14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18
from cStringIO import StringIO
20
20
import bzrlib.errors
21
from bzrlib.trace import mutter, note
21
from bzrlib.trace import mutter, note, warning
22
22
from bzrlib.branch import Branch
23
23
from bzrlib.progress import ProgressBar
26
def greedy_fetch(to_branch, from_branch, revision, pb):
24
from bzrlib.xml5 import serializer_v5
25
from bzrlib.osutils import sha_string, split_lines
26
from bzrlib.errors import NoSuchRevision
28
"""Copying of history from one branch to another.
30
The basic plan is that every branch knows the history of everything
31
that has merged into it. As the first step of a merge, pull, or
32
branch operation we copy history from the source into the destination
35
The copying is done in a slightly complicated order. We don't want to
36
add a revision to the store until everything it refers to is also
37
stored, so that if a revision is present we can totally recreate it.
38
However, we can't know what files are included in a revision until we
39
read its inventory. Therefore, we first pull the XML and hold it in
40
memory until we've updated all of the files referenced.
43
# TODO: Avoid repeatedly opening weaves so many times.
45
# XXX: This doesn't handle ghost (not present in branch) revisions at
46
# all yet. I'm not sure they really should be supported.
48
# NOTE: This doesn't copy revisions which may be present but not
49
# merged into the last revision. I'm not sure we want to do that.
51
# - get a list of revisions that need to be pulled in
52
# - for each one, pull in that revision file
53
# and get the inventory, and store the inventory with right
55
# - and get the ancestry, and store that with right parents too
56
# - and keep a note of all file ids and version seen
57
# - then go through all files; for each one get the weave,
58
# and add in all file versions
62
def greedy_fetch(to_branch, from_branch, revision=None, pb=None):
27
63
f = Fetcher(to_branch, from_branch, revision, pb)
28
64
return f.count_copied, f.failed_revisions
31
68
class Fetcher(object):
32
"""Pull history from one branch to another."""
33
def __init__(self, to_branch, from_branch, revision_limit=None, pb=None):
69
"""Pull revisions and texts from one branch to another.
71
This doesn't update the destination's history; that can be done
72
separately if desired.
75
If set, pull only up to this revision_id.
79
last_revision -- if last_revision
80
is given it will be that, otherwise the last revision of
83
count_copied -- number of revisions copied
85
count_texts -- number of file texts copied
87
def __init__(self, to_branch, from_branch, last_revision=None, pb=None):
34
88
self.to_branch = to_branch
89
self.to_weaves = to_branch.weave_store
90
self.to_control = to_branch.control_weaves
35
91
self.from_branch = from_branch
36
self.revision_limit = revision_limit
92
self.from_weaves = from_branch.weave_store
93
self.from_control = from_branch.control_weaves
94
self.failed_revisions = []
38
99
self.pb = bzrlib.ui.ui_factory.progress_bar()
41
self._scan_histories()
42
self.failed_revisions = []
47
def _scan_histories(self):
48
self.from_history = from_branch.revision_history()
49
self.required_revisions = set(from_history)
50
self.to_history = to_branch.revision_history()
51
if self.revision_limit:
52
raise NotImplementedError('sorry, revision_limit not handled yet')
53
self.need_revisions = []
54
for rev_id in self.from_history:
55
if not has_revision(self.to_branch):
56
self.need_revisions.append(rev_id)
102
self.last_revision = self._find_last_revision(last_revision)
103
mutter('fetch up to rev {%s}', self.last_revision)
104
revs_to_fetch = self._compare_ancestries()
105
self._copy_revisions(revs_to_fetch)
106
self.new_ancestry = revs_to_fetch
110
def _find_last_revision(self, last_revision):
111
"""Find the limiting source revision.
113
Every ancestor of that revision will be merged across.
115
Returns the revision_id, or returns None if there's no history
116
in the source branch."""
117
self.pb.update('get source history')
118
from_history = self.from_branch.revision_history()
119
self.pb.update('get destination history')
121
if last_revision not in from_history:
122
raise NoSuchRevision(self.from_branch, last_revision)
126
return from_history[-1]
128
return None # no history in the source branch
131
def _compare_ancestries(self):
132
"""Get a list of revisions that must be copied.
134
That is, every revision that's in the ancestry of the source
135
branch and not in the destination branch."""
136
self.pb.update('get source ancestry')
137
self.from_ancestry = self.from_branch.get_ancestry(self.last_revision)
139
dest_last_rev = self.to_branch.last_revision()
140
self.pb.update('get destination ancestry')
142
dest_ancestry = self.to_branch.get_ancestry(dest_last_rev)
145
ss = set(dest_ancestry)
147
for rev_id in self.from_ancestry:
149
to_fetch.append(rev_id)
57
150
mutter('need to get revision {%s}', rev_id)
61
while self.need_revisions:
62
rev_id = self.need_revisions.pop()
63
mutter('try to get revision {%s}', rev_id)
151
mutter('need to get %d revisions in total', len(to_fetch))
152
self.count_total = len(to_fetch)
157
def _copy_revisions(self, revs_to_fetch):
159
for rev_id in revs_to_fetch:
163
if self.to_branch.has_revision(rev_id):
165
self.pb.update('fetch revision', i, self.count_total)
166
self._copy_one_revision(rev_id)
167
self.count_copied += 1
170
def _copy_one_revision(self, rev_id):
171
"""Copy revision and everything referenced by it."""
