Completed
Push — master ( e21d19...ee2f36 )
by Ionel Cristian
06:01
created

BenchmarkSession   B

Complexity

Total Complexity 52

Size/Duplication

Total Lines 239
Duplicated Lines 14.64 %

Importance

Changes 16
Bugs 0 Features 0
Metric Value
c 16
b 0
f 0
dl 35
loc 239
rs 7.9487
wmc 52

10 Methods

Rating   Name   Duplication   Size   Complexity  
D __init__() 0 63 9
A machine_info() 0 8 1
B check_regressions() 0 9 5
A finish() 0 9 2
A display() 0 15 2
B display_cprofile() 0 17 6
C handle_loading() 0 29 7
C prepare_benchmarks() 0 22 8
D handle_saving() 35 42 8
A next_num() 0 11 4

How to fix   Duplicated Code    Complexity   

Duplicated Code

Duplicate code is one of the most pungent code smells. A rule that is often used is to re-structure code once it is duplicated in three or more places.

Common duplication problems, and corresponding solutions are:

Complex Class

 Tip:   Before tackling complexity, make sure that you eliminate any duplication first. This often can reduce the size of classes significantly.

Complex classes like BenchmarkSession often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes.

Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.

1
from __future__ import division
2
from __future__ import print_function
3
4
import pytest
5
6
from .fixture import statistics
7
from .fixture import statistics_error
8
from .logger import Logger
9
from .storage import Storage
10
from .table import TableResults
11
from .utils import NAME_FORMATTERS
12
from .utils import SecondsDecimal
13
from .utils import cached_property
14
from .utils import first_or_value
15
from .utils import get_machine_id
16
from .utils import load_timer
17
from .utils import safe_dumps
18
from .utils import short_filename
19
20
21
class PerformanceRegression(pytest.UsageError):
22
    pass
23
24
25
class BenchmarkSession(object):
26
    compared_mapping = None
27
    groups = None
28
29
    def __init__(self, config):
30
        self.verbose = config.getoption("benchmark_verbose")
31
        self.logger = Logger(self.verbose, config)
32
        self.config = config
33
        self.performance_regressions = []
34
        self.benchmarks = []
35
        self.machine_id = get_machine_id()
36
37
        self.options = dict(
38
            min_time=SecondsDecimal(config.getoption("benchmark_min_time")),
39
            min_rounds=config.getoption("benchmark_min_rounds"),
40
            max_time=SecondsDecimal(config.getoption("benchmark_max_time")),
41
            timer=load_timer(config.getoption("benchmark_timer")),
42
            calibration_precision=config.getoption("benchmark_calibration_precision"),
43
            disable_gc=config.getoption("benchmark_disable_gc"),
44
            warmup=config.getoption("benchmark_warmup"),
45
            warmup_iterations=config.getoption("benchmark_warmup_iterations"),
46
            use_cprofile=bool(config.getoption("benchmark_cprofile")),
47
        )
48
        self.skip = config.getoption("benchmark_skip")
49
        self.disabled = config.getoption("benchmark_disable") and not config.getoption("benchmark_enable")
50
        self.cprofile_sort_by = config.getoption("benchmark_cprofile")
51
52
        if config.getoption("dist", "no") != "no" and not self.skip:
53
            self.logger.warn(
54
                "BENCHMARK-U2",
55
                "Benchmarks are automatically disabled because xdist plugin is active."
56
                "Benchmarks cannot be performed reliably in a parallelized environment.",
57
                fslocation="::"
58
            )
59
            self.disabled = True
60
        if hasattr(config, "slaveinput"):
61
            self.disabled = True
62
        if not statistics:
63
            self.logger.warn(
64
                "BENCHMARK-U3",
65
                "Benchmarks are automatically disabled because we could not import `statistics`\n\n%s" %
66
                statistics_error,
67
                fslocation="::"
68
            )
69
            self.disabled = True
70
71
        self.only = config.getoption("benchmark_only")
72
        self.sort = config.getoption("benchmark_sort")
73
        self.columns = config.getoption("benchmark_columns")
74
        if self.skip and self.only:
75
            raise pytest.UsageError("Can't have both --benchmark-only and --benchmark-skip options.")
76
        if self.disabled and self.only:
77
            raise pytest.UsageError(
78
                "Can't have both --benchmark-only and --benchmark-disable options. Note that --benchmark-disable is "
79
                "automatically activated if xdist is on or you're missing the statistics dependency.")
