##// END OF EJS Templates
bundle2: grab kwarg using sysstr...
Augie Fackler -
r39792:4a8bfec9 default
parent child Browse files
Show More
@@ -1,2268 +1,2268
1 1 # bundle2.py - generic container format to transmit arbitrary data.
2 2 #
3 3 # Copyright 2013 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 """Handling of the new bundle2 format
8 8
9 9 The goal of bundle2 is to act as an atomically packet to transmit a set of
10 10 payloads in an application agnostic way. It consist in a sequence of "parts"
11 11 that will be handed to and processed by the application layer.
12 12
13 13
14 14 General format architecture
15 15 ===========================
16 16
17 17 The format is architectured as follow
18 18
19 19 - magic string
20 20 - stream level parameters
21 21 - payload parts (any number)
22 22 - end of stream marker.
23 23
24 24 the Binary format
25 25 ============================
26 26
27 27 All numbers are unsigned and big-endian.
28 28
29 29 stream level parameters
30 30 ------------------------
31 31
32 32 Binary format is as follow
33 33
34 34 :params size: int32
35 35
36 36 The total number of Bytes used by the parameters
37 37
38 38 :params value: arbitrary number of Bytes
39 39
40 40 A blob of `params size` containing the serialized version of all stream level
41 41 parameters.
42 42
43 43 The blob contains a space separated list of parameters. Parameters with value
44 44 are stored in the form `<name>=<value>`. Both name and value are urlquoted.
45 45
46 46 Empty name are obviously forbidden.
47 47
48 48 Name MUST start with a letter. If this first letter is lower case, the
49 49 parameter is advisory and can be safely ignored. However when the first
50 50 letter is capital, the parameter is mandatory and the bundling process MUST
51 51 stop if he is not able to proceed it.
52 52
53 53 Stream parameters use a simple textual format for two main reasons:
54 54
55 55 - Stream level parameters should remain simple and we want to discourage any
56 56 crazy usage.
57 57 - Textual data allow easy human inspection of a bundle2 header in case of
58 58 troubles.
59 59
60 60 Any Applicative level options MUST go into a bundle2 part instead.
61 61
62 62 Payload part
63 63 ------------------------
64 64
65 65 Binary format is as follow
66 66
67 67 :header size: int32
68 68
69 69 The total number of Bytes used by the part header. When the header is empty
70 70 (size = 0) this is interpreted as the end of stream marker.
71 71
72 72 :header:
73 73
74 74 The header defines how to interpret the part. It contains two piece of
75 75 data: the part type, and the part parameters.
76 76
77 77 The part type is used to route an application level handler, that can
78 78 interpret payload.
79 79
80 80 Part parameters are passed to the application level handler. They are
81 81 meant to convey information that will help the application level object to
82 82 interpret the part payload.
83 83
84 84 The binary format of the header is has follow
85 85
86 86 :typesize: (one byte)
87 87
88 88 :parttype: alphanumerical part name (restricted to [a-zA-Z0-9_:-]*)
89 89
90 90 :partid: A 32bits integer (unique in the bundle) that can be used to refer
91 91 to this part.
92 92
93 93 :parameters:
94 94
95 95 Part's parameter may have arbitrary content, the binary structure is::
96 96
97 97 <mandatory-count><advisory-count><param-sizes><param-data>
98 98
99 99 :mandatory-count: 1 byte, number of mandatory parameters
100 100
101 101 :advisory-count: 1 byte, number of advisory parameters
102 102
103 103 :param-sizes:
104 104
105 105 N couple of bytes, where N is the total number of parameters. Each
106 106 couple contains (<size-of-key>, <size-of-value) for one parameter.
107 107
108 108 :param-data:
109 109
110 110 A blob of bytes from which each parameter key and value can be
111 111 retrieved using the list of size couples stored in the previous
112 112 field.
113 113
114 114 Mandatory parameters comes first, then the advisory ones.
115 115
116 116 Each parameter's key MUST be unique within the part.
117 117
118 118 :payload:
119 119
120 120 payload is a series of `<chunksize><chunkdata>`.
121 121
122 122 `chunksize` is an int32, `chunkdata` are plain bytes (as much as
123 123 `chunksize` says)` The payload part is concluded by a zero size chunk.
124 124
125 125 The current implementation always produces either zero or one chunk.
126 126 This is an implementation limitation that will ultimately be lifted.
127 127
128 128 `chunksize` can be negative to trigger special case processing. No such
129 129 processing is in place yet.
130 130
131 131 Bundle processing
132 132 ============================
133 133
134 134 Each part is processed in order using a "part handler". Handler are registered
135 135 for a certain part type.
136 136
137 137 The matching of a part to its handler is case insensitive. The case of the
138 138 part type is used to know if a part is mandatory or advisory. If the Part type
139 139 contains any uppercase char it is considered mandatory. When no handler is
140 140 known for a Mandatory part, the process is aborted and an exception is raised.
141 141 If the part is advisory and no handler is known, the part is ignored. When the
142 142 process is aborted, the full bundle is still read from the stream to keep the
143 143 channel usable. But none of the part read from an abort are processed. In the
144 144 future, dropping the stream may become an option for channel we do not care to
145 145 preserve.
146 146 """
147 147
148 148 from __future__ import absolute_import, division
149 149
150 150 import collections
151 151 import errno
152 152 import os
153 153 import re
154 154 import string
155 155 import struct
156 156 import sys
157 157
158 158 from .i18n import _
159 159 from . import (
160 160 bookmarks,
161 161 changegroup,
162 162 encoding,
163 163 error,
164 164 node as nodemod,
165 165 obsolete,
166 166 phases,
167 167 pushkey,
168 168 pycompat,
169 169 streamclone,
170 170 tags,
171 171 url,
172 172 util,
173 173 )
174 174 from .utils import (
175 175 stringutil,
176 176 )
177 177
178 178 urlerr = util.urlerr
179 179 urlreq = util.urlreq
180 180
181 181 _pack = struct.pack
182 182 _unpack = struct.unpack
183 183
184 184 _fstreamparamsize = '>i'
185 185 _fpartheadersize = '>i'
186 186 _fparttypesize = '>B'
187 187 _fpartid = '>I'
188 188 _fpayloadsize = '>i'
189 189 _fpartparamcount = '>BB'
190 190
191 191 preferedchunksize = 32768
192 192
193 193 _parttypeforbidden = re.compile('[^a-zA-Z0-9_:-]')
194 194
195 195 def outdebug(ui, message):
196 196 """debug regarding output stream (bundling)"""
197 197 if ui.configbool('devel', 'bundle2.debug'):
198 198 ui.debug('bundle2-output: %s\n' % message)
199 199
200 200 def indebug(ui, message):
201 201 """debug on input stream (unbundling)"""
202 202 if ui.configbool('devel', 'bundle2.debug'):
203 203 ui.debug('bundle2-input: %s\n' % message)
204 204
205 205 def validateparttype(parttype):
206 206 """raise ValueError if a parttype contains invalid character"""
207 207 if _parttypeforbidden.search(parttype):
208 208 raise ValueError(parttype)
209 209
210 210 def _makefpartparamsizes(nbparams):
211 211 """return a struct format to read part parameter sizes
212 212
213 213 The number parameters is variable so we need to build that format
214 214 dynamically.
215 215 """
216 216 return '>'+('BB'*nbparams)
217 217
218 218 parthandlermapping = {}
219 219
220 220 def parthandler(parttype, params=()):
221 221 """decorator that register a function as a bundle2 part handler
222 222
223 223 eg::
224 224
225 225 @parthandler('myparttype', ('mandatory', 'param', 'handled'))
226 226 def myparttypehandler(...):
227 227 '''process a part of type "my part".'''
228 228 ...
229 229 """
230 230 validateparttype(parttype)
231 231 def _decorator(func):
232 232 lparttype = parttype.lower() # enforce lower case matching.
233 233 assert lparttype not in parthandlermapping
234 234 parthandlermapping[lparttype] = func
235 235 func.params = frozenset(params)
236 236 return func
237 237 return _decorator
238 238
239 239 class unbundlerecords(object):
240 240 """keep record of what happens during and unbundle
241 241
242 242 New records are added using `records.add('cat', obj)`. Where 'cat' is a
243 243 category of record and obj is an arbitrary object.
244 244
245 245 `records['cat']` will return all entries of this category 'cat'.
246 246
247 247 Iterating on the object itself will yield `('category', obj)` tuples
248 248 for all entries.
249 249
250 250 All iterations happens in chronological order.
251 251 """
252 252
253 253 def __init__(self):
254 254 self._categories = {}
255 255 self._sequences = []
256 256 self._replies = {}
257 257
258 258 def add(self, category, entry, inreplyto=None):
259 259 """add a new record of a given category.
260 260
261 261 The entry can then be retrieved in the list returned by
262 262 self['category']."""
263 263 self._categories.setdefault(category, []).append(entry)
264 264 self._sequences.append((category, entry))
265 265 if inreplyto is not None:
266 266 self.getreplies(inreplyto).add(category, entry)
267 267
268 268 def getreplies(self, partid):
269 269 """get the records that are replies to a specific part"""
270 270 return self._replies.setdefault(partid, unbundlerecords())
271 271
272 272 def __getitem__(self, cat):
273 273 return tuple(self._categories.get(cat, ()))
274 274
275 275 def __iter__(self):
276 276 return iter(self._sequences)
277 277
278 278 def __len__(self):
279 279 return len(self._sequences)
280 280
281 281 def __nonzero__(self):
282 282 return bool(self._sequences)
283 283
284 284 __bool__ = __nonzero__
285 285
286 286 class bundleoperation(object):
287 287 """an object that represents a single bundling process
288 288
289 289 Its purpose is to carry unbundle-related objects and states.
290 290
291 291 A new object should be created at the beginning of each bundle processing.
292 292 The object is to be returned by the processing function.
293 293
294 294 The object has very little content now it will ultimately contain:
295 295 * an access to the repo the bundle is applied to,
296 296 * a ui object,
297 297 * a way to retrieve a transaction to add changes to the repo,
298 298 * a way to record the result of processing each part,
299 299 * a way to construct a bundle response when applicable.
300 300 """
301 301
302 302 def __init__(self, repo, transactiongetter, captureoutput=True, source=''):
303 303 self.repo = repo
304 304 self.ui = repo.ui
305 305 self.records = unbundlerecords()
306 306 self.reply = None
307 307 self.captureoutput = captureoutput
308 308 self.hookargs = {}
309 309 self._gettransaction = transactiongetter
310 310 # carries value that can modify part behavior
311 311 self.modes = {}
312 312 self.source = source
313 313
314 314 def gettransaction(self):
315 315 transaction = self._gettransaction()
316 316
317 317 if self.hookargs:
318 318 # the ones added to the transaction supercede those added
319 319 # to the operation.
320 320 self.hookargs.update(transaction.hookargs)
321 321 transaction.hookargs = self.hookargs
322 322
323 323 # mark the hookargs as flushed. further attempts to add to
324 324 # hookargs will result in an abort.
325 325 self.hookargs = None
326 326
327 327 return transaction
328 328
329 329 def addhookargs(self, hookargs):
330 330 if self.hookargs is None:
331 331 raise error.ProgrammingError('attempted to add hookargs to '
332 332 'operation after transaction started')
333 333 self.hookargs.update(hookargs)
334 334
335 335 class TransactionUnavailable(RuntimeError):
336 336 pass
337 337
338 338 def _notransaction():
339 339 """default method to get a transaction while processing a bundle
340 340
341 341 Raise an exception to highlight the fact that no transaction was expected
342 342 to be created"""
343 343 raise TransactionUnavailable()
344 344
345 345 def applybundle(repo, unbundler, tr, source, url=None, **kwargs):
346 346 # transform me into unbundler.apply() as soon as the freeze is lifted
347 347 if isinstance(unbundler, unbundle20):
348 348 tr.hookargs['bundle2'] = '1'
349 349 if source is not None and 'source' not in tr.hookargs:
350 350 tr.hookargs['source'] = source
351 351 if url is not None and 'url' not in tr.hookargs:
352 352 tr.hookargs['url'] = url
353 353 return processbundle(repo, unbundler, lambda: tr, source=source)
354 354 else:
355 355 # the transactiongetter won't be used, but we might as well set it
356 356 op = bundleoperation(repo, lambda: tr, source=source)
357 357 _processchangegroup(op, unbundler, tr, source, url, **kwargs)
358 358 return op
359 359
360 360 class partiterator(object):
361 361 def __init__(self, repo, op, unbundler):
362 362 self.repo = repo
363 363 self.op = op
364 364 self.unbundler = unbundler
365 365 self.iterator = None
366 366 self.count = 0
367 367 self.current = None
368 368
369 369 def __enter__(self):
370 370 def func():
371 371 itr = enumerate(self.unbundler.iterparts())
372 372 for count, p in itr:
373 373 self.count = count
374 374 self.current = p
375 375 yield p
376 376 p.consume()
377 377 self.current = None
378 378 self.iterator = func()
379 379 return self.iterator
380 380
381 381 def __exit__(self, type, exc, tb):
382 382 if not self.iterator:
383 383 return
384 384
385 385 # Only gracefully abort in a normal exception situation. User aborts
386 386 # like Ctrl+C throw a KeyboardInterrupt which is not a base Exception,
387 387 # and should not gracefully cleanup.
