~bzr-pqm/bzr/bzr.dev

6341.1.3 by Jelmer Vernooij
Move search result code to vf_search module.
1
# Copyright (C) 2007-2011 Canonical Ltd
2
#
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
7
#
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
11
# GNU General Public License for more details.
12
#
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
16
6379.6.7 by Jelmer Vernooij
Move importing from future until after doc string, otherwise the doc string will disappear.
17
"""Searching in versioned file repositories."""
18
6379.6.3 by Jelmer Vernooij
Use absolute_import.
19
from __future__ import absolute_import
20
6341.1.3 by Jelmer Vernooij
Move search result code to vf_search module.
21
from bzrlib import (
22
    debug,
23
    revision,
24
    trace,
25
    )
26
27
from bzrlib.graph import (
28
    DictParentsProvider,
29
    Graph,
30
    invert_parent_map,
31
    )
32
33
34
class AbstractSearchResult(object):
35
    """The result of a search, describing a set of keys.
36
    
37
    Search results are typically used as the 'fetch_spec' parameter when
38
    fetching revisions.
39
40
    :seealso: AbstractSearch
41
    """
42
43
    def get_recipe(self):
44
        """Return a recipe that can be used to replay this search.
45
46
        The recipe allows reconstruction of the same results at a later date.
47
48
        :return: A tuple of `(search_kind_str, *details)`.  The details vary by
49
            kind of search result.
50
        """
51
        raise NotImplementedError(self.get_recipe)
52
53
    def get_network_struct(self):
54
        """Return a tuple that can be transmitted via the HPSS protocol."""
55
        raise NotImplementedError(self.get_network_struct)
56
57
    def get_keys(self):
58
        """Return the keys found in this search.
59
60
        :return: A set of keys.
61
        """
62
        raise NotImplementedError(self.get_keys)
63
64
    def is_empty(self):
65
        """Return false if the search lists 1 or more revisions."""
66
        raise NotImplementedError(self.is_empty)
67
68
    def refine(self, seen, referenced):
69
        """Create a new search by refining this search.
70
71
        :param seen: Revisions that have been satisfied.
72
        :param referenced: Revision references observed while satisfying some
73
            of this search.
74
        :return: A search result.
75
        """
76
        raise NotImplementedError(self.refine)
77
78
79
class AbstractSearch(object):
80
    """A search that can be executed, producing a search result.
81
82
    :seealso: AbstractSearchResult
83
    """
84
85
    def execute(self):
86
        """Construct a network-ready search result from this search description.
87
88
        This may take some time to search repositories, etc.
89
90
        :return: A search result (an object that implements
91
            AbstractSearchResult's API).
92
        """
93
        raise NotImplementedError(self.execute)
94
95
96
class SearchResult(AbstractSearchResult):
97
    """The result of a breadth first search.
98
99
    A SearchResult provides the ability to reconstruct the search or access a
100
    set of the keys the search found.
101
    """
102
103
    def __init__(self, start_keys, exclude_keys, key_count, keys):
104
        """Create a SearchResult.
105
106
        :param start_keys: The keys the search started at.
107
        :param exclude_keys: The keys the search excludes.
108
        :param key_count: The total number of keys (from start to but not
109
            including exclude).
110
        :param keys: The keys the search found. Note that in future we may get
111
            a SearchResult from a smart server, in which case the keys list is
112
            not necessarily immediately available.
113
        """
114
        self._recipe = ('search', start_keys, exclude_keys, key_count)
115
        self._keys = frozenset(keys)
116
117
    def __repr__(self):
118
        kind, start_keys, exclude_keys, key_count = self._recipe
119
        if len(start_keys) > 5:
120
            start_keys_repr = repr(list(start_keys)[:5])[:-1] + ', ...]'
121
        else:
122
            start_keys_repr = repr(start_keys)
123
        if len(exclude_keys) > 5:
124
            exclude_keys_repr = repr(list(exclude_keys)[:5])[:-1] + ', ...]'