80
        self.group_by = config.getoption("benchmark_group_by")
81
        self.save = config.getoption("benchmark_save")
82
        self.autosave = config.getoption("benchmark_autosave")
83
        self.save_data = config.getoption("benchmark_save_data")
84
        self.json = config.getoption("benchmark_json")
85
        self.compare = config.getoption("benchmark_compare")
86
        self.compare_fail = config.getoption("benchmark_compare_fail")
87
        self.name_format = NAME_FORMATTERS[config.getoption("benchmark_name")]
88
89
        self.storage = Storage(config.getoption("benchmark_storage"),
90
                               default_machine_id=self.machine_id, logger=self.logger)
91
        self.histogram = first_or_value(config.getoption("benchmark_histogram"), False)
92
93
    @cached_property
94
    def machine_info(self):
95
        obj = self.config.hook.pytest_benchmark_generate_machine_info(config=self.config)
96
        self.config.hook.pytest_benchmark_update_machine_info(
97
            config=self.config,
98
            machine_info=obj
99
        )
100
        return obj
101
102
    def prepare_benchmarks(self):
103
        for bench in self.benchmarks:
104
            if bench:
105
                compared = False
106
                for path, compared_mapping in self.compared_mapping.items():
107
                    if bench.fullname in compared_mapping:
108
                        compared = compared_mapping[bench.fullname]
109
                        source = short_filename(path, self.machine_id)
110
                        flat_bench = bench.as_dict(include_data=False, stats=False, cprofile=self.cprofile_sort_by)
111
                        flat_bench.update(compared["stats"])
112
                        flat_bench["path"] = str(path)
113
                        flat_bench["source"] = source
114
                        if self.compare_fail:
115
                            for check in self.compare_fail:
116
                                fail = check.fails(bench, flat_bench)
117
                                if fail:
118
                                    self.performance_regressions.append((self.name_format(flat_bench), fail))
119
                        yield flat_bench
120
                flat_bench = bench.as_dict(include_data=False, flat=True, cprofile=self.cprofile_sort_by)
121
                flat_bench["path"] = None
122
                flat_bench["source"] = compared and "NOW"
123
                yield flat_bench
124
125
    @property
126
    def next_num(self):
127
        files = self.storage.query("[0-9][0-9][0-9][0-9]_*")
128
        files.sort(reverse=True)
129
        if not files:
130
            return "0001"
131
        for f in files:
132
            try:
133
                return "%04i" % (int(str(f.name).split('_')[0]) + 1)
134
            except ValueError:
135
                raise
136
137
    def handle_saving(self):
138
        save = self.benchmarks and self.save or self.autosave
139
        if save or self.json:
140
            commit_info = self.config.hook.pytest_benchmark_generate_commit_info(config=self.config)
141
            self.config.hook.pytest_benchmark_update_commit_info(config=self.config, commit_info=commit_info)
142
143 View Code Duplication
        if self.json:
0 ignored issues
show
Duplication introduced by
This code seems to be duplicated in your project.
Loading history...
144
            output_json = self.config.hook.pytest_benchmark_generate_json(
145
                config=self.config,
146
                benchmarks=self.benchmarks,
147
                include_data=True,
148
                machine_info=self.machine_info,
149
                commit_info=commit_info,
150
            )
151
            self.config.hook.pytest_benchmark_update_json(
152
                config=self.config,
153
                benchmarks=self.benchmarks,
154
                output_json=output_json,
155
            )
156
            with self.json as fh:
157
                fh.write(safe_dumps(output_json, ensure_ascii=True, indent=4).encode())
158
            self.logger.info("Wrote benchmark data in: %s" % self.json, purple=True)
159
160 View Code Duplication
        if save:
0 ignored issues
show
Duplication introduced by
This code seems to be duplicated in your project.
Loading history...
161
            output_json = self.config.hook.pytest_benchmark_generate_json(
162
                config=self.config,
163
                benchmarks=self.benchmarks,
164
                include_data=self.save_data,
165
                machine_info=self.machine_info,
166