388 388 if isinstance(exc, Exception):
389 389 # Any exceptions seeking to the end of the bundle at this point are
390 390 # almost certainly related to the underlying stream being bad.
391 391 # And, chances are that the exception we're handling is related to
392 392 # getting in that bad state. So, we swallow the seeking error and
393 393 # re-raise the original error.
394 394 seekerror = False
395 395 try:
396 396 if self.current:
397 397 # consume the part content to not corrupt the stream.
398 398 self.current.consume()
399 399
400 400 for part in self.iterator:
401 401 # consume the bundle content
402 402 part.consume()
403 403 except Exception:
404 404 seekerror = True
405 405
406 406 # Small hack to let caller code distinguish exceptions from bundle2
407 407 # processing from processing the old format. This is mostly needed
408 408 # to handle different return codes to unbundle according to the type
409 409 # of bundle. We should probably clean up or drop this return code
410 410 # craziness in a future version.
411 411 exc.duringunbundle2 = True
412 412 salvaged = []
413 413 replycaps = None
414 414 if self.op.reply is not None:
415 415 salvaged = self.op.reply.salvageoutput()
416 416 replycaps = self.op.reply.capabilities
417 417 exc._replycaps = replycaps
418 418 exc._bundle2salvagedoutput = salvaged
419 419
420 420 # Re-raising from a variable loses the original stack. So only use
421 421 # that form if we need to.
422 422 if seekerror:
423 423 raise exc
424 424
425 425 self.repo.ui.debug('bundle2-input-bundle: %i parts total\n' %
426 426 self.count)
427 427
428 428 def processbundle(repo, unbundler, transactiongetter=None, op=None, source=''):
429 429 """This function process a bundle, apply effect to/from a repo
430 430
431 431 It iterates over each part then searches for and uses the proper handling
432 432 code to process the part. Parts are processed in order.
433 433
434 434 Unknown Mandatory part will abort the process.
435 435
436 436 It is temporarily possible to provide a prebuilt bundleoperation to the
437 437 function. This is used to ensure output is properly propagated in case of
438 438 an error during the unbundling. This output capturing part will likely be
439 439 reworked and this ability will probably go away in the process.
440 440 """
441 441 if op is None:
442 442 if transactiongetter is None:
443 443 transactiongetter = _notransaction
444 444 op = bundleoperation(repo, transactiongetter, source=source)
445 445 # todo:
446 446 # - replace this is a init function soon.
447 447 # - exception catching
448 448 unbundler.params
449 449 if repo.ui.debugflag:
450 450 msg = ['bundle2-input-bundle:']
451 451 if unbundler.params:
452 452 msg.append(' %i params' % len(unbundler.params))
453 453 if op._gettransaction is None or op._gettransaction is _notransaction:
454 454 msg.append(' no-transaction')
455 455 else:
456 456 msg.append(' with-transaction')
457 457 msg.append('\n')
458 458 repo.ui.debug(''.join(msg))
459 459
460 460 processparts(repo, op, unbundler)
461 461
462 462 return op
463 463
464 464 def processparts(repo, op, unbundler):
465 465 with partiterator(repo, op, unbundler) as parts:
466 466 for part in parts:
467 467 _processpart(op, part)
468 468
469 469 def _processchangegroup(op, cg, tr, source, url, **kwargs):
470 470 ret = cg.apply(op.repo, tr, source, url, **kwargs)
471 471 op.records.add('changegroup', {
472 472 'return': ret,
473 473 })
474 474 return ret
475 475
476 476 def _gethandler(op, part):
477 477 status = 'unknown' # used by debug output
478 478 try:
479 479 handler = parthandlermapping.get(part.type)
480 480 if handler is None:
481 481 status = 'unsupported-type'
482 482 raise error.BundleUnknownFeatureError(parttype=part.type)
483 483 indebug(op.ui, 'found a handler for part %s' % part.type)
484 484 unknownparams = part.mandatorykeys - handler.params
485 485 if unknownparams:
486 486 unknownparams = list(unknownparams)
487 487 unknownparams.sort()
488 488 status = 'unsupported-params (%s)' % ', '.join(unknownparams)
489 489 raise error.BundleUnknownFeatureError(parttype=part.type,
490 490 params=unknownparams)
491 491 status = 'supported'
492 492 except error.BundleUnknownFeatureError as exc:
493 493 if part.mandatory: # mandatory parts
494 494 raise
495 495 indebug(op.ui, 'ignoring unsupported advisory part %s' % exc)
496 496 return # skip to part processing
497 497 finally:
498 498 if op.ui.debugflag:
499 499 msg = ['bundle2-input-part: "%s"' % part.type]
500 500 if not part.mandatory:
501 501 msg.append(' (advisory)')
502 502 nbmp = len(part.mandatorykeys)
503 503 nbap = len(part.params) - nbmp
504 504 if nbmp or nbap:
505 505 msg.append(' (params:')
506 506 if nbmp:
507 507 msg.append(' %i mandatory' % nbmp)
508 508 if nbap:
509 509 msg.append(' %i advisory' % nbmp)
510 510 msg.append(')')
511 511 msg.append(' %s\n' % status)
512 512 op.ui.debug(''.join(msg))
513 513
514 514 return handler
515 515
516 516 def _processpart(op, part):
517 517 """process a single part from a bundle
518 518
519 519 The part is guaranteed to have been fully consumed when the function exits
520 520 (even if an exception is raised)."""
521 521 handler = _gethandler(op, part)
522 522 if handler is None:
523 523 return
524 524
525 525 # handler is called outside the above try block so that we don't
526 526 # risk catching KeyErrors from anything other than the
527 527 # parthandlermapping lookup (any KeyError raised by handler()
528 528 # itself represents a defect of a different variety).
529 529 output = None
530 530 if op.captureoutput and op.reply is not None:
531 531 op.ui.pushbuffer(error=True, subproc=True)
532 532 output = ''
533 533 try:
534 534 handler(op, part)
535 535 finally:
536 536 if output is not None:
537 537 output = op.ui.popbuffer()
538 538 if output:
539 539 outpart = op.reply.newpart('output', data=output,
540 540 mandatory=False)
541 541 outpart.addparam(
542 542 'in-reply-to', pycompat.bytestr(part.id), mandatory=False)
543 543
544 544 def decodecaps(blob):
545 545 """decode a bundle2 caps bytes blob into a dictionary
546 546
547 547 The blob is a list of capabilities (one per line)
548 548 Capabilities may have values using a line of the form::
549 549
550 550 capability=value1,value2,value3
551 551
552 552 The values are always a list."""
553 553 caps = {}
554 554 for line in blob.splitlines():
555 555 if not line:
556 556 continue
557 557 if '=' not in line:
558 558 key, vals = line, ()
559 559 else:
560 560 key, vals = line.split('=', 1)
561 561 vals = vals.split(',')
562 562 key = urlreq.unquote(key)
563 563 vals = [urlreq.unquote(v) for v in vals]
564 564 caps[key] = vals
565 565 return caps
566 566
567 567 def encodecaps(caps):
568 568 """encode a bundle2 caps dictionary into a bytes blob"""
569 569 chunks = []
570 570 for ca in sorted(caps):
571 571 vals = caps[ca]
572 572 ca = urlreq.quote(ca)
573 573 vals = [urlreq.quote(v) for v in vals]
574 574 if vals:
575 575 ca = "%s=%s" % (ca, ','.join(vals))
576 576 chunks.append(ca)
577 577 return '\n'.join(chunks)
578 578
579 579 bundletypes = {
580 580 "": ("", 'UN'), # only when using unbundle on ssh and old http servers
581 581 # since the unification ssh accepts a header but there
582 582 # is no capability signaling it.
583 583 "HG20": (), # special-cased below
584 584 "HG10UN": ("HG10UN", 'UN'),
585 585 "HG10BZ": ("HG10", 'BZ'),
586 586 "HG10GZ": ("HG10GZ", 'GZ'),
587 587 }
588 588
589 589 # hgweb uses this list to communicate its preferred type
590 590 bundlepriority = ['HG10GZ', 'HG10BZ', 'HG10UN']
591 591
592 592 class bundle20(object):
593 593 """represent an outgoing bundle2 container
594 594
595 595 Use the `addparam` method to add stream level parameter. and `newpart` to
596 596 populate it. Then call `getchunks` to retrieve all the binary chunks of
597 597 data that compose the bundle2 container."""
598 598
599 599 _magicstring = 'HG20'
600 600
601 601 def __init__(self, ui, capabilities=()):
602 602 self.ui = ui
603 603 self._params = []
604 604 self._parts = []
605 605 self.capabilities = dict(capabilities)
606 606 self._compengine = util.compengines.forbundletype('UN')
607 607 self._compopts = None
608 608 # If compression is being handled by a consumer of the raw
609 609 # data (e.g. the wire protocol), unsetting this flag tells
610 610 # consumers that the bundle is best left uncompressed.
611 611 self.prefercompressed = True
612 612
613 613 def setcompression(self, alg, compopts=None):
614 614 """setup core part compression to <alg>"""
615 615 if alg in (None, 'UN'):
616 616 return
617 617 assert not any(n.lower() == 'compression' for n, v in self._params)
618 618 self.addparam('Compression', alg)
619 619 self._compengine = util.compengines.forbundletype(alg)
620 620 self._compopts = compopts
621 621
622 622 @property
623 623 def nbparts(self):
624 624 """total number of parts added to the bundler"""
625 625 return len(self._parts)
626 626
627 627 # methods used to defines the bundle2 content
628 628 def addparam(self, name, value=None):
629 629 """add a stream level parameter"""
630 630 if not name:
631 631 raise error.ProgrammingError(b'empty parameter name')
632 632 if name[0:1] not in pycompat.bytestr(string.ascii_letters):
633 633 raise error.ProgrammingError(b'non letter first character: %s'
634 634 % name)
635 635 self._params.append((name, value))
636 636
637 637 def addpart(self, part):
638 638 """add a new part to the bundle2 container
639 639
640 640 Parts contains the actual applicative payload."""
641 641 assert part.id is None
642 642 part.id = len(self._parts) # very cheap counter
643 643 self._parts.append(part)
644 644
645 645 def newpart(self, typeid, *args, **kwargs):
646 646 """create a new part and add it to the containers
647 647
648 648 As the part is directly added to the containers. For now, this means
649 649 that any failure to properly initialize the part after calling
650 650 ``newpart`` should result in a failure of the whole bundling process.
651 651
652 652 You can still fall back to manually create and add if you need better
653 653 control."""