125
        else:
126
            exclude_keys_repr = repr(exclude_keys)
127
        return '<%s %s:(%s, %s, %d)>' % (self.__class__.__name__,
128
            kind, start_keys_repr, exclude_keys_repr, key_count)
129
130
    def get_recipe(self):
131
        """Return a recipe that can be used to replay this search.
132
133
        The recipe allows reconstruction of the same results at a later date
134
        without knowing all the found keys. The essential elements are a list
135
        of keys to start and to stop at. In order to give reproducible
136
        results when ghosts are encountered by a search they are automatically
137
        added to the exclude list (or else ghost filling may alter the
138
        results).
139
140
        :return: A tuple ('search', start_keys_set, exclude_keys_set,
141
            revision_count). To recreate the results of this search, create a
142
            breadth first searcher on the same graph starting at start_keys.
143
            Then call next() (or next_with_ghosts()) repeatedly, and on every
144
            result, call stop_searching_any on any keys from the exclude_keys
145
            set. The revision_count value acts as a trivial cross-check - the
146
            found revisions of the new search should have as many elements as
147
            revision_count. If it does not, then additional revisions have been
148
            ghosted since the search was executed the first time and the second
149
            time.
150
        """
151
        return self._recipe
152
153
    def get_network_struct(self):
154
        start_keys = ' '.join(self._recipe[1])
155
        stop_keys = ' '.join(self._recipe[2])
156
        count = str(self._recipe[3])
157
        return (self._recipe[0], '\n'.join((start_keys, stop_keys, count)))
158
159
    def get_keys(self):
160
        """Return the keys found in this search.
161
162
        :return: A set of keys.
163
        """
164
        return self._keys
165
166
    def is_empty(self):
167
        """Return false if the search lists 1 or more revisions."""
168
        return self._recipe[3] == 0
169
170
    def refine(self, seen, referenced):
171
        """Create a new search by refining this search.
172
173
        :param seen: Revisions that have been satisfied.
174
        :param referenced: Revision references observed while satisfying some
175
            of this search.
176
        """
177
        start = self._recipe[1]
178
        exclude = self._recipe[2]
179
        count = self._recipe[3]
180
        keys = self.get_keys()
181
        # New heads = referenced + old heads - seen things - exclude
182
        pending_refs = set(referenced)
183
        pending_refs.update(start)
184
        pending_refs.difference_update(seen)
185
        pending_refs.difference_update(exclude)
186
        # New exclude = old exclude + satisfied heads
187
        seen_heads = start.intersection(seen)
188
        exclude.update(seen_heads)
189
        # keys gets seen removed
190
        keys = keys - seen
191
        # length is reduced by len(seen)
192
        count -= len(seen)
193
        return SearchResult(pending_refs, exclude, count, keys)
194
195
196
class PendingAncestryResult(AbstractSearchResult):
197
    """A search result that will reconstruct the ancestry for some graph heads.
198
199
    Unlike SearchResult, this doesn't hold the complete search result in
200
    memory, it just holds a description of how to generate it.
201
    """
202
203
    def __init__(self, heads, repo):
204
        """Constructor.
205
206
        :param heads: an iterable of graph heads.
207
        :param repo: a repository to use to generate the ancestry for the given
208
            heads.
209
        """
210
        self.heads = frozenset(heads)
211
        self.repo = repo
212
213
    def __repr__(self):
214
        if len(self.heads) > 5:
215
            heads_repr = repr(list(self.heads)[:5])[:-1]
216
            heads_repr += ', <%d more>...]' % (len(self.heads) - 5,)
217
        else:
218
            heads_repr = repr(self.heads)
219
        return '<%s heads:%s repo:%r>' % (
220
            self.__class__.__name__, heads_repr, self.repo)
221
222
    def get_recipe(self):
223
        """Return a recipe that can be used to replay this search.
