1
0
forked from Mirror/wren
Files
wren/util/test.py

423 lines
12 KiB
Python
Raw Permalink Normal View History

#!/usr/bin/env python
from __future__ import print_function
Improve Travis Build & Test Coverage Build Wren for more targets, and run the test suite on both 32 and 64 bit builds. * Update the build config to test both with and without NAN_TAGGING defined. * Updatest `util/test.py` to take the executable suffix as a parameter. This allows the makefile to control which binaries will be tested. Adds a new target to the makefile to be run by travis, this runs the test suite against all of the configurations it builds. * Gcc on some 32 bit platforms was complaining about numeric overflows when -INFINITY was used. Update the logic for converting a double to a string to not explicitly check against the literal values. * Make CI builds run the tests on both 64 _and_ 32 bit builds. * If I limit the number of CPUs on my MBP I can get some of the tests to time out, I'm imagining that the specs of the Travis Macs means that the same is happening there too. Updated the test script to allow an extra few seconds for the test to complete successfully before killing it. * Due to slight differences in accuracy in some computations tests were failing on 32 bit builds. Stop comparing things quite as exactly in the cases where it is causing issues. For some reason 12.34 was refusing to compare equal to itself. Bad show 12.34 :-/. I've also updated the test so it doesn't leak handles even if the assertions fail. * Double-cast from `double` to `uint32_t` to prevent undefined behaviour on overflow of basic integers. This should hopefully prevent 32 bit test failures on Linux. * Move to a version of LibUV with a fix for the 32 bit build error on Travis.
2016-10-09 09:18:27 +01:00
from argparse import ArgumentParser
from collections import defaultdict
from os import listdir
from os.path import abspath, basename, dirname, isdir, isfile, join, realpath, relpath, splitext
import re
from subprocess import Popen, PIPE
import sys
from threading import Timer
import platform
# Runs the tests.
Improve Travis Build & Test Coverage Build Wren for more targets, and run the test suite on both 32 and 64 bit builds. * Update the build config to test both with and without NAN_TAGGING defined. * Updatest `util/test.py` to take the executable suffix as a parameter. This allows the makefile to control which binaries will be tested. Adds a new target to the makefile to be run by travis, this runs the test suite against all of the configurations it builds. * Gcc on some 32 bit platforms was complaining about numeric overflows when -INFINITY was used. Update the logic for converting a double to a string to not explicitly check against the literal values. * Make CI builds run the tests on both 64 _and_ 32 bit builds. * If I limit the number of CPUs on my MBP I can get some of the tests to time out, I'm imagining that the specs of the Travis Macs means that the same is happening there too. Updated the test script to allow an extra few seconds for the test to complete successfully before killing it. * Due to slight differences in accuracy in some computations tests were failing on 32 bit builds. Stop comparing things quite as exactly in the cases where it is causing issues. For some reason 12.34 was refusing to compare equal to itself. Bad show 12.34 :-/. I've also updated the test so it doesn't leak handles even if the assertions fail. * Double-cast from `double` to `uint32_t` to prevent undefined behaviour on overflow of basic integers. This should hopefully prevent 32 bit test failures on Linux. * Move to a version of LibUV with a fix for the 32 bit build error on Travis.
