tests.py

from collections import namedtuple
import shutil
from typing import Optional, List, Tuple, T, Union, Dict
import collections
from mjtest.environment import Environment, TestMode, TEST_MODES
from os.path import join, exists, basename
import logging
import os
import multiprocessing
from mjtest.util.parallelism import available_cpu_count
from mjtest.util.utils import cprint, colored
from pprint import pprint
import shutil
import difflib

_LOG = logging.getLogger("tests")


RunResult = namedtuple("RunResult", ['count', 'failed'])


class TestSuite:
    """
    The whole set of tests.
    """

    def __init__(self, env: Environment):
        self.env = env
        self.test_cases = {} # type: Dict[str, List[TestCase]]
        self.correct_test_cases = collections.defaultdict(set) # type: Dict[str, Set[str]]
        self._load_test_cases()

    def _load_test_cases(self):
        types = [self.env.mode]#TEST_MODES#TEST_MODES[TEST_MODES.index(self.env.mode):]
        if self.env.ci_testing:
            types = TEST_MODES
        elif self.env.mode in TestMode.USE_TESTS_OF_OTHER:
            types += TestMode.USE_TESTS_OF_OTHER[self.env.mode]
        for type in types:
            self._load_test_case_type(type)


    def _load_test_case_type(self, type: str):
        dir = join(self.env.test_dir, type)
        if exists(dir):
            self._load_test_case_dir(type, dir)
        else:
            _LOG.info("Test folder {} doesn't exist".format(dir))

    def _load_test_case_dir(self, mode: str, dir: str):
        correct_test_cases = set()
        log_file = self._log_file_for_type(mode)
        if exists(log_file):
            with open(log_file) as f:
                correct_test_cases = set()
                for t in f.readlines():
                    t = t.strip()
                    if len(t) > 0:
                        self.correct_test_cases[mode].add(t)
                        correct_test_cases.add(t)
        m = mode
        if m != self.env.mode and self.env.mode in TestMode.USE_TESTS_OF_OTHER and \
            m in TestMode.USE_TESTS_OF_OTHER[self.env.mode]:
            m = self.env.mode
        file_names = []
        for root, dirs, files in os.walk(dir):
            base = os.path.relpath(root, dir)
            if dir == root:
                file_names.extend(files)
            elif base == ".preprocessed":
                continue
            else:
                file_names.extend(join(base, file) for file in files)
        for file in sorted(file_names):
            if not TestCase.has_valid_file_ending(self.env.mode, file):
                _LOG.debug("Skip file " + file)
            elif self.env.only_incorrect_tests and file in correct_test_cases:
                _LOG.info("Skip file {} as its test case was executed correctly the last run")
            else:
                file_path = join(dir, file)
                if self.env.has_to_preprocess(file_path) and self.env.is_lib_file(file_path):
                    _LOG.debug("Skip lib file '{}'".format(file))
                    continue
                preprocessed = self.env.preprocess(join(dir, file))
                test_case = TestCase.create_from_file(self.env, m, join(dir, file), preprocessed)
                if not test_case:
                    pass
                elif not test_case.can_run():
                    _LOG.debug("Skip test case '{}' because it isn't suited".format(test_case.name()))
                else:
                    if m not in self.test_cases:
                        self.test_cases[m] = []
                    self.test_cases[m].append(test_case)
        if m in self.test_cases and len(self.test_cases[m]) == 0:
            del self.test_cases[m]

    def _log_file_for_type(self, type: str):
        return join(self.env.test_dir, type, ".mjtest_correct_testcases_" + self.env.mode)

    def _add_correct_test_case(self, test_case: 'TestCase'):
        self.correct_test_cases[test_case.type].add(basename(test_case.file))

    def run(self) -> RunResult:
        ret = RunResult(0, 0)
        try:
            for mode in self.test_cases.keys():
                single_ret = RunResult(0, 0)
                if self.env.parallel:
                    single_ret = self._run_parallel(mode, available_cpu_count())
                else:
                    single_ret = self._run_sequential(mode)
                ret = RunResult(ret.count + single_ret.count, ret.failed + single_ret.failed)
        except BaseException:
            logging.exception("")
        finally:
            print("-" * 40)
            if ret.failed > 0:  # some tests failed
                print(colored("Ran {} tests, of which ".format(ret.count), "red") +
                      colored("{} failed.".format(ret.failed), "red", attrs=["bold"]))
            else:
                cprint("All {} run tests succeeded".format(ret.count), "green")
            if self.env.produce_reports and (self.env.produce_all_reports or ret.failed > 0):
                report_dir = self.env.report_dir + "." + ("successful" if ret.failed == 0 else "failed")
                try:
                    os.rename(self.env.report_dir, report_dir)
                except IOError:
                    pass
                print("A full report for each test can be found at {}".format(
                    os.path.relpath(report_dir)))

            return ret

    def _run_sequential(self, mode: str) -> RunResult:
        failed = 0
        count = 0
        for test_case in self.test_cases[mode]:
            try:
                ret = self._run_test_case(test_case)
                if ret is False or not ret.is_correct():
                    failed += 1
                else:
                    self._add_correct_test_case(test_case)
                count += 1
            except KeyboardInterrupt:
                return RunResult(count, failed)
        return RunResult(count, failed)