224
225
        The recipe allows reconstruction of the same results at a later date.
226
227
        :seealso SearchResult.get_recipe:
228
229
        :return: A tuple ('proxy-search', start_keys_set, set(), -1)
230
            To recreate this result, create a PendingAncestryResult with the
231
            start_keys_set.
232
        """
233
        return ('proxy-search', self.heads, set(), -1)
234
235
    def get_network_struct(self):
236
        parts = ['ancestry-of']
237
        parts.extend(self.heads)
238
        return parts
239
240
    def get_keys(self):
241
        """See SearchResult.get_keys.
242
243
        Returns all the keys for the ancestry of the heads, excluding
244
        NULL_REVISION.
245
        """
246
        return self._get_keys(self.repo.get_graph())
247
248
    def _get_keys(self, graph):
249
        NULL_REVISION = revision.NULL_REVISION
250
        keys = [key for (key, parents) in graph.iter_ancestry(self.heads)
251
                if key != NULL_REVISION and parents is not None]
252
        return keys
253
254
    def is_empty(self):
255
        """Return false if the search lists 1 or more revisions."""
256
        if revision.NULL_REVISION in self.heads:
257
            return len(self.heads) == 1
258
        else:
259
            return len(self.heads) == 0
260
261
    def refine(self, seen, referenced):
262
        """Create a new search by refining this search.
263
264
        :param seen: Revisions that have been satisfied.
265
        :param referenced: Revision references observed while satisfying some
266
            of this search.
267
        """
268
        referenced = self.heads.union(referenced)
269
        return PendingAncestryResult(referenced - seen, self.repo)
270
271
272
class EmptySearchResult(AbstractSearchResult):
273
    """An empty search result."""
274
275
    def is_empty(self):
276
        return True
277
278
279
class EverythingResult(AbstractSearchResult):
280
    """A search result that simply requests everything in the repository."""
281
282
    def __init__(self, repo):
283
        self._repo = repo
284
285
    def __repr__(self):
286
        return '%s(%r)' % (self.__class__.__name__, self._repo)
287
288
    def get_recipe(self):
289
        raise NotImplementedError(self.get_recipe)
290
291
    def get_network_struct(self):
292
        return ('everything',)
293
294
    def get_keys(self):
295
        if 'evil' in debug.debug_flags:
296
            from bzrlib import remote
297
            if isinstance(self._repo, remote.RemoteRepository):
298
                # warn developers (not users) not to do this
299
                trace.mutter_callsite(
300
                    2, "EverythingResult(RemoteRepository).get_keys() is slow.")
301
        return self._repo.all_revision_ids()
302
303
    def is_empty(self):
304
        # It's ok for this to wrongly return False: the worst that can happen
305
        # is that RemoteStreamSource will initiate a get_stream on an empty
306
        # repository.  And almost all repositories are non-empty.
307
        return False
308
309
    def refine(self, seen, referenced):
310
        heads = set(self._repo.all_revision_ids())
311
        heads.difference_update(seen)
312
        heads.update(referenced)
313
        return PendingAncestryResult(heads, self._repo)
314
315
316
class EverythingNotInOther(AbstractSearch):
317
    """Find all revisions in that are in one repo but not the other."""
318
319
    def __init__(self, to_repo, from_repo, find_ghosts=False):
320
        self.to_repo = to_repo
321
        self.from_repo = from_repo
322
        self.find_ghosts = find_ghosts
323
324
    def execute(self):
325
        return self.to_repo.search_missing_revision_ids(
326
            self.from_repo, find_ghosts=self.find_ghosts)
327
328
329
class NotInOtherForRevs(AbstractSearch):
330
    """Find all revisions missing in one repo for a some specific heads."""
331
332
    def __init__(self, to_repo, from_repo, required_ids, if_present_ids=None,
333
            find_ghosts=False, limit=None):
334
        """Constructor.