2016-10-09 09:18:27 +01:00
parser = ArgumentParser()
parser.add_argument('--suffix', default='')
Improve Travis Build & Test Coverage Build Wren for more targets, and run the test suite on both 32 and 64 bit builds. * Update the build config to test both with and without NAN_TAGGING defined. * Updatest `util/test.py` to take the executable suffix as a parameter. This allows the makefile to control which binaries will be tested. Adds a new target to the makefile to be run by travis, this runs the test suite against all of the configurations it builds. * Gcc on some 32 bit platforms was complaining about numeric overflows when -INFINITY was used. Update the logic for converting a double to a string to not explicitly check against the literal values. * Make CI builds run the tests on both 64 _and_ 32 bit builds. * If I limit the number of CPUs on my MBP I can get some of the tests to time out, I'm imagining that the specs of the Travis Macs means that the same is happening there too. Updated the test script to allow an extra few seconds for the test to complete successfully before killing it. * Due to slight differences in accuracy in some computations tests were failing on 32 bit builds. Stop comparing things quite as exactly in the cases where it is causing issues. For some reason 12.34 was refusing to compare equal to itself. Bad show 12.34 :-/. I've also updated the test so it doesn't leak handles even if the assertions fail. * Double-cast from `double` to `uint32_t` to prevent undefined behaviour on overflow of basic integers. This should hopefully prevent 32 bit test failures on Linux. * Move to a version of LibUV with a fix for the 32 bit build error on Travis.
2016-10-09 09:18:27 +01:00
parser.add_argument('suite', nargs='?')
args = parser.parse_args(sys.argv[1:])
config = args.suffix.lstrip('_d')
is_debug = args.suffix.startswith('_d')
2014-01-23 23:29:50 -08:00
WREN_DIR = dirname(dirname(realpath(__file__)))
WREN_APP = join(WREN_DIR, 'bin', 'wren_test' + args.suffix)
WREN_APP_WITH_EXT = WREN_APP
if platform.system() == "Windows":
WREN_APP_WITH_EXT += ".exe"
if not isfile(WREN_APP_WITH_EXT):
print("The binary file 'wren_test' was not found, expected it to be at " + WREN_APP)
print("In order to run the tests, you need to build Wren first!")
sys.exit(1)
# print("Wren Test Directory - " + WREN_DIR)
# print("Wren Test App - " + WREN_APP)
Improve Travis Build & Test Coverage Build Wren for more targets, and run the test suite on both 32 and 64 bit builds. * Update the build config to test both with and without NAN_TAGGING defined. * Updatest `util/test.py` to take the executable suffix as a parameter. This allows the makefile to control which binaries will be tested. Adds a new target to the makefile to be run by travis, this runs the test suite against all of the configurations it builds. * Gcc on some 32 bit platforms was complaining about numeric overflows when -INFINITY was used. Update the logic for converting a double to a string to not explicitly check against the literal values. * Make CI builds run the tests on both 64 _and_ 32 bit builds. * If I limit the number of CPUs on my MBP I can get some of the tests to time out, I'm imagining that the specs of the Travis Macs means that the same is happening there too. Updated the test script to allow an extra few seconds for the test to complete successfully before killing it. * Due to slight differences in accuracy in some computations tests were failing on 32 bit builds. Stop comparing things quite as exactly in the cases where it is causing issues. For some reason 12.34 was refusing to compare equal to itself. Bad show 12.34 :-/. I've also updated the test so it doesn't leak handles even if the assertions fail. * Double-cast from `double` to `uint32_t` to prevent undefined behaviour on overflow of basic integers. This should hopefully prevent 32 bit test failures on Linux. * Move to a version of LibUV with a fix for the 32 bit build error on Travis.
2016-10-09 09:18:27 +01:00
EXPECT_PATTERN = re.compile(r'// expect: ?(.*)')
EXPECT_ERROR_PATTERN = re.compile(r'// expect error(?! line)')
2013-11-09 21:32:57 -08:00
EXPECT_ERROR_LINE_PATTERN = re.compile(r'// expect error line (\d+)')
EXPECT_RUNTIME_ERROR_PATTERN = re.compile(r'// expect (handled )?runtime error: (.+)')
ERROR_PATTERN = re.compile(r'\[.* line (\d+)\] Error')
STACK_TRACE_PATTERN = re.compile(r'(?:\[\./)?test/.* line (\d+)\] in')
2015-01-11 21:47:29 -08:00
STDIN_PATTERN = re.compile(r'// stdin: (.*)')
SKIP_PATTERN = re.compile(r'// skip: (.*)')
NONTEST_PATTERN = re.compile(r'// nontest')
passed = 0
failed = 0
num_skipped = 0
skipped = defaultdict(int)
expectations = 0
class Test:
def __init__(self, path):
self.path = path
self.output = []
self.compile_errors = set()
self.runtime_error_line = 0
self.runtime_error_message = None
self.exit_code = 0
self.input_bytes = None
self.failures = []
def parse(self):
global num_skipped
global skipped
global expectations
input_lines = []