654 654 part = bundlepart(typeid, *args, **kwargs)
655 655 self.addpart(part)
656 656 return part
657 657
658 658 # methods used to generate the bundle2 stream
659 659 def getchunks(self):
660 660 if self.ui.debugflag:
661 661 msg = ['bundle2-output-bundle: "%s",' % self._magicstring]
662 662 if self._params:
663 663 msg.append(' (%i params)' % len(self._params))
664 664 msg.append(' %i parts total\n' % len(self._parts))
665 665 self.ui.debug(''.join(msg))
666 666 outdebug(self.ui, 'start emission of %s stream' % self._magicstring)
667 667 yield self._magicstring
668 668 param = self._paramchunk()
669 669 outdebug(self.ui, 'bundle parameter: %s' % param)
670 670 yield _pack(_fstreamparamsize, len(param))
671 671 if param:
672 672 yield param
673 673 for chunk in self._compengine.compressstream(self._getcorechunk(),
674 674 self._compopts):
675 675 yield chunk
676 676
677 677 def _paramchunk(self):
678 678 """return a encoded version of all stream parameters"""
679 679 blocks = []
680 680 for par, value in self._params:
681 681 par = urlreq.quote(par)
682 682 if value is not None:
683 683 value = urlreq.quote(value)
684 684 par = '%s=%s' % (par, value)
685 685 blocks.append(par)
686 686 return ' '.join(blocks)
687 687
688 688 def _getcorechunk(self):
689 689 """yield chunk for the core part of the bundle
690 690
691 691 (all but headers and parameters)"""
692 692 outdebug(self.ui, 'start of parts')
693 693 for part in self._parts:
694 694 outdebug(self.ui, 'bundle part: "%s"' % part.type)
695 695 for chunk in part.getchunks(ui=self.ui):
696 696 yield chunk
697 697 outdebug(self.ui, 'end of bundle')
698 698 yield _pack(_fpartheadersize, 0)
699 699
700 700
701 701 def salvageoutput(self):
702 702 """return a list with a copy of all output parts in the bundle
703 703
704 704 This is meant to be used during error handling to make sure we preserve
705 705 server output"""
706 706 salvaged = []
707 707 for part in self._parts:
708 708 if part.type.startswith('output'):
709 709 salvaged.append(part.copy())
710 710 return salvaged
711 711
712 712
713 713 class unpackermixin(object):
714 714 """A mixin to extract bytes and struct data from a stream"""
715 715
716 716 def __init__(self, fp):
717 717 self._fp = fp
718 718
719 719 def _unpack(self, format):
720 720 """unpack this struct format from the stream
721 721
722 722 This method is meant for internal usage by the bundle2 protocol only.
723 723 They directly manipulate the low level stream including bundle2 level
724 724 instruction.
725 725
726 726 Do not use it to implement higher-level logic or methods."""
727 727 data = self._readexact(struct.calcsize(format))
728 728 return _unpack(format, data)
729 729
730 730 def _readexact(self, size):
731 731 """read exactly <size> bytes from the stream
732 732
733 733 This method is meant for internal usage by the bundle2 protocol only.
734 734 They directly manipulate the low level stream including bundle2 level
735 735 instruction.
736 736
737 737 Do not use it to implement higher-level logic or methods."""
738 738 return changegroup.readexactly(self._fp, size)
739 739
740 740 def getunbundler(ui, fp, magicstring=None):
741 741 """return a valid unbundler object for a given magicstring"""
742 742 if magicstring is None:
743 743 magicstring = changegroup.readexactly(fp, 4)
744 744 magic, version = magicstring[0:2], magicstring[2:4]
745 745 if magic != 'HG':
746 746 ui.debug(
747 747 "error: invalid magic: %r (version %r), should be 'HG'\n"
748 748 % (magic, version))
749 749 raise error.Abort(_('not a Mercurial bundle'))
750 750 unbundlerclass = formatmap.get(version)
751 751 if unbundlerclass is None:
752 752 raise error.Abort(_('unknown bundle version %s') % version)
753 753 unbundler = unbundlerclass(ui, fp)
754 754 indebug(ui, 'start processing of %s stream' % magicstring)
755 755 return unbundler
756 756
757 757 class unbundle20(unpackermixin):
758 758 """interpret a bundle2 stream
759 759
760 760 This class is fed with a binary stream and yields parts through its
761 761 `iterparts` methods."""
762 762
763 763 _magicstring = 'HG20'
764 764
765 765 def __init__(self, ui, fp):
766 766 """If header is specified, we do not read it out of the stream."""
767 767 self.ui = ui
768 768 self._compengine = util.compengines.forbundletype('UN')
769 769 self._compressed = None
770 770 super(unbundle20, self).__init__(fp)
771 771
772 772 @util.propertycache
773 773 def params(self):
774 774 """dictionary of stream level parameters"""
775 775 indebug(self.ui, 'reading bundle2 stream parameters')
776 776 params = {}
777 777 paramssize = self._unpack(_fstreamparamsize)[0]
778 778 if paramssize < 0:
779 779 raise error.BundleValueError('negative bundle param size: %i'
780 780 % paramssize)
781 781 if paramssize:
782 782 params = self._readexact(paramssize)
783 783 params = self._processallparams(params)
784 784 return params
785 785
786 786 def _processallparams(self, paramsblock):
787 787 """"""
788 788 params = util.sortdict()
789 789 for p in paramsblock.split(' '):
790 790 p = p.split('=', 1)
791 791 p = [urlreq.unquote(i) for i in p]
792 792 if len(p) < 2:
793 793 p.append(None)
794 794 self._processparam(*p)
795 795 params[p[0]] = p[1]
796 796 return params
797 797
798 798
799 799 def _processparam(self, name, value):
800 800 """process a parameter, applying its effect if needed
801 801
802 802 Parameter starting with a lower case letter are advisory and will be
803 803 ignored when unknown. Those starting with an upper case letter are
804 804 mandatory and will this function will raise a KeyError when unknown.
805 805
806 806 Note: no option are currently supported. Any input will be either
807 807 ignored or failing.
808 808 """
809 809 if not name:
810 810 raise ValueError(r'empty parameter name')
811 811 if name[0:1] not in pycompat.bytestr(string.ascii_letters):
812 812 raise ValueError(r'non letter first character: %s' % name)
813 813 try:
814 814 handler = b2streamparamsmap[name.lower()]
815 815 except KeyError:
816 816 if name[0:1].islower():
817 817 indebug(self.ui, "ignoring unknown parameter %s" % name)
818 818 else:
819 819 raise error.BundleUnknownFeatureError(params=(name,))
820 820 else:
821 821 handler(self, name, value)
822 822
823 823 def _forwardchunks(self):
824 824 """utility to transfer a bundle2 as binary
825 825
826 826 This is made necessary by the fact the 'getbundle' command over 'ssh'
827 827 have no way to know then the reply end, relying on the bundle to be
828 828 interpreted to know its end. This is terrible and we are sorry, but we
829 829 needed to move forward to get general delta enabled.
830 830 """
831 831 yield self._magicstring
832 832 assert 'params' not in vars(self)
833 833 paramssize = self._unpack(_fstreamparamsize)[0]
834 834 if paramssize < 0:
835 835 raise error.BundleValueError('negative bundle param size: %i'
836 836 % paramssize)
837 837 yield _pack(_fstreamparamsize, paramssize)
838 838 if paramssize:
839 839 params = self._readexact(paramssize)
840 840 self._processallparams(params)
841 841 yield params
842 842 assert self._compengine.bundletype == 'UN'
843 843 # From there, payload might need to be decompressed
844 844 self._fp = self._compengine.decompressorreader(self._fp)
845 845 emptycount = 0
846 846 while emptycount < 2:
847 847 # so we can brainlessly loop
848 848 assert _fpartheadersize == _fpayloadsize
849 849 size = self._unpack(_fpartheadersize)[0]
850 850 yield _pack(_fpartheadersize, size)
851 851 if size:
852 852 emptycount = 0
853 853 else:
854 854 emptycount += 1
855 855 continue
856 856 if size == flaginterrupt:
857 857 continue
858 858 elif size < 0:
859 859 raise error.BundleValueError('negative chunk size: %i')
860 860 yield self._readexact(size)
861 861
862 862
863 863 def iterparts(self, seekable=False):
864 864 """yield all parts contained in the stream"""
865 865 cls = seekableunbundlepart if seekable else unbundlepart
866 866 # make sure param have been loaded
867 867 self.params
868 868 # From there, payload need to be decompressed
869 869 self._fp = self._compengine.decompressorreader(self._fp)
870 870 indebug(self.ui, 'start extraction of bundle2 parts')
871 871 headerblock = self._readpartheader()
872 872 while headerblock is not None:
873 873 part = cls(self.ui, headerblock, self._fp)
874 874 yield part
875 875 # Ensure part is fully consumed so we can start reading the next
876 876 # part.
877 877 part.consume()
878 878
879 879 headerblock = self._readpartheader()
880 880 indebug(self.ui, 'end of bundle2 stream')
881 881
882 882 def _readpartheader(self):
883 883 """reads a part header size and return the bytes blob
884 884
885 885 returns None if empty"""
886 886 headersize = self._unpack(_fpartheadersize)[0]
887 887 if headersize < 0:
888 888 raise error.BundleValueError('negative part header size: %i'
889 889 % headersize)
890 890 indebug(self.ui, 'part header size: %i' % headersize)
891 891 if headersize:
892 892 return self._readexact(headersize)
893 893 return None
894 894
895 895 def compressed(self):
896 896 self.params # load params
897 897 return self._compressed
898 898
899 899 def close(self):
900 900 """close underlying file"""
901 901 if util.safehasattr(self._fp, 'close'):
902 902 return self._fp.close()
903 903
904 904 formatmap = {'20': unbundle20}
905 905
906 906 b2streamparamsmap = {}
907 907
908 908 def b2streamparamhandler(name):
909 909 """register a handler for a stream level parameter"""
910 910 def decorator(func):
911 911 assert name not in formatmap
912 912 b2streamparamsmap[name] = func
913 913 return func
914 914 return decorator
915 915
916 916 @b2streamparamhandler('compression')
917 917 def processcompression(unbundler, param, value):
918 918 """read compression parameter and install payload decompression"""
919 919 if value not in util.compengines.supportedbundletypes:
920 920 raise error.BundleUnknownFeatureError(params=(param,),
921 921 values=(value,))
922 922 unbundler._compengine = util.compengines.forbundletype(value)
923 923 if value is not None:
924 924 unbundler._compressed = True
925 925
926 926 class bundlepart(object):
927 927 """A bundle2 part contains application level payload
928 928
929 929 The part `type` is used to route the part to the application level
930 930 handler.
931 931
932 932 The part payload is contained in ``part.data``. It could be raw bytes or a
933 933 generator of byte chunks.
934 934
935 935 You can add parameters to the part using the ``addparam`` method.
936 936 Parameters can be either mandatory (default) or advisory. Remote side
937 937 should be able to safely ignore the advisory ones.
938 938
939 939 Both data and parameters cannot be modified after the generation has begun.
940 940 """
941 941
942 942 def __init__(self, parttype, mandatoryparams=(), advisoryparams=(),
943 943 data='', mandatory=True):
944 944 validateparttype(parttype)
945 945 self.id = None
946 946 self.type = parttype
947 947 self._data = data
948 948 self._mandatoryparams = list(mandatoryparams)
949 949 self._advisoryparams = list(advisoryparams)
950 950 # checking for duplicated entries
951 951 self._seenparams = set()
952 952 for pname, __ in self._mandatoryparams + self._advisoryparams:
953 953 if pname in self._seenparams:
954 954 raise error.ProgrammingError('duplicated params: %s' % pname)
955 955 self._seenparams.add(pname)
956 956 # status of the part's generation:
957 957 # - None: not started,
958 958 # - False: currently generated,
959 959 # - True: generation done.
960 960 self._generated = None
961 961 self.mandatory = mandatory
962 962
963 963 def __repr__(self):
964 964 cls = "%s.%s" % (self.__class__.__module__, self.__class__.__name__)
965 965 return ('<%s object at %x; id: %s; type: %s; mandatory: %s>'
966 966 % (cls, id(self), self.id, self.type, self.mandatory))
967 967
968 968 def copy(self):
969 969 """return a copy of the part
970 970
971 971 The new part have the very same content but no partid assigned yet.
972 972 Parts with generated data cannot be copied."""
973 973 assert not util.safehasattr(self.data, 'next')
974 974 return self.__class__(self.type, self._mandatoryparams,
975 975 self._advisoryparams, self._data, self.mandatory)
976 976
977 977 # methods used to defines the part content
978 978 @property
979 979 def data(self):
980 980 return self._data
981 981
982 982 @data.setter
983 983 def data(self, data):
984 984 if self._generated is not None:
985 985 raise error.ReadOnlyPartError('part is being generated')
986 986 self._data = data
987 987
988 988 @property
989 989 def mandatoryparams(self):
990 990 # make it an immutable tuple to force people through ``addparam``
991 991 return tuple(self._mandatoryparams)
992 992
993 993 @property
994 994 def advisoryparams(self):
995 995 # make it an immutable tuple to force people through ``addparam``
996 996 return tuple(self._advisoryparams)
997 997
998 998 def addparam(self, name, value='', mandatory=True):
999 999 """add a parameter to the part
1000 1000
1001 1001 If 'mandatory' is set to True, the remote handler must claim support
1002 1002 for this parameter or the unbundling will be aborted.
