[benchmark] Python 3 Support

Author: palimondo

benchmark/scripts/Benchmark_Driver

@@ -34,6 +34,7 @@ import re
 import subprocess
 import sys
 import time
+from functools import reduce
 
 from compare_perf_tests import LogParser
 
@@ -65,7 +66,7 @@ class BenchmarkDriver(object):
 
     def _invoke(self, cmd):
         return self._subprocess.check_output(
-            cmd, stderr=self._subprocess.STDOUT)
+            cmd, stderr=self._subprocess.STDOUT, universal_newlines=True)
 
     @property
     def test_harness(self):
@@ -144,7 +145,7 @@ class BenchmarkDriver(object):
             verbose, measure_memory, quantile, gather_metadata)
         output = self._invoke(cmd)
         results = self.parser.results_from_string(output)
-        return results.items()[0][1] if test else results
+        return list(results.items())[0][1] if test else results
 
     def _cmd_run(self, test, num_samples, num_iters, sample_time, min_samples,
                  verbose, measure_memory, quantile, gather_metadata):
@@ -219,9 +220,9 @@ class BenchmarkDriver(object):
             print(format(values))
 
         def result_values(r):
-            return map(str, [r.test_num, r.name, r.num_samples, r.min,
-                             r.samples.q1, r.median, r.samples.q3, r.max,
-                             r.max_rss])
+            return [str(value) for value in
+                    [r.test_num, r.name, r.num_samples, r.min,
+                     r.samples.q1, r.median, r.samples.q3, r.max, r.max_rss]]
 
         header = ['#', 'TEST', 'SAMPLES', 'MIN(μs)', 'Q1(μs)', 'MEDIAN(μs)',
                   'Q3(μs)', 'MAX(μs)', 'MAX_RSS(B)']
@@ -303,7 +304,11 @@ class MarkdownReportHandler(logging.StreamHandler):
         msg = self.format(record)
         stream = self.stream
         try:
-            if (isinstance(msg, unicode) and
+            unicode_type = unicode  # Python 2
+        except NameError:
+            unicode_type = str  # Python 3
+        try:
+            if (isinstance(msg, unicode_type) and
                     getattr(stream, 'encoding', None)):
                 stream.write(msg.encode(stream.encoding))
             else:
@@ -415,10 +420,10 @@ class BenchmarkDoctor(object):
         setup, ratio = BenchmarkDoctor._setup_overhead(measurements)
         setup = 0 if ratio < 0.05 else setup
         runtime = min(
-            [(result.samples.min - correction) for i_series in
-             [BenchmarkDoctor._select(measurements, num_iters=i)
-              for correction in [(setup / i) for i in [1, 2]]
-              ] for result in i_series])
+            [(result.samples.min - correction) for correction, i_series in
+             [(correction, BenchmarkDoctor._select(measurements, num_iters=i))
+              for i, correction in [(i, setup // i) for i in [1, 2]]]
+             for result in i_series])
 
         threshold = 1000
         if threshold < runtime:
@@ -473,7 +478,7 @@ class BenchmarkDoctor(object):
 
     @staticmethod
     def _reasonable_setup_time(measurements):
-        setup = min([result.setup
+        setup = min([result.setup or 0
                      for result in BenchmarkDoctor._select(measurements)])
         if 200000 < setup:  # 200 ms
             BenchmarkDoctor.log_runtime.error(
@@ -537,7 +542,7 @@ class BenchmarkDoctor(object):
 
         def capped(s):
             return min(s, 200)
-        run_args = [(capped(num_samples), 1), (capped(num_samples / 2), 2)]
+        run_args = [(capped(num_samples), 1), (capped(num_samples // 2), 2)]
         opts = self.driver.args.optimization
         opts = opts if isinstance(opts, list) else [opts]
         self.log.debug(
@@ -691,6 +696,7 @@ def parse_args(args):
     subparsers = parser.add_subparsers(
         title='Swift benchmark driver commands',
         help='See COMMAND -h for additional arguments', metavar='COMMAND')
+    subparsers.required = True
 
     shared_benchmarks_parser = argparse.ArgumentParser(add_help=False)
     benchmarks_group = shared_benchmarks_parser.add_mutually_exclusive_group()

benchmark/scripts/compare_perf_tests.py

@@ -1,4 +1,4 @@
-#!/usr/bin/python
+#!/usr/bin/env python
 # -*- coding: utf-8 -*-
 
