##// END OF EJS Templates
bundle2: make source a mandatory argument for bundle2.applybundle() (API)...
Pulkit Goyal -
r37255:684a6a26 default
parent child Browse files
Show More
@@ -1,2263 +1,2263
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 def applybundle(repo, unbundler, tr, source=None, url=None, **kwargs):
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 ValueError(r'empty parameter name')
632 632 if name[0:1] not in pycompat.bytestr(string.ascii_letters):
633 633 raise ValueError(r'non letter first character: %s' % name)
634 634 self._params.append((name, value))
635 635
636 636 def addpart(self, part):
637 637 """add a new part to the bundle2 container
638 638
639 639 Parts contains the actual applicative payload."""
640 640 assert part.id is None
641 641 part.id = len(self._parts) # very cheap counter
642 642 self._parts.append(part)
643 643
644 644 def newpart(self, typeid, *args, **kwargs):
645 645 """create a new part and add it to the containers
646 646
647 647 As the part is directly added to the containers. For now, this means
648 648 that any failure to properly initialize the part after calling
649 649 ``newpart`` should result in a failure of the whole bundling process.
650 650
651 651 You can still fall back to manually create and add if you need better
652 652 control."""
653 653 part = bundlepart(typeid, *args, **kwargs)
654 654 self.addpart(part)
655 655 return part
656 656
657 657 # methods used to generate the bundle2 stream
658 658 def getchunks(self):
659 659 if self.ui.debugflag:
660 660 msg = ['bundle2-output-bundle: "%s",' % self._magicstring]
661 661 if self._params:
662 662 msg.append(' (%i params)' % len(self._params))
663 663 msg.append(' %i parts total\n' % len(self._parts))
664 664 self.ui.debug(''.join(msg))
665 665 outdebug(self.ui, 'start emission of %s stream' % self._magicstring)
666 666 yield self._magicstring
667 667 param = self._paramchunk()
668 668 outdebug(self.ui, 'bundle parameter: %s' % param)
669 669 yield _pack(_fstreamparamsize, len(param))
670 670 if param:
671 671 yield param
672 672 for chunk in self._compengine.compressstream(self._getcorechunk(),
673 673 self._compopts):
674 674 yield chunk
675 675
676 676 def _paramchunk(self):
677 677 """return a encoded version of all stream parameters"""
678 678 blocks = []
679 679 for par, value in self._params:
680 680 par = urlreq.quote(par)
681 681 if value is not None:
682 682 value = urlreq.quote(value)
683 683 par = '%s=%s' % (par, value)
684 684 blocks.append(par)
685 685 return ' '.join(blocks)
686 686
687 687 def _getcorechunk(self):
688 688 """yield chunk for the core part of the bundle
689 689
690 690 (all but headers and parameters)"""
691 691 outdebug(self.ui, 'start of parts')
692 692 for part in self._parts:
693 693 outdebug(self.ui, 'bundle part: "%s"' % part.type)
694 694 for chunk in part.getchunks(ui=self.ui):
695 695 yield chunk
696 696 outdebug(self.ui, 'end of bundle')
697 697 yield _pack(_fpartheadersize, 0)
698 698
699 699
700 700 def salvageoutput(self):
701 701 """return a list with a copy of all output parts in the bundle
702 702
703 703 This is meant to be used during error handling to make sure we preserve
704 704 server output"""
705 705 salvaged = []
706 706 for part in self._parts:
707 707 if part.type.startswith('output'):
708 708 salvaged.append(part.copy())
709 709 return salvaged
710 710
711 711
712 712 class unpackermixin(object):
713 713 """A mixin to extract bytes and struct data from a stream"""
714 714
715 715 def __init__(self, fp):
716 716 self._fp = fp
717 717
718 718 def _unpack(self, format):
719 719 """unpack this struct format from the stream
720 720
721 721 This method is meant for internal usage by the bundle2 protocol only.
722 722 They directly manipulate the low level stream including bundle2 level
723 723 instruction.
724 724
725 725 Do not use it to implement higher-level logic or methods."""
726 726 data = self._readexact(struct.calcsize(format))
727 727 return _unpack(format, data)
728 728
729 729 def _readexact(self, size):
730 730 """read exactly <size> bytes from the stream
731 731
732 732 This method is meant for internal usage by the bundle2 protocol only.
733 733 They directly manipulate the low level stream including bundle2 level
734 734 instruction.
735 735
736 736 Do not use it to implement higher-level logic or methods."""
737 737 return changegroup.readexactly(self._fp, size)
738 738
739 739 def getunbundler(ui, fp, magicstring=None):
740 740 """return a valid unbundler object for a given magicstring"""
741 741 if magicstring is None:
742 742 magicstring = changegroup.readexactly(fp, 4)
743 743 magic, version = magicstring[0:2], magicstring[2:4]
744 744 if magic != 'HG':
745 745 ui.debug(
746 746 "error: invalid magic: %r (version %r), should be 'HG'\n"
747 747 % (magic, version))
748 748 raise error.Abort(_('not a Mercurial bundle'))
749 749 unbundlerclass = formatmap.get(version)
750 750 if unbundlerclass is None:
751 751 raise error.Abort(_('unknown bundle version %s') % version)
752 752 unbundler = unbundlerclass(ui, fp)
753 753 indebug(ui, 'start processing of %s stream' % magicstring)
754 754 return unbundler
755 755
756 756 class unbundle20(unpackermixin):
757 757 """interpret a bundle2 stream
758 758
759 759 This class is fed with a binary stream and yields parts through its
760 760 `iterparts` methods."""
761 761
762 762 _magicstring = 'HG20'
763 763
764 764 def __init__(self, ui, fp):
765 765 """If header is specified, we do not read it out of the stream."""
766 766 self.ui = ui
767 767 self._compengine = util.compengines.forbundletype('UN')
768 768 self._compressed = None
769 769 super(unbundle20, self).__init__(fp)
770 770
771 771 @util.propertycache
772 772 def params(self):
773 773 """dictionary of stream level parameters"""
774 774 indebug(self.ui, 'reading bundle2 stream parameters')
775 775 params = {}
776 776 paramssize = self._unpack(_fstreamparamsize)[0]
777 777 if paramssize < 0:
778 778 raise error.BundleValueError('negative bundle param size: %i'
779 779 % paramssize)
780 780 if paramssize:
781 781 params = self._readexact(paramssize)
782 782 params = self._processallparams(params)
783 783 return params
784 784
785 785 def _processallparams(self, paramsblock):
786 786 """"""
787 787 params = util.sortdict()
788 788 for p in paramsblock.split(' '):
789 789 p = p.split('=', 1)
790 790 p = [urlreq.unquote(i) for i in p]
791 791 if len(p) < 2:
792 792 p.append(None)
793 793 self._processparam(*p)
794 794 params[p[0]] = p[1]
795 795 return params
796 796
797 797
798 798 def _processparam(self, name, value):
799 799 """process a parameter, applying its effect if needed
800 800
801 801 Parameter starting with a lower case letter are advisory and will be
802 802 ignored when unknown. Those starting with an upper case letter are
803 803 mandatory and will this function will raise a KeyError when unknown.
804 804
805 805 Note: no option are currently supported. Any input will be either
806 806 ignored or failing.
807 807 """
808 808 if not name:
809 809 raise ValueError(r'empty parameter name')
810 810 if name[0:1] not in pycompat.bytestr(string.ascii_letters):
811 811 raise ValueError(r'non letter first character: %s' % name)
812 812 try:
813 813 handler = b2streamparamsmap[name.lower()]
814 814 except KeyError:
815 815 if name[0:1].islower():
816 816 indebug(self.ui, "ignoring unknown parameter %s" % name)
817 817 else:
818 818 raise error.BundleUnknownFeatureError(params=(name,))
819 819 else:
820 820 handler(self, name, value)
821 821
822 822 def _forwardchunks(self):
823 823 """utility to transfer a bundle2 as binary
824 824
825 825 This is made necessary by the fact the 'getbundle' command over 'ssh'
826 826 have no way to know then the reply end, relying on the bundle to be
827 827 interpreted to know its end. This is terrible and we are sorry, but we
828 828 needed to move forward to get general delta enabled.
829 829 """
830 830 yield self._magicstring
831 831 assert 'params' not in vars(self)
832 832 paramssize = self._unpack(_fstreamparamsize)[0]
833 833 if paramssize < 0:
834 834 raise error.BundleValueError('negative bundle param size: %i'
835 835 % paramssize)
836 836 yield _pack(_fstreamparamsize, paramssize)
837 837 if paramssize:
838 838 params = self._readexact(paramssize)
839 839 self._processallparams(params)
840 840 yield params
841 841 assert self._compengine.bundletype == 'UN'
842 842 # From there, payload might need to be decompressed
843 843 self._fp = self._compengine.decompressorreader(self._fp)
844 844 emptycount = 0
845 845 while emptycount < 2:
846 846 # so we can brainlessly loop
847 847 assert _fpartheadersize == _fpayloadsize
848 848 size = self._unpack(_fpartheadersize)[0]
849 849 yield _pack(_fpartheadersize, size)
850 850 if size:
851 851 emptycount = 0
852 852 else:
853 853 emptycount += 1
854 854 continue
855 855 if size == flaginterrupt:
856 856 continue
857 857 elif size < 0:
858 858 raise error.BundleValueError('negative chunk size: %i')
859 859 yield self._readexact(size)
860 860
861 861
862 862 def iterparts(self, seekable=False):
863 863 """yield all parts contained in the stream"""
864 864 cls = seekableunbundlepart if seekable else unbundlepart
865 865 # make sure param have been loaded
866 866 self.params
867 867 # From there, payload need to be decompressed
868 868 self._fp = self._compengine.decompressorreader(self._fp)
869 869 indebug(self.ui, 'start extraction of bundle2 parts')
870 870 headerblock = self._readpartheader()
871 871 while headerblock is not None:
872 872 part = cls(self.ui, headerblock, self._fp)
873 873 yield part
874 874 # Ensure part is fully consumed so we can start reading the next
875 875 # part.