1003 1003
1004 1004 The 'name' and 'value' cannot exceed 255 bytes each.
1005 1005 """
1006 1006 if self._generated is not None:
1007 1007 raise error.ReadOnlyPartError('part is being generated')
1008 1008 if name in self._seenparams:
1009 1009 raise ValueError('duplicated params: %s' % name)
1010 1010 self._seenparams.add(name)
1011 1011 params = self._advisoryparams
1012 1012 if mandatory:
1013 1013 params = self._mandatoryparams
1014 1014 params.append((name, value))
1015 1015
1016 1016 # methods used to generates the bundle2 stream
1017 1017 def getchunks(self, ui):
1018 1018 if self._generated is not None:
1019 1019 raise error.ProgrammingError('part can only be consumed once')
1020 1020 self._generated = False
1021 1021
1022 1022 if ui.debugflag:
1023 1023 msg = ['bundle2-output-part: "%s"' % self.type]
1024 1024 if not self.mandatory:
1025 1025 msg.append(' (advisory)')
1026 1026 nbmp = len(self.mandatoryparams)
1027 1027 nbap = len(self.advisoryparams)
1028 1028 if nbmp or nbap:
1029 1029 msg.append(' (params:')
1030 1030 if nbmp:
1031 1031 msg.append(' %i mandatory' % nbmp)
1032 1032 if nbap:
1033 1033 msg.append(' %i advisory' % nbmp)
1034 1034 msg.append(')')
1035 1035 if not self.data:
1036 1036 msg.append(' empty payload')
1037 1037 elif (util.safehasattr(self.data, 'next')
1038 1038 or util.safehasattr(self.data, '__next__')):
1039 1039 msg.append(' streamed payload')
1040 1040 else:
1041 1041 msg.append(' %i bytes payload' % len(self.data))
1042 1042 msg.append('\n')
1043 1043 ui.debug(''.join(msg))
1044 1044
1045 1045 #### header
1046 1046 if self.mandatory:
1047 1047 parttype = self.type.upper()
1048 1048 else:
1049 1049 parttype = self.type.lower()
1050 1050 outdebug(ui, 'part %s: "%s"' % (pycompat.bytestr(self.id), parttype))
1051 1051 ## parttype
1052 1052 header = [_pack(_fparttypesize, len(parttype)),
1053 1053 parttype, _pack(_fpartid, self.id),
1054 1054 ]
1055 1055 ## parameters
1056 1056 # count
1057 1057 manpar = self.mandatoryparams
1058 1058 advpar = self.advisoryparams
1059 1059 header.append(_pack(_fpartparamcount, len(manpar), len(advpar)))
1060 1060 # size
1061 1061 parsizes = []
1062 1062 for key, value in manpar:
1063 1063 parsizes.append(len(key))
1064 1064 parsizes.append(len(value))
1065 1065 for key, value in advpar:
1066 1066 parsizes.append(len(key))
1067 1067 parsizes.append(len(value))
1068 1068 paramsizes = _pack(_makefpartparamsizes(len(parsizes) // 2), *parsizes)
1069 1069 header.append(paramsizes)
1070 1070 # key, value
1071 1071 for key, value in manpar:
1072 1072 header.append(key)
1073 1073 header.append(value)
1074 1074 for key, value in advpar:
1075 1075 header.append(key)
1076 1076 header.append(value)
1077 1077 ## finalize header
1078 1078 try:
1079 1079 headerchunk = ''.join(header)
1080 1080 except TypeError:
1081 1081 raise TypeError(r'Found a non-bytes trying to '
1082 1082 r'build bundle part header: %r' % header)
1083 1083 outdebug(ui, 'header chunk size: %i' % len(headerchunk))
1084 1084 yield _pack(_fpartheadersize, len(headerchunk))
1085 1085 yield headerchunk
1086 1086 ## payload
1087 1087 try:
1088 1088 for chunk in self._payloadchunks():
1089 1089 outdebug(ui, 'payload chunk size: %i' % len(chunk))
1090 1090 yield _pack(_fpayloadsize, len(chunk))
1091 1091 yield chunk
1092 1092 except GeneratorExit:
1093 1093 # GeneratorExit means that nobody is listening for our
1094 1094 # results anyway, so just bail quickly rather than trying
1095 1095 # to produce an error part.
1096 1096 ui.debug('bundle2-generatorexit\n')
1097 1097 raise
1098 1098 except BaseException as exc:
1099 1099 bexc = stringutil.forcebytestr(exc)
1100 1100 # backup exception data for later
1101 1101 ui.debug('bundle2-input-stream-interrupt: encoding exception %s'
1102 1102 % bexc)
1103 1103 tb = sys.exc_info()[2]
1104 1104 msg = 'unexpected error: %s' % bexc
1105 1105 interpart = bundlepart('error:abort', [('message', msg)],
1106 1106 mandatory=False)
1107 1107 interpart.id = 0
1108 1108 yield _pack(_fpayloadsize, -1)
1109 1109 for chunk in interpart.getchunks(ui=ui):
1110 1110 yield chunk
1111 1111 outdebug(ui, 'closing payload chunk')
1112 1112 # abort current part payload
1113 1113 yield _pack(_fpayloadsize, 0)
1114 1114 pycompat.raisewithtb(exc, tb)
1115 1115 # end of payload
1116 1116 outdebug(ui, 'closing payload chunk')
1117 1117 yield _pack(_fpayloadsize, 0)
1118 1118 self._generated = True
1119 1119
1120 1120 def _payloadchunks(self):
1121 1121 """yield chunks of a the part payload
1122 1122
1123 1123 Exists to handle the different methods to provide data to a part."""
1124 1124 # we only support fixed size data now.
1125 1125 # This will be improved in the future.
1126 1126 if (util.safehasattr(self.data, 'next')
1127 1127 or util.safehasattr(self.data, '__next__')):
1128 1128 buff = util.chunkbuffer(self.data)
1129 1129 chunk = buff.read(preferedchunksize)
1130 1130 while chunk:
1131 1131 yield chunk
1132 1132 chunk = buff.read(preferedchunksize)
1133 1133 elif len(self.data):
1134 1134 yield self.data
1135 1135
1136 1136
1137 1137 flaginterrupt = -1
1138 1138
1139 1139 class interrupthandler(unpackermixin):
1140 1140 """read one part and process it with restricted capability
1141 1141
1142 1142 This allows to transmit exception raised on the producer size during part
1143 1143 iteration while the consumer is reading a part.
1144 1144
1145 1145 Part processed in this manner only have access to a ui object,"""
1146 1146
1147 1147 def __init__(self, ui, fp):
1148 1148 super(interrupthandler, self).__init__(fp)
1149 1149 self.ui = ui
1150 1150
1151 1151 def _readpartheader(self):
1152 1152 """reads a part header size and return the bytes blob
1153 1153
1154 1154 returns None if empty"""
1155 1155 headersize = self._unpack(_fpartheadersize)[0]
1156 1156 if headersize < 0:
1157 1157 raise error.BundleValueError('negative part header size: %i'
1158 1158 % headersize)
1159 1159 indebug(self.ui, 'part header size: %i\n' % headersize)
1160 1160 if headersize:
1161 1161 return self._readexact(headersize)
1162 1162 return None
1163 1163
1164 1164 def __call__(self):
1165 1165
1166 1166 self.ui.debug('bundle2-input-stream-interrupt:'
1167 1167 ' opening out of band context\n')
1168 1168 indebug(self.ui, 'bundle2 stream interruption, looking for a part.')
1169 1169 headerblock = self._readpartheader()
1170 1170 if headerblock is None:
1171 1171 indebug(self.ui, 'no part found during interruption.')
1172 1172 return
1173 1173 part = unbundlepart(self.ui, headerblock, self._fp)
1174 1174 op = interruptoperation(self.ui)
1175 1175 hardabort = False
1176 1176 try:
1177 1177 _processpart(op, part)
1178 1178 except (SystemExit, KeyboardInterrupt):
1179 1179 hardabort = True
1180 1180 raise
1181 1181 finally:
1182 1182 if not hardabort:
1183 1183 part.consume()
1184 1184 self.ui.debug('bundle2-input-stream-interrupt:'
1185 1185 ' closing out of band context\n')
1186 1186
1187 1187 class interruptoperation(object):
1188 1188 """A limited operation to be use by part handler during interruption
1189 1189
1190 1190 It only have access to an ui object.
1191 1191 """
1192 1192
1193 1193 def __init__(self, ui):
1194 1194 self.ui = ui
1195 1195 self.reply = None
1196 1196 self.captureoutput = False
1197 1197
1198 1198 @property
1199 1199 def repo(self):
1200 1200 raise error.ProgrammingError('no repo access from stream interruption')
1201 1201
1202 1202 def gettransaction(self):
1203 1203 raise TransactionUnavailable('no repo access from stream interruption')
1204 1204
1205 1205 def decodepayloadchunks(ui, fh):
1206 1206 """Reads bundle2 part payload data into chunks.
1207 1207
1208 1208 Part payload data consists of framed chunks. This function takes
1209 1209 a file handle and emits those chunks.
1210 1210 """
1211 1211 dolog = ui.configbool('devel', 'bundle2.debug')
1212 1212 debug = ui.debug
1213 1213
1214 1214 headerstruct = struct.Struct(_fpayloadsize)
1215 1215 headersize = headerstruct.size
1216 1216 unpack = headerstruct.unpack
1217 1217
1218 1218 readexactly = changegroup.readexactly
1219 1219 read = fh.read
1220 1220
1221 1221 chunksize = unpack(readexactly(fh, headersize))[0]
1222 1222 indebug(ui, 'payload chunk size: %i' % chunksize)
1223 1223
1224 1224 # changegroup.readexactly() is inlined below for performance.
1225 1225 while chunksize:
1226 1226 if chunksize >= 0:
1227 1227 s = read(chunksize)
1228 1228 if len(s) < chunksize:
1229 1229 raise error.Abort(_('stream ended unexpectedly '
1230 1230 ' (got %d bytes, expected %d)') %
1231 1231 (len(s), chunksize))
1232 1232
1233 1233 yield s
1234 1234 elif chunksize == flaginterrupt:
1235 1235 # Interrupt "signal" detected. The regular stream is interrupted
1236 1236 # and a bundle2 part follows. Consume it.
1237 1237 interrupthandler(ui, fh)()
1238 1238 else:
1239 1239 raise error.BundleValueError(
1240 1240 'negative payload chunk size: %s' % chunksize)
1241 1241
1242 1242 s = read(headersize)
1243 1243 if len(s) < headersize:
1244 1244 raise error.Abort(_('stream ended unexpectedly '
1245 1245 ' (got %d bytes, expected %d)') %
1246 1246 (len(s), chunksize))
1247 1247
1248 1248 chunksize = unpack(s)[0]
1249 1249
1250 1250 # indebug() inlined for performance.
1251 1251 if dolog:
1252 1252 debug('bundle2-input: payload chunk size: %i\n' % chunksize)
1253 1253
1254 1254 class unbundlepart(unpackermixin):
1255 1255 """a bundle part read from a bundle"""
1256 1256
1257 1257 def __init__(self, ui, header, fp):
1258 1258 super(unbundlepart, self).__init__(fp)
1259 1259 self._seekable = (util.safehasattr(fp, 'seek') and
1260 1260 util.safehasattr(fp, 'tell'))
1261 1261 self.ui = ui
1262 1262 # unbundle state attr
1263 1263 self._headerdata = header
1264 1264 self._headeroffset = 0
1265 1265 self._initialized = False
1266 1266 self.consumed = False
1267 1267 # part data
1268 1268 self.id = None
1269 1269 self.type = None
1270 1270 self.mandatoryparams = None
1271 1271 self.advisoryparams = None
1272 1272 self.params = None
1273 1273 self.mandatorykeys = ()
1274 1274 self._readheader()
1275 1275 self._mandatory = None
1276 1276 self._pos = 0
1277 1277
1278 1278 def _fromheader(self, size):
1279 1279 """return the next <size> byte from the header"""
1280 1280 offset = self._headeroffset
1281 1281 data = self._headerdata[offset:(offset + size)]
1282 1282 self._headeroffset = offset + size
1283 1283 return data
1284 1284
1285 1285 def _unpackheader(self, format):
1286 1286 """read given format from header
1287 1287
1288 1288 This automatically compute the size of the format to read."""
