |
|
230 | 230 | def _checkforservererror(self, pointers, responses, action): |
|
231 | 231 | """Scans errors from objects |
|
232 | 232 | |
|
233 | 233 | Raises LfsRemoteError if any objects have an error""" |
|
234 | 234 | for response in responses: |
|
235 | 235 | # The server should return 404 when objects cannot be found. Some |
|
236 | 236 | # server implementation (ex. lfs-test-server) does not set "error" |
|
237 | 237 | # but just removes "download" from "actions". Treat that case |
|
238 | 238 | # as the same as 404 error. |
|
239 | 239 | notfound = (response.get('error', {}).get('code') == 404 |
|
240 | 240 | or (action == 'download' |
|
241 | 241 | and action not in response.get('actions', []))) |
|
242 | 242 | if notfound: |
|
243 | 243 | ptrmap = {p.oid(): p for p in pointers} |
|
244 | 244 | p = ptrmap.get(response['oid'], None) |
|
245 | 245 | if p: |
|
246 | 246 | filename = getattr(p, 'filename', 'unknown') |
|
247 | 247 | raise LfsRemoteError( |
|
248 | 248 | _(('LFS server error. Remote object ' |
|
249 | 249 | 'for "%s" not found: %r')) % (filename, response)) |
|
250 | 250 | else: |
|
251 | 251 | raise LfsRemoteError( |
|
252 | 252 | _('LFS server error. Unsolicited response for oid %s') |
|
253 | 253 | % response['oid']) |
|
254 | 254 | if 'error' in response: |
|
255 | 255 | raise LfsRemoteError(_('LFS server error: %r') % response) |
|
256 | 256 | |
|
257 | 257 | def _extractobjects(self, response, pointers, action): |
|
258 | 258 | """extract objects from response of the batch API |
|
259 | 259 | |
|
260 | 260 | response: parsed JSON object returned by batch API |
|
261 | 261 | return response['objects'] filtered by action |
|
262 | 262 | raise if any object has an error |
|
263 | 263 | """ |
|
264 | 264 | # Scan errors from objects - fail early |
|
265 | 265 | objects = response.get('objects', []) |
|
266 | 266 | self._checkforservererror(pointers, objects, action) |
|
267 | 267 | |
|
268 | 268 | # Filter objects with given action. Practically, this skips uploading |
|
269 | 269 | # objects which exist in the server. |
|
270 | 270 | filteredobjects = [o for o in objects if action in o.get('actions', [])] |
|
271 | 271 | |
|
272 | 272 | return filteredobjects |
|
273 | 273 | |
|
274 | 274 | def _basictransfer(self, obj, action, localstore): |
|
275 | 275 | """Download or upload a single object using basic transfer protocol |
|
276 | 276 | |
|
277 | 277 | obj: dict, an object description returned by batch API |
|
278 | 278 | action: string, one of ['upload', 'download'] |
|
279 | 279 | localstore: blobstore.local |
|
280 | 280 | |
|
281 | 281 | See https://github.com/git-lfs/git-lfs/blob/master/docs/api/\ |
|
282 | 282 | basic-transfers.md |
|
283 | 283 | """ |
|
284 | 284 | oid = str(obj['oid']) |
|
285 | 285 | |
|
286 | 286 | href = str(obj['actions'][action].get('href')) |
|
287 | 287 | headers = obj['actions'][action].get('header', {}).items() |
|
288 | 288 | |
|
289 | 289 | request = util.urlreq.request(href) |
|
290 | 290 | if action == 'upload': |
|
291 | 291 | # If uploading blobs, read data from local blobstore. |
|
292 | 292 | with localstore.open(oid) as fp: |
|
293 | 293 | _verifyfile(oid, fp) |
|
294 | 294 | request.data = filewithprogress(localstore.open(oid), None) |
|
295 | 295 | request.get_method = lambda: 'PUT' |
|
296 | 296 | |
|
297 | 297 | for k, v in headers: |
|
298 | 298 | request.add_header(k, v) |
|
299 | 299 | |
|
300 | 300 | response = b'' |
|
301 | 301 | try: |
|
302 | 302 | req = self.urlopener.open(request) |
|
303 | 303 | if action == 'download': |
|
304 | 304 | # If downloading blobs, store downloaded data to local blobstore |
|
305 | 305 | localstore.download(oid, req) |
|
306 | 306 | else: |
|