172
mutter('copying revision {%s}', rev_id)
173
rev_xml = self.from_branch.get_revision_xml(rev_id)
174
inv_xml = self.from_branch.get_inventory_xml(rev_id)
175
rev = serializer_v5.read_revision_from_string(rev_xml)
176
inv = serializer_v5.read_inventory_from_string(inv_xml)
177
assert rev.revision_id == rev_id
178
assert rev.inventory_sha1 == sha_string(inv_xml)
179
mutter(' commiter %s, %d parents',
182
self._copy_new_texts(rev_id, inv)
183
self._copy_inventory(rev_id, inv_xml, rev.parent_ids)
184
self._copy_ancestry(rev_id, rev.parent_ids)
185
self.to_branch.revision_store.add(StringIO(rev_xml), rev_id)
188
def _copy_inventory(self, rev_id, inv_xml, parent_ids):
189
self.to_control.add_text('inventory', rev_id,
190
split_lines(inv_xml), parent_ids)
193
def _copy_ancestry(self, rev_id, parent_ids):
194
ancestry_lines = self.from_control.get_lines('ancestry', rev_id)
195
self.to_control.add_text('ancestry', rev_id, ancestry_lines,
69
def has_revision(branch, revision_id):
71
branch.get_revision_xml(revision_id)
73
except bzrlib.errors.NoSuchRevision:
77
def old_greedy_fetch(to_branch, from_branch, revision=None, pb=None):
78
"""Copy all history from one branch to another.
81
If set, copy only up to this point in the source branch.
83
@returns: number copied, missing ids
85
from_history = from_branch.revision_history()
86
required_revisions = set(from_history)
88
if revision is not None:
89
required_revisions.add(revision)
91
rev_index = from_history.index(revision)
94
if rev_index is not None:
95
from_history = from_history[:rev_index + 1]
97
from_history = [revision]
98
to_history = to_branch.revision_history()
100
for rev_id in from_history:
101
if not has_revision(to_branch, rev_id):
102
missing.append(rev_id)
104
# recurse down through the revision graph, looking for things that
107
while len(missing) > 0:
108
installed, failed = to_branch.install_revisions(from_branch,
109
revision_ids=missing,
112
required_failed = failed.intersection(required_revisions)
113
if len(required_failed) > 0:
114
raise bzrlib.errors.InstallFailed(required_failed)
115
for rev_id in failed:
116
note("Failed to install %s" % rev_id)
117
all_failed.update(failed)
119
for rev_id in missing:
121
revision = from_branch.get_revision(rev_id)
122
except bzrlib.errors.NoSuchRevision:
123
if revision in from_history:
127
for parent in [p.revision_id for p in revision.parents]:
128
if not has_revision(to_branch, parent):
129
new_missing.add(parent)
130
missing = new_missing
131
return count, all_failed
134
def old_install_revisions(branch, other, revision_ids, pb):
135
"""Copy revisions from other branch into branch.
137
This is a lower-level function used by a pull or a merge. It
138
incorporates some history from one branch into another, but
139
does not update the revision history or operate on the working
143
Sequence of revisions to copy.
146
Progress bar for copying.
149
if hasattr(other.revision_store, "prefetch"):
150
other.revision_store.prefetch(revision_ids)
151
if hasattr(other.inventory_store, "prefetch"):
152
other.inventory_store.prefetch(revision_ids)
155
pb = bzrlib.ui.ui_factory.progress_bar()
162
for i, rev_id in enumerate(revision_ids):
163
pb.update('fetching revision', i+1, len(revision_ids))
165
rev = other.get_revision(rev_id)
166
except bzrlib.errors.NoSuchRevision:
170
revisions.append(rev)
171
inv = other.get_inventory(rev_id)
172
for key, entry in inv.iter_entries():
173
if entry.text_id is None:
175
if entry.text_id not in branch.text_store:
176
needed_texts.add(entry.text_id)
180
count, cp_fail = branch.text_store.copy_multi(other.text_store,
182
count, cp_fail = branch.inventory_store.copy_multi(other.inventory_store,
184
count, cp_fail = branch.revision_store.copy_multi(other.revision_store,
187
assert len(cp_fail) == 0
188
return count, failures
199
def _copy_new_texts(self, rev_id, inv):
200
"""Copy any new texts occuring in this revision."""
201
# TODO: Rather than writing out weaves every time, hold them
202
# in memory until everything's done? But this way is nicer
203
# if it's interrupted.
204
for path, ie in inv.iter_entries():
205
if ie.kind != 'file':
207
if ie.text_version != rev_id:
209
mutter('%s {%s} is changed in this revision',
211
self._copy_one_text(rev_id, ie.file_id)
214
def _copy_one_text(self, rev_id, file_id):
215
"""Copy one file text."""
216
mutter('copy text version {%s} of file {%s}',
218
from_weave = self.from_weaves.get_weave(file_id)
219
from_idx = from_weave.lookup(rev_id)
220
from_parents = map(from_weave.idx_to_name, from_weave.parents(from_idx))
221
text_lines = from_weave.get(from_idx)
222
to_weave = self.to_weaves.get_weave_or_empty(file_id)
223
to_parents = map(to_weave.lookup, from_parents)
224
# it's ok to add even if the text is already there
225
to_weave.add(rev_id, to_parents, text_lines)
226
self.to_weaves.put_weave(file_id, to_weave)
227
self.count_texts += 1
added
removed
bzrlib/fetch.py