876 876 part.consume()
877 877
878 878 headerblock = self._readpartheader()
879 879 indebug(self.ui, 'end of bundle2 stream')
880 880
881 881 def _readpartheader(self):
882 882 """reads a part header size and return the bytes blob
883 883
884 884 returns None if empty"""
885 885 headersize = self._unpack(_fpartheadersize)[0]
886 886 if headersize < 0:
887 887 raise error.BundleValueError('negative part header size: %i'
888 888 % headersize)
889 889 indebug(self.ui, 'part header size: %i' % headersize)
890 890 if headersize:
891 891 return self._readexact(headersize)
892 892 return None
893 893
894 894 def compressed(self):
895 895 self.params # load params
896 896 return self._compressed
897 897
898 898 def close(self):
899 899 """close underlying file"""
900 900 if util.safehasattr(self._fp, 'close'):
901 901 return self._fp.close()
902 902
903 903 formatmap = {'20': unbundle20}
904 904
905 905 b2streamparamsmap = {}
906 906
907 907 def b2streamparamhandler(name):
908 908 """register a handler for a stream level parameter"""
909 909 def decorator(func):
910 910 assert name not in formatmap
911 911 b2streamparamsmap[name] = func
912 912 return func
913 913 return decorator
914 914
915 915 @b2streamparamhandler('compression')
916 916 def processcompression(unbundler, param, value):
917 917 """read compression parameter and install payload decompression"""
918 918 if value not in util.compengines.supportedbundletypes:
919 919 raise error.BundleUnknownFeatureError(params=(param,),
920 920 values=(value,))
921 921 unbundler._compengine = util.compengines.forbundletype(value)
922 922 if value is not None:
923 923 unbundler._compressed = True
924 924
925 925 class bundlepart(object):
926 926 """A bundle2 part contains application level payload
927 927
928 928 The part `type` is used to route the part to the application level
929 929 handler.
930 930
931 931 The part payload is contained in ``part.data``. It could be raw bytes or a
932 932 generator of byte chunks.
933 933
934 934 You can add parameters to the part using the ``addparam`` method.
935 935 Parameters can be either mandatory (default) or advisory. Remote side
936 936 should be able to safely ignore the advisory ones.
937 937
938 938 Both data and parameters cannot be modified after the generation has begun.
939 939 """
940 940
941 941 def __init__(self, parttype, mandatoryparams=(), advisoryparams=(),
942 942 data='', mandatory=True):
943 943 validateparttype(parttype)
944 944 self.id = None
945 945 self.type = parttype
946 946 self._data = data
947 947 self._mandatoryparams = list(mandatoryparams)
948 948 self._advisoryparams = list(advisoryparams)
949 949 # checking for duplicated entries
950 950 self._seenparams = set()
951 951 for pname, __ in self._mandatoryparams + self._advisoryparams:
952 952 if pname in self._seenparams:
953 953 raise error.ProgrammingError('duplicated params: %s' % pname)
954 954 self._seenparams.add(pname)
955 955 # status of the part's generation:
956 956 # - None: not started,
957 957 # - False: currently generated,
958 958 # - True: generation done.
959 959 self._generated = None
960 960 self.mandatory = mandatory
961 961
962 962 def __repr__(self):
963 963 cls = "%s.%s" % (self.__class__.__module__, self.__class__.__name__)
964 964 return ('<%s object at %x; id: %s; type: %s; mandatory: %s>'
965 965 % (cls, id(self), self.id, self.type, self.mandatory))
966 966
967 967 def copy(self):
968 968 """return a copy of the part
969 969
970 970 The new part have the very same content but no partid assigned yet.
971 971 Parts with generated data cannot be copied."""
972 972 assert not util.safehasattr(self.data, 'next')
973 973 return self.__class__(self.type, self._mandatoryparams,
974 974 self._advisoryparams, self._data, self.mandatory)
975 975
976 976 # methods used to defines the part content
977 977 @property
978 978 def data(self):
979 979 return self._data
980 980
981 981 @data.setter
982 982 def data(self, data):
983 983 if self._generated is not None:
984 984 raise error.ReadOnlyPartError('part is being generated')
985 985 self._data = data
986 986
987 987 @property
988 988 def mandatoryparams(self):
989 989 # make it an immutable tuple to force people through ``addparam``
990 990 return tuple(self._mandatoryparams)
991 991
992 992 @property
993 993 def advisoryparams(self):
994 994 # make it an immutable tuple to force people through ``addparam``
995 995 return tuple(self._advisoryparams)
996 996
997 997 def addparam(self, name, value='', mandatory=True):
998 998 """add a parameter to the part
999 999
1000 1000 If 'mandatory' is set to True, the remote handler must claim support
1001 1001 for this parameter or the unbundling will be aborted.
1002 1002
1003 1003 The 'name' and 'value' cannot exceed 255 bytes each.
1004 1004 """
1005 1005 if self._generated is not None:
1006 1006 raise error.ReadOnlyPartError('part is being generated')
1007 1007 if name in self._seenparams:
1008 1008 raise ValueError('duplicated params: %s' % name)
1009 1009 self._seenparams.add(name)
1010 1010 params = self._advisoryparams
1011 1011 if mandatory:
1012 1012 params = self._mandatoryparams
1013 1013 params.append((name, value))
1014 1014
1015 1015 # methods used to generates the bundle2 stream
1016 1016 def getchunks(self, ui):
1017 1017 if self._generated is not None:
1018 1018 raise error.ProgrammingError('part can only be consumed once')
1019 1019 self._generated = False
1020 1020
1021 1021 if ui.debugflag:
1022 1022 msg = ['bundle2-output-part: "%s"' % self.type]
1023 1023 if not self.mandatory:
1024 1024 msg.append(' (advisory)')
1025 1025 nbmp = len(self.mandatoryparams)
1026 1026 nbap = len(self.advisoryparams)
1027 1027 if nbmp or nbap:
1028 1028 msg.append(' (params:')
1029 1029 if nbmp:
1030 1030 msg.append(' %i mandatory' % nbmp)
1031 1031 if nbap:
1032 1032 msg.append(' %i advisory' % nbmp)
1033 1033 msg.append(')')
1034 1034 if not self.data:
1035 1035 msg.append(' empty payload')
1036 1036 elif (util.safehasattr(self.data, 'next')
1037 1037 or util.safehasattr(self.data, '__next__')):
1038 1038 msg.append(' streamed payload')
1039 1039 else:
1040 1040 msg.append(' %i bytes payload' % len(self.data))
1041 1041 msg.append('\n')
1042 1042 ui.debug(''.join(msg))
1043 1043
1044 1044 #### header
1045 1045 if self.mandatory:
1046 1046 parttype = self.type.upper()
1047 1047 else:
1048 1048 parttype = self.type.lower()
1049 1049 outdebug(ui, 'part %s: "%s"' % (pycompat.bytestr(self.id), parttype))
1050 1050 ## parttype
1051 1051 header = [_pack(_fparttypesize, len(parttype)),
1052 1052 parttype, _pack(_fpartid, self.id),
1053 1053 ]
1054 1054 ## parameters
1055 1055 # count
1056 1056 manpar = self.mandatoryparams
1057 1057 advpar = self.advisoryparams
1058 1058 header.append(_pack(_fpartparamcount, len(manpar), len(advpar)))
1059 1059 # size
1060 1060 parsizes = []
1061 1061 for key, value in manpar:
1062 1062 parsizes.append(len(key))
1063 1063 parsizes.append(len(value))
1064 1064 for key, value in advpar:
1065 1065 parsizes.append(len(key))
1066 1066 parsizes.append(len(value))
1067 1067 paramsizes = _pack(_makefpartparamsizes(len(parsizes) // 2), *parsizes)
1068 1068 header.append(paramsizes)
1069 1069 # key, value
1070 1070 for key, value in manpar:
1071 1071 header.append(key)
1072 1072 header.append(value)
1073 1073 for key, value in advpar:
1074 1074 header.append(key)
1075 1075 header.append(value)
1076 1076 ## finalize header
1077 1077 try:
1078 1078 headerchunk = ''.join(header)
1079 1079 except TypeError:
1080 1080 raise TypeError(r'Found a non-bytes trying to '
1081 1081 r'build bundle part header: %r' % header)
1082 1082 outdebug(ui, 'header chunk size: %i' % len(headerchunk))
1083 1083 yield _pack(_fpartheadersize, len(headerchunk))
1084 1084 yield headerchunk
1085 1085 ## payload
1086 1086 try:
1087 1087 for chunk in self._payloadchunks():
1088 1088 outdebug(ui, 'payload chunk size: %i' % len(chunk))
1089 1089 yield _pack(_fpayloadsize, len(chunk))
1090 1090 yield chunk
1091 1091 except GeneratorExit:
1092 1092 # GeneratorExit means that nobody is listening for our
1093 1093 # results anyway, so just bail quickly rather than trying
1094 1094 # to produce an error part.