335
336
        :param required_ids: revision IDs of heads that must be found, or else
337
            the search will fail with NoSuchRevision.  All revisions in their
338
            ancestry not already in the other repository will be included in
339
            the search result.
340
        :param if_present_ids: revision IDs of heads that may be absent in the
341
            source repository.  If present, then their ancestry not already
342
            found in other will be included in the search result.
343
        :param limit: maximum number of revisions to fetch
344
        """
345
        self.to_repo = to_repo
346
        self.from_repo = from_repo
347
        self.find_ghosts = find_ghosts
348
        self.required_ids = required_ids
349
        self.if_present_ids = if_present_ids
350
        self.limit = limit
351
352
    def __repr__(self):
353
        if len(self.required_ids) > 5:
354
            reqd_revs_repr = repr(list(self.required_ids)[:5])[:-1] + ', ...]'
355
        else:
356
            reqd_revs_repr = repr(self.required_ids)
357
        if self.if_present_ids and len(self.if_present_ids) > 5:
358
            ifp_revs_repr = repr(list(self.if_present_ids)[:5])[:-1] + ', ...]'
359
        else:
360
            ifp_revs_repr = repr(self.if_present_ids)
361
362
        return ("<%s from:%r to:%r find_ghosts:%r req'd:%r if-present:%r"
363
                "limit:%r>") % (
364
                self.__class__.__name__, self.from_repo, self.to_repo,
365
                self.find_ghosts, reqd_revs_repr, ifp_revs_repr,
366
                self.limit)
367
368
    def execute(self):
369
        return self.to_repo.search_missing_revision_ids(
370
            self.from_repo, revision_ids=self.required_ids,
371
            if_present_ids=self.if_present_ids, find_ghosts=self.find_ghosts,
372
            limit=self.limit)
373
374
375
def search_result_from_parent_map(parent_map, missing_keys):
376
    """Transform a parent_map into SearchResult information."""
377
    if not parent_map:
378
        # parent_map is empty or None, simple search result
379
        return [], [], 0
380
    # start_set is all the keys in the cache
381
    start_set = set(parent_map)
382
    # result set is all the references to keys in the cache
383
    result_parents = set()
384
    for parents in parent_map.itervalues():
385
        result_parents.update(parents)
386
    stop_keys = result_parents.difference(start_set)
387
    # We don't need to send ghosts back to the server as a position to
388
    # stop either.
389
    stop_keys.difference_update(missing_keys)
390
    key_count = len(parent_map)
391
    if (revision.NULL_REVISION in result_parents
392
        and revision.NULL_REVISION in missing_keys):
393
        # If we pruned NULL_REVISION from the stop_keys because it's also
394
        # in our cache of "missing" keys we need to increment our key count
395
        # by 1, because the reconsitituted SearchResult on the server will
396
        # still consider NULL_REVISION to be an included key.
397
        key_count += 1
398
    included_keys = start_set.intersection(result_parents)
399
    start_set.difference_update(included_keys)
400
    return start_set, stop_keys, key_count
401
402
403
def _run_search(parent_map, heads, exclude_keys):
404
    """Given a parent map, run a _BreadthFirstSearcher on it.
405
406
    Start at heads, walk until you hit exclude_keys. As a further improvement,
407
    watch for any heads that you encounter while walking, which means they were
408
    not heads of the search.
409
410
    This is mostly used to generate a succinct recipe for how to walk through
411
    most of parent_map.