line_num = 1
# Note #1: we have unicode tests that require utf-8 decoding.
# Note #2: python `open` on 3.x modifies contents regarding newlines.
# To prevent this, we specify newline='' and we don't use the
# readlines/splitlines/etc family of functions, these
# employ the universal newlines concept which does this.
# We have tests that embed \r and \r\n for validation, all of which
# get manipulated in a not helpful way by these APIs.
with open(self.path, 'r', encoding="utf-8", newline='', errors='replace') as file:
data = file.read()
lines = re.split('\n|\r\n', data)
for line in lines:
if len(line) <= 0:
line_num += 1
continue
match = EXPECT_PATTERN.search(line)
if match:
self.output.append((match.group(1), line_num))
expectations += 1
match = EXPECT_ERROR_PATTERN.search(line)
if match:
self.compile_errors.add(line_num)
# If we expect a compile error, it should exit with WREN_EX_DATAERR.
self.exit_code = 65
expectations += 1
match = EXPECT_ERROR_LINE_PATTERN.search(line)
if match:
self.compile_errors.add(int(match.group(1)))
# If we expect a compile error, it should exit with WREN_EX_DATAERR.
self.exit_code = 65
expectations += 1
match = EXPECT_RUNTIME_ERROR_PATTERN.search(line)
if match:
self.runtime_error_line = line_num
self.runtime_error_message = match.group(2)
# If the runtime error isn't handled, it should exit with WREN_EX_SOFTWARE.
if match.group(1) != "handled ":
self.exit_code = 70
expectations += 1
match = STDIN_PATTERN.search(line)
if match:
input_lines.append(match.group(1))
match = SKIP_PATTERN.search(line)
if match:
num_skipped += 1
skipped[match.group(1)] += 1
return False
# Not a test file at all, so ignore it.
match = NONTEST_PATTERN.search(line)
if match:
return False
line_num += 1
# If any input is fed to the test in stdin, concatenate it into one string.
if input_lines:
self.input_bytes = "\n".join(input_lines).encode("utf-8")
# If we got here, it's a valid test.
return True
def run(self, app, type):
# Invoke wren and run the test.
test_arg = self.path
proc = Popen([app, test_arg], stdin=PIPE, stdout=PIPE, stderr=PIPE)
Improve Travis Build & Test Coverage Build Wren for more targets, and run the test suite on both 32 and 64 bit builds. * Update the build config to test both with and without NAN_TAGGING defined. * Updatest `util/test.py` to take the executable suffix as a parameter. This allows the makefile to control which binaries will be tested. Adds a new target to the makefile to be run by travis, this runs the test suite against all of the configurations it builds. * Gcc on some 32 bit platforms was complaining about numeric overflows when -INFINITY was used. Update the logic for converting a double to a string to not explicitly check against the literal values. * Make CI builds run the tests on both 64 _and_ 32 bit builds. * If I limit the number of CPUs on my MBP I can get some of the tests to time out, I'm imagining that the specs of the Travis Macs means that the same is happening there too. Updated the test script to allow an extra few seconds for the test to complete successfully before killing it. * Due to slight differences in accuracy in some computations tests were failing on 32 bit builds. Stop comparing things quite as exactly in the cases where it is causing issues. For some reason 12.34 was refusing to compare equal to itself. Bad show 12.34 :-/. I've also updated the test so it doesn't leak handles even if the assertions fail. * Double-cast from `double` to `uint32_t` to prevent undefined behaviour on overflow of basic integers. This should hopefully prevent 32 bit test failures on Linux. * Move to a version of LibUV with a fix for the 32 bit build error on Travis.