1095 1095 ui.debug('bundle2-generatorexit\n')
1096 1096 raise
1097 1097 except BaseException as exc:
1098 1098 bexc = stringutil.forcebytestr(exc)
1099 1099 # backup exception data for later
1100 1100 ui.debug('bundle2-input-stream-interrupt: encoding exception %s'
1101 1101 % bexc)
1102 1102 tb = sys.exc_info()[2]
1103 1103 msg = 'unexpected error: %s' % bexc
1104 1104 interpart = bundlepart('error:abort', [('message', msg)],
1105 1105 mandatory=False)
1106 1106 interpart.id = 0
1107 1107 yield _pack(_fpayloadsize, -1)
1108 1108 for chunk in interpart.getchunks(ui=ui):
1109 1109 yield chunk
1110 1110 outdebug(ui, 'closing payload chunk')
1111 1111 # abort current part payload
1112 1112 yield _pack(_fpayloadsize, 0)
1113 1113 pycompat.raisewithtb(exc, tb)
1114 1114 # end of payload
1115 1115 outdebug(ui, 'closing payload chunk')
1116 1116 yield _pack(_fpayloadsize, 0)
1117 1117 self._generated = True
1118 1118
1119 1119 def _payloadchunks(self):
1120 1120 """yield chunks of a the part payload
1121 1121
1122 1122 Exists to handle the different methods to provide data to a part."""
1123 1123 # we only support fixed size data now.
1124 1124 # This will be improved in the future.
1125 1125 if (util.safehasattr(self.data, 'next')
1126 1126 or util.safehasattr(self.data, '__next__')):
1127 1127 buff = util.chunkbuffer(self.data)
1128 1128 chunk = buff.read(preferedchunksize)
1129 1129 while chunk:
1130 1130 yield chunk
1131 1131 chunk = buff.read(preferedchunksize)
1132 1132 elif len(self.data):
1133 1133 yield self.data
1134 1134
1135 1135
1136 1136 flaginterrupt = -1
1137 1137
1138 1138 class interrupthandler(unpackermixin):
1139 1139 """read one part and process it with restricted capability
1140 1140
1141 1141 This allows to transmit exception raised on the producer size during part
1142 1142 iteration while the consumer is reading a part.
1143 1143
1144 1144 Part processed in this manner only have access to a ui object,"""
1145 1145
1146 1146 def __init__(self, ui, fp):
1147 1147 super(interrupthandler, self).__init__(fp)
1148 1148 self.ui = ui
1149 1149
1150 1150 def _readpartheader(self):
1151 1151 """reads a part header size and return the bytes blob
1152 1152
1153 1153 returns None if empty"""
1154 1154 headersize = self._unpack(_fpartheadersize)[0]
1155 1155 if headersize < 0:
1156 1156 raise error.BundleValueError('negative part header size: %i'
1157 1157 % headersize)
1158 1158 indebug(self.ui, 'part header size: %i\n' % headersize)
1159 1159 if headersize:
1160 1160 return self._readexact(headersize)
1161 1161 return None
1162 1162
1163 1163 def __call__(self):
1164 1164
1165 1165 self.ui.debug('bundle2-input-stream-interrupt:'
1166 1166 ' opening out of band context\n')
1167 1167 indebug(self.ui, 'bundle2 stream interruption, looking for a part.')
1168 1168 headerblock = self._readpartheader()
1169 1169 if headerblock is None:
1170 1170 indebug(self.ui, 'no part found during interruption.')
1171 1171 return
1172 1172 part = unbundlepart(self.ui, headerblock, self._fp)
1173 1173 op = interruptoperation(self.ui)
1174 1174 hardabort = False
1175 1175 try:
1176 1176 _processpart(op, part)
1177 1177 except (SystemExit, KeyboardInterrupt):
1178 1178 hardabort = True
1179 1179 raise
1180 1180 finally:
1181 1181 if not hardabort:
1182 1182 part.consume()
1183 1183 self.ui.debug('bundle2-input-stream-interrupt:'
1184 1184 ' closing out of band context\n')
1185 1185
1186 1186 class interruptoperation(object):
1187 1187 """A limited operation to be use by part handler during interruption
1188 1188
1189 1189 It only have access to an ui object.
1190 1190 """
1191 1191
1192 1192 def __init__(self, ui):
1193 1193 self.ui = ui
1194 1194 self.reply = None
1195 1195 self.captureoutput = False
1196 1196
1197 1197 @property
1198 1198 def repo(self):
1199 1199 raise error.ProgrammingError('no repo access from stream interruption')
1200 1200
1201 1201 def gettransaction(self):
1202 1202 raise TransactionUnavailable('no repo access from stream interruption')
1203 1203
1204 1204 def decodepayloadchunks(ui, fh):
1205 1205 """Reads bundle2 part payload data into chunks.
1206 1206
1207 1207 Part payload data consists of framed chunks. This function takes
1208 1208 a file handle and emits those chunks.
1209 1209 """
1210 1210 dolog = ui.configbool('devel', 'bundle2.debug')
1211 1211 debug = ui.debug
1212 1212
1213 1213 headerstruct = struct.Struct(_fpayloadsize)
1214 1214 headersize = headerstruct.size
1215 1215 unpack = headerstruct.unpack
1216 1216
1217 1217 readexactly = changegroup.readexactly
1218 1218 read = fh.read
1219 1219
1220 1220 chunksize = unpack(readexactly(fh, headersize))[0]
1221 1221 indebug(ui, 'payload chunk size: %i' % chunksize)
1222 1222
1223 1223 # changegroup.readexactly() is inlined below for performance.
1224 1224 while chunksize:
1225 1225 if chunksize >= 0:
1226 1226 s = read(chunksize)
1227 1227 if len(s) < chunksize:
1228 1228 raise error.Abort(_('stream ended unexpectedly '
1229 1229 ' (got %d bytes, expected %d)') %
1230 1230 (len(s), chunksize))
1231 1231
1232 1232 yield s
1233 1233 elif chunksize == flaginterrupt:
1234 1234 # Interrupt "signal" detected. The regular stream is interrupted
1235 1235 # and a bundle2 part follows. Consume it.
1236 1236 interrupthandler(ui, fh)()
1237 1237 else:
1238 1238 raise error.BundleValueError(
1239 1239 'negative payload chunk size: %s' % chunksize)
1240 1240
1241 1241 s = read(headersize)
1242 1242 if len(s) < headersize:
1243 1243 raise error.Abort(_('stream ended unexpectedly '
1244 1244 ' (got %d bytes, expected %d)') %
1245 1245 (len(s), chunksize))
1246 1246
1247 1247 chunksize = unpack(s)[0]
1248 1248
1249 1249 # indebug() inlined for performance.
1250 1250 if dolog:
1251 1251 debug('bundle2-input: payload chunk size: %i\n' % chunksize)
1252 1252
1253 1253 class unbundlepart(unpackermixin):
1254 1254 """a bundle part read from a bundle"""
1255 1255
1256 1256 def __init__(self, ui, header, fp):
1257 1257 super(unbundlepart, self).__init__(fp)
1258 1258 self._seekable = (util.safehasattr(fp, 'seek') and
1259 1259 util.safehasattr(fp, 'tell'))
1260 1260 self.ui = ui
1261 1261 # unbundle state attr
1262 1262 self._headerdata = header
1263 1263 self._headeroffset = 0
1264 1264 self._initialized = False
1265 1265 self.consumed = False
1266 1266 # part data
1267 1267 self.id = None
1268 1268 self.type = None
1269 1269 self.mandatoryparams = None
1270 1270 self.advisoryparams = None
1271 1271 self.params = None
1272 1272 self.mandatorykeys = ()
1273 1273 self._readheader()
1274 1274 self._mandatory = None
1275 1275 self._pos = 0
1276 1276
1277 1277 def _fromheader(self, size):
1278 1278 """return the next <size> byte from the header"""
1279 1279 offset = self._headeroffset
1280 1280 data = self._headerdata[offset:(offset + size)]
1281 1281 self._headeroffset = offset + size
1282 1282 return data
1283 1283
1284 1284 def _unpackheader(self, format):
1285 1285 """read given format from header
1286 1286
1287 1287 This automatically compute the size of the format to read."""
1288 1288 data = self._fromheader(struct.calcsize(format))
1289 1289 return _unpack(format, data)
1290 1290
1291 1291 def _initparams(self, mandatoryparams, advisoryparams):
1292 1292 """internal function to setup all logic related parameters"""
1293 1293 # make it read only to prevent people touching it by mistake.