412
413
    :return: (_BreadthFirstSearcher, set(heads_encountered_by_walking))
414
    """
415
    g = Graph(DictParentsProvider(parent_map))
416
    s = g._make_breadth_first_searcher(heads)
417
    found_heads = set()
418
    while True:
419
        try:
420
            next_revs = s.next()
421
        except StopIteration:
422
            break
423
        for parents in s._current_parents.itervalues():
424
            f_heads = heads.intersection(parents)
425
            if f_heads:
426
                found_heads.update(f_heads)
427
        stop_keys = exclude_keys.intersection(next_revs)
428
        if stop_keys:
429
            s.stop_searching_any(stop_keys)
430
    for parents in s._current_parents.itervalues():
431
        f_heads = heads.intersection(parents)
432
        if f_heads:
433
            found_heads.update(f_heads)
434
    return s, found_heads
435
436
437
def _find_possible_heads(parent_map, tip_keys, depth):
438
    """Walk backwards (towards children) through the parent_map.
439
440
    This finds 'heads' that will hopefully succinctly describe our search
441
    graph.
442
    """
443
    child_map = invert_parent_map(parent_map)
444
    heads = set()
445
    current_roots = tip_keys
446
    walked = set(current_roots)
447
    while current_roots and depth > 0:
448
        depth -= 1
449
        children = set()
450
        children_update = children.update
451
        for p in current_roots:
452
            # Is it better to pre- or post- filter the children?
453
            try:
454
                children_update(child_map[p])
455
            except KeyError:
456
                heads.add(p)
457
        # If we've seen a key before, we don't want to walk it again. Note that
458
        # 'children' stays relatively small while 'walked' grows large. So
459
        # don't use 'difference_update' here which has to walk all of 'walked'.
460
        # '.difference' is smart enough to walk only children and compare it to
461
        # walked.
462
        children = children.difference(walked)
463
        walked.update(children)
464
        current_roots = children
465
    if current_roots:
466
        # We walked to the end of depth, so these are the new tips.
467
        heads.update(current_roots)
468
    return heads
469
470
471
def limited_search_result_from_parent_map(parent_map, missing_keys, tip_keys,
472
                                          depth):
473
    """Transform a parent_map that is searching 'tip_keys' into an
474
    approximate SearchResult.
475
476
    We should be able to generate a SearchResult from a given set of starting
477
    keys, that covers a subset of parent_map that has the last step pointing at
478
    tip_keys. This is to handle the case that really-long-searches shouldn't be
479
    started from scratch on each get_parent_map request, but we *do* want to
480
    filter out some of the keys that we've already seen, so we don't get
481
    information that we already know about on every request.
482
483
    The server will validate the search (that starting at start_keys and
484
    stopping at stop_keys yields the exact key_count), so we have to be careful
485
    to give an exact recipe.
486
487
    Basic algorithm is:
488
        1) Invert parent_map to get child_map (todo: have it cached and pass it
489
           in)
490
        2) Starting at tip_keys, walk towards children for 'depth' steps.
491
        3) At that point, we have the 'start' keys.
492
        4) Start walking parent_map from 'start' keys, counting how many keys
493
           are seen, and generating stop_keys for anything that would walk
494
           outside of the parent_map.
495
496
    :param parent_map: A map from {child_id: (parent_ids,)}
497
    :param missing_keys: parent_ids that we know are unavailable
498
    :param tip_keys: the revision_ids that we are searching
499
    :param depth: How far back to walk.
500
    """
501
    if not parent_map:
502
        # No search to send, because we haven't done any searching yet.
503
        return [], [], 0
504
    heads = _find_possible_heads(parent_map, tip_keys, depth)
505
    s, found_heads = _run_search(parent_map, heads, set(tip_keys))
6341.1.5 by Jelmer Vernooij
Fix get_state().
506
    start_keys, exclude_keys, keys = s.get_state()
6341.1.3 by Jelmer Vernooij
Move search result code to vf_search module.
507
    if found_heads:
508
        # Anything in found_heads are redundant start_keys, we hit them while
509
        # walking, so we can exclude them from the start list.
510
        start_keys = set(start_keys).difference(found_heads)
6341.1.5 by Jelmer Vernooij
Fix get_state().
511
    return start_keys, exclude_keys, len(keys)