1289 1289 data = self._fromheader(struct.calcsize(format))
1290 1290 return _unpack(format, data)
1291 1291
1292 1292 def _initparams(self, mandatoryparams, advisoryparams):
1293 1293 """internal function to setup all logic related parameters"""
1294 1294 # make it read only to prevent people touching it by mistake.
1295 1295 self.mandatoryparams = tuple(mandatoryparams)
1296 1296 self.advisoryparams = tuple(advisoryparams)
1297 1297 # user friendly UI
1298 1298 self.params = util.sortdict(self.mandatoryparams)
1299 1299 self.params.update(self.advisoryparams)
1300 1300 self.mandatorykeys = frozenset(p[0] for p in mandatoryparams)
1301 1301
1302 1302 def _readheader(self):
1303 1303 """read the header and setup the object"""
1304 1304 typesize = self._unpackheader(_fparttypesize)[0]
1305 1305 self.type = self._fromheader(typesize)
1306 1306 indebug(self.ui, 'part type: "%s"' % self.type)
1307 1307 self.id = self._unpackheader(_fpartid)[0]
1308 1308 indebug(self.ui, 'part id: "%s"' % pycompat.bytestr(self.id))
1309 1309 # extract mandatory bit from type
1310 1310 self.mandatory = (self.type != self.type.lower())
1311 1311 self.type = self.type.lower()
1312 1312 ## reading parameters
1313 1313 # param count
1314 1314 mancount, advcount = self._unpackheader(_fpartparamcount)
1315 1315 indebug(self.ui, 'part parameters: %i' % (mancount + advcount))
1316 1316 # param size
1317 1317 fparamsizes = _makefpartparamsizes(mancount + advcount)
1318 1318 paramsizes = self._unpackheader(fparamsizes)
1319 1319 # make it a list of couple again
1320 1320 paramsizes = list(zip(paramsizes[::2], paramsizes[1::2]))
1321 1321 # split mandatory from advisory
1322 1322 mansizes = paramsizes[:mancount]
1323 1323 advsizes = paramsizes[mancount:]
1324 1324 # retrieve param value
1325 1325 manparams = []
1326 1326 for key, value in mansizes:
1327 1327 manparams.append((self._fromheader(key), self._fromheader(value)))
1328 1328 advparams = []
1329 1329 for key, value in advsizes:
1330 1330 advparams.append((self._fromheader(key), self._fromheader(value)))
1331 1331 self._initparams(manparams, advparams)
1332 1332 ## part payload
1333 1333 self._payloadstream = util.chunkbuffer(self._payloadchunks())
1334 1334 # we read the data, tell it
1335 1335 self._initialized = True
1336 1336
1337 1337 def _payloadchunks(self):
1338 1338 """Generator of decoded chunks in the payload."""
1339 1339 return decodepayloadchunks(self.ui, self._fp)
1340 1340
1341 1341 def consume(self):
1342 1342 """Read the part payload until completion.
1343 1343
1344 1344 By consuming the part data, the underlying stream read offset will
1345 1345 be advanced to the next part (or end of stream).
1346 1346 """
1347 1347 if self.consumed:
1348 1348 return
1349 1349
1350 1350 chunk = self.read(32768)
1351 1351 while chunk:
1352 1352 self._pos += len(chunk)
1353 1353 chunk = self.read(32768)
1354 1354
1355 1355 def read(self, size=None):
1356 1356 """read payload data"""
1357 1357 if not self._initialized:
1358 1358 self._readheader()
1359 1359 if size is None:
1360 1360 data = self._payloadstream.read()
1361 1361 else:
1362 1362 data = self._payloadstream.read(size)
1363 1363 self._pos += len(data)
1364 1364 if size is None or len(data) < size:
1365 1365 if not self.consumed and self._pos:
1366 1366 self.ui.debug('bundle2-input-part: total payload size %i\n'
1367 1367 % self._pos)
1368 1368 self.consumed = True
1369 1369 return data
1370 1370
1371 1371 class seekableunbundlepart(unbundlepart):
1372 1372 """A bundle2 part in a bundle that is seekable.
1373 1373
1374 1374 Regular ``unbundlepart`` instances can only be read once. This class
1375 1375 extends ``unbundlepart`` to enable bi-directional seeking within the
1376 1376 part.
1377 1377
1378 1378 Bundle2 part data consists of framed chunks. Offsets when seeking
1379 1379 refer to the decoded data, not the offsets in the underlying bundle2
1380 1380 stream.
1381 1381
1382 1382 To facilitate quickly seeking within the decoded data, instances of this
1383 1383 class maintain a mapping between offsets in the underlying stream and
1384 1384 the decoded payload. This mapping will consume memory in proportion
1385 1385 to the number of chunks within the payload (which almost certainly
1386 1386 increases in proportion with the size of the part).
1387 1387 """
1388 1388 def __init__(self, ui, header, fp):
1389 1389 # (payload, file) offsets for chunk starts.
1390 1390 self._chunkindex = []
1391 1391
1392 1392 super(seekableunbundlepart, self).__init__(ui, header, fp)
1393 1393
1394 1394 def _payloadchunks(self, chunknum=0):
1395 1395 '''seek to specified chunk and start yielding data'''
1396 1396 if len(self._chunkindex) == 0:
1397 1397 assert chunknum == 0, 'Must start with chunk 0'
1398 1398 self._chunkindex.append((0, self._tellfp()))
1399 1399 else:
1400 1400 assert chunknum < len(self._chunkindex), \
1401 1401 'Unknown chunk %d' % chunknum
1402 1402 self._seekfp(self._chunkindex[chunknum][1])
1403 1403
1404 1404 pos = self._chunkindex[chunknum][0]
1405 1405
1406 1406 for chunk in decodepayloadchunks(self.ui, self._fp):
1407 1407 chunknum += 1
1408 1408 pos += len(chunk)
1409 1409 if chunknum == len(self._chunkindex):
1410 1410 self._chunkindex.append((pos, self._tellfp()))
1411 1411
1412 1412 yield chunk
1413 1413
1414 1414 def _findchunk(self, pos):
1415 1415 '''for a given payload position, return a chunk number and offset'''
1416 1416 for chunk, (ppos, fpos) in enumerate(self._chunkindex):
1417 1417 if ppos == pos:
1418 1418 return chunk, 0
1419 1419 elif ppos > pos:
1420 1420 return chunk - 1, pos - self._chunkindex[chunk - 1][0]
1421 1421 raise ValueError('Unknown chunk')
1422 1422
1423 1423 def tell(self):
1424 1424 return self._pos
1425 1425
1426 1426 def seek(self, offset, whence=os.SEEK_SET):
1427 1427 if whence == os.SEEK_SET:
1428 1428 newpos = offset
1429 1429 elif whence == os.SEEK_CUR:
1430 1430 newpos = self._pos + offset
1431 1431 elif whence == os.SEEK_END:
1432 1432 if not self.consumed:
1433 1433 # Can't use self.consume() here because it advances self._pos.
1434 1434 chunk = self.read(32768)
1435 1435 while chunk:
1436 1436 chunk = self.read(32768)
1437 1437 newpos = self._chunkindex[-1][0] - offset
1438 1438 else:
1439 1439 raise ValueError('Unknown whence value: %r' % (whence,))
1440 1440
1441 1441 if newpos > self._chunkindex[-1][0] and not self.consumed:
1442 1442 # Can't use self.consume() here because it advances self._pos.
1443 1443 chunk = self.read(32768)
1444 1444 while chunk:
1445 1445 chunk = self.read(32668)
1446 1446
1447 1447 if not 0 <= newpos <= self._chunkindex[-1][0]:
1448 1448 raise ValueError('Offset out of range')
1449 1449
1450 1450 if self._pos != newpos:
1451 1451 chunk, internaloffset = self._findchunk(newpos)
1452 1452 self._payloadstream = util.chunkbuffer(self._payloadchunks(chunk))
1453 1453 adjust = self.read(internaloffset)
1454 1454 if len(adjust) != internaloffset:
1455 1455 raise error.Abort(_('Seek failed\n'))
1456 1456 self._pos = newpos
1457 1457
1458 1458 def _seekfp(self, offset, whence=0):
1459 1459 """move the underlying file pointer
1460 1460
1461 1461 This method is meant for internal usage by the bundle2 protocol only.
1462 1462 They directly manipulate the low level stream including bundle2 level
1463 1463 instruction.
1464 1464
1465 1465 Do not use it to implement higher-level logic or methods."""
1466 1466 if self._seekable:
1467 1467 return self._fp.seek(offset, whence)
1468 1468 else:
1469 1469 raise NotImplementedError(_('File pointer is not seekable'))
1470 1470
1471 1471 def _tellfp(self):
1472 1472 """return the file offset, or None if file is not seekable
1473 1473
1474 1474 This method is meant for internal usage by the bundle2 protocol only.
1475 1475 They directly manipulate the low level stream including bundle2 level
1476 1476 instruction.
1477 1477
1478 1478 Do not use it to implement higher-level logic or methods."""
1479 1479 if self._seekable:
1480 1480 try:
1481 1481 return self._fp.tell()
1482 1482 except IOError as e:
1483 1483 if e.errno == errno.ESPIPE:
1484 1484 self._seekable = False
1485 1485 else:
1486 1486 raise
1487 1487 return None
1488 1488
1489 1489 # These are only the static capabilities.
1490 1490 # Check the 'getrepocaps' function for the rest.
1491 1491 capabilities = {'HG20': (),
1492 1492 'bookmarks': (),
1493 1493 'error': ('abort', 'unsupportedcontent', 'pushraced',
1494 1494 'pushkey'),
1495 1495 'listkeys': (),
1496 1496 'pushkey': (),
1497 1497 'digests': tuple(sorted(util.DIGESTS.keys())),
1498 1498 'remote-changegroup': ('http', 'https'),
1499 1499 'hgtagsfnodes': (),
1500 1500 'rev-branch-cache': (),
1501 1501 'phases': ('heads',),
1502 1502 'stream': ('v2',),
1503 1503 }
1504 1504
1505 1505 def getrepocaps(repo, allowpushback=False, role=None):
1506 1506 """return the bundle2 capabilities for a given repo
1507 1507
1508 1508 Exists to allow extensions (like evolution) to mutate the capabilities.
1509 1509
1510 1510 The returned value is used for servers advertising their capabilities as
1511 1511 well as clients advertising their capabilities to servers as part of
1512 1512 bundle2 requests. The ``role`` argument specifies which is which.
1513 1513 """
1514 1514 if role not in ('client', 'server'):
1515 1515 raise error.ProgrammingError('role argument must be client or server')
1516 1516
1517 1517 caps = capabilities.copy()
1518 1518 caps['changegroup'] = tuple(sorted(
1519 1519 changegroup.supportedincomingversions(repo)))
1520 1520 if obsolete.isenabled(repo, obsolete.exchangeopt):
1521 1521 supportedformat = tuple('V%i' % v for v in obsolete.formats)
1522 1522 caps['obsmarkers'] = supportedformat
1523 1523 if allowpushback:
1524 1524 caps['pushback'] = ()
1525 1525 cpmode = repo.ui.config('server', 'concurrent-push-mode')
1526 1526 if cpmode == 'check-related':
1527 1527 caps['checkheads'] = ('related',)
1528 1528 if 'phases' in repo.ui.configlist('devel', 'legacy.exchange'):
1529 1529 caps.pop('phases')
1530 1530
1531 1531 # Don't advertise stream clone support in server mode if not configured.
1532 1532 if role == 'server':
1533 1533 streamsupported = repo.ui.configbool('server', 'uncompressed',
1534 1534 untrusted=True)
1535 1535 featuresupported = repo.ui.configbool('server', 'bundle2.stream')
1536 1536
1537 1537 if not streamsupported or not featuresupported:
1538 1538 caps.pop('stream')
1539 1539 # Else always advertise support on client, because payload support
1540 1540 # should always be advertised.