1294 1294 self.mandatoryparams = tuple(mandatoryparams)
1295 1295 self.advisoryparams = tuple(advisoryparams)
1296 1296 # user friendly UI
1297 1297 self.params = util.sortdict(self.mandatoryparams)
1298 1298 self.params.update(self.advisoryparams)
1299 1299 self.mandatorykeys = frozenset(p[0] for p in mandatoryparams)
1300 1300
1301 1301 def _readheader(self):
1302 1302 """read the header and setup the object"""
1303 1303 typesize = self._unpackheader(_fparttypesize)[0]
1304 1304 self.type = self._fromheader(typesize)
1305 1305 indebug(self.ui, 'part type: "%s"' % self.type)
1306 1306 self.id = self._unpackheader(_fpartid)[0]
1307 1307 indebug(self.ui, 'part id: "%s"' % pycompat.bytestr(self.id))
1308 1308 # extract mandatory bit from type
1309 1309 self.mandatory = (self.type != self.type.lower())
1310 1310 self.type = self.type.lower()
1311 1311 ## reading parameters
1312 1312 # param count
1313 1313 mancount, advcount = self._unpackheader(_fpartparamcount)
1314 1314 indebug(self.ui, 'part parameters: %i' % (mancount + advcount))
1315 1315 # param size
1316 1316 fparamsizes = _makefpartparamsizes(mancount + advcount)
1317 1317 paramsizes = self._unpackheader(fparamsizes)
1318 1318 # make it a list of couple again
1319 1319 paramsizes = list(zip(paramsizes[::2], paramsizes[1::2]))
1320 1320 # split mandatory from advisory
1321 1321 mansizes = paramsizes[:mancount]
1322 1322 advsizes = paramsizes[mancount:]
1323 1323 # retrieve param value
1324 1324 manparams = []
1325 1325 for key, value in mansizes:
1326 1326 manparams.append((self._fromheader(key), self._fromheader(value)))
1327 1327 advparams = []
1328 1328 for key, value in advsizes:
1329 1329 advparams.append((self._fromheader(key), self._fromheader(value)))
1330 1330 self._initparams(manparams, advparams)
1331 1331 ## part payload
1332 1332 self._payloadstream = util.chunkbuffer(self._payloadchunks())
1333 1333 # we read the data, tell it
1334 1334 self._initialized = True
1335 1335
1336 1336 def _payloadchunks(self):
1337 1337 """Generator of decoded chunks in the payload."""
1338 1338 return decodepayloadchunks(self.ui, self._fp)
1339 1339
1340 1340 def consume(self):
1341 1341 """Read the part payload until completion.
1342 1342
1343 1343 By consuming the part data, the underlying stream read offset will
1344 1344 be advanced to the next part (or end of stream).
1345 1345 """
1346 1346 if self.consumed:
1347 1347 return
1348 1348
1349 1349 chunk = self.read(32768)
1350 1350 while chunk:
1351 1351 self._pos += len(chunk)
1352 1352 chunk = self.read(32768)
1353 1353
1354 1354 def read(self, size=None):
1355 1355 """read payload data"""
1356 1356 if not self._initialized:
1357 1357 self._readheader()
1358 1358 if size is None:
1359 1359 data = self._payloadstream.read()
1360 1360 else:
1361 1361 data = self._payloadstream.read(size)
1362 1362 self._pos += len(data)
1363 1363 if size is None or len(data) < size:
1364 1364 if not self.consumed and self._pos:
1365 1365 self.ui.debug('bundle2-input-part: total payload size %i\n'
1366 1366 % self._pos)
1367 1367 self.consumed = True
1368 1368 return data
1369 1369
1370 1370 class seekableunbundlepart(unbundlepart):
1371 1371 """A bundle2 part in a bundle that is seekable.
1372 1372
1373 1373 Regular ``unbundlepart`` instances can only be read once. This class
1374 1374 extends ``unbundlepart`` to enable bi-directional seeking within the
1375 1375 part.
1376 1376
1377 1377 Bundle2 part data consists of framed chunks. Offsets when seeking
1378 1378 refer to the decoded data, not the offsets in the underlying bundle2
1379 1379 stream.
1380 1380
1381 1381 To facilitate quickly seeking within the decoded data, instances of this
1382 1382 class maintain a mapping between offsets in the underlying stream and
1383 1383 the decoded payload. This mapping will consume memory in proportion
1384 1384 to the number of chunks within the payload (which almost certainly
1385 1385 increases in proportion with the size of the part).
1386 1386 """
1387 1387 def __init__(self, ui, header, fp):
1388 1388 # (payload, file) offsets for chunk starts.
1389 1389 self._chunkindex = []
1390 1390
1391 1391 super(seekableunbundlepart, self).__init__(ui, header, fp)
1392 1392
1393 1393 def _payloadchunks(self, chunknum=0):
1394 1394 '''seek to specified chunk and start yielding data'''
1395 1395 if len(self._chunkindex) == 0:
1396 1396 assert chunknum == 0, 'Must start with chunk 0'
1397 1397 self._chunkindex.append((0, self._tellfp()))
1398 1398 else:
1399 1399 assert chunknum < len(self._chunkindex), \
1400 1400 'Unknown chunk %d' % chunknum
1401 1401 self._seekfp(self._chunkindex[chunknum][1])
1402 1402
1403 1403 pos = self._chunkindex[chunknum][0]
1404 1404
1405 1405 for chunk in decodepayloadchunks(self.ui, self._fp):
1406 1406 chunknum += 1
1407 1407 pos += len(chunk)
1408 1408 if chunknum == len(self._chunkindex):
1409 1409 self._chunkindex.append((pos, self._tellfp()))
1410 1410
1411 1411 yield chunk
1412 1412
1413 1413 def _findchunk(self, pos):
1414 1414 '''for a given payload position, return a chunk number and offset'''
1415 1415 for chunk, (ppos, fpos) in enumerate(self._chunkindex):
1416 1416 if ppos == pos:
1417 1417 return chunk, 0
1418 1418 elif ppos > pos:
1419 1419 return chunk - 1, pos - self._chunkindex[chunk - 1][0]
1420 1420 raise ValueError('Unknown chunk')
1421 1421
1422 1422 def tell(self):
1423 1423 return self._pos
1424 1424
1425 1425 def seek(self, offset, whence=os.SEEK_SET):
1426 1426 if whence == os.SEEK_SET:
1427 1427 newpos = offset
1428 1428 elif whence == os.SEEK_CUR:
1429 1429 newpos = self._pos + offset
1430 1430 elif whence == os.SEEK_END:
1431 1431 if not self.consumed:
1432 1432 # Can't use self.consume() here because it advances self._pos.
1433 1433 chunk = self.read(32768)
1434 1434 while chunk:
1435 1435 chunk = self.read(32768)
1436 1436 newpos = self._chunkindex[-1][0] - offset
1437 1437 else:
1438 1438 raise ValueError('Unknown whence value: %r' % (whence,))
1439 1439
1440 1440 if newpos > self._chunkindex[-1][0] and not self.consumed:
1441 1441 # Can't use self.consume() here because it advances self._pos.
1442 1442 chunk = self.read(32768)
1443 1443 while chunk:
1444 1444 chunk = self.read(32668)
1445 1445
1446 1446 if not 0 <= newpos <= self._chunkindex[-1][0]:
1447 1447 raise ValueError('Offset out of range')
1448 1448
1449 1449 if self._pos != newpos:
1450 1450 chunk, internaloffset = self._findchunk(newpos)
1451 1451 self._payloadstream = util.chunkbuffer(self._payloadchunks(chunk))
1452 1452 adjust = self.read(internaloffset)
1453 1453 if len(adjust) != internaloffset:
1454 1454 raise error.Abort(_('Seek failed\n'))
1455 1455 self._pos = newpos
1456 1456
1457 1457 def _seekfp(self, offset, whence=0):
1458 1458 """move the underlying file pointer
1459 1459
1460 1460 This method is meant for internal usage by the bundle2 protocol only.
1461 1461 They directly manipulate the low level stream including bundle2 level
1462 1462 instruction.
1463 1463
1464 1464 Do not use it to implement higher-level logic or methods."""
1465 1465 if self._seekable:
1466 1466 return self._fp.seek(offset, whence)
1467 1467 else:
1468 1468 raise NotImplementedError(_('File pointer is not seekable'))
1469 1469
1470 1470 def _tellfp(self):
1471 1471 """return the file offset, or None if file is not seekable
1472 1472
1473 1473 This method is meant for internal usage by the bundle2 protocol only.
1474 1474 They directly manipulate the low level stream including bundle2 level
1475 1475 instruction.
1476 1476
1477 1477 Do not use it to implement higher-level logic or methods."""
1478 1478 if self._seekable:
1479 1479 try:
1480 1480 return self._fp.tell()
1481 1481 except IOError as e:
1482 1482 if e.errno == errno.ESPIPE:
1483 1483 self._seekable = False
1484 1484 else:
1485 1485 raise
1486 1486 return None
1487 1487
1488 1488 # These are only the static capabilities.
1489 1489 # Check the 'getrepocaps' function for the rest.
1490 1490 capabilities = {'HG20': (),
1491 1491 'bookmarks': (),
1492 1492 'error': ('abort', 'unsupportedcontent', 'pushraced',
1493 1493 'pushkey'),
1494 1494 'listkeys': (),
1495 1495 'pushkey': (),
1496 1496 'digests': tuple(sorted(util.DIGESTS.keys())),
1497 1497 'remote-changegroup': ('http', 'https'),
1498 1498 'hgtagsfnodes': (),
1499 1499 'rev-branch-cache': (),
1500 1500 'phases': ('heads',),
1501 1501 'stream': ('v2',),
1502 1502 }
1503 1503
1504 1504 def getrepocaps(repo, allowpushback=False, role=None):
1505 1505 """return the bundle2 capabilities for a given repo
1506 1506
1507 1507 Exists to allow extensions (like evolution) to mutate the capabilities.
1508 1508
1509 1509 The returned value is used for servers advertising their capabilities as
1510 1510 well as clients advertising their capabilities to servers as part of
1511 1511 bundle2 requests. The ``role`` argument specifies which is which.