2016-10-09 09:18:27 +01:00
# If a test takes longer than five seconds, kill it.
#
# This is mainly useful for running the tests while stress testing the GC,
# which can make a few pathological tests much slower.
timed_out = [False]
def kill_process(p):
timed_out[0] = True
p.kill()
Improve Travis Build & Test Coverage Build Wren for more targets, and run the test suite on both 32 and 64 bit builds. * Update the build config to test both with and without NAN_TAGGING defined. * Updatest `util/test.py` to take the executable suffix as a parameter. This allows the makefile to control which binaries will be tested. Adds a new target to the makefile to be run by travis, this runs the test suite against all of the configurations it builds. * Gcc on some 32 bit platforms was complaining about numeric overflows when -INFINITY was used. Update the logic for converting a double to a string to not explicitly check against the literal values. * Make CI builds run the tests on both 64 _and_ 32 bit builds. * If I limit the number of CPUs on my MBP I can get some of the tests to time out, I'm imagining that the specs of the Travis Macs means that the same is happening there too. Updated the test script to allow an extra few seconds for the test to complete successfully before killing it. * Due to slight differences in accuracy in some computations tests were failing on 32 bit builds. Stop comparing things quite as exactly in the cases where it is causing issues. For some reason 12.34 was refusing to compare equal to itself. Bad show 12.34 :-/. I've also updated the test so it doesn't leak handles even if the assertions fail. * Double-cast from `double` to `uint32_t` to prevent undefined behaviour on overflow of basic integers. This should hopefully prevent 32 bit test failures on Linux. * Move to a version of LibUV with a fix for the 32 bit build error on Travis.
2016-10-09 09:18:27 +01:00
timer = Timer(5, kill_process, [proc])
try:
timer.start()
out, err = proc.communicate(self.input_bytes)
if timed_out[0]:
self.fail("Timed out.")
else:
self.validate(type == "example", proc.returncode, out, err)
finally:
timer.cancel()
def validate(self, is_example, exit_code, out, err):
if self.compile_errors and self.runtime_error_message:
self.fail("Test error: Cannot expect both compile and runtime errors.")
return
try:
out = out.decode("utf-8").replace('\r\n', '\n')
err = err.decode("utf-8").replace('\r\n', '\n')
except:
self.fail('Error decoding output.')
error_lines = err.split('\n')
# Validate that an expected runtime error occurred.
if self.runtime_error_message:
self.validate_runtime_error(error_lines)
else:
self.validate_compile_errors(error_lines)
self.validate_exit_code(exit_code, error_lines)
# Ignore output from examples.
if is_example: return
self.validate_output(out)
def validate_runtime_error(self, error_lines):
if len(error_lines) < 2:
self.fail('Expected runtime error "{0}" and got none.',
self.runtime_error_message)
return
# Skip any compile errors. This can happen if there is a compile error in
# a module loaded by the module being tested.
line = 0
while ERROR_PATTERN.search(error_lines[line]):
line += 1
if error_lines[line] != self.runtime_error_message:
self.fail('Expected runtime error "{0}" and got:',
self.runtime_error_message)
self.fail(error_lines[line])
2015-08-08 07:17:30 -07:00
# Make sure the stack trace has the right line. Skip over any lines that
# come from builtin libraries.
match = False
stack_lines = error_lines[line + 1:]
2015-08-08 07:17:30 -07:00
for stack_line in stack_lines:
match = STACK_TRACE_PATTERN.search(stack_line)
if match: break
if not match:
self.fail('Expected stack trace and got:')
2015-08-08 07:17:30 -07:00
for stack_line in stack_lines:
self.fail(stack_line)
else:
stack_line = int(match.group(1))
if stack_line != self.runtime_error_line:
self.fail('Expected runtime error on line {0} but was on line {1}.',
self.runtime_error_line, stack_line)
def validate_compile_errors(self, error_lines):
