1
# Copyright (C) 2009, 2010 Canonical Ltd
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.
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.
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
17
"""Functionality for doing annotations in the 'optimal' way"""
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
22
annotate, # Must be lazy to avoid circular importing
34
class Annotator(object):
35
"""Class that drives performing annotations."""
37
def __init__(self, vf):
38
"""Create a new Annotator from a VersionedFile."""
42
# Map from key => number of nexts that will be built from this key
43
self._num_needed_children = {}
44
self._annotations_cache = {}
45
self._heads_provider = None
46
self._ann_tuple_cache = {}
48
def _update_needed_children(self, key, parent_keys):
49
for parent_key in parent_keys:
50
if parent_key in self._num_needed_children:
51
self._num_needed_children[parent_key] += 1
53
self._num_needed_children[parent_key] = 1
55
def _get_needed_keys(self, key):
56
"""Determine the texts we need to get from the backing vf.
58
:return: (vf_keys_needed, ann_keys_needed)
59
vf_keys_needed These are keys that we need to get from the vf
60
ann_keys_needed Texts which we have in self._text_cache but we
61
don't have annotations for. We need to yield these
62
in the proper order so that we can get proper
65
parent_map = self._parent_map
66
# We need 1 extra copy of the node we will be looking at when we are
68
self._num_needed_children[key] = 1
69
vf_keys_needed = set()
70
ann_keys_needed = set()
71
needed_keys = set([key])
75
for key in needed_keys:
76
if key in self._parent_map:
77
# We don't need to lookup this key in the vf
78
if key not in self._text_cache:
79
# Extract this text from the vf
80
vf_keys_needed.add(key)
81
elif key not in self._annotations_cache:
82
# We do need to annotate
83
ann_keys_needed.add(key)
84
next_parent_map[key] = self._parent_map[key]
86
parent_lookup.append(key)
87
vf_keys_needed.add(key)
89
next_parent_map.update(self._vf.get_parent_map(parent_lookup))
90
for key, parent_keys in next_parent_map.iteritems():
91
if parent_keys is None: # No graph versionedfile
93
next_parent_map[key] = ()
94
self._update_needed_children(key, parent_keys)
95
needed_keys.update([key for key in parent_keys
96
if key not in parent_map])
97
parent_map.update(next_parent_map)
98
# _heads_provider does some graph caching, so it is only valid while
99
# self._parent_map hasn't changed
100
self._heads_provider = None
101
return vf_keys_needed, ann_keys_needed
103
def _get_needed_texts(self, key, pb=None):
104
"""Get the texts we need to properly annotate key.
106
:param key: A Key that is present in self._vf
107
:return: Yield (this_key, text, num_lines)
108
'text' is an opaque object that just has to work with whatever
109
matcher object we are using. Currently it is always 'lines' but
110
future improvements may change this to a simple text string.
112
keys, ann_keys = self._get_needed_keys(key)
114
pb.update('getting stream', 0, len(keys))
115
stream = self._vf.get_record_stream(keys, 'topological', True)
116
for idx, record in enumerate(stream):
118
pb.update('extracting', 0, len(keys))
119
if record.storage_kind == 'absent':
120
raise errors.RevisionNotPresent(record.key, self._vf)
121
this_key = record.key
122
lines = osutils.chunks_to_lines(record.get_bytes_as('chunked'))
123
num_lines = len(lines)
124
self._text_cache[this_key] = lines
125
yield this_key, lines, num_lines
127
lines = self._text_cache[key]
128
num_lines = len(lines)
129
yield key, lines, num_lines
131
def _get_parent_annotations_and_matches(self, key, text, parent_key):
132
"""Get the list of annotations for the parent, and the matching lines.
134
:param text: The opaque value given by _get_needed_texts
135
:param parent_key: The key for the parent text
136
:return: (parent_annotations, matching_blocks)
137
parent_annotations is a list as long as the number of lines in
139
matching_blocks is a list of (parent_idx, text_idx, len) tuples
140
indicating which lines match between the two texts
142
parent_lines = self._text_cache[parent_key]
143
parent_annotations = self._annotations_cache[parent_key]
144
# PatienceSequenceMatcher should probably be part of Policy
145
matcher = patiencediff.PatienceSequenceMatcher(None,
147
matching_blocks = matcher.get_matching_blocks()
148
return parent_annotations, matching_blocks
150
def _update_from_first_parent(self, key, annotations, lines, parent_key):
151
"""Reannotate this text relative to its first parent."""
153
matching_blocks) = self._get_parent_annotations_and_matches(
154
key, lines, parent_key)
156
for parent_idx, lines_idx, match_len in matching_blocks:
157
# For all matching regions we copy across the parent annotations
158
annotations[lines_idx:lines_idx + match_len] = \
159
parent_annotations[parent_idx:parent_idx + match_len]
161
def _update_from_other_parents(self, key, annotations, lines,
162
this_annotation, parent_key):
163
"""Reannotate this text relative to a second (or more) parent."""