1512 1512 """
1513 1513 if role not in ('client', 'server'):
1514 1514 raise error.ProgrammingError('role argument must be client or server')
1515 1515
1516 1516 caps = capabilities.copy()
1517 1517 caps['changegroup'] = tuple(sorted(
1518 1518 changegroup.supportedincomingversions(repo)))
1519 1519 if obsolete.isenabled(repo, obsolete.exchangeopt):
1520 1520 supportedformat = tuple('V%i' % v for v in obsolete.formats)
1521 1521 caps['obsmarkers'] = supportedformat
1522 1522 if allowpushback:
1523 1523 caps['pushback'] = ()
1524 1524 cpmode = repo.ui.config('server', 'concurrent-push-mode')
1525 1525 if cpmode == 'check-related':
1526 1526 caps['checkheads'] = ('related',)
1527 1527 if 'phases' in repo.ui.configlist('devel', 'legacy.exchange'):
1528 1528 caps.pop('phases')
1529 1529
1530 1530 # Don't advertise stream clone support in server mode if not configured.
1531 1531 if role == 'server':
1532 1532 streamsupported = repo.ui.configbool('server', 'uncompressed',
1533 1533 untrusted=True)
1534 1534 featuresupported = repo.ui.configbool('experimental', 'bundle2.stream')
1535 1535
1536 1536 if not streamsupported or not featuresupported:
1537 1537 caps.pop('stream')
1538 1538 # Else always advertise support on client, because payload support
1539 1539 # should always be advertised.
1540 1540
1541 1541 return caps
1542 1542
1543 1543 def bundle2caps(remote):
1544 1544 """return the bundle capabilities of a peer as dict"""
1545 1545 raw = remote.capable('bundle2')
1546 1546 if not raw and raw != '':
1547 1547 return {}
1548 1548 capsblob = urlreq.unquote(remote.capable('bundle2'))
1549 1549 return decodecaps(capsblob)
1550 1550
1551 1551 def obsmarkersversion(caps):
1552 1552 """extract the list of supported obsmarkers versions from a bundle2caps dict
1553 1553 """
1554 1554 obscaps = caps.get('obsmarkers', ())
1555 1555 return [int(c[1:]) for c in obscaps if c.startswith('V')]
1556 1556
1557 1557 def writenewbundle(ui, repo, source, filename, bundletype, outgoing, opts,
1558 1558 vfs=None, compression=None, compopts=None):
1559 1559 if bundletype.startswith('HG10'):
1560 1560 cg = changegroup.makechangegroup(repo, outgoing, '01', source)
1561 1561 return writebundle(ui, cg, filename, bundletype, vfs=vfs,
1562 1562 compression=compression, compopts=compopts)
1563 1563 elif not bundletype.startswith('HG20'):
1564 1564 raise error.ProgrammingError('unknown bundle type: %s' % bundletype)
1565 1565
1566 1566 caps = {}
1567 1567 if 'obsolescence' in opts:
1568 1568 caps['obsmarkers'] = ('V1',)
1569 1569 bundle = bundle20(ui, caps)
1570 1570 bundle.setcompression(compression, compopts)
1571 1571 _addpartsfromopts(ui, repo, bundle, source, outgoing, opts)
1572 1572 chunkiter = bundle.getchunks()
1573 1573
1574 1574 return changegroup.writechunks(ui, chunkiter, filename, vfs=vfs)
1575 1575
1576 1576 def _addpartsfromopts(ui, repo, bundler, source, outgoing, opts):
1577 1577 # We should eventually reconcile this logic with the one behind
1578 1578 # 'exchange.getbundle2partsgenerator'.
1579 1579 #
1580 1580 # The type of input from 'getbundle' and 'writenewbundle' are a bit
1581 1581 # different right now. So we keep them separated for now for the sake of
1582 1582 # simplicity.
1583 1583
1584 1584 # we might not always want a changegroup in such bundle, for example in
1585 1585 # stream bundles
1586 1586 if opts.get('changegroup', True):
1587 1587 cgversion = opts.get('cg.version')
1588 1588 if cgversion is None:
1589 1589 cgversion = changegroup.safeversion(repo)
1590 1590 cg = changegroup.makechangegroup(repo, outgoing, cgversion, source)
1591 1591 part = bundler.newpart('changegroup', data=cg.getchunks())
1592 1592 part.addparam('version', cg.version)
1593 1593 if 'clcount' in cg.extras:
1594 1594 part.addparam('nbchanges', '%d' % cg.extras['clcount'],
1595 1595 mandatory=False)
1596 1596 if opts.get('phases') and repo.revs('%ln and secret()',
1597 1597 outgoing.missingheads):
1598 1598 part.addparam('targetphase', '%d' % phases.secret, mandatory=False)
1599 1599
1600 1600 if opts.get('streamv2', False):
1601 1601 addpartbundlestream2(bundler, repo, stream=True)
1602 1602
1603 1603 if opts.get('tagsfnodescache', True):
1604 1604 addparttagsfnodescache(repo, bundler, outgoing)
1605 1605
1606 1606 if opts.get('revbranchcache', True):
1607 1607 addpartrevbranchcache(repo, bundler, outgoing)
1608 1608
1609 1609 if opts.get('obsolescence', False):
1610 1610 obsmarkers = repo.obsstore.relevantmarkers(outgoing.missing)
1611 1611 buildobsmarkerspart(bundler, obsmarkers)
1612 1612
1613 1613 if opts.get('phases', False):
1614 1614 headsbyphase = phases.subsetphaseheads(repo, outgoing.missing)
1615 1615 phasedata = phases.binaryencode(headsbyphase)
1616 1616 bundler.newpart('phase-heads', data=phasedata)
1617 1617
1618 1618 def addparttagsfnodescache(repo, bundler, outgoing):
1619 1619 # we include the tags fnode cache for the bundle changeset
1620 1620 # (as an optional parts)
1621 1621 cache = tags.hgtagsfnodescache(repo.unfiltered())
1622 1622 chunks = []
1623 1623
1624 1624 # .hgtags fnodes are only relevant for head changesets. While we could
1625 1625 # transfer values for all known nodes, there will likely be little to
1626 1626 # no benefit.
1627 1627 #
1628 1628 # We don't bother using a generator to produce output data because
1629 1629 # a) we only have 40 bytes per head and even esoteric numbers of heads
1630 1630 # consume little memory (1M heads is 40MB) b) we don't want to send the
1631 1631 # part if we don't have entries and knowing if we have entries requires
1632 1632 # cache lookups.
1633 1633 for node in outgoing.missingheads:
1634 1634 # Don't compute missing, as this may slow down serving.
1635 1635 fnode = cache.getfnode(node, computemissing=False)
1636 1636 if fnode is not None:
1637 1637 chunks.extend([node, fnode])
1638 1638
1639 1639 if chunks:
1640 1640 bundler.newpart('hgtagsfnodes', data=''.join(chunks))
1641 1641
1642 1642 def addpartrevbranchcache(repo, bundler, outgoing):
1643 1643 # we include the rev branch cache for the bundle changeset
1644 1644 # (as an optional parts)
1645 1645 cache = repo.revbranchcache()
1646 1646 cl = repo.unfiltered().changelog
1647 1647 branchesdata = collections.defaultdict(lambda: (set(), set()))
1648 1648 for node in outgoing.missing:
1649 1649 branch, close = cache.branchinfo(cl.rev(node))
1650 1650 branchesdata[branch][close].add(node)
1651 1651
1652 1652 def generate():
1653 1653 for branch, (nodes, closed) in sorted(branchesdata.items()):
1654 1654 utf8branch = encoding.fromlocal(branch)
1655 1655 yield rbcstruct.pack(len(utf8branch), len(nodes), len(closed))
1656 1656 yield utf8branch
1657 1657 for n in sorted(nodes):
1658 1658 yield n
1659 1659 for n in sorted(closed):
1660 1660 yield n
1661 1661
1662 1662 bundler.newpart('cache:rev-branch-cache', data=generate())
1663 1663
1664 1664 def _formatrequirementsspec(requirements):
1665 1665 return urlreq.quote(','.join(sorted(requirements)))
1666 1666
1667 1667 def _formatrequirementsparams(requirements):
1668 1668 requirements = _formatrequirementsspec(requirements)
1669 1669 params = "%s%s" % (urlreq.quote("requirements="), requirements)
1670 1670 return params
1671 1671
1672 1672 def addpartbundlestream2(bundler, repo, **kwargs):
1673 1673 if not kwargs.get('stream', False):
1674 1674 return
1675 1675
1676 1676 if not streamclone.allowservergeneration(repo):
1677 1677 raise error.Abort(_('stream data requested but server does not allow '
1678 1678 'this feature'),
1679 1679 hint=_('well-behaved clients should not be '
1680 1680 'requesting stream data from servers not '
1681 1681 'advertising it; the client may be buggy'))
1682 1682
1683 1683 # Stream clones don't compress well. And compression undermines a
1684 1684 # goal of stream clones, which is to be fast. Communicate the desire
1685 1685 # to avoid compression to consumers of the bundle.
1686 1686 bundler.prefercompressed = False
1687 1687
1688 1688 filecount, bytecount, it = streamclone.generatev2(repo)
1689 1689 requirements = _formatrequirementsspec(repo.requirements)
1690 1690 part = bundler.newpart('stream2', data=it)
1691 1691 part.addparam('bytecount', '%d' % bytecount, mandatory=True)
1692 1692 part.addparam('filecount', '%d' % filecount, mandatory=True)
1693 1693 part.addparam('requirements', requirements, mandatory=True)
1694 1694
1695 1695 def buildobsmarkerspart(bundler, markers):
1696 1696 """add an obsmarker part to the bundler with <markers>
1697 1697
1698 1698 No part is created if markers is empty.