307 | 307 | while True: |
|
308 | 308 | data = req.read(1048576) |
|
309 | 309 | if not data: |
|
310 | 310 | break |
|
311 | 311 | response += data |
|
312 | 312 | if response: |
|
313 | 313 | self.ui.debug('lfs %s response: %s' % (action, response)) |
|
314 | 314 | except util.urlerr.httperror as ex: |
|
315 | 315 | if self.ui.debugflag: |
|
316 | 316 | self.ui.debug('%s: %s\n' % (oid, ex.read())) |
|
317 | 317 | raise LfsRemoteError(_('HTTP error: %s (oid=%s, action=%s)') |
|
318 | 318 | % (ex, oid, action)) |
|
319 | 319 | |
|
320 | 320 | def _batch(self, pointers, localstore, action): |
|
321 | 321 | if action not in ['upload', 'download']: |
|
322 | 322 | raise error.ProgrammingError('invalid Git-LFS action: %s' % action) |
|
323 | 323 | |
|
324 | 324 | response = self._batchrequest(pointers, action) |
|
325 | 325 | objects = self._extractobjects(response, pointers, action) |
|
326 | 326 | total = sum(x.get('size', 0) for x in objects) |
|
327 | 327 | sizes = {} |
|
328 | 328 | for obj in objects: |
|
329 | 329 | sizes[obj.get('oid')] = obj.get('size', 0) |
|
330 | 330 | topic = {'upload': _('lfs uploading'), |
|
331 | 331 | 'download': _('lfs downloading')}[action] |
|
332 | 332 | if len(objects) > 1: |
|
333 | 333 | self.ui.note(_('lfs: need to transfer %d objects (%s)\n') |
|
334 | 334 | % (len(objects), util.bytecount(total))) |
|
335 | 335 | self.ui.progress(topic, 0, total=total) |
|
336 | 336 | def transfer(chunk): |
|
337 | 337 | for obj in chunk: |
|
338 | 338 | objsize = obj.get('size', 0) |
|
339 | 339 | if self.ui.verbose: |
|
340 | 340 | if action == 'download': |
|
341 | 341 | msg = _('lfs: downloading %s (%s)\n') |
|
342 | 342 | elif action == 'upload': |
|
343 | 343 | msg = _('lfs: uploading %s (%s)\n') |
|
344 | 344 | self.ui.note(msg % (obj.get('oid'), |
|
345 | 345 | util.bytecount(objsize))) |
|
346 | 346 | retry = self.retry |
|
347 | 347 | while True: |
|
348 | 348 | try: |
|
349 | 349 | self._basictransfer(obj, action, localstore) |
|
350 | 350 | yield 1, obj.get('oid') |
|
351 | 351 | break |
|
352 | 352 | except socket.error as ex: |
|
353 | 353 | if retry > 0: |
|
354 | 354 | self.ui.note( |
|
355 | 355 | _('lfs: failed: %r (remaining retry %d)\n') |
|
356 | 356 | % (ex, retry)) |
|
357 | 357 | retry -= 1 |
|
358 | 358 | continue |
|
359 | 359 | raise |
|
360 | 360 | |
|
361 | 361 | # Until https multiplexing gets sorted out |
|
362 | 362 | if self.ui.configbool('experimental', 'lfs.worker-enable'): |
|
363 | 363 | oids = worker.worker(self.ui, 0.1, transfer, (), |
|
364 | 364 | sorted(objects, key=lambda o: o.get('oid'))) |
|
365 | 365 | else: |
|
366 | 366 | oids = transfer(sorted(objects, key=lambda o: o.get('oid'))) |
|
367 | 367 | |
|
368 | 368 | processed = 0 |
|
369 | 369 | blobs = 0 |
|
370 | 370 | for _one, oid in oids: |
|
371 | 371 | processed += sizes[oid] |
|
372 | 372 | blobs += 1 |
|
373 | 373 | self.ui.progress(topic, processed, total=total) |
|
374 | 374 | self.ui.note(_('lfs: processed: %s\n') % oid) |
|
375 | 375 | self.ui.progress(topic, pos=None, total=total) |
|
376 | 376 | |
|
377 | 377 | if blobs > 0: |
|
378 | 378 | if action == 'upload': |
|
379 | 379 | self.ui.status(_('lfs: uploaded %d files (%s)\n') |
|
380 | 380 | % (blobs, util.bytecount(processed))) |
|
381 | 381 | # TODO: coalesce the download requests, and comment this in |
|
382 | 382 | #elif action == 'download': |
|
383 | 383 | # self.ui.status(_('lfs: downloaded %d files (%s)\n') |
|
384 | 384 | # % (blobs, util.bytecount(processed))) |
|
385 | 385 | |
|
386 | 386 | def __del__(self): |
|
387 | 387 | # copied from mercurial/httppeer.py |
|
388 | 388 | urlopener = getattr(self, 'urlopener', None) |
|
389 | 389 | if urlopener: |
|