                commit_info=commit_info,
167
            )
168
            self.config.hook.pytest_benchmark_update_json(
169
                config=self.config,
170
                benchmarks=self.benchmarks,
171
                output_json=output_json,
172
            )
173
            output_file = self.storage.get("%s_%s.json" % (self.next_num, save))
174
            assert not output_file.exists()
175
176
            with output_file.open('wb') as fh:
177
                fh.write(safe_dumps(output_json, ensure_ascii=True, indent=4).encode())
178
            self.logger.info("Saved benchmark data in: %s" % output_file)
179
180
    def handle_loading(self):
181
        self.compared_mapping = {}
182
        if self.compare:
183
            if self.compare is True:
184
                compared_benchmarks = list(self.storage.load('[0-9][0-9][0-9][0-9]_'))[-1:]
185
            else:
186
                compared_benchmarks = list(self.storage.load(self.compare))
187
188
            if not compared_benchmarks:
189
                msg = "Can't compare. No benchmark files in %r" % str(self.storage)
190
                if self.compare is True:
191
                    msg += ". Can't load the previous benchmark."
192
                    code = "BENCHMARK-C2"
193
                else:
194
                    msg += " match %r." % self.compare
195
                    code = "BENCHMARK-C1"
196
                self.logger.warn(code, msg, fslocation=self.storage.location)
197
198
            for path, compared_benchmark in compared_benchmarks:
199
                self.config.hook.pytest_benchmark_compare_machine_info(
200
                    config=self.config,
201
                    benchmarksession=self,
202
                    machine_info=self.machine_info,
203
                    compared_benchmark=compared_benchmark,
204
                )
205
                self.compared_mapping[path] = dict(
206
                    (bench['fullname'], bench) for bench in compared_benchmark['benchmarks']
207
                )
208
                self.logger.info("Comparing against benchmarks from: %s" % path)
209
210
    def finish(self):
211
        self.handle_saving()
212
        self.handle_loading()
213
        prepared_benchmarks = list(self.prepare_benchmarks())
214
        if prepared_benchmarks:
215
            self.groups = self.config.hook.pytest_benchmark_group_stats(
216
                config=self.config,
217
                benchmarks=prepared_benchmarks,
218
                group_by=self.group_by
219
            )
220
221
    def display(self, tr):
222
        if not self.groups:
223
            return
224
225
        tr.ensure_newline()
226
        results_table = TableResults(
227
            columns=self.columns,
228
            sort=self.sort,
229
            histogram=self.histogram,
230
            name_format=self.name_format,
231
            logger=self.logger
232
        )
233
        results_table.display(tr, self.groups)
234
        self.check_regressions()
235
        self.display_cprofile(tr)
236
237
    def check_regressions(self):
238
        if self.compare_fail and not self.compared_mapping:
239
            raise pytest.UsageError("--benchmark-compare-fail requires valid --benchmark-compare.")
240
241
        if self.performance_regressions:
242
            self.logger.error("Performance has regressed:\n%s" % "\n".join(
243
                "\t%s - %s" % line for line in self.performance_regressions
244
            ))
245
            raise PerformanceRegression("Performance has regressed.")
246
247
    def display_cprofile(self, tr):
248
        if self.options["use_cprofile"]:
249
            tr.section("cProfile information")
250
            tr.write_line("Time in s")
251
            for group in self.groups:
252
                group_name, benchmarks = group
253
                for benchmark in benchmarks:
254
                    tr.write(benchmark["fullname"], yellow=True)
255
                    if benchmark["source"]:
256
                        tr.write_line(" ({})".format((benchmark["source"])))
257
                    else:
258
                        tr.write("\n")
259
                    tr.write_line("ncalls\ttottime\tpercall\tcumtime\tpercall\tfilename:lineno(function)")
260
                    for function_info in benchmark["cprofile"]:
261
                        line = "{ncalls_recursion}\t{tottime:.{prec}f}\t{tottime_per:.{prec}f}\t{cumtime:.{prec}f}\t{cumtime_per:.{prec}f}\t{function_name}".format(prec=4, **function_info)
262
                        tr.write_line(line)
263
                    tr.write("\n")
264