1541 1541
1542 1542 return caps
1543 1543
1544 1544 def bundle2caps(remote):
1545 1545 """return the bundle capabilities of a peer as dict"""
1546 1546 raw = remote.capable('bundle2')
1547 1547 if not raw and raw != '':
1548 1548 return {}
1549 1549 capsblob = urlreq.unquote(remote.capable('bundle2'))
1550 1550 return decodecaps(capsblob)
1551 1551
1552 1552 def obsmarkersversion(caps):
1553 1553 """extract the list of supported obsmarkers versions from a bundle2caps dict
1554 1554 """
1555 1555 obscaps = caps.get('obsmarkers', ())
1556 1556 return [int(c[1:]) for c in obscaps if c.startswith('V')]
1557 1557
1558 1558 def writenewbundle(ui, repo, source, filename, bundletype, outgoing, opts,
1559 1559 vfs=None, compression=None, compopts=None):
1560 1560 if bundletype.startswith('HG10'):
1561 1561 cg = changegroup.makechangegroup(repo, outgoing, '01', source)
1562 1562 return writebundle(ui, cg, filename, bundletype, vfs=vfs,
1563 1563 compression=compression, compopts=compopts)
1564 1564 elif not bundletype.startswith('HG20'):
1565 1565 raise error.ProgrammingError('unknown bundle type: %s' % bundletype)
1566 1566
1567 1567 caps = {}
1568 1568 if 'obsolescence' in opts:
1569 1569 caps['obsmarkers'] = ('V1',)
1570 1570 bundle = bundle20(ui, caps)
1571 1571 bundle.setcompression(compression, compopts)
1572 1572 _addpartsfromopts(ui, repo, bundle, source, outgoing, opts)
1573 1573 chunkiter = bundle.getchunks()
1574 1574
1575 1575 return changegroup.writechunks(ui, chunkiter, filename, vfs=vfs)
1576 1576
1577 1577 def _addpartsfromopts(ui, repo, bundler, source, outgoing, opts):
1578 1578 # We should eventually reconcile this logic with the one behind
1579 1579 # 'exchange.getbundle2partsgenerator'.
1580 1580 #
1581 1581 # The type of input from 'getbundle' and 'writenewbundle' are a bit
1582 1582 # different right now. So we keep them separated for now for the sake of
1583 1583 # simplicity.
1584 1584
1585 1585 # we might not always want a changegroup in such bundle, for example in
1586 1586 # stream bundles
1587 1587 if opts.get('changegroup', True):
1588 1588 cgversion = opts.get('cg.version')
1589 1589 if cgversion is None:
1590 1590 cgversion = changegroup.safeversion(repo)
1591 1591 cg = changegroup.makechangegroup(repo, outgoing, cgversion, source)
1592 1592 part = bundler.newpart('changegroup', data=cg.getchunks())
1593 1593 part.addparam('version', cg.version)
1594 1594 if 'clcount' in cg.extras:
1595 1595 part.addparam('nbchanges', '%d' % cg.extras['clcount'],
1596 1596 mandatory=False)
1597 1597 if opts.get('phases') and repo.revs('%ln and secret()',
1598 1598 outgoing.missingheads):
1599 1599 part.addparam('targetphase', '%d' % phases.secret, mandatory=False)
1600 1600
1601 1601 if opts.get('streamv2', False):
1602 1602 addpartbundlestream2(bundler, repo, stream=True)
1603 1603
1604 1604 if opts.get('tagsfnodescache', True):
1605 1605 addparttagsfnodescache(repo, bundler, outgoing)
1606 1606
1607 1607 if opts.get('revbranchcache', True):
1608 1608 addpartrevbranchcache(repo, bundler, outgoing)
1609 1609
1610 1610 if opts.get('obsolescence', False):
1611 1611 obsmarkers = repo.obsstore.relevantmarkers(outgoing.missing)
1612 1612 buildobsmarkerspart(bundler, obsmarkers)
1613 1613
1614 1614 if opts.get('phases', False):
1615 1615 headsbyphase = phases.subsetphaseheads(repo, outgoing.missing)
1616 1616 phasedata = phases.binaryencode(headsbyphase)
1617 1617 bundler.newpart('phase-heads', data=phasedata)
1618 1618
1619 1619 def addparttagsfnodescache(repo, bundler, outgoing):
1620 1620 # we include the tags fnode cache for the bundle changeset
1621 1621 # (as an optional parts)
1622 1622 cache = tags.hgtagsfnodescache(repo.unfiltered())
1623 1623 chunks = []
1624 1624
1625 1625 # .hgtags fnodes are only relevant for head changesets. While we could
1626 1626 # transfer values for all known nodes, there will likely be little to
1627 1627 # no benefit.
1628 1628 #
1629 1629 # We don't bother using a generator to produce output data because
1630 1630 # a) we only have 40 bytes per head and even esoteric numbers of heads
1631 1631 # consume little memory (1M heads is 40MB) b) we don't want to send the
1632 1632 # part if we don't have entries and knowing if we have entries requires
1633 1633 # cache lookups.
1634 1634 for node in outgoing.missingheads:
1635 1635 # Don't compute missing, as this may slow down serving.
1636 1636 fnode = cache.getfnode(node, computemissing=False)
1637 1637 if fnode is not None:
1638 1638 chunks.extend([node, fnode])
1639 1639
1640 1640 if chunks:
1641 1641 bundler.newpart('hgtagsfnodes', data=''.join(chunks))
1642 1642
1643 1643 def addpartrevbranchcache(repo, bundler, outgoing):
1644 1644 # we include the rev branch cache for the bundle changeset
1645 1645 # (as an optional parts)
1646 1646 cache = repo.revbranchcache()
1647 1647 cl = repo.unfiltered().changelog
1648 1648 branchesdata = collections.defaultdict(lambda: (set(), set()))
1649 1649 for node in outgoing.missing:
1650 1650 branch, close = cache.branchinfo(cl.rev(node))
1651 1651 branchesdata[branch][close].add(node)
1652 1652
1653 1653 def generate():
1654 1654 for branch, (nodes, closed) in sorted(branchesdata.items()):
1655 1655 utf8branch = encoding.fromlocal(branch)
1656 1656 yield rbcstruct.pack(len(utf8branch), len(nodes), len(closed))
1657 1657 yield utf8branch
1658 1658 for n in sorted(nodes):
1659 1659 yield n
1660 1660 for n in sorted(closed):
1661 1661 yield n
1662 1662
1663 1663 bundler.newpart('cache:rev-branch-cache', data=generate(),
1664 1664 mandatory=False)
1665 1665
1666 1666 def _formatrequirementsspec(requirements):
1667 1667 return urlreq.quote(','.join(sorted(requirements)))
1668 1668
1669 1669 def _formatrequirementsparams(requirements):
1670 1670 requirements = _formatrequirementsspec(requirements)
1671 1671 params = "%s%s" % (urlreq.quote("requirements="), requirements)
1672 1672 return params
1673 1673
1674 1674 def addpartbundlestream2(bundler, repo, **kwargs):
1675 if not kwargs.get('stream', False):
1675 if not kwargs.get(r'stream', False):
1676 1676 return
1677 1677
1678 1678 if not streamclone.allowservergeneration(repo):
1679 1679 raise error.Abort(_('stream data requested but server does not allow '
1680 1680 'this feature'),
1681 1681 hint=_('well-behaved clients should not be '
1682 1682 'requesting stream data from servers not '
1683 1683 'advertising it; the client may be buggy'))
1684 1684
1685 1685 # Stream clones don't compress well. And compression undermines a
1686 1686 # goal of stream clones, which is to be fast. Communicate the desire
1687 1687 # to avoid compression to consumers of the bundle.
1688 1688 bundler.prefercompressed = False
1689 1689
1690 1690 filecount, bytecount, it = streamclone.generatev2(repo)
1691 1691 requirements = _formatrequirementsspec(repo.requirements)
1692 1692 part = bundler.newpart('stream2', data=it)
1693 1693 part.addparam('bytecount', '%d' % bytecount, mandatory=True)
1694 1694 part.addparam('filecount', '%d' % filecount, mandatory=True)
1695 1695 part.addparam('requirements', requirements, mandatory=True)
1696 1696
1697 1697 def buildobsmarkerspart(bundler, markers):
1698 1698 """add an obsmarker part to the bundler with <markers>
1699 1699
1700 1700 No part is created if markers is empty.
1701 1701 Raises ValueError if the bundler doesn't support any known obsmarker format.
1702 1702 """
1703 1703 if not markers:
1704 1704 return None
1705 1705
1706 1706 remoteversions = obsmarkersversion(bundler.capabilities)
1707 1707 version = obsolete.commonversion(remoteversions)
1708 1708 if version is None:
1709 1709 raise ValueError('bundler does not support common obsmarker format')
1710 1710 stream = obsolete.encodemarkers(markers, True, version=version)
1711 1711 return bundler.newpart('obsmarkers', data=stream)
1712 1712
1713 1713 def writebundle(ui, cg, filename, bundletype, vfs=None, compression=None,
1714 1714 compopts=None):
1715 1715 """Write a bundle file and return its filename.
1716 1716
1717 1717 Existing files will not be overwritten.
1718 1718 If no filename is specified, a temporary file is created.
1719 1719 bz2 compression can be turned off.
1720 1720 The bundle file will be deleted in case of errors.
1721 1721 """
1722 1722
1723 1723 if bundletype == "HG20":
1724 1724 bundle = bundle20(ui)
1725 1725 bundle.setcompression(compression, compopts)
1726 1726 part = bundle.newpart('changegroup', data=cg.getchunks())
1727 1727 part.addparam('version', cg.version)
1728 1728 if 'clcount' in cg.extras:
1729 1729 part.addparam('nbchanges', '%d' % cg.extras['clcount'],
1730 1730 mandatory=False)
1731 1731 chunkiter = bundle.getchunks()
1732 1732 else:
1733 1733 # compression argument is only for the bundle2 case
1734 1734 assert compression is None
1735 1735 if cg.version != '01':
1736 1736 raise error.Abort(_('old bundle types only supports v1 '
1737 1737 'changegroups'))
1738 1738 header, comp = bundletypes[bundletype]
1739 1739 if comp not in util.compengines.supportedbundletypes:
1740 1740 raise error.Abort(_('unknown stream compression type: %s')
1741 1741 % comp)
1742 1742 compengine = util.compengines.forbundletype(comp)
1743 1743 def chunkiter():
1744 1744 yield header
1745 1745 for chunk in compengine.compressstream(cg.getchunks(), compopts):
1746 1746 yield chunk
1747 1747 chunkiter = chunkiter()
1748 1748
1749 1749 # parse the changegroup data, otherwise we will block
1750 1750 # in case of sshrepo because we don't know the end of the stream
1751 1751 return changegroup.writechunks(ui, chunkiter, filename, vfs=vfs)
1752 1752
1753 1753 def combinechangegroupresults(op):
1754 1754 """logic to combine 0 or more addchangegroup results into one"""
1755 1755 results = [r.get('return', 0)
1756 1756 for r in op.records['changegroup']]
1757 1757 changedheads = 0
1758 1758 result = 1
1759 1759 for ret in results:
1760 1760 # If any changegroup result is 0, return 0
1761 1761 if ret == 0:
1762 1762 result = 0
1763 1763 break
1764 1764 if ret < -1:
1765 1765 changedheads += ret + 1
1766 1766 elif ret > 1:
1767 1767 changedheads += ret - 1
1768 1768 if changedheads > 0:
1769 1769 result = 1 + changedheads
1770 1770 elif changedheads < 0:
1771 1771 result = -1 + changedheads
1772 1772 return result
1773 1773
1774 1774 @parthandler('changegroup', ('version', 'nbchanges', 'treemanifest',
1775 1775 'targetphase'))
1776 1776 def handlechangegroup(op, inpart):
1777 1777 """apply a changegroup part on the repo
1778 1778
1779 1779 This is a very early implementation that will massive rework before being
1780 1780 inflicted to any end-user.