1699 1699 Raises ValueError if the bundler doesn't support any known obsmarker format.
1700 1700 """
1701 1701 if not markers:
1702 1702 return None
1703 1703
1704 1704 remoteversions = obsmarkersversion(bundler.capabilities)
1705 1705 version = obsolete.commonversion(remoteversions)
1706 1706 if version is None:
1707 1707 raise ValueError('bundler does not support common obsmarker format')
1708 1708 stream = obsolete.encodemarkers(markers, True, version=version)
1709 1709 return bundler.newpart('obsmarkers', data=stream)
1710 1710
1711 1711 def writebundle(ui, cg, filename, bundletype, vfs=None, compression=None,
1712 1712 compopts=None):
1713 1713 """Write a bundle file and return its filename.
1714 1714
1715 1715 Existing files will not be overwritten.
1716 1716 If no filename is specified, a temporary file is created.
1717 1717 bz2 compression can be turned off.
1718 1718 The bundle file will be deleted in case of errors.
1719 1719 """
1720 1720
1721 1721 if bundletype == "HG20":
1722 1722 bundle = bundle20(ui)
1723 1723 bundle.setcompression(compression, compopts)
1724 1724 part = bundle.newpart('changegroup', data=cg.getchunks())
1725 1725 part.addparam('version', cg.version)
1726 1726 if 'clcount' in cg.extras:
1727 1727 part.addparam('nbchanges', '%d' % cg.extras['clcount'],
1728 1728 mandatory=False)
1729 1729 chunkiter = bundle.getchunks()
1730 1730 else:
1731 1731 # compression argument is only for the bundle2 case
1732 1732 assert compression is None
1733 1733 if cg.version != '01':
1734 1734 raise error.Abort(_('old bundle types only supports v1 '
1735 1735 'changegroups'))
1736 1736 header, comp = bundletypes[bundletype]
1737 1737 if comp not in util.compengines.supportedbundletypes:
1738 1738 raise error.Abort(_('unknown stream compression type: %s')
1739 1739 % comp)
1740 1740 compengine = util.compengines.forbundletype(comp)
1741 1741 def chunkiter():
1742 1742 yield header
1743 1743 for chunk in compengine.compressstream(cg.getchunks(), compopts):
1744 1744 yield chunk
1745 1745 chunkiter = chunkiter()
1746 1746
1747 1747 # parse the changegroup data, otherwise we will block
1748 1748 # in case of sshrepo because we don't know the end of the stream
1749 1749 return changegroup.writechunks(ui, chunkiter, filename, vfs=vfs)
1750 1750
1751 1751 def combinechangegroupresults(op):
1752 1752 """logic to combine 0 or more addchangegroup results into one"""
1753 1753 results = [r.get('return', 0)
1754 1754 for r in op.records['changegroup']]
1755 1755 changedheads = 0
1756 1756 result = 1
1757 1757 for ret in results:
1758 1758 # If any changegroup result is 0, return 0
1759 1759 if ret == 0:
1760 1760 result = 0
1761 1761 break
1762 1762 if ret < -1:
1763 1763 changedheads += ret + 1
1764 1764 elif ret > 1:
1765 1765 changedheads += ret - 1
1766 1766 if changedheads > 0:
1767 1767 result = 1 + changedheads
1768 1768 elif changedheads < 0:
1769 1769 result = -1 + changedheads
1770 1770 return result
1771 1771
1772 1772 @parthandler('changegroup', ('version', 'nbchanges', 'treemanifest',
1773 1773 'targetphase'))
1774 1774 def handlechangegroup(op, inpart):
1775 1775 """apply a changegroup part on the repo
1776 1776
1777 1777 This is a very early implementation that will massive rework before being
1778 1778 inflicted to any end-user.
1779 1779 """
1780 1780 tr = op.gettransaction()
1781 1781 unpackerversion = inpart.params.get('version', '01')
1782 1782 # We should raise an appropriate exception here
1783 1783 cg = changegroup.getunbundler(unpackerversion, inpart, None)
1784 1784 # the source and url passed here are overwritten by the one contained in
1785 1785 # the transaction.hookargs argument. So 'bundle2' is a placeholder
1786 1786 nbchangesets = None
1787 1787 if 'nbchanges' in inpart.params:
1788 1788 nbchangesets = int(inpart.params.get('nbchanges'))
1789 1789 if ('treemanifest' in inpart.params and
1790 1790 'treemanifest' not in op.repo.requirements):
1791 1791 if len(op.repo.changelog) != 0:
1792 1792 raise error.Abort(_(
1793 1793 "bundle contains tree manifests, but local repo is "
1794 1794 "non-empty and does not use tree manifests"))
1795 1795 op.repo.requirements.add('treemanifest')
1796 1796 op.repo._applyopenerreqs()
1797 1797 op.repo._writerequirements()
1798 1798 extrakwargs = {}
1799 1799 targetphase = inpart.params.get('targetphase')
1800 1800 if targetphase is not None:
1801 1801 extrakwargs[r'targetphase'] = int(targetphase)
1802 1802 ret = _processchangegroup(op, cg, tr, 'bundle2', 'bundle2',
1803 1803 expectedtotal=nbchangesets, **extrakwargs)
1804 1804 if op.reply is not None:
1805 1805 # This is definitely not the final form of this
1806 1806 # return. But one need to start somewhere.
1807 1807 part = op.reply.newpart('reply:changegroup', mandatory=False)
1808 1808 part.addparam(
1809 1809 'in-reply-to', pycompat.bytestr(inpart.id), mandatory=False)
1810 1810 part.addparam('return', '%i' % ret, mandatory=False)
1811 1811 assert not inpart.read()
1812 1812
1813 1813 _remotechangegroupparams = tuple(['url', 'size', 'digests'] +
1814 1814 ['digest:%s' % k for k in util.DIGESTS.keys()])
1815 1815 @parthandler('remote-changegroup', _remotechangegroupparams)
1816 1816 def handleremotechangegroup(op, inpart):
1817 1817 """apply a bundle10 on the repo, given an url and validation information
1818 1818
1819 1819 All the information about the remote bundle to import are given as
1820 1820 parameters. The parameters include:
1821 1821 - url: the url to the bundle10.
1822 1822 - size: the bundle10 file size. It is used to validate what was
1823 1823 retrieved by the client matches the server knowledge about the bundle.
1824 1824 - digests: a space separated list of the digest types provided as
1825 1825 parameters.
1826 1826 - digest:<digest-type>: the hexadecimal representation of the digest with
1827 1827 that name. Like the size, it is used to validate what was retrieved by
1828 1828 the client matches what the server knows about the bundle.
1829 1829
1830 1830 When multiple digest types are given, all of them are checked.
1831 1831 """
1832 1832 try:
1833 1833 raw_url = inpart.params['url']
1834 1834 except KeyError:
1835 1835 raise error.Abort(_('remote-changegroup: missing "%s" param') % 'url')
1836 1836 parsed_url = util.url(raw_url)
1837 1837 if parsed_url.scheme not in capabilities['remote-changegroup']:
1838 1838 raise error.Abort(_('remote-changegroup does not support %s urls') %
1839 1839 parsed_url.scheme)
1840 1840
1841 1841 try:
1842 1842 size = int(inpart.params['size'])
1843 1843 except ValueError:
1844 1844 raise error.Abort(_('remote-changegroup: invalid value for param "%s"')
1845 1845 % 'size')
1846 1846 except KeyError:
1847 1847 raise error.Abort(_('remote-changegroup: missing "%s" param') % 'size')
1848 1848
1849 1849 digests = {}
1850 1850 for typ in inpart.params.get('digests', '').split():
1851 1851 param = 'digest:%s' % typ
1852 1852 try:
1853 1853 value = inpart.params[param]
1854 1854 except KeyError:
1855 1855 raise error.Abort(_('remote-changegroup: missing "%s" param') %
1856 1856 param)
1857 1857 digests[typ] = value
1858 1858
1859 1859 real_part = util.digestchecker(url.open(op.ui, raw_url), size, digests)
1860 1860
1861 1861 tr = op.gettransaction()
1862 1862 from . import exchange
1863 1863 cg = exchange.readbundle(op.repo.ui, real_part, raw_url)
1864 1864 if not isinstance(cg, changegroup.cg1unpacker):
1865 1865 raise error.Abort(_('%s: not a bundle version 1.0') %
1866 1866 util.hidepassword(raw_url))
1867 1867 ret = _processchangegroup(op, cg, tr, 'bundle2', 'bundle2')
1868 1868 if op.reply is not None:
1869 1869 # This is definitely not the final form of this
1870 1870 # return. But one need to start somewhere.
