1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
|
import py
class PerfResult:
"""Holds information about a benchmark run of a particular test run."""
def __init__(self, date=0.0, test_id="", revision=0.0,
revision_id="NONE", timestamp=0.0,
revision_date=0.0, elapsed_time=-1,
committer="", message="", nick=""):
self.__dict__.update(locals())
del self.self
class PerfResultCollection(object):
"""Holds information about several PerfResult objects. The
objects should have the same test_id and revision_id"""
def __init__(self, results=None):
if results is None:
self.results = []
else:
self.results = results[:]
#self.check()
def __repr__(self):
self.check()
if not self.results:
return "<PerfResultCollection EMPTY>"
sample = self.results[0]
return "<PerfResultCollection test_id=%s, revno=%s>" %(
sample.test_id, sample.revision)
@property
def min_elapsed(self):
return self.getfastest().elapsed_time
def getfastest(self):
x = None
for res in self.results:
if x is None or res.elapsed_time < x.elapsed_time:
x = res
return x
@property
def test_id(self):
# check for empty results?
return self.results[0].test_id
@property
def revision_id(self):
# check for empty results?
return self.results[0].revision_id
@property
def revision(self):
# check for empty results?
return self.results[0].revision
def check(self):
for s1, s2 in zip(self.results, self.results[1:]):
assert s1.revision_id == s2.revision_id
assert s1.test_id == s2.test_id
assert s1.revision == s2.revision
assert s1.date != s2.date
def append(self, sample):
self.results.append(sample)
self.check()
def extend(self, results):
self.results.extend(results)
self.check()
def __len__(self):
return len(self.results)
class PerfResultDelta:
"""represents the difference of two PerfResultCollections"""
def __init__(self, _from, _to=None):
if _from is None:
_from = _to
if _to is None:
_to = _from
if isinstance(_from, list):
_from = PerfResultCollection(_from)
if isinstance(_to, list):
_to = PerfResultCollection(_to)
assert isinstance(_from, PerfResultCollection)
assert isinstance(_to, PerfResultCollection)
assert _from.test_id == _to.test_id, (_from.test_id, _to.test_id)
self._from = _from
self._to = _to
self.test_id = self._to.test_id
self.delta = self._to.min_elapsed - self._from.min_elapsed
# percentage
m1 = self._from.min_elapsed
m2 = self._to.min_elapsed
if m1 == 0:
self.percent = 0.0
else:
self.percent = float(m2-m1) / float(m1)
class PerfTable:
"""parses performance history data files and yields PerfResult objects
through the get_results method.
if an branch is given, it is used to get more information for each
revision we have data from.
"""
branch = None
def __init__(self, iterlines = []):
""":param iterline: lines of performance history data,
e.g., history_file.realdlines()
"""
self._revision_cache = {}
self.results = list(self.parse(iterlines))
def parse(self, iterlines):
"""parse lines like
--date 1152625530.0 hacker@canonical.com-20..6dc
1906ms bzrlib....one_add_kernel_like_tree
"""
date = None
revision_id = None
for line in iterlines:
line = line.strip()
if not line:
continue
if line.startswith('--date'):
_, date, revision_id = line.split(None, 2)
date = float(date)
continue
perfresult = PerfResult(date=date, revision_id=revision_id)
elapsed_time, test_id = line.split(None, 1)
perfresult.elapsed_time = int(elapsed_time[:-2])
perfresult.test_id = test_id.strip()
yield self.annotate(perfresult)
def add_lines(self, lines):
"""add lines of performance history data """
self.results += list(self.parse(lines))
def get_time_for_revision_id(self, revision_id):
"""return the data of the revision or 0"""
if revision_id in self._revision_cache:
return self._revision_cache[revision_id][1].timestamp
return 0
def get_time(self, revision_id):
"""return revision date or the date of recording the
performance history data"""
t = self.get_time_for_revision_id(revision_id)
if t:
return t
result = list(self.get_results(revision_ids=[revision_id],
sorted_by_rev_date=False))[0]
return result.date
count = py.std.itertools.count()
def annotate(self, result):
"""Try to put extra information for each revision on the
PerfResult objects. These information are retrieved from a
branch object.
"""
#if self.branch is None:
# return result
class Branch:
revision_id = result.revision_id
nick = "fake"
self.branch = Branch()
result.revision = self.count.next()
result.revision_date = "01/01/2007"
result.message = "fake log message"
result.timestamp = 1231231.0
return result
revision_id = result.revision_id
if revision_id in self._revision_cache:
revision, rev, nick = self._revision_cache[revision_id]
else:
revision = self.branch.revision_id_to_revno(revision_id)
rev = self.branch.repository.get_revision(revision_id)
nick = self.branch._get_nick()
self._revision_cache[revision_id] = (revision, rev, nick)
result.revision = revision
result.committer = rev.committer
result.message = rev.message
result.timstamp = rev.timestamp
# XXX no format_date, but probably this whole function
# goes away soon
result.revision_date = format_date(rev.timestamp, rev.timezone or 0)
result.nick = nick
return result
def get_results(self, test_ids=None, revision_ids=None,
sorted_by_rev_date=True):
# XXX we might want to build indexes for speed
for result in self.results:
if test_ids and result.test_id not in test_ids:
continue
if revision_ids and result.revision_id not in revision_ids:
continue
yield result
def list_values_of(self, attr):
"""return a list of unique values of the specified attribute
of PerfResult objects"""
return dict.fromkeys((getattr(r, attr) for r in self.results)).keys()
def get_testid2collections(self):
"""return a mapping of test_id to list of PerfResultCollection
sorted by revision"""
test_ids = self.list_values_of('test_id')
testid2resultcollections = {}
for test_id in test_ids:
revnos = {}
for result in self.get_results(test_ids=[test_id]):
revnos.setdefault(result.revision, []).append(result)
for revno, results in revnos.iteritems():
collection = PerfResultCollection(results)
l = testid2resultcollections.setdefault(test_id, [])
l.append(collection)
# sort collection list by revision number
for collections in testid2resultcollections.itervalues():
collections.sort(lambda x,y: cmp(x.revision, y.revision))
return testid2resultcollections
|