1781 1781 """
1782 1782 from . import localrepo
1783 1783
1784 1784 tr = op.gettransaction()
1785 1785 unpackerversion = inpart.params.get('version', '01')
1786 1786 # We should raise an appropriate exception here
1787 1787 cg = changegroup.getunbundler(unpackerversion, inpart, None)
1788 1788 # the source and url passed here are overwritten by the one contained in
1789 1789 # the transaction.hookargs argument. So 'bundle2' is a placeholder
1790 1790 nbchangesets = None
1791 1791 if 'nbchanges' in inpart.params:
1792 1792 nbchangesets = int(inpart.params.get('nbchanges'))
1793 1793 if ('treemanifest' in inpart.params and
1794 1794 'treemanifest' not in op.repo.requirements):
1795 1795 if len(op.repo.changelog) != 0:
1796 1796 raise error.Abort(_(
1797 1797 "bundle contains tree manifests, but local repo is "
1798 1798 "non-empty and does not use tree manifests"))
1799 1799 op.repo.requirements.add('treemanifest')
1800 1800 op.repo.svfs.options = localrepo.resolvestorevfsoptions(
1801 1801 op.repo.ui, op.repo.requirements)
1802 1802 op.repo._writerequirements()
1803 1803 extrakwargs = {}
1804 1804 targetphase = inpart.params.get('targetphase')
1805 1805 if targetphase is not None:
1806 1806 extrakwargs[r'targetphase'] = int(targetphase)
1807 1807 ret = _processchangegroup(op, cg, tr, 'bundle2', 'bundle2',
1808 1808 expectedtotal=nbchangesets, **extrakwargs)
1809 1809 if op.reply is not None:
1810 1810 # This is definitely not the final form of this
1811 1811 # return. But one need to start somewhere.
1812 1812 part = op.reply.newpart('reply:changegroup', mandatory=False)
1813 1813 part.addparam(
1814 1814 'in-reply-to', pycompat.bytestr(inpart.id), mandatory=False)
1815 1815 part.addparam('return', '%i' % ret, mandatory=False)
1816 1816 assert not inpart.read()
1817 1817
1818 1818 _remotechangegroupparams = tuple(['url', 'size', 'digests'] +
1819 1819 ['digest:%s' % k for k in util.DIGESTS.keys()])
1820 1820 @parthandler('remote-changegroup', _remotechangegroupparams)
1821 1821 def handleremotechangegroup(op, inpart):
1822 1822 """apply a bundle10 on the repo, given an url and validation information
1823 1823
1824 1824 All the information about the remote bundle to import are given as
1825 1825 parameters. The parameters include:
1826 1826 - url: the url to the bundle10.
1827 1827 - size: the bundle10 file size. It is used to validate what was
1828 1828 retrieved by the client matches the server knowledge about the bundle.
1829 1829 - digests: a space separated list of the digest types provided as
1830 1830 parameters.
1831 1831 - digest:<digest-type>: the hexadecimal representation of the digest with
1832 1832 that name. Like the size, it is used to validate what was retrieved by
1833 1833 the client matches what the server knows about the bundle.
1834 1834
1835 1835 When multiple digest types are given, all of them are checked.
1836 1836 """
1837 1837 try:
1838 1838 raw_url = inpart.params['url']
1839 1839 except KeyError:
1840 1840 raise error.Abort(_('remote-changegroup: missing "%s" param') % 'url')
1841 1841 parsed_url = util.url(raw_url)
1842 1842 if parsed_url.scheme not in capabilities['remote-changegroup']:
1843 1843 raise error.Abort(_('remote-changegroup does not support %s urls') %
1844 1844 parsed_url.scheme)
1845 1845
1846 1846 try:
1847 1847 size = int(inpart.params['size'])
1848 1848 except ValueError:
1849 1849 raise error.Abort(_('remote-changegroup: invalid value for param "%s"')
1850 1850 % 'size')
1851 1851 except KeyError:
1852 1852 raise error.Abort(_('remote-changegroup: missing "%s" param') % 'size')
1853 1853
1854 1854 digests = {}
1855 1855 for typ in inpart.params.get('digests', '').split():
1856 1856 param = 'digest:%s' % typ
1857 1857 try:
1858 1858 value = inpart.params[param]
1859 1859 except KeyError:
1860 1860 raise error.Abort(_('remote-changegroup: missing "%s" param') %
1861 1861 param)
1862 1862 digests[typ] = value
1863 1863
1864 1864 real_part = util.digestchecker(url.open(op.ui, raw_url), size, digests)
1865 1865
1866 1866 tr = op.gettransaction()
1867 1867 from . import exchange
1868 1868 cg = exchange.readbundle(op.repo.ui, real_part, raw_url)
1869 1869 if not isinstance(cg, changegroup.cg1unpacker):
1870 1870 raise error.Abort(_('%s: not a bundle version 1.0') %
1871 1871 util.hidepassword(raw_url))
1872 1872 ret = _processchangegroup(op, cg, tr, 'bundle2', 'bundle2')
1873 1873 if op.reply is not None:
1874 1874 # This is definitely not the final form of this
1875 1875 # return. But one need to start somewhere.
1876 1876 part = op.reply.newpart('reply:changegroup')
1877 1877 part.addparam(
1878 1878 'in-reply-to', pycompat.bytestr(inpart.id), mandatory=False)
1879 1879 part.addparam('return', '%i' % ret, mandatory=False)
1880 1880 try:
1881 1881 real_part.validate()
1882 1882 except error.Abort as e:
1883 1883 raise error.Abort(_('bundle at %s is corrupted:\n%s') %
1884 1884 (util.hidepassword(raw_url), bytes(e)))
1885 1885 assert not inpart.read()
1886 1886
1887 1887 @parthandler('reply:changegroup', ('return', 'in-reply-to'))
1888 1888 def handlereplychangegroup(op, inpart):
1889 1889 ret = int(inpart.params['return'])
1890 1890 replyto = int(inpart.params['in-reply-to'])
1891 1891 op.records.add('changegroup', {'return': ret}, replyto)
1892 1892
1893 1893 @parthandler('check:bookmarks')
1894 1894 def handlecheckbookmarks(op, inpart):
1895 1895 """check location of bookmarks
1896 1896
1897 1897 This part is to be used to detect push race regarding bookmark, it
1898 1898 contains binary encoded (bookmark, node) tuple. If the local state does
1899 1899 not marks the one in the part, a PushRaced exception is raised
1900 1900 """
1901 1901 bookdata = bookmarks.binarydecode(inpart)
1902 1902
1903 1903 msgstandard = ('repository changed while pushing - please try again '
1904 1904 '(bookmark "%s" move from %s to %s)')
1905 1905 msgmissing = ('repository changed while pushing - please try again '
1906 1906 '(bookmark "%s" is missing, expected %s)')
1907 1907 msgexist = ('repository changed while pushing - please try again '
1908 1908 '(bookmark "%s" set on %s, expected missing)')
1909 1909 for book, node in bookdata:
1910 1910 currentnode = op.repo._bookmarks.get(book)
1911 1911 if currentnode != node:
1912 1912 if node is None:
1913 1913 finalmsg = msgexist % (book, nodemod.short(currentnode))
1914 1914 elif currentnode is None:
1915 1915 finalmsg = msgmissing % (book, nodemod.short(node))
1916 1916 else:
1917 1917 finalmsg = msgstandard % (book, nodemod.short(node),
1918 1918 nodemod.short(currentnode))
1919 1919 raise error.PushRaced(finalmsg)
1920 1920
1921 1921 @parthandler('check:heads')
1922 1922 def handlecheckheads(op, inpart):
1923 1923 """check that head of the repo did not change
1924 1924
1925 1925 This is used to detect a push race when using unbundle.
1926 1926 This replaces the "heads" argument of unbundle."""
1927 1927 h = inpart.read(20)
1928 1928 heads = []
1929 1929 while len(h) == 20:
1930 1930 heads.append(h)
1931 1931 h = inpart.read(20)
1932 1932 assert not h
1933 1933 # Trigger a transaction so that we are guaranteed to have the lock now.
1934 1934 if op.ui.configbool('experimental', 'bundle2lazylocking'):
1935 1935 op.gettransaction()
1936 1936 if sorted(heads) != sorted(op.repo.heads()):
1937 1937 raise error.PushRaced('repository changed while pushing - '
1938 1938 'please try again')
1939 1939
1940 1940 @parthandler('check:updated-heads')
1941 1941 def handlecheckupdatedheads(op, inpart):
1942 1942 """check for race on the heads touched by a push
1943 1943
1944 1944 This is similar to 'check:heads' but focus on the heads actually updated
1945 1945 during the push. If other activities happen on unrelated heads, it is
1946 1946 ignored.
1947 1947
1948 1948 This allow server with high traffic to avoid push contention as long as
1949 1949 unrelated parts of the graph are involved."""
1950 1950 h = inpart.read(20)
1951 1951 heads = []
1952 1952 while len(h) == 20:
1953 1953 heads.append(h)
1954 1954 h = inpart.read(20)
1955 1955 assert not h
1956 1956 # trigger a transaction so that we are guaranteed to have the lock now.
1957 1957 if op.ui.configbool('experimental', 'bundle2lazylocking'):
1958 1958 op.gettransaction()
1959 1959
1960 1960 currentheads = set()
1961 1961 for ls in op.repo.branchmap().itervalues():
1962 1962 currentheads.update(ls)
1963 1963
1964 1964 for h in heads:
1965 1965 if h not in currentheads:
1966 1966 raise error.PushRaced('repository changed while pushing - '
1967 1967 'please try again')
1968 1968
1969 1969 @parthandler('check:phases')
1970 1970 def handlecheckphases(op, inpart):
1971 1971 """check that phase boundaries of the repository did not change
1972 1972
1973 1973 This is used to detect a push race.
1974 1974 """
1975 1975 phasetonodes = phases.binarydecode(inpart)
1976 1976 unfi = op.repo.unfiltered()
1977 1977 cl = unfi.changelog
1978 1978 phasecache = unfi._phasecache
1979 1979 msg = ('repository changed while pushing - please try again '
1980 1980 '(%s is %s expected %s)')
1981 1981 for expectedphase, nodes in enumerate(phasetonodes):
1982 1982 for n in nodes:
1983 1983 actualphase = phasecache.phase(unfi, cl.rev(n))
1984 1984 if actualphase != expectedphase:
1985 1985 finalmsg = msg % (nodemod.short(n),
1986 1986 phases.phasenames[actualphase],
1987 1987 phases.phasenames[expectedphase])
1988 1988 raise error.PushRaced(finalmsg)
1989 1989
1990 1990 @parthandler('output')
1991 1991 def handleoutput(op, inpart):
1992 1992 """forward output captured on the server to the client"""
1993 1993 for line in inpart.read().splitlines():
1994 1994 op.ui.status(_('remote: %s\n') % line)
1995 1995
1996 1996 @parthandler('replycaps')
1997 1997 def handlereplycaps(op, inpart):
1998 1998 """Notify that a reply bundle should be created
1999 1999
2000 2000 The payload contains the capabilities information for the reply"""
2001 2001 caps = decodecaps(inpart.read())
2002 2002 if op.reply is None:
2003 2003 op.reply = bundle20(op.ui, caps)
2004 2004
2005 2005 class AbortFromPart(error.Abort):
2006 2006 """Sub-class of Abort that denotes an error from a bundle2 part."""