1871 1871 part = op.reply.newpart('reply:changegroup')
1872 1872 part.addparam(
1873 1873 'in-reply-to', pycompat.bytestr(inpart.id), mandatory=False)
1874 1874 part.addparam('return', '%i' % ret, mandatory=False)
1875 1875 try:
1876 1876 real_part.validate()
1877 1877 except error.Abort as e:
1878 1878 raise error.Abort(_('bundle at %s is corrupted:\n%s') %
1879 1879 (util.hidepassword(raw_url), str(e)))
1880 1880 assert not inpart.read()
1881 1881
1882 1882 @parthandler('reply:changegroup', ('return', 'in-reply-to'))
1883 1883 def handlereplychangegroup(op, inpart):
1884 1884 ret = int(inpart.params['return'])
1885 1885 replyto = int(inpart.params['in-reply-to'])
1886 1886 op.records.add('changegroup', {'return': ret}, replyto)
1887 1887
1888 1888 @parthandler('check:bookmarks')
1889 1889 def handlecheckbookmarks(op, inpart):
1890 1890 """check location of bookmarks
1891 1891
1892 1892 This part is to be used to detect push race regarding bookmark, it
1893 1893 contains binary encoded (bookmark, node) tuple. If the local state does
1894 1894 not marks the one in the part, a PushRaced exception is raised
1895 1895 """
1896 1896 bookdata = bookmarks.binarydecode(inpart)
1897 1897
1898 1898 msgstandard = ('repository changed while pushing - please try again '
1899 1899 '(bookmark "%s" move from %s to %s)')
1900 1900 msgmissing = ('repository changed while pushing - please try again '
1901 1901 '(bookmark "%s" is missing, expected %s)')
1902 1902 msgexist = ('repository changed while pushing - please try again '
1903 1903 '(bookmark "%s" set on %s, expected missing)')
1904 1904 for book, node in bookdata:
1905 1905 currentnode = op.repo._bookmarks.get(book)
1906 1906 if currentnode != node:
1907 1907 if node is None:
1908 1908 finalmsg = msgexist % (book, nodemod.short(currentnode))
1909 1909 elif currentnode is None:
1910 1910 finalmsg = msgmissing % (book, nodemod.short(node))
1911 1911 else:
1912 1912 finalmsg = msgstandard % (book, nodemod.short(node),
1913 1913 nodemod.short(currentnode))
1914 1914 raise error.PushRaced(finalmsg)
1915 1915
1916 1916 @parthandler('check:heads')
1917 1917 def handlecheckheads(op, inpart):
1918 1918 """check that head of the repo did not change
1919 1919
1920 1920 This is used to detect a push race when using unbundle.
1921 1921 This replaces the "heads" argument of unbundle."""
1922 1922 h = inpart.read(20)
1923 1923 heads = []
1924 1924 while len(h) == 20:
1925 1925 heads.append(h)
1926 1926 h = inpart.read(20)
1927 1927 assert not h
1928 1928 # Trigger a transaction so that we are guaranteed to have the lock now.
1929 1929 if op.ui.configbool('experimental', 'bundle2lazylocking'):
1930 1930 op.gettransaction()
1931 1931 if sorted(heads) != sorted(op.repo.heads()):
1932 1932 raise error.PushRaced('repository changed while pushing - '
1933 1933 'please try again')
1934 1934
1935 1935 @parthandler('check:updated-heads')
1936 1936 def handlecheckupdatedheads(op, inpart):
1937 1937 """check for race on the heads touched by a push
1938 1938
1939 1939 This is similar to 'check:heads' but focus on the heads actually updated
1940 1940 during the push. If other activities happen on unrelated heads, it is
1941 1941 ignored.
1942 1942
1943 1943 This allow server with high traffic to avoid push contention as long as
1944 1944 unrelated parts of the graph are involved."""
1945 1945 h = inpart.read(20)
1946 1946 heads = []
1947 1947 while len(h) == 20:
1948 1948 heads.append(h)
1949 1949 h = inpart.read(20)
1950 1950 assert not h
1951 1951 # trigger a transaction so that we are guaranteed to have the lock now.
1952 1952 if op.ui.configbool('experimental', 'bundle2lazylocking'):
1953 1953 op.gettransaction()
1954 1954
1955 1955 currentheads = set()
1956 1956 for ls in op.repo.branchmap().itervalues():
1957 1957 currentheads.update(ls)
1958 1958
1959 1959 for h in heads:
1960 1960 if h not in currentheads:
1961 1961 raise error.PushRaced('repository changed while pushing - '
1962 1962 'please try again')
1963 1963
1964 1964 @parthandler('check:phases')
1965 1965 def handlecheckphases(op, inpart):
1966 1966 """check that phase boundaries of the repository did not change
1967 1967
1968 1968 This is used to detect a push race.
1969 1969 """
1970 1970 phasetonodes = phases.binarydecode(inpart)
1971 1971 unfi = op.repo.unfiltered()
1972 1972 cl = unfi.changelog
1973 1973 phasecache = unfi._phasecache
1974 1974 msg = ('repository changed while pushing - please try again '
1975 1975 '(%s is %s expected %s)')
1976 1976 for expectedphase, nodes in enumerate(phasetonodes):
1977 1977 for n in nodes:
1978 1978 actualphase = phasecache.phase(unfi, cl.rev(n))
1979 1979 if actualphase != expectedphase:
1980 1980 finalmsg = msg % (nodemod.short(n),
1981 1981 phases.phasenames[actualphase],
1982 1982 phases.phasenames[expectedphase])
1983 1983 raise error.PushRaced(finalmsg)
1984 1984
1985 1985 @parthandler('output')
1986 1986 def handleoutput(op, inpart):
1987 1987 """forward output captured on the server to the client"""
1988 1988 for line in inpart.read().splitlines():
1989 1989 op.ui.status(_('remote: %s\n') % line)
1990 1990
1991 1991 @parthandler('replycaps')
1992 1992 def handlereplycaps(op, inpart):
1993 1993 """Notify that a reply bundle should be created
1994 1994
1995 1995 The payload contains the capabilities information for the reply"""
1996 1996 caps = decodecaps(inpart.read())
1997 1997 if op.reply is None:
1998 1998 op.reply = bundle20(op.ui, caps)
1999 1999
2000 2000 class AbortFromPart(error.Abort):
2001 2001 """Sub-class of Abort that denotes an error from a bundle2 part."""
2002 2002
2003 2003 @parthandler('error:abort', ('message', 'hint'))
2004 2004 def handleerrorabort(op, inpart):
2005 2005 """Used to transmit abort error over the wire"""
2006 2006 raise AbortFromPart(inpart.params['message'],
2007 2007 hint=inpart.params.get('hint'))
2008 2008
2009 2009 @parthandler('error:pushkey', ('namespace', 'key', 'new', 'old', 'ret',
2010 2010 'in-reply-to'))
2011 2011 def handleerrorpushkey(op, inpart):
2012 2012 """Used to transmit failure of a mandatory pushkey over the wire"""
2013 2013 kwargs = {}
2014 2014 for name in ('namespace', 'key', 'new', 'old', 'ret'):
2015 2015 value = inpart.params.get(name)
2016 2016 if value is not None:
2017 2017 kwargs[name] = value
2018 2018 raise error.PushkeyFailed(inpart.params['in-reply-to'],
2019 2019 **pycompat.strkwargs(kwargs))
2020 2020
2021 2021 @parthandler('error:unsupportedcontent', ('parttype', 'params'))
2022 2022 def handleerrorunsupportedcontent(op, inpart):
2023 2023 """Used to transmit unknown content error over the wire"""
2024 2024 kwargs = {}
2025 2025 parttype = inpart.params.get('parttype')
2026 2026 if parttype is not None:
2027 2027 kwargs['parttype'] = parttype
2028 2028 params = inpart.params.get('params')
2029 2029 if params is not None:
2030 2030 kwargs['params'] = params.split('\0')
2031 2031
2032 2032 raise error.BundleUnknownFeatureError(**pycompat.strkwargs(kwargs))
2033 2033
2034 2034 @parthandler('error:pushraced', ('message',))
2035 2035 def handleerrorpushraced(op, inpart):
2036 2036 """Used to transmit push race error over the wire"""
2037 2037 raise error.ResponseError(_('push failed:'), inpart.params['message'])
2038 2038
2039 2039 @parthandler('listkeys', ('namespace',))
2040 2040 def handlelistkeys(op, inpart):
2041 2041 """retrieve pushkey namespace content stored in a bundle2"""
2042 2042 namespace = inpart.params['namespace']
2043 2043 r = pushkey.decodekeys(inpart.read())
2044 2044 op.records.add('listkeys', (namespace, r))
2045 2045
2046 2046 @parthandler('pushkey', ('namespace', 'key', 'old', 'new'))
2047 2047 def handlepushkey(op, inpart):
2048 2048 """process a pushkey request"""
2049 2049 dec = pushkey.decode
2050 2050 namespace = dec(inpart.params['namespace'])
2051 2051 key = dec(inpart.params['key'])
2052 2052 old = dec(inpart.params['old'])
2053 2053 new = dec(inpart.params['new'])
2054 2054 # Grab the transaction to ensure that we have the lock before performing the
2055 2055 # pushkey.
2056 2056 if op.ui.configbool('experimental', 'bundle2lazylocking'):
2057 2057 op.gettransaction()
2058 2058 ret = op.repo.pushkey(namespace, key, old, new)
2059 2059 record = {'namespace': namespace,
2060 2060 'key': key,
2061 2061 'old': old,
2062 2062 'new': new}
2063 2063 op.records.add('pushkey', record)
2064 2064 if op.reply is not None:
2065 2065 rpart = op.reply.newpart('reply:pushkey')
2066 2066 rpart.addparam(
2067 2067 'in-reply-to', pycompat.bytestr(inpart.id), mandatory=False)
2068 2068 rpart.addparam('return', '%i' % ret, mandatory=False)
2069 2069 if inpart.mandatory and not ret:
2070 2070 kwargs = {}
2071 2071 for key in ('namespace', 'key', 'new', 'old', 'ret'):
2072 2072 if key in inpart.params:
2073 2073 kwargs[key] = inpart.params[key]
2074 2074 raise error.PushkeyFailed(partid='%d' % inpart.id,
2075 2075 **pycompat.strkwargs(kwargs))
2076 2076
2077 2077 @parthandler('bookmarks')
2078 2078 def handlebookmark(op, inpart):
2079 2079 """transmit bookmark information
2080 2080
2081 2081 The part contains binary encoded bookmark information.