 # ===--- compare_perf_tests.py -------------------------------------------===//
@@ -33,6 +33,7 @@ class `ReportFormatter` creates the test comparison report in specified format.
 import sys
 from bisect import bisect_left, bisect_right
 from collections import namedtuple
+from functools import reduce
 from math import ceil, sqrt
 
 
@@ -164,13 +165,14 @@ def sd(self):
                 sqrt(self.S_runtime / (self.count - 1)))
 
     @staticmethod
-    def running_mean_variance((k, M_, S_), x):
+    def running_mean_variance(stats, x):
         """Compute running variance, B. P. Welford's method.
 
         See Knuth TAOCP vol 2, 3rd edition, page 232, or
         https://www.johndcook.com/blog/standard_deviation/
         M is mean, Standard Deviation is defined as sqrt(S/k-1)
         """
+        (k, M_, S_) = stats
         k = float(k + 1)
         M = M_ + (x - M_) / k
         S = S_ + (x - M_) * (x - M)
@@ -662,7 +664,7 @@ def _column_widths(self):
         def max_widths(maximum, widths):
             return map(max, zip(maximum, widths))
 
-        return reduce(max_widths, widths, [0] * 4)
+        return list(reduce(max_widths, widths, [0] * 4))
 
     def _formatted_text(self, label_formatter, ventile_formatter,
                         COLUMN_SEPARATOR, DELIMITER_ROW, SEPARATOR, SECTION):
@@ -679,7 +681,8 @@ def row(contents):
 
         def header(title, column_labels):
             labels = (column_labels if not self.single_table else
-                      map(label_formatter, (title, ) + column_labels[1:]))
+                      [label_formatter(c)
+                       for c in (title, ) + column_labels[1:]])
             h = (('' if not self.header_printed else SEPARATOR) +
                  row(labels) +
                  (row(DELIMITER_ROW) if not self.header_printed else ''))
@@ -852,8 +855,12 @@ def main():
     print(report)
 
     if args.output:
-        with open(args.output, 'w') as f:
-            f.write(report)
+        if sys.version_info < (3, 0):
+            with open(args.output, 'w') as f:
+                f.write(report)
+        else:
+            with open(args.output, 'w', encoding='utf-8') as f:
+                f.write(report)
 
 
 if __name__ == '__main__':

benchmark/scripts/test_Benchmark_Driver.py

@@ -1,4 +1,4 @@
-#!/usr/bin/python
+#!/usr/bin/env python
 # -*- coding: utf-8 -*-
 
 # ===--- test_Benchmark_Driver.py ----------------------------------------===//
@@ -17,12 +17,11 @@
 import os
 import time
 import unittest
-from StringIO import StringIO
-from imp import load_source
 
 from compare_perf_tests import PerformanceTestResult
 
 from test_utils import Mock, MockLoggingHandler, Stub, captured_output
+from test_utils import StringIO, load_source
 
 # import Benchmark_Driver  # doesn't work because it misses '.py' extension
 Benchmark_Driver = load_source(
@@ -45,7 +44,7 @@ def assert_contains(self, texts, output):
     def test_requires_command_argument(self):
         with captured_output() as (_, err):
             self.assertRaises(SystemExit, parse_args, [])
-        self.assert_contains(['usage:', 'COMMAND', 'too few arguments'],
+        self.assert_contains(['usage:', 'COMMAND', 'error:', 'arguments'],
                              err.getvalue())
 
     def test_command_help_lists_commands(self):
@@ -150,7 +149,7 @@ def __init__(self, responses=None):
         super(SubprocessMock, self).__init__(responses)
 
         def _check_output(args, stdin=None, stdout=None, stderr=None,
-                          shell=False):
+                          shell=False, universal_newlines=False):
             return self.record_and_respond(args, stdin, stdout, stderr, shell)
         self.check_output = _check_output
 
@@ -387,7 +386,7 @@ def test_log_results(self):
         def assert_log_written(out, log_file, content):
             self.assertEqual(out.getvalue(),
                              'Logging results to: ' + log_file + '\n')
-            with open(log_file, 'rU') as f:
+            with open(log_file, 'r') as f:
                 text = f.read()
             self.assertEqual(text, "formatted output")
 

benchmark/scripts/test_compare_perf_tests.py

@@ -1,4 +1,4 @@
-#!/usr/bin/python
+#!/usr/bin/env python
 # -*- coding: utf-8 -*-
 