2007 2007
2008 2008 @parthandler('error:abort', ('message', 'hint'))
2009 2009 def handleerrorabort(op, inpart):
2010 2010 """Used to transmit abort error over the wire"""
2011 2011 raise AbortFromPart(inpart.params['message'],
2012 2012 hint=inpart.params.get('hint'))
2013 2013
2014 2014 @parthandler('error:pushkey', ('namespace', 'key', 'new', 'old', 'ret',
2015 2015 'in-reply-to'))
2016 2016 def handleerrorpushkey(op, inpart):
2017 2017 """Used to transmit failure of a mandatory pushkey over the wire"""
2018 2018 kwargs = {}
2019 2019 for name in ('namespace', 'key', 'new', 'old', 'ret'):
2020 2020 value = inpart.params.get(name)
2021 2021 if value is not None:
2022 2022 kwargs[name] = value
2023 2023 raise error.PushkeyFailed(inpart.params['in-reply-to'],
2024 2024 **pycompat.strkwargs(kwargs))
2025 2025
2026 2026 @parthandler('error:unsupportedcontent', ('parttype', 'params'))
2027 2027 def handleerrorunsupportedcontent(op, inpart):
2028 2028 """Used to transmit unknown content error over the wire"""
2029 2029 kwargs = {}
2030 2030 parttype = inpart.params.get('parttype')
2031 2031 if parttype is not None:
2032 2032 kwargs['parttype'] = parttype
2033 2033 params = inpart.params.get('params')
2034 2034 if params is not None:
2035 2035 kwargs['params'] = params.split('\0')
2036 2036
2037 2037 raise error.BundleUnknownFeatureError(**pycompat.strkwargs(kwargs))
2038 2038
2039 2039 @parthandler('error:pushraced', ('message',))
2040 2040 def handleerrorpushraced(op, inpart):
2041 2041 """Used to transmit push race error over the wire"""
2042 2042 raise error.ResponseError(_('push failed:'), inpart.params['message'])
2043 2043
2044 2044 @parthandler('listkeys', ('namespace',))
2045 2045 def handlelistkeys(op, inpart):
2046 2046 """retrieve pushkey namespace content stored in a bundle2"""
2047 2047 namespace = inpart.params['namespace']
2048 2048 r = pushkey.decodekeys(inpart.read())
2049 2049 op.records.add('listkeys', (namespace, r))
2050 2050
2051 2051 @parthandler('pushkey', ('namespace', 'key', 'old', 'new'))
2052 2052 def handlepushkey(op, inpart):
2053 2053 """process a pushkey request"""
2054 2054 dec = pushkey.decode
2055 2055 namespace = dec(inpart.params['namespace'])
2056 2056 key = dec(inpart.params['key'])
2057 2057 old = dec(inpart.params['old'])
2058 2058 new = dec(inpart.params['new'])
2059 2059 # Grab the transaction to ensure that we have the lock before performing the
2060 2060 # pushkey.
2061 2061 if op.ui.configbool('experimental', 'bundle2lazylocking'):
2062 2062 op.gettransaction()
2063 2063 ret = op.repo.pushkey(namespace, key, old, new)
2064 2064 record = {'namespace': namespace,
2065 2065 'key': key,
2066 2066 'old': old,
2067 2067 'new': new}
2068 2068 op.records.add('pushkey', record)
2069 2069 if op.reply is not None:
2070 2070 rpart = op.reply.newpart('reply:pushkey')
2071 2071 rpart.addparam(
2072 2072 'in-reply-to', pycompat.bytestr(inpart.id), mandatory=False)
2073 2073 rpart.addparam('return', '%i' % ret, mandatory=False)
2074 2074 if inpart.mandatory and not ret:
2075 2075 kwargs = {}
2076 2076 for key in ('namespace', 'key', 'new', 'old', 'ret'):
2077 2077 if key in inpart.params:
2078 2078 kwargs[key] = inpart.params[key]
2079 2079 raise error.PushkeyFailed(partid='%d' % inpart.id,
2080 2080 **pycompat.strkwargs(kwargs))
2081 2081
2082 2082 @parthandler('bookmarks')
2083 2083 def handlebookmark(op, inpart):
2084 2084 """transmit bookmark information
2085 2085
2086 2086 The part contains binary encoded bookmark information.
2087 2087
2088 2088 The exact behavior of this part can be controlled by the 'bookmarks' mode
2089 2089 on the bundle operation.
2090 2090
2091 2091 When mode is 'apply' (the default) the bookmark information is applied as
2092 2092 is to the unbundling repository. Make sure a 'check:bookmarks' part is
2093 2093 issued earlier to check for push races in such update. This behavior is
2094 2094 suitable for pushing.
2095 2095
2096 2096 When mode is 'records', the information is recorded into the 'bookmarks'
2097 2097 records of the bundle operation. This behavior is suitable for pulling.
2098 2098 """
2099 2099 changes = bookmarks.binarydecode(inpart)
2100 2100
2101 2101 pushkeycompat = op.repo.ui.configbool('server', 'bookmarks-pushkey-compat')
2102 2102 bookmarksmode = op.modes.get('bookmarks', 'apply')
2103 2103
2104 2104 if bookmarksmode == 'apply':
2105 2105 tr = op.gettransaction()
2106 2106 bookstore = op.repo._bookmarks
2107 2107 if pushkeycompat:
2108 2108 allhooks = []
2109 2109 for book, node in changes:
2110 2110 hookargs = tr.hookargs.copy()
2111 2111 hookargs['pushkeycompat'] = '1'
2112 2112 hookargs['namespace'] = 'bookmarks'
2113 2113 hookargs['key'] = book
2114 2114 hookargs['old'] = nodemod.hex(bookstore.get(book, ''))
2115 2115 hookargs['new'] = nodemod.hex(node if node is not None else '')
2116 2116 allhooks.append(hookargs)
2117 2117
2118 2118 for hookargs in allhooks:
2119 2119 op.repo.hook('prepushkey', throw=True,
2120 2120 **pycompat.strkwargs(hookargs))
2121 2121
2122 2122 bookstore.applychanges(op.repo, op.gettransaction(), changes)
2123 2123
2124 2124 if pushkeycompat:
2125 2125 def runhook():
2126 2126 for hookargs in allhooks:
2127 2127 op.repo.hook('pushkey', **pycompat.strkwargs(hookargs))
2128 2128 op.repo._afterlock(runhook)
2129 2129
2130 2130 elif bookmarksmode == 'records':
2131 2131 for book, node in changes:
2132 2132 record = {'bookmark': book, 'node': node}
2133 2133 op.records.add('bookmarks', record)
2134 2134 else:
2135 2135 raise error.ProgrammingError('unkown bookmark mode: %s' % bookmarksmode)
2136 2136
2137 2137 @parthandler('phase-heads')
2138 2138 def handlephases(op, inpart):
2139 2139 """apply phases from bundle part to repo"""
2140 2140 headsbyphase = phases.binarydecode(inpart)
2141 2141 phases.updatephases(op.repo.unfiltered(), op.gettransaction, headsbyphase)
2142 2142
2143 2143 @parthandler('reply:pushkey', ('return', 'in-reply-to'))
2144 2144 def handlepushkeyreply(op, inpart):
2145 2145 """retrieve the result of a pushkey request"""
2146 2146 ret = int(inpart.params['return'])
2147 2147 partid = int(inpart.params['in-reply-to'])
2148 2148 op.records.add('pushkey', {'return': ret}, partid)
2149 2149
2150 2150 @parthandler('obsmarkers')
2151 2151 def handleobsmarker(op, inpart):
2152 2152 """add a stream of obsmarkers to the repo"""
2153 2153 tr = op.gettransaction()
2154 2154 markerdata = inpart.read()
2155 2155 if op.ui.config('experimental', 'obsmarkers-exchange-debug'):
2156 2156 op.ui.write(('obsmarker-exchange: %i bytes received\n')
2157 2157 % len(markerdata))
2158 2158 # The mergemarkers call will crash if marker creation is not enabled.
2159 2159 # we want to avoid this if the part is advisory.
2160 2160 if not inpart.mandatory and op.repo.obsstore.readonly:
2161 2161 op.repo.ui.debug('ignoring obsolescence markers, feature not enabled\n')
2162 2162 return
2163 2163 new = op.repo.obsstore.mergemarkers(tr, markerdata)
2164 2164 op.repo.invalidatevolatilesets()
2165 2165 if new:
2166 2166 op.repo.ui.status(_('%i new obsolescence markers\n') % new)
2167 2167 op.records.add('obsmarkers', {'new': new})
2168 2168 if op.reply is not None:
2169 2169 rpart = op.reply.newpart('reply:obsmarkers')
2170 2170 rpart.addparam(
2171 2171 'in-reply-to', pycompat.bytestr(inpart.id), mandatory=False)
2172 2172 rpart.addparam('new', '%i' % new, mandatory=False)
2173 2173
2174 2174
2175 2175 @parthandler('reply:obsmarkers', ('new', 'in-reply-to'))
2176 2176 def handleobsmarkerreply(op, inpart):
2177 2177 """retrieve the result of a pushkey request"""
2178 2178 ret = int(inpart.params['new'])
2179 2179 partid = int(inpart.params['in-reply-to'])
2180 2180 op.records.add('obsmarkers', {'new': ret}, partid)
2181 2181
2182 2182 @parthandler('hgtagsfnodes')
2183 2183 def handlehgtagsfnodes(op, inpart):
2184 2184 """Applies .hgtags fnodes cache entries to the local repo.
2185 2185
2186 2186 Payload is pairs of 20 byte changeset nodes and filenodes.
2187 2187 """
2188 2188 # Grab the transaction so we ensure that we have the lock at this point.
2189 2189 if op.ui.configbool('experimental', 'bundle2lazylocking'):
2190 2190 op.gettransaction()
2191 2191 cache = tags.hgtagsfnodescache(op.repo.unfiltered())
2192 2192
2193 2193 count = 0
2194 2194 while True:
2195 2195 node = inpart.read(20)
2196 2196 fnode = inpart.read(20)
2197 2197 if len(node) < 20 or len(fnode) < 20:
2198 2198 op.ui.debug('ignoring incomplete received .hgtags fnodes data\n')
2199 2199 break
2200 2200 cache.setfnode(node, fnode)
2201 2201 count += 1
2202 2202
2203 2203 cache.write()
2204 2204 op.ui.debug('applied %i hgtags fnodes cache entries\n' % count)
2205 2205
2206 2206 rbcstruct = struct.Struct('>III')
2207 2207
2208 2208 @parthandler('cache:rev-branch-cache')
2209 2209 def handlerbc(op, inpart):
2210 2210 """receive a rev-branch-cache payload and update the local cache
2211 2211
2212 2212 The payload is a series of data related to each branch
2213 2213
2214 2214 1) branch name length
2215 2215 2) number of open heads
2216 2216 3) number of closed heads
2217 2217 4) open heads nodes
2218 2218 5) closed heads nodes
2219 2219 """
2220 2220 total = 0
2221 2221 rawheader = inpart.read(rbcstruct.size)
2222 2222 cache = op.repo.revbranchcache()
2223 2223 cl = op.repo.unfiltered().changelog
2224 2224 while rawheader:
2225 2225 header = rbcstruct.unpack(rawheader)
2226 2226 total += header[1] + header[2]
2227 2227 utf8branch = inpart.read(header[0])
2228 2228 branch = encoding.tolocal(utf8branch)
2229 2229 for x in pycompat.xrange(header[1]):
2230 2230 node = inpart.read(20)
2231 2231 rev = cl.rev(node)
2232 2232 cache.setdata(branch, rev, node, False)
2233 2233 for x in pycompat.xrange(header[2]):
2234 2234 node = inpart.read(20)
2235 2235 rev = cl.rev(node)
2236 2236 cache.setdata(branch, rev, node, True)
2237 2237 rawheader = inpart.read(rbcstruct.size)
2238 2238 cache.write()
2239 2239
2240 2240 @parthandler('pushvars')
2241 2241 def bundle2getvars(op, part):
2242 2242 '''unbundle a bundle2 containing shellvars on the server'''
2243 2243 # An option to disable unbundling on server-side for security reasons
2244 2244 if op.ui.configbool('push', 'pushvars.server'):
2245 2245 hookargs = {}
2246 2246 for key, value in part.advisoryparams:
2247 2247 key = key.upper()
2248 2248 # We want pushed variables to have USERVAR_ prepended so we know
2249 2249 # they came from the --pushvar flag.
2250 2250 key = "USERVAR_" + key
2251 2251 hookargs[key] = value
2252 2252 op.addhookargs(hookargs)
2253 2253
2254 2254 @parthandler('stream2', ('requirements', 'filecount', 'bytecount'))
2255 2255 def handlestreamv2bundle(op, part):
2256 2256
2257 2257 requirements = urlreq.unquote(part.params['requirements']).split(',')
2258 2258 filecount = int(part.params['filecount'])
2259 2259 bytecount = int(part.params['bytecount'])
2260 2260
2261 2261 repo = op.repo
2262 2262 if len(repo):
2263 2263 msg = _('cannot apply stream clone to non empty repository')
2264 2264 raise error.Abort(msg)
2265 2265
2266 2266 repo.ui.debug('applying stream bundle\n')
2267 2267 streamclone.applybundlev2(repo, part, filecount, bytecount,
2268 2268 requirements)
General Comments 0
You need to be logged in to leave comments. Login now