2082 2082
2083 2083 The exact behavior of this part can be controlled by the 'bookmarks' mode
2084 2084 on the bundle operation.
2085 2085
2086 2086 When mode is 'apply' (the default) the bookmark information is applied as
2087 2087 is to the unbundling repository. Make sure a 'check:bookmarks' part is
2088 2088 issued earlier to check for push races in such update. This behavior is
2089 2089 suitable for pushing.
2090 2090
2091 2091 When mode is 'records', the information is recorded into the 'bookmarks'
2092 2092 records of the bundle operation. This behavior is suitable for pulling.
2093 2093 """
2094 2094 changes = bookmarks.binarydecode(inpart)
2095 2095
2096 2096 pushkeycompat = op.repo.ui.configbool('server', 'bookmarks-pushkey-compat')
2097 2097 bookmarksmode = op.modes.get('bookmarks', 'apply')
2098 2098
2099 2099 if bookmarksmode == 'apply':
2100 2100 tr = op.gettransaction()
2101 2101 bookstore = op.repo._bookmarks
2102 2102 if pushkeycompat:
2103 2103 allhooks = []
2104 2104 for book, node in changes:
2105 2105 hookargs = tr.hookargs.copy()
2106 2106 hookargs['pushkeycompat'] = '1'
2107 2107 hookargs['namespace'] = 'bookmarks'
2108 2108 hookargs['key'] = book
2109 2109 hookargs['old'] = nodemod.hex(bookstore.get(book, ''))
2110 2110 hookargs['new'] = nodemod.hex(node if node is not None else '')
2111 2111 allhooks.append(hookargs)
2112 2112
2113 2113 for hookargs in allhooks:
2114 2114 op.repo.hook('prepushkey', throw=True,
2115 2115 **pycompat.strkwargs(hookargs))
2116 2116
2117 2117 bookstore.applychanges(op.repo, op.gettransaction(), changes)
2118 2118
2119 2119 if pushkeycompat:
2120 2120 def runhook():
2121 2121 for hookargs in allhooks:
2122 2122 op.repo.hook('pushkey', **pycompat.strkwargs(hookargs))
2123 2123 op.repo._afterlock(runhook)
2124 2124
2125 2125 elif bookmarksmode == 'records':
2126 2126 for book, node in changes:
2127 2127 record = {'bookmark': book, 'node': node}
2128 2128 op.records.add('bookmarks', record)
2129 2129 else:
2130 2130 raise error.ProgrammingError('unkown bookmark mode: %s' % bookmarksmode)
2131 2131
2132 2132 @parthandler('phase-heads')
2133 2133 def handlephases(op, inpart):
2134 2134 """apply phases from bundle part to repo"""
2135 2135 headsbyphase = phases.binarydecode(inpart)
2136 2136 phases.updatephases(op.repo.unfiltered(), op.gettransaction, headsbyphase)
2137 2137
2138 2138 @parthandler('reply:pushkey', ('return', 'in-reply-to'))
2139 2139 def handlepushkeyreply(op, inpart):
2140 2140 """retrieve the result of a pushkey request"""
2141 2141 ret = int(inpart.params['return'])
2142 2142 partid = int(inpart.params['in-reply-to'])
2143 2143 op.records.add('pushkey', {'return': ret}, partid)
2144 2144
2145 2145 @parthandler('obsmarkers')
2146 2146 def handleobsmarker(op, inpart):
2147 2147 """add a stream of obsmarkers to the repo"""
2148 2148 tr = op.gettransaction()
2149 2149 markerdata = inpart.read()
2150 2150 if op.ui.config('experimental', 'obsmarkers-exchange-debug'):
2151 2151 op.ui.write(('obsmarker-exchange: %i bytes received\n')
2152 2152 % len(markerdata))
2153 2153 # The mergemarkers call will crash if marker creation is not enabled.
2154 2154 # we want to avoid this if the part is advisory.
2155 2155 if not inpart.mandatory and op.repo.obsstore.readonly:
2156 2156 op.repo.ui.debug('ignoring obsolescence markers, feature not enabled\n')
2157 2157 return
2158 2158 new = op.repo.obsstore.mergemarkers(tr, markerdata)
2159 2159 op.repo.invalidatevolatilesets()
2160 2160 if new:
2161 2161 op.repo.ui.status(_('%i new obsolescence markers\n') % new)
2162 2162 op.records.add('obsmarkers', {'new': new})
2163 2163 if op.reply is not None:
2164 2164 rpart = op.reply.newpart('reply:obsmarkers')
2165 2165 rpart.addparam(
2166 2166 'in-reply-to', pycompat.bytestr(inpart.id), mandatory=False)
2167 2167 rpart.addparam('new', '%i' % new, mandatory=False)
2168 2168
2169 2169
2170 2170 @parthandler('reply:obsmarkers', ('new', 'in-reply-to'))
2171 2171 def handleobsmarkerreply(op, inpart):
2172 2172 """retrieve the result of a pushkey request"""
2173 2173 ret = int(inpart.params['new'])
2174 2174 partid = int(inpart.params['in-reply-to'])
2175 2175 op.records.add('obsmarkers', {'new': ret}, partid)
2176 2176
2177 2177 @parthandler('hgtagsfnodes')
2178 2178 def handlehgtagsfnodes(op, inpart):
2179 2179 """Applies .hgtags fnodes cache entries to the local repo.
2180 2180
2181 2181 Payload is pairs of 20 byte changeset nodes and filenodes.
2182 2182 """
2183 2183 # Grab the transaction so we ensure that we have the lock at this point.
2184 2184 if op.ui.configbool('experimental', 'bundle2lazylocking'):
2185 2185 op.gettransaction()
2186 2186 cache = tags.hgtagsfnodescache(op.repo.unfiltered())
2187 2187
2188 2188 count = 0
2189 2189 while True:
2190 2190 node = inpart.read(20)
2191 2191 fnode = inpart.read(20)
2192 2192 if len(node) < 20 or len(fnode) < 20:
2193 2193 op.ui.debug('ignoring incomplete received .hgtags fnodes data\n')
2194 2194 break
2195 2195 cache.setfnode(node, fnode)
2196 2196 count += 1
2197 2197
2198 2198 cache.write()
2199 2199 op.ui.debug('applied %i hgtags fnodes cache entries\n' % count)
2200 2200
2201 2201 rbcstruct = struct.Struct('>III')
2202 2202
2203 2203 @parthandler('cache:rev-branch-cache')
2204 2204 def handlerbc(op, inpart):
2205 2205 """receive a rev-branch-cache payload and update the local cache
2206 2206
2207 2207 The payload is a series of data related to each branch
2208 2208
2209 2209 1) branch name length
2210 2210 2) number of open heads
2211 2211 3) number of closed heads
2212 2212 4) open heads nodes
2213 2213 5) closed heads nodes
2214 2214 """
2215 2215 total = 0
2216 2216 rawheader = inpart.read(rbcstruct.size)
2217 2217 cache = op.repo.revbranchcache()
2218 2218 cl = op.repo.unfiltered().changelog
2219 2219 while rawheader:
2220 2220 header = rbcstruct.unpack(rawheader)
2221 2221 total += header[1] + header[2]
2222 2222 utf8branch = inpart.read(header[0])
2223 2223 branch = encoding.tolocal(utf8branch)
2224 2224 for x in xrange(header[1]):
2225 2225 node = inpart.read(20)
2226 2226 rev = cl.rev(node)
2227 2227 cache.setdata(branch, rev, node, False)
2228 2228 for x in xrange(header[2]):
2229 2229 node = inpart.read(20)
2230 2230 rev = cl.rev(node)
2231 2231 cache.setdata(branch, rev, node, True)
2232 2232 rawheader = inpart.read(rbcstruct.size)
2233 2233 cache.write()
2234 2234
2235 2235 @parthandler('pushvars')
2236 2236 def bundle2getvars(op, part):
2237 2237 '''unbundle a bundle2 containing shellvars on the server'''
2238 2238 # An option to disable unbundling on server-side for security reasons
2239 2239 if op.ui.configbool('push', 'pushvars.server'):
2240 2240 hookargs = {}
2241 2241 for key, value in part.advisoryparams:
2242 2242 key = key.upper()
2243 2243 # We want pushed variables to have USERVAR_ prepended so we know
2244 2244 # they came from the --pushvar flag.
2245 2245 key = "USERVAR_" + key
2246 2246 hookargs[key] = value
2247 2247 op.addhookargs(hookargs)
2248 2248
2249 2249 @parthandler('stream2', ('requirements', 'filecount', 'bytecount'))
2250 2250 def handlestreamv2bundle(op, part):
2251 2251
2252 2252 requirements = urlreq.unquote(part.params['requirements']).split(',')
2253 2253 filecount = int(part.params['filecount'])
2254 2254 bytecount = int(part.params['bytecount'])
2255 2255
2256 2256 repo = op.repo
2257 2257 if len(repo):
2258 2258 msg = _('cannot apply stream clone to non empty repository')
2259 2259 raise error.Abort(msg)
2260 2260
2261 2261 repo.ui.debug('applying stream bundle\n')
2262 2262 streamclone.applybundlev2(repo, part, filecount, bytecount,
2263 2263 requirements)
General Comments 0
You need to be logged in to leave comments. Login now