Пример #1
0
def test_service():
    global cnt
    settings = baseSettings
    settings["service"] = True
    settings["sleep"] = 1
    settings["onlyOnce"] = True
    cnt = 0
    mId = 3
    cycles = 10

    def callback(file, id, processingCrawler):
        global cnt
        cnt = cnt + 1
        assert id == mId, "Argurments doesn't match the expected. Got {} instead of {}".format(
            id, mId)
        if cnt >= cycles:
            processingCrawler.settings["service"] = False

    mCrawler = crawler.createCrawler(settings, callback)
    startTime = datetime.datetime.now()
    mCrawler.process(mId, mCrawler)
    endTime = datetime.datetime.now()
    diffTime = endTime - startTime

    def checkTime(seconds):
        if seconds > (cycles - 1) * settings["sleep"] and seconds < (
                cycles + 1) * settings["sleep"]:
            return True
        return False

    assert checkTime(
        diffTime.seconds), "Test took {}s, expceted time would be {}s".format(
            diffTime.seconds, cycles * settings["sleep"])
    assert cnt == cycles, "Wrong number of cycles. Got {} instead of {}".format(
        cnt, cycles)
Пример #2
0
def test_save():
    dummyName = "dummyMemo"
    mCrawler = crawler.createCrawler(baseSettings)
    mCrawler.memo.append(dummyName)
    mCrawler.save()
    with open(baseSettings["memo"], 'rb') as f:
        data = json.load(f)
    assert dummyName in data, "Didn't found {} in {}".format(
        dummyName, baseSettings["memo"])
Пример #3
0
def test_load():
    dummyName = "dummyLoad"
    data = [dummyName]
    with open(baseSettings['memo'], 'w') as f:
        json.dump(data, f, indent=4)
    mCrawler = crawler.createCrawler(baseSettings)
    assert len(
        mCrawler.memo) == 1, "Crawler memo contains not exactly one item"
    assert mCrawler.memo[
        0] == dummyName, "Crawlers memo contains {} instead of {}".format(
            mCrawler.memo[0], dummyName)
Пример #4
0
def test_createCrawlerFactory():
    settings = baseSettings

    tests = [["local", crawler.localCrawler],
             ["google", crawler.googleCrawler], ["git", crawler.gitCrawler]]
    for test in tests:
        settings["type"] = test[0]
        myCrawler = crawler.createCrawler(settings)
        testCrawler = test[1](settings)
        assert type(myCrawler) == type(
            testCrawler
        ), "Wrong crawler type was created. Created crawler was: {}".format(
            type(myCrawler))
Пример #5
0
def find_list(baseSettings, tests):
    for i, test in enumerate(tests):
        settings = baseSettings
        for key, val in test[0].items():
            settings[key] = val
        myCrawler = crawler.createCrawler(settings)
        myCrawler.memo = []
        try:
            results = myCrawler.getList()
            assert len(
                results) == test[1], "Found {} instead of {} files".format(
                    len(results), test[1])
            if len(test[2]) > 0:
                for name in results:
                    assert name in test[
                        2], "Unexpected file ({}) in Test {} appeared in found files. Expected {}".format(
                            name, i, test[2])
        except ValueError as VE:
            assert settings["onlyOnce"] == False, "Unexpected exeption raises"
Пример #6
0
def callback_singleReturn(baseSettings, tests):
    global singleReturnCnt
    settings = baseSettings
    settings["onlyOnce"] = False

    for test in tests:
        for key, val in test[0].items():
            settings[key] = val
        singleReturnCnt = 0

        def callback(file):
            global singleReturnCnt
            if len(test[2]) > 0:
                assert file in test[2], "Couldn't find file ({}) in {}".format(
                    file, test[2])
            singleReturnCnt += 1

        myCrawler = crawler.createCrawler(settings, callback)
        myCrawler.process()
        assert singleReturnCnt == test[
            1], "Found {} instead of {} files".format(singleReturnCnt, test[1])
Пример #7
0
def find_gen(baseSettings, tests):
    for i, test in enumerate(tests):
        settings = baseSettings
        for key, val in test[0].items():
            settings[key] = val
        myCrawler = crawler.createCrawler(settings)
        myCrawler.memo = []
        gen = myCrawler.generator()
        cnt = 0
        results = []
        try:
            while True:
                name = next(gen)
                results.append(name)
                assert name in test[
                    2], "Unexpected file ({}) appeared in found files. During Test: {}".format(
                        name, i)
                cnt += 1
        except StopIteration:
            assert cnt == test[1], "Found {} instead of {} {} files".format(
                cnt, test[1], test[0])
Пример #8
0
def callback_listReturn(baseSettings, tests):
    settings = baseSettings
    settings["singleReturn"] = False

    for test in tests:
        for key, val in test[0].items():
            settings[key] = val
        settings["onlyOnce"] = True

        def callback(files):
            if len(test[2]) > 0:
                for file in files:
                    assert file in test[
                        2], "Couldn't find file ({}) in {}".format(
                            file, test[2])
            assert len(
                files) == test[1], "Found {} instead of {} files".format(
                    len(files), test[1])

        myCrawler = crawler.createCrawler(settings, callback)
        myCrawler.memo = []
        myCrawler.process()
Пример #9
0
def test_create_google_crawler():
    settings = baseSettings
    crawler.createCrawler(settings)