    def _func(self, test_case: 'TestCase'):
        ret = self._run_test_case(test_case)
        if ret is not False and ret.is_correct():
            return 0, test_case
        return 1, test_case

    def _run_parallel(self, mode: str, parallel_jobs: int) -> RunResult:
        pool = multiprocessing.Pool(parallel_jobs)
        rets = pool.map(self._func, self.test_cases[mode])
        result = RunResult(len(rets), sum(map(lambda x: x[0], rets)))
        for (suc, test_case) in rets:
            if suc == 0:
                self._add_correct_test_case(test_case)
        return result


    def _run_test_case(self, test_case: 'TestCase') -> Optional['TestResult']:
        try:
            ret = test_case.run()
            color = "green" if ret.is_correct() else "red"
            print(colored("[{result:7s}] {tc:40s}".format(
                result="SUCCESS" if ret.is_correct() else "FAIL",
                tc=test_case.name()), color, attrs=["bold"]) +
                  colored("" if ret.is_correct() else ret.short_message(), color))
            try:
                if self.env.produce_reports and (self.env.produce_all_reports or not ret.is_correct()):
                    if not exists(self.env.report_dir):
                        os.mkdir(self.env.report_dir)
                    rep_dir = join(self.env.report_dir, test_case.type)
                    if not exists(rep_dir):
                        try:
                            os.mkdir(rep_dir)
                        except IOError:
                            pass
                    suffix = ".correct" if ret.is_correct() else ".incorrect"
                    ret.store_at(join(rep_dir, test_case.short_name() + suffix))
                if self.env.output_incorrect_reports and not ret.is_correct():
                    print(colored("Report for failing test case {}".format(test_case.short_name()), "red", attrs=["bold"]))
                    print(colored(ret.long_message(), "red"))
                return ret
            except IOError:
                _LOG.exception("Caught i/o error while trying to store the report for '{}'"
                               .format(test_case.name()))
                return False
        except KeyboardInterrupt:
            raise
        except BaseException:
            _LOG.exception("At test case '{}'".format(test_case.short_name()))
            return False

    def store(self):
        for mode in self.correct_test_cases.keys():
            log_file = self._log_file_for_type(mode)
            try:
                try:
                    os.mkdir(os.path.dirname(log_file))
                except IOError:
                    pass
                with open(log_file, "w") as f:
                    f.write("\n".join(self.correct_test_cases[mode]))
            except IOError as e:
                _LOG.exception("Caught i/o error while storing {}".format(log_file))


class TestCase:
    """
    A single test case.
    """

    TEST_CASE_CLASSES = dict((k, []) for k in TEST_MODES)
    FILE_ENDINGS = []
    INVALID_FILE_ENDINGS = []

    def __init__(self, env: Environment, type: str, file: str, preprocessed_file: str):
        self.env = env
        self.type = type
        self.file = file
        self.preprocessed_file = preprocessed_file

    def should_succeed(self) -> bool:
        raise NotImplementedError()

    def can_run(self, mode: str = "") -> bool:
        mode = mode or self.env.mode
        same_mode = self.type == mode
        types = TEST_MODES[TEST_MODES.index(self.env.mode):]
        if self.env.ci_testing:
            return same_mode or \
                   (self.type in types and self.should_succeed()) or \
                   (self.type not in types and not self.should_succeed())
        else:
            return same_mode

    def run(self) -> 'TestResult':
        raise NotImplementedError()

    @classmethod
    def create_from_file(cls, env: Environment, mode: str, file: str, preprocessed_file: str) -> Optional['TestCase']:
        if cls.has_valid_file_ending(env.mode, file):
            return cls._test_case_class_for_file(env.mode, file)(env, mode, file, preprocessed_file)
        return None

    def name(self):
        return "{}:{}".format(self.type, self.short_name())

    def short_name(self) -> str:
        raise NotImplementedError()

    @classmethod
    def _test_case_class_for_file(cls, type: str, file: str):
        for t in cls.TEST_CASE_CLASSES[type]:
            if any(file.endswith(e) for e in t.FILE_ENDINGS) and \
                    not any(file.endswith(e) for e in t.INVALID_FILE_ENDINGS):
                return t
        return False