390 | 390 | for h in urlopener.handlers: |
|
391 | 391 | h.close() |
|
392 | 392 | getattr(h, "close_all", lambda : None)() |
|
393 | 393 | |
|
394 | 394 | class _dummyremote(object): |
|
395 | 395 | """Dummy store storing blobs to temp directory.""" |
|
396 | 396 | |
|
397 | 397 | def __init__(self, repo, url): |
|
398 | 398 | fullpath = repo.vfs.join('lfs', url.path) |
|
399 | 399 | self.vfs = lfsvfs(fullpath) |
|
400 | 400 | |
|
401 | 401 | def writebatch(self, pointers, fromstore): |
|
402 | 402 | for p in _deduplicate(pointers): |
|
403 | 403 | content = fromstore.read(p.oid(), verify=True) |
|
404 | 404 | with self.vfs(p.oid(), 'wb', atomictemp=True) as fp: |
|
405 | 405 | fp.write(content) |
|
406 | 406 | |
|
407 | 407 | def readbatch(self, pointers, tostore): |
|
408 | 408 | for p in _deduplicate(pointers): |
|
409 | 409 | with self.vfs(p.oid(), 'rb') as fp: |
|
410 | 410 | tostore.download(p.oid(), fp) |
|
411 | 411 | |
|
412 | 412 | class _nullremote(object): |
|
413 | 413 | """Null store storing blobs to /dev/null.""" |
|
414 | 414 | |
|
415 | 415 | def __init__(self, repo, url): |
|
416 | 416 | pass |
|
417 | 417 | |
|
418 | 418 | def writebatch(self, pointers, fromstore): |
|
419 | 419 | pass |
|
420 | 420 | |
|
421 | 421 | def readbatch(self, pointers, tostore): |
|
422 | 422 | pass |
|
423 | 423 | |
|
424 | 424 | class _promptremote(object): |
|
425 | 425 | """Prompt user to set lfs.url when accessed.""" |
|
426 | 426 | |
|
427 | 427 | def __init__(self, repo, url): |
|
428 | 428 | pass |
|
429 | 429 | |
|
430 | 430 | def writebatch(self, pointers, fromstore, ui=None): |
|
431 | 431 | self._prompt() |
|
432 | 432 | |
|
433 | 433 | def readbatch(self, pointers, tostore, ui=None): |
|
434 | 434 | self._prompt() |
|
435 | 435 | |
|
436 | 436 | def _prompt(self): |
|
437 | 437 | raise error.Abort(_('lfs.url needs to be configured')) |
|
438 | 438 | |
|
439 | 439 | _storemap = { |
|
440 | 440 | 'https': _gitlfsremote, |
|
441 | 441 | 'http': _gitlfsremote, |
|
442 | 442 | 'file': _dummyremote, |
|
443 | 443 | 'null': _nullremote, |
|
444 | 444 | None: _promptremote, |
|
445 | 445 | } |
|
446 | 446 | |
|
447 | 447 | def _deduplicate(pointers): |
|
448 | 448 | """Remove any duplicate oids that exist in the list""" |
|
449 | 449 | reduced = util.sortdict() |
|
450 | 450 | for p in pointers: |
|
451 | 451 | reduced[p.oid()] = p |
|
452 | 452 | return reduced.values() |
|
453 | 453 | |
|
454 | 454 | def _verify(oid, content): |
|
455 | 455 | realoid = hashlib.sha256(content).hexdigest() |
|
456 | 456 | if realoid != oid: |
|
457 | 457 | raise error.Abort(_('detected corrupt lfs object: %s') % oid, |
|
458 | 458 | hint=_('run hg verify')) |
|
459 | 459 | |
|
460 | 460 | def _verifyfile(oid, fp): |
|
461 | 461 | sha256 = hashlib.sha256() |
|
462 | 462 | while True: |
|
463 | 463 | data = fp.read(1024 * 1024) |
|
464 | 464 | if not data: |
|
465 | 465 | break |
|
466 | 466 | sha256.update(data) |
|
467 | 467 | realoid = sha256.hexdigest() |
|
468 | 468 | if realoid != oid: |
|
469 | 469 | raise error.Abort(_('detected corrupt lfs object: %s') % oid, |
|
470 | 470 | hint=_('run hg verify')) |
|
471 | 471 | |
|
472 | 472 | def remote(repo): |
|
473 | 473 | """remotestore factory. return a store in _storemap depending on config""" |
|
474 | 474 | url = util.url(repo.ui.config('lfs', 'url') or '') |
|
475 | 475 | scheme = url.scheme |
|
476 | 476 | if scheme not in _storemap: |
|
477 | 477 | raise error.Abort(_('lfs: unknown url scheme: %s') % scheme) |
|
478 | 478 | return _storemap[scheme](repo, url) |
|
479 | 479 | |
|
480 | 480 | class LfsRemoteError(error.RevlogError): |
|
481 | 481 | pass |
General Comments 0
You need to be logged in to leave comments.
Login now