# Validate that every compile error was expected.
found_errors = set()
for line in error_lines:
match = ERROR_PATTERN.search(line)
if match:
error_line = float(match.group(1))
if error_line in self.compile_errors:
found_errors.add(error_line)
else:
self.fail('Unexpected error:')
self.fail(line)
elif line != '':
self.fail('Unexpected output on stderr:')
self.fail(line)
# Validate that every expected error occurred.
for line in self.compile_errors - found_errors:
self.fail('Missing expected error on line {0}.', line)
def validate_exit_code(self, exit_code, error_lines):
if exit_code == self.exit_code: return
self.fail('Expected return code {0} and got {1}. Stderr:',
self.exit_code, exit_code)
self.failures += error_lines
def validate_output(self, out):
# Remove the trailing last empty line.
out_lines = out.split('\n')
if out_lines[-1] == '':
del out_lines[-1]
index = 0
for line in out_lines:
if sys.version_info < (3, 0):
line = line.encode('utf-8')
if index >= len(self.output):
self.fail('Got output "{0}" when none was expected.', line)
elif self.output[index][0] != line:
self.fail('Expected output "{0}" on line {1} and got "{2}".',
self.output[index][0], self.output[index][1], line)
index += 1
while index < len(self.output):
self.fail('Missing expected output "{0}" on line {1}.',
self.output[index][0], self.output[index][1])
index += 1
def fail(self, message, *args):
if args:
message = message.format(*args)
self.failures.append(message)
def color_text(text, color):
"""Converts text to a string and wraps it in the ANSI escape sequence for
color, if supported."""
# No ANSI escapes on Windows.
if sys.platform == 'win32':
2015-07-19 10:31:55 -07:00
return str(text)
return color + str(text) + '\033[0m'
def green(text): return color_text(text, '\033[32m')
def pink(text): return color_text(text, '\033[91m')
def red(text): return color_text(text, '\033[31m')
def yellow(text): return color_text(text, '\033[33m')
def walk(dir, callback, ignored=None):
"""
Walks [dir], and executes [callback] on each file unless it is [ignored].
"""
if not ignored:
ignored = []
ignored += [".",".."]
dir = abspath(dir)
for file in [file for file in listdir(dir) if not file in ignored]:
nfile = join(dir, file)
if isdir(nfile):
walk(nfile, callback)
else:
callback(nfile)
def print_line(line=None):
# Erase the line.
print('\033[2K', end='')
# Move the cursor to the beginning.
print('\r', end='')
if line:
print(line, end='')
sys.stdout.flush()
def run_script(app, path, type):
global passed
global failed
global num_skipped
if (splitext(path)[1] != '.wren'):
return
# Check if we are just running a subset of the tests.
Improve Travis Build & Test Coverage Build Wren for more targets, and run the test suite on both 32 and 64 bit builds. * Update the build config to test both with and without NAN_TAGGING defined. * Updatest `util/test.py` to take the executable suffix as a parameter. This allows the makefile to control which binaries will be tested. Adds a new target to the makefile to be run by travis, this runs the test suite against all of the configurations it builds. * Gcc on some 32 bit platforms was complaining about numeric overflows when -INFINITY was used. Update the logic for converting a double to a string to not explicitly check against the literal values. * Make CI builds run the tests on both 64 _and_ 32 bit builds. * If I limit the number of CPUs on my MBP I can get some of the tests to time out, I'm imagining that the specs of the Travis Macs means that the same is happening there too. Updated the test script to allow an extra few seconds for the test to complete successfully before killing it. * Due to slight differences in accuracy in some computations tests were failing on 32 bit builds. Stop comparing things quite as exactly in the cases where it is causing issues. For some reason 12.34 was refusing to compare equal to itself. Bad show 12.34 :-/. I've also updated the test so it doesn't leak handles even if the assertions fail. * Double-cast from `double` to `uint32_t` to prevent undefined behaviour on overflow of basic integers. This should hopefully prevent 32 bit test failures on Linux. * Move to a version of LibUV with a fix for the 32 bit build error on Travis.