    @classmethod
    def has_valid_file_ending(cls, type: str, file: str):
        return cls._test_case_class_for_file(type, file) != False


class TestResult:

    def __init__(self, test_case: TestCase, error_code: int):
        self.test_case = test_case
        self.error_code = error_code

    def is_correct(self) -> bool:
        return self.succeeded() == self.test_case.should_succeed()

    def succeeded(self) -> bool:
        return self.error_code == 0

    def store_at(self, file: str):
        with open(file, "w") as f:
            print(self.long_message(), file=f)

    def short_message(self) -> str:
        raise NotImplementedError()

    def long_message(self) -> str:
        raise NotImplementedError()


class BasicTestResult(TestResult):

    def __init__(self, test_case: TestCase, error_code: int, output: str = None, error_output: str = None,
                 incorrect_msg: str = "incorrect return code"):
        super().__init__(test_case, error_code)
        self._incorrect_msg = incorrect_msg
        self._contains_error_str = error_output is not None and "error" in error_output
        self.error_output = error_output
        self.output = output
        self.other_texts = [] # type: List[Tuple[str, str, bool]]
        if output:
            self.add_additional_text("Output", output)
        if error_output:
            self.add_additional_text("Error output", error_output)

    def is_correct(self):
        if self.succeeded():
            return super().is_correct()
        else:
            return super().is_correct() and self._contains_error_str

    def short_message(self) -> str:
        if self.is_correct():
            return "correct"
        else:
            if not self.succeeded() and not self.test_case.should_succeed() and not self._contains_error_str:
                return "the error output doesn't contain the word \"error\""
            return self._incorrect_msg

    def long_message(self) -> str:
        file_content = []
        with open(self.test_case.preprocessed_file, "r") as f:
            file_content = [line.rstrip() for line in f]
        others = []
        for title, content, long_text in self.other_texts:
            if long_text:
                others.append("""
{}:

{}

""".format(title, self._ident(content)))
            else:
                others.append("""{}: {}\n""".format(title, content))
        return """{}

Source file:

{}

Return code: {}

{}
""".format(self.short_message().capitalize(), self._ident(file_content), self.error_code,
           "\n".join(others))

    def add_additional_text(self, title: str, content: str):
        self.other_texts.append((title, content, True))

    def add_additional_text_line(self, title: str, content: str):
        self.other_texts.append((title, content, False))

    def _ident(self, text: Union[str,List[str]]) -> str:
        arr = text if isinstance(text, list) else text.split("\n")
        if len(arr) == 0 or text == "":
            return ""
        arr = ["[{:04d}]    {:s}".format(i + 1, l) for (i, l) in enumerate(arr)]
        return "\n".join(arr)


class BasicDiffTestResult(BasicTestResult):

    def __init__(self, test_case: TestCase, error_code: int, output: str, error_output: str, expected_output: str):
        super().__init__(test_case, error_code, output, error_output)
        self.expected_output = expected_output
        self._is_output_correct = self.expected_output.strip() == self.output.strip()
        if self.is_correct():
            self.add_additional_text("Expected and actual output", self.output)
        elif self.succeeded() and self.test_case.should_succeed():
            self.add_additional_text("Diff[expected output, actual output]", self._output_diff())
            self.add_additional_text("Expected output", self.expected_output)
            #self.add_additional_text("Actual output", self.output)

    def is_correct(self):
        if self.succeeded():
            return super().is_correct() and self.is_output_correct()
        else:
            return super().is_correct() and self._contains_error_str

    def _output_diff(self) -> str:
        return "".join(difflib.Differ().compare(self.expected_output.splitlines(True), self.output.splitlines(True)))

    def is_output_correct(self) -> str:
        return self._is_output_correct

    def short_message(self) -> str:
        if self.is_correct():
            return "correct"
        else:
            if not self.succeeded() and not self.test_case.should_succeed() and not self._contains_error_str:
                return "the error output doesn't contain the word \"error\""
            if self.succeeded() and self.test_case.should_succeed():
                return "the actual output differs from the expected"
            return "incorrect return code"


class DiffTest(TestCase):

    FILE_ENDINGS = [".invalid.mj", ".valid.mj", ".mj"]
    OUTPUT_FILE_ENDING = ".out"
    MODE = TestMode.ast

    def __init__(self, env: Environment, type: str, file: str, preprocessed_file: str):
        super().__init__(env, type, file, preprocessed_file)
        self._should_succeed = not file.endswith(".invalid.mj")
        self._expected_output_file = file + self.OUTPUT_FILE_ENDING
        self._has_expected_output_file = exists(self._expected_output_file)

    def should_succeed(self) -> bool:
        return self._should_succeed

    def short_name(self) -> str:
        return basename(self.file)

    def run(self) -> BasicDiffTestResult:
        out, err, rtcode = self.env.run_mj_command(self.MODE, self.file)
        exp_out = ""
        if rtcode == 0 and self.should_succeed():
            if self._has_expected_output_file and self.type == self.MODE and self.env.mode == self.MODE:
                with open(self._expected_output_file, "r") as f:
                    exp_out = f.read()
            #else:
            #    _LOG.error("Expected output file for test case {}:{} is missing.".format(self.MODE, self.short_name()))
        if self.type == self.MODE and self.env.mode == self.MODE:
            return BasicDiffTestResult(self, rtcode, out.decode(), err.decode(), exp_out)
        return BasicTestResult(self, rtcode, out.decode(), err.decode())


class LexerDiffTest(DiffTest):

    MODE = TestMode.lexer

TestCase.TEST_CASE_CLASSES[TestMode.lexer].append(LexerDiffTest)

import mjtest.test.syntax_tests
import mjtest.test.ast_tests
import mjtest.test.semantic_tests
import mjtest.test.exec_tests