 # ===--- test_compare_perf_tests.py --------------------------------------===//
@@ -377,8 +377,7 @@ def test_merge(self):
 1,AngryPhonebook,1,12325,12325,12325,0,12325,10510336
 1,AngryPhonebook,1,11616,11616,11616,0,11616,10502144
 1,AngryPhonebook,1,12270,12270,12270,0,12270,10498048""".split('\n')[1:]
-        results = map(PerformanceTestResult,
-                      [line.split(',') for line in tests])
+        results = [PerformanceTestResult(line.split(',')) for line in tests]
 
         def as_tuple(r):
             return (r.num_samples, r.min, r.max, round(r.mean, 2),
@@ -524,17 +523,13 @@ class OldAndNewLog(unittest.TestCase):
 
     old_results = dict([(r.name, r)
                         for r in
-                        map(PerformanceTestResult,
-                            [line.split(',')
-                             for line in
-                             old_log_content.splitlines()])])
+                        [PerformanceTestResult(line.split(','))
+                         for line in old_log_content.splitlines()]])
 
     new_results = dict([(r.name, r)
                         for r in
-                        map(PerformanceTestResult,
-                            [line.split(',')
-                             for line in
-                             new_log_content.splitlines()])])
+                        [PerformanceTestResult(line.split(','))
+                         for line in new_log_content.splitlines()]])
 
     old_results['D'] = PerformanceTestResult(
         '184,D,200,648,4,1,5,9,5,3,45,40,3,1,,,,1,1,,4,4,4,268'.split(','),
@@ -721,7 +716,7 @@ def test_results_from_merge(self):
         concatenated_logs = """4,ArrayAppend,20,23641,29000,24990,0,24990
 4,ArrayAppend,1,20000,20000,20000,0,20000"""
         results = LogParser.results_from_string(concatenated_logs)
-        self.assertEqual(results.keys(), ['ArrayAppend'])
+        self.assertEqual(list(results.keys()), ['ArrayAppend'])
         result = results['ArrayAppend']
         self.assertTrue(isinstance(result, PerformanceTestResult))
         self.assertEqual(result.min, 20000)
@@ -743,7 +738,7 @@ def test_results_from_merge_verbose(self):
     Sample 3,364245
 3,Array2D,4,363094,376131,368159,5931,369169"""
         results = LogParser.results_from_string(concatenated_logs)
-        self.assertEqual(results.keys(), ['Array2D'])
+        self.assertEqual(list(results.keys()), ['Array2D'])
         result = results['Array2D']
         self.assertTrue(isinstance(result, PerformanceTestResult))
         self.assertEqual(result.min, 350815)
@@ -1135,8 +1130,12 @@ def execute_main_with_format(self, report_format, test_output=False):
         report_out = out.getvalue()
 
         if test_output:
-            with open(report_file, 'r') as f:
-                report = f.read()
+            if sys.version_info < (3, 0):
+                with open(report_file, 'r') as f:
+                    report = f.read()
+            else:
+                with open(report_file, 'r', encoding='utf-8') as f:
+                    report = f.read()
             # because print adds newline, add one here, too:
             report_file = str(report + '\n')
         else:

benchmark/scripts/test_utils.py

@@ -1,4 +1,4 @@
-#!/usr/bin/python
+#!/usr/bin/env python
 # -*- coding: utf-8 -*-
 
 # ===--- test_utils.py ---------------------------------------------------===//
@@ -24,9 +24,21 @@
 
 import logging
 import sys
-from StringIO import StringIO
+
+# Cross-version compatibility layer
+try:
+    from StringIO import StringIO  # for Python 2
+except ImportError:
+    from io import StringIO  # for Python 3
 from contextlib import contextmanager
 
+if sys.version_info < (3, 4):  # imp.load_source is deprecated in Python 3.4
+    from imp import load_source
+else:
+    def load_source(name, path):
+        from importlib.machinery import SourceFileLoader
+        return SourceFileLoader(name, path).load_module()
+
 
 @contextmanager
 def captured_output():