2016-10-09 09:18:27 +01:00
if args.suite:
this_test = relpath(path, join(WREN_DIR, 'test'))
Improve Travis Build & Test Coverage Build Wren for more targets, and run the test suite on both 32 and 64 bit builds. * Update the build config to test both with and without NAN_TAGGING defined. * Updatest `util/test.py` to take the executable suffix as a parameter. This allows the makefile to control which binaries will be tested. Adds a new target to the makefile to be run by travis, this runs the test suite against all of the configurations it builds. * Gcc on some 32 bit platforms was complaining about numeric overflows when -INFINITY was used. Update the logic for converting a double to a string to not explicitly check against the literal values. * Make CI builds run the tests on both 64 _and_ 32 bit builds. * If I limit the number of CPUs on my MBP I can get some of the tests to time out, I'm imagining that the specs of the Travis Macs means that the same is happening there too. Updated the test script to allow an extra few seconds for the test to complete successfully before killing it. * Due to slight differences in accuracy in some computations tests were failing on 32 bit builds. Stop comparing things quite as exactly in the cases where it is causing issues. For some reason 12.34 was refusing to compare equal to itself. Bad show 12.34 :-/. I've also updated the test so it doesn't leak handles even if the assertions fail. * Double-cast from `double` to `uint32_t` to prevent undefined behaviour on overflow of basic integers. This should hopefully prevent 32 bit test failures on Linux. * Move to a version of LibUV with a fix for the 32 bit build error on Travis.
2016-10-09 09:18:27 +01:00
if not this_test.startswith(args.suite):
return
# Update the status line.
2017-01-12 21:38:06 -08:00
print_line('({}) Passed: {} Failed: {} Skipped: {} '.format(
relpath(app, WREN_DIR), green(passed), red(failed), yellow(num_skipped)))
# Make a nice short path relative to the working directory.
# Normalize it to use "/" since, among other things, wren expects its argument
# to use that.
path = relpath(path).replace("\\", "/")
# Read the test and parse out the expectations.
test = Test(path)
if not test.parse():
# It's a skipped or non-test file.
return
test.run(app, type)
# Display the results.
if len(test.failures) == 0:
passed += 1
else:
failed += 1
print_line(red('FAIL') + ': ' + path)
2015-02-27 07:22:27 -08:00
print('')
for failure in test.failures:
print(' ' + pink(failure))
2015-02-27 07:22:27 -08:00
print('')
def run_test(path, example=False):
run_script(WREN_APP, path, "test")
def run_api_test(path):
run_script(WREN_APP, path, "api test")
def run_example(path):
2015-11-29 09:43:47 -08:00
# Don't run examples that require user input.
if "animals" in path: return
if "guess_number" in path: return
2015-11-29 09:43:47 -08:00
# This one is annoyingly slow.
if "skynet" in path: return
run_script(WREN_APP, path, "example")
walk(join(WREN_DIR, 'test'), run_test, ignored=['api', 'benchmark'])
walk(join(WREN_DIR, 'test', 'api'), run_api_test)
walk(join(WREN_DIR, 'example'), run_example)
print_line()
if failed == 0:
print('All ' + green(passed) + ' tests passed (' + str(expectations) +
' expectations).')
else:
print(green(passed) + ' tests passed. ' + red(failed) + ' tests failed.')
for key in sorted(skipped.keys()):
print('Skipped ' + yellow(skipped[key]) + ' tests: ' + key)
2015-01-02 08:13:51 -08:00
if failed != 0:
sys.exit(1)