165
matching_blocks) = self._get_parent_annotations_and_matches(
166
key, lines, parent_key)
171
# TODO: consider making all annotations unique and then using 'is'
172
# everywhere. Current results claim that isn't any faster,
173
# because of the time spent deduping
174
# deduping also saves a bit of memory. For NEWS it saves ~1MB,
175
# but that is out of 200-300MB for extracting everything, so a
176
# fairly trivial amount
177
for parent_idx, lines_idx, match_len in matching_blocks:
178
# For lines which match this parent, we will now resolve whether
179
# this parent wins over the current annotation
180
ann_sub = annotations[lines_idx:lines_idx + match_len]
181
par_sub = parent_annotations[parent_idx:parent_idx + match_len]
182
if ann_sub == par_sub:
184
for idx in xrange(match_len):
186
par_ann = par_sub[idx]
187
ann_idx = lines_idx + idx
191
if ann == this_annotation:
192
# Originally claimed 'this', but it was really in this
194
annotations[ann_idx] = par_ann
196
# Resolve the fact that both sides have a different value for
198
if ann == last_ann and par_ann == last_parent:
199
annotations[ann_idx] = last_res
202
new_ann.update(par_ann)
203
new_ann = tuple(sorted(new_ann))
204
annotations[ann_idx] = new_ann
206
last_parent = par_ann
209
def _record_annotation(self, key, parent_keys, annotations):
210
self._annotations_cache[key] = annotations
211
for parent_key in parent_keys:
212
num = self._num_needed_children[parent_key]
215
del self._text_cache[parent_key]
216
del self._annotations_cache[parent_key]
217
# Do we want to clean up _num_needed_children at this point as
219
self._num_needed_children[parent_key] = num
221
def _annotate_one(self, key, text, num_lines):
222
this_annotation = (key,)
223
# Note: annotations will be mutated by calls to _update_from*
224
annotations = [this_annotation] * num_lines
225
parent_keys = self._parent_map[key]
227
self._update_from_first_parent(key, annotations, text,
229
for parent in parent_keys[1:]:
230
self._update_from_other_parents(key, annotations, text,
231
this_annotation, parent)
232
self._record_annotation(key, parent_keys, annotations)
234
def add_special_text(self, key, parent_keys, text):
235
"""Add a specific text to the graph.
237
This is used to add a text which is not otherwise present in the
238
versioned file. (eg. a WorkingTree injecting 'current:' into the
239
graph to annotate the edited content.)
241
:param key: The key to use to request this text be annotated
242
:param parent_keys: The parents of this text
243
:param text: A string containing the content of the text
245
self._parent_map[key] = parent_keys
246
self._text_cache[key] = osutils.split_lines(text)
247
self._heads_provider = None
249
def annotate(self, key):
250
"""Return annotated fulltext for the given key.
252
:param key: A tuple defining the text to annotate
253
:return: ([annotations], [lines])
254
annotations is a list of tuples of keys, one for each line in lines
255
each key is a possible source for the given line.
256
lines the text of "key" as a list of lines
258
pb = ui.ui_factory.nested_progress_bar()
260
for text_key, text, num_lines in self._get_needed_texts(key, pb=pb):
261
self._annotate_one(text_key, text, num_lines)
265
annotations = self._annotations_cache[key]
267
raise errors.RevisionNotPresent(key, self._vf)
268
return annotations, self._text_cache[key]
270
def _get_heads_provider(self):
271
if self._heads_provider is None:
272
self._heads_provider = _mod_graph.KnownGraph(self._parent_map)
273
return self._heads_provider
275
def _resolve_annotation_tie(self, the_heads, line, tiebreaker):
276
if tiebreaker is None:
277
head = sorted(the_heads)[0]
279
# Backwards compatibility, break up the heads into pairs and
281
next_head = iter(the_heads)
282
head = next_head.next()
283
for possible_head in next_head:
284
annotated_lines = ((head, line), (possible_head, line))
285
head = tiebreaker(annotated_lines)[0]
288
def annotate_flat(self, key):
289
"""Determine the single-best-revision to source for each line.
291
This is meant as a compatibility thunk to how annotate() used to work.
292
:return: [(ann_key, line)]
293
A list of tuples with a single annotation key for each line.
295
custom_tiebreaker = annotate._break_annotation_tie
296
annotations, lines = self.annotate(key)
298
heads = self._get_heads_provider().heads
300
for annotation, line in zip(annotations, lines):
301
if len(annotation) == 1:
304
the_heads = heads(annotation)
305
if len(the_heads) == 1:
306
for head in the_heads: break # get the item out of the set
308
head = self._resolve_annotation_tie(the_heads, line,
added
removed
bzrlib/_annotator_py.py
