summaryrefslogtreecommitdiffstats
path: root/src/3rdparty/v8/test/message/testcfg.py
diff options
context:
space:
mode:
Diffstat (limited to 'src/3rdparty/v8/test/message/testcfg.py')
-rw-r--r--src/3rdparty/v8/test/message/testcfg.py218
1 files changed, 0 insertions, 218 deletions
diff --git a/src/3rdparty/v8/test/message/testcfg.py b/src/3rdparty/v8/test/message/testcfg.py
deleted file mode 100644
index 1b788d5..0000000
--- a/src/3rdparty/v8/test/message/testcfg.py
+++ /dev/null
@@ -1,218 +0,0 @@
-# Copyright 2008 the V8 project authors. All rights reserved.
-# Redistribution and use in source and binary forms, with or without
-# modification, are permitted provided that the following conditions are
-# met:
-#
-# * Redistributions of source code must retain the above copyright
-# notice, this list of conditions and the following disclaimer.
-# * Redistributions in binary form must reproduce the above
-# copyright notice, this list of conditions and the following
-# disclaimer in the documentation and/or other materials provided
-# with the distribution.
-# * Neither the name of Google Inc. nor the names of its
-# contributors may be used to endorse or promote products derived
-# from this software without specific prior written permission.
-#
-# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
-# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
-# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
-# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
-# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
-# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
-# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
-# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
-# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-
-import itertools
-import os
-import re
-
-from testrunner.local import testsuite
-from testrunner.local import utils
-from testrunner.objects import testcase
-
-
-FLAGS_PATTERN = re.compile(r"//\s+Flags:(.*)")
-
-
-class MessageTestSuite(testsuite.TestSuite):
- def __init__(self, name, root):
- super(MessageTestSuite, self).__init__(name, root)
-
- def ListTests(self, context):
- tests = []
- for dirname, dirs, files in os.walk(self.root):
- for dotted in [x for x in dirs if x.startswith('.')]:
- dirs.remove(dotted)
- dirs.sort()
- files.sort()
- for filename in files:
- if filename.endswith(".js"):
- testname = join(dirname[len(self.root) + 1:], filename[:-3])
- test = testcase.TestCase(self, testname)
- tests.append(test)
- return tests
-
- def GetFlagsForTestCase(self, testcase, context):
- source = self.GetSourceForTest(testcase)
- result = []
- flags_match = re.findall(FLAGS_PATTERN, source)
- for match in flags_match:
- result += match.strip().split()
- result += context.mode_flags
- result.append(os.path.join(self.root, testcase.path + ".js"))
- return testcase.flags + result
-
- def GetSourceForTest(self, testcase):
- filename = os.path.join(self.root, testcase.path + self.suffix())
- with open(filename) as f:
- return f.read()
-
- def _IgnoreLine(self, string):
- """Ignore empty lines, valgrind output and Android output."""
- if not string: return True
- return (string.startswith("==") or string.startswith("**") or
- string.startswith("ANDROID"))
-
- def IsFailureOutput(self, output, testpath):
- expected_path = os.path.join(self.root, testpath + ".out")
- expected_lines = []
- # Can't use utils.ReadLinesFrom() here because it strips whitespace.
- with open(expected_path) as f:
- for line in f:
- if line.startswith("#") or not line.strip(): continue
- expected_lines.append(line)
- raw_lines = output.stdout.splitlines()
- actual_lines = [ s for s in raw_lines if not self._IgnoreLine(s) ]
- env = { "basename": os.path.basename(testpath + ".js") }
- if len(expected_lines) != len(actual_lines):
- return True
- for (expected, actual) in itertools.izip(expected_lines, actual_lines):
- pattern = re.escape(expected.rstrip() % env)
- pattern = pattern.replace("\\*", ".*")
- pattern = "^%s$" % pattern
- if not re.match(pattern, actual):
- return True
- return False
-
- def StripOutputForTransmit(self, testcase):
- pass
-
-
-def GetSuite(name, root):
- return MessageTestSuite(name, root)
-
-
-# Deprecated definitions below.
-# TODO(jkummerow): Remove when SCons is no longer supported.
-
-
-import test
-from os.path import join, exists, basename, isdir
-
-class MessageTestCase(test.TestCase):
-
- def __init__(self, path, file, expected, mode, context, config):
- super(MessageTestCase, self).__init__(context, path, mode)
- self.file = file
- self.expected = expected
- self.config = config
-
- def IgnoreLine(self, str):
- """Ignore empty lines, valgrind output and Android output."""
- if not str: return True
- return (str.startswith('==') or str.startswith('**') or
- str.startswith('ANDROID'))
-
- def IsFailureOutput(self, output):
- f = file(self.expected)
- # Skip initial '#' comment and spaces
- for line in f:
- if (not line.startswith('#')) and (not line.strip()):
- break
- # Convert output lines to regexps that we can match
- env = { 'basename': basename(self.file) }
- patterns = [ ]
- for line in f:
- if not line.strip():
- continue
- pattern = re.escape(line.rstrip() % env)
- pattern = pattern.replace('\\*', '.*')
- pattern = '^%s$' % pattern
- patterns.append(pattern)
- # Compare actual output with the expected
- raw_lines = output.stdout.splitlines()
- outlines = [ s for s in raw_lines if not self.IgnoreLine(s) ]
- if len(outlines) != len(patterns):
- return True
- for i in xrange(len(patterns)):
- if not re.match(patterns[i], outlines[i]):
- return True
- return False
-
- def GetLabel(self):
- return "%s %s" % (self.mode, self.GetName())
-
- def GetName(self):
- return self.path[-1]
-
- def GetCommand(self):
- result = self.config.context.GetVmCommand(self, self.mode)
- source = open(self.file).read()
- flags_match = re.findall(FLAGS_PATTERN, source)
- for match in flags_match:
- result += match.strip().split()
- result.append(self.file)
- return result
-
- def GetSource(self):
- return (open(self.file).read()
- + "\n--- expected output ---\n"
- + open(self.expected).read())
-
-
-class MessageTestConfiguration(test.TestConfiguration):
-
- def __init__(self, context, root):
- super(MessageTestConfiguration, self).__init__(context, root)
-
- def Ls(self, path):
- if isdir(path):
- return [f[:-3] for f in os.listdir(path) if f.endswith('.js')]
- else:
- return []
-
- def ListTests(self, current_path, path, mode, variant_flags):
- mjsunit = [current_path + [t] for t in self.Ls(self.root)]
- regress = [current_path + ['regress', t] for t in self.Ls(join(self.root, 'regress'))]
- bugs = [current_path + ['bugs', t] for t in self.Ls(join(self.root, 'bugs'))]
- mjsunit.sort()
- regress.sort()
- bugs.sort()
- all_tests = mjsunit + regress + bugs
- result = []
- for test in all_tests:
- if self.Contains(path, test):
- file_prefix = join(self.root, reduce(join, test[1:], ""))
- file_path = file_prefix + ".js"
- output_path = file_prefix + ".out"
- if not exists(output_path):
- print "Could not find %s" % output_path
- continue
- result.append(MessageTestCase(test, file_path, output_path, mode,
- self.context, self))
- return result
-
- def GetBuildRequirements(self):
- return ['d8']
-
- def GetTestStatus(self, sections, defs):
- status_file = join(self.root, 'message.status')
- if exists(status_file):
- test.ReadConfigurationInto(status_file, sections, defs)
-
-
-def GetConfiguration(context, root):
- return MessageTestConfiguration(context, root)