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 __future__ import absolute_import
21
from .lazy_import import lazy_import
22
lazy_import(globals(), """
27
annotate, # Must be lazy to avoid circular importing
42
class Annotator(object):
43
"""Class that drives performing annotations."""
45
def __init__(self, vf):
46
"""Create a new Annotator from a VersionedFile."""
50
# Map from key => number of nexts that will be built from this key
51
self._num_needed_children = {}
52
self._annotations_cache = {}
53
self._heads_provider = None
54
self._ann_tuple_cache = {}
56
def _update_needed_children(self, key, parent_keys):
57
for parent_key in parent_keys:
58
if parent_key in self._num_needed_children:
59
self._num_needed_children[parent_key] += 1
61
self._num_needed_children[parent_key] = 1
63
def _get_needed_keys(self, key):
64
"""Determine the texts we need to get from the backing vf.
66
:return: (vf_keys_needed, ann_keys_needed)
67
vf_keys_needed These are keys that we need to get from the vf
68
ann_keys_needed Texts which we have in self._text_cache but we
69
don't have annotations for. We need to yield these
70
in the proper order so that we can get proper
73
parent_map = self._parent_map
74
# We need 1 extra copy of the node we will be looking at when we are
76
self._num_needed_children[key] = 1
77
vf_keys_needed = set()
78
ann_keys_needed = set()
83
for key in needed_keys:
84
if key in self._parent_map:
85
# We don't need to lookup this key in the vf
86
if key not in self._text_cache:
87
# Extract this text from the vf
88
vf_keys_needed.add(key)
89
elif key not in self._annotations_cache:
90
# We do need to annotate
91
ann_keys_needed.add(key)
92
next_parent_map[key] = self._parent_map[key]
94
parent_lookup.append(key)
95
vf_keys_needed.add(key)
97
next_parent_map.update(self._vf.get_parent_map(parent_lookup))
98
for key, parent_keys in viewitems(next_parent_map):
99
if parent_keys is None: # No graph versionedfile
101
next_parent_map[key] = ()
102
self._update_needed_children(key, parent_keys)
103
needed_keys.update([key for key in parent_keys
104
if key not in parent_map])
105
parent_map.update(next_parent_map)
106
# _heads_provider does some graph caching, so it is only valid
107
# while self._parent_map hasn't changed
108
self._heads_provider = None
109
return vf_keys_needed, ann_keys_needed
111
def _get_needed_texts(self, key, pb=None):
112
"""Get the texts we need to properly annotate key.
114
:param key: A Key that is present in self._vf
115
:return: Yield (this_key, text, num_lines)
116
'text' is an opaque object that just has to work with whatever
117
matcher object we are using. Currently it is always 'lines' but
118
future improvements may change this to a simple text string.
120
keys, ann_keys = self._get_needed_keys(key)
122
pb.update('getting stream', 0, len(keys))
123
stream = self._vf.get_record_stream(keys, 'topological', True)
124
for idx, record in enumerate(stream):
126
pb.update('extracting', 0, len(keys))
127
if record.storage_kind == 'absent':
128
raise errors.RevisionNotPresent(record.key, self._vf)
129
this_key = record.key
130
lines = record.get_bytes_as('lines')
131
num_lines = len(lines)
132
self._text_cache[this_key] = lines
133
yield this_key, lines, num_lines
135
lines = self._text_cache[key]
136
num_lines = len(lines)
137
yield key, lines, num_lines
139
def _get_parent_annotations_and_matches(self, key, text, parent_key):
140
"""Get the list of annotations for the parent, and the matching lines.
142
:param text: The opaque value given by _get_needed_texts
143
:param parent_key: The key for the parent text
144
:return: (parent_annotations, matching_blocks)
145
parent_annotations is a list as long as the number of lines in
147
matching_blocks is a list of (parent_idx, text_idx, len) tuples
148
indicating which lines match between the two texts
150
parent_lines = self._text_cache[parent_key]
151
parent_annotations = self._annotations_cache[parent_key]
152
# PatienceSequenceMatcher should probably be part of Policy
153
matcher = patiencediff.PatienceSequenceMatcher(
154
None, parent_lines, text)
155
matching_blocks = matcher.get_matching_blocks()
156
return parent_annotations, matching_blocks
158
def _update_from_first_parent(self, key, annotations, lines, parent_key):
159
"""Reannotate this text relative to its first parent."""
161
matching_blocks) = self._get_parent_annotations_and_matches(
162
key, lines, parent_key)
164
for parent_idx, lines_idx, match_len in matching_blocks:
165
# For all matching regions we copy across the parent annotations
166
annotations[lines_idx:lines_idx + match_len] = \
167
parent_annotations[parent_idx:parent_idx + match_len]
169
def _update_from_other_parents(self, key, annotations, lines,
170
this_annotation, parent_key):
171
"""Reannotate this text relative to a second (or more) parent."""
173
matching_blocks) = self._get_parent_annotations_and_matches(
174
key, lines, parent_key)
179
# TODO: consider making all annotations unique and then using 'is'
180
# everywhere. Current results claim that isn't any faster,
181
# because of the time spent deduping
182
# deduping also saves a bit of memory. For NEWS it saves ~1MB,
183
# but that is out of 200-300MB for extracting everything, so a
184
# fairly trivial amount
185
for parent_idx, lines_idx, match_len in matching_blocks:
186
# For lines which match this parent, we will now resolve whether
187
# this parent wins over the current annotation
188
ann_sub = annotations[lines_idx:lines_idx + match_len]
189
par_sub = parent_annotations[parent_idx:parent_idx + match_len]
190
if ann_sub == par_sub:
192
for idx in range(match_len):
194
par_ann = par_sub[idx]
195
ann_idx = lines_idx + idx
199
if ann == this_annotation:
200
# Originally claimed 'this', but it was really in this
202
annotations[ann_idx] = par_ann
204
# Resolve the fact that both sides have a different value for
206
if ann == last_ann and par_ann == last_parent:
207
annotations[ann_idx] = last_res
210
new_ann.update(par_ann)
211
new_ann = tuple(sorted(new_ann))
212
annotations[ann_idx] = new_ann
214
last_parent = par_ann
217
def _record_annotation(self, key, parent_keys, annotations):
218
self._annotations_cache[key] = annotations
219
for parent_key in parent_keys:
220
num = self._num_needed_children[parent_key]
223
del self._text_cache[parent_key]
224
del self._annotations_cache[parent_key]
225
# Do we want to clean up _num_needed_children at this point as
227
self._num_needed_children[parent_key] = num
229
def _annotate_one(self, key, text, num_lines):
230
this_annotation = (key,)
231
# Note: annotations will be mutated by calls to _update_from*
232
annotations = [this_annotation] * num_lines
233
parent_keys = self._parent_map[key]
235
self._update_from_first_parent(key, annotations, text,
237
for parent in parent_keys[1:]:
238
self._update_from_other_parents(key, annotations, text,
239
this_annotation, parent)
240
self._record_annotation(key, parent_keys, annotations)
242
def add_special_text(self, key, parent_keys, text):
243
"""Add a specific text to the graph.
245
This is used to add a text which is not otherwise present in the
246
versioned file. (eg. a WorkingTree injecting 'current:' into the
247
graph to annotate the edited content.)
249
:param key: The key to use to request this text be annotated
250
:param parent_keys: The parents of this text
251
:param text: A string containing the content of the text
253
self._parent_map[key] = parent_keys
254
self._text_cache[key] = osutils.split_lines(text)
255
self._heads_provider = None
257
def annotate(self, key):
258
"""Return annotated fulltext for the given key.
260
:param key: A tuple defining the text to annotate
261
:return: ([annotations], [lines])
262
annotations is a list of tuples of keys, one for each line in lines
263
each key is a possible source for the given line.
264
lines the text of "key" as a list of lines
266
with ui.ui_factory.nested_progress_bar() as pb:
267
for text_key, text, num_lines in self._get_needed_texts(
269
self._annotate_one(text_key, text, num_lines)
271
annotations = self._annotations_cache[key]
273
raise errors.RevisionNotPresent(key, self._vf)
274
return annotations, self._text_cache[key]
276
def _get_heads_provider(self):
277
if self._heads_provider is None:
278
self._heads_provider = _mod_graph.KnownGraph(self._parent_map)
279
return self._heads_provider
281
def _resolve_annotation_tie(self, the_heads, line, tiebreaker):
282
if tiebreaker is None:
283
head = sorted(the_heads)[0]
285
# Backwards compatibility, break up the heads into pairs and
287
next_head = iter(the_heads)
288
head = next(next_head)
289
for possible_head in next_head:
290
annotated_lines = ((head, line), (possible_head, line))
291
head = tiebreaker(annotated_lines)[0]
294
def annotate_flat(self, key):
295
"""Determine the single-best-revision to source for each line.
297
This is meant as a compatibility thunk to how annotate() used to work.
298
:return: [(ann_key, line)]
299
A list of tuples with a single annotation key for each line.
301
custom_tiebreaker = annotate._break_annotation_tie
302
annotations, lines = self.annotate(key)
304
heads = self._get_heads_provider().heads
306
for annotation, line in zip(annotations, lines):
307
if len(annotation) == 1:
310
the_heads = heads(annotation)
311
if len(the_heads) == 1:
312
for head in the_heads:
313
break # get the item out of the set
315
head = self._resolve_annotation_tie(the_heads, line,