コード例 #1
0
ファイル: fake.py プロジェクト: hellysmile/fake-useragent
    def __getattr__(self, attr):
        if attr in self.safe_attrs:
            return super(UserAgent, self).__getattr__(attr)

        try:
            for value, replacement in settings.REPLACEMENTS.items():
                attr = attr.replace(value, replacement)

            attr = attr.lower()

            if attr == 'random':
                browser = random.choice(self.data_randomize)
            else:
                browser = settings.SHORTCUTS.get(attr, attr)

            return random.choice(self.data_browsers[browser])
        except (KeyError, IndexError):
            if self.fallback is None:
                raise FakeUserAgentError('Error occurred during getting browser')  # noqa
            else:
                logger.warning(
                    'Error occurred during getting browser, '
                    'but was suppressed with fallback.',
                )

                return self.fallback
コード例 #2
0
ファイル: fake.py プロジェクト: hellysmile/fake-useragent
    def load(self):
        try:
            with self.load.lock:
                if self.cache:
                    self.data = load_cached(
                        self.path,
                        use_cache_server=self.use_cache_server,
                        verify_ssl=self.verify_ssl,
                    )
                else:
                    self.data = load(
                        use_cache_server=self.use_cache_server,
                        verify_ssl=self.verify_ssl,
                    )

                # TODO: change source file format
                # version 0.1.4+ migration tool
                self.data_randomize = list(self.data['randomize'].values())
                self.data_browsers = self.data['browsers']
        except FakeUserAgentError:
            if self.fallback is None:
                raise
            else:
                logger.warning(
                    'Error occurred during fetching data, '
                    'but was suppressed with fallback.',
                )
コード例 #3
0
ファイル: fake.py プロジェクト: chenji123/SpiderProjects
    def load(self):
        try:
            with self.load.lock:
                if self.cache:
                    self.data = load_cached(
                        self.path,
                        use_cache_server=self.use_cache_server,
                        verify_ssl=self.verify_ssl,
                    )
                else:
                    self.data = load(
                        use_cache_server=self.use_cache_server,
                        verify_ssl=self.verify_ssl,
                    )

                # TODO: change source file format
                # version 0.1.4+ migration tool
                self.data_randomize = list(self.data['randomize'].values())
                self.data_browsers = self.data['browsers']
        except FakeUserAgentError:
            if self.fallback is None:
                raise
            else:
                logger.warning(
                    'Error occurred during fetching data, '
                    'but was suppressed with fallback.', )
コード例 #4
0
ファイル: fake.py プロジェクト: chenji123/SpiderProjects
    def __getattr__(self, attr):
        if attr in self.safe_attrs:
            return super(UserAgent, self).__getattr__(attr)

        try:
            for value, replacement in settings.REPLACEMENTS.items():
                attr = attr.replace(value, replacement)

            attr = attr.lower()

            if attr == 'random':
                browser = random.choice(self.data_randomize)
            else:
                browser = settings.SHORTCUTS.get(attr, attr)

            return random.choice(self.data_browsers[browser])
        except (KeyError, IndexError):
            if self.fallback is None:
                raise FakeUserAgentError(
                    'Error occurred during getting browser')  # noqa
            else:
                logger.warning(
                    'Error occurred during getting browser, '
                    'but was suppressed with fallback.', )

                return self.fallback
コード例 #5
0
def load(use_cache_server=True):
    browsers_dict = {}
    randomize_dict = {}

    try:
        for item in get_browsers():
            browser, percent = item

            browser_key = browser

            for value, replacement in settings.REPLACEMENTS.items():
                browser_key = browser_key.replace(value, replacement)

            browser_key = browser_key.lower()

            browsers_dict[browser_key] = get_browser_versions(browser)

            # it is actually so bad way for randomizing, simple list with
            # browser_key's is event better
            # I've failed so much a lot of years ago.
            # Ideas for refactoring
            # {'chrome': <percantage|int>, 'firefox': '<percatage|int>'}
            for _ in range(int(float(percent) * 10)):
                randomize_dict[str(len(randomize_dict))] = browser_key
    except Exception as exc:
        if not use_cache_server:
            raise exc

        logger.warning(
            'Error occurred during loading data. '
            'Trying to use cache server %s',
            settings.CACHE_SERVER,
            exc_info=exc,
        )
        try:
            ret = json.loads(get(settings.CACHE_SERVER).decode('utf-8'))
        except (TypeError, ValueError):
            raise FakeUserAgentError('Can not load data from cache server')
    else:
        ret = {
            'browsers': browsers_dict,
            'randomize': randomize_dict,
        }

    if not isinstance(ret, dict):
        raise FakeUserAgentError('Data is not dictionary ', ret)

    for param in ['browsers', 'randomize']:
        if param not in ret:
            raise FakeUserAgentError('Missing data param: ', param)

        if not isinstance(ret[param], dict):
            raise FakeUserAgentError('Data param is not dictionary', ret[param])  # noqa

        if not ret[param]:
            raise FakeUserAgentError('Data param is empty', ret[param])

    return ret
コード例 #6
0
def load():
    browsers_dict = {}
    randomize_dict = {}

    try:
        for item in get_browsers():
            browser, percent = item

            browser_key = browser

            for value, replacement in settings.REPLACEMENTS.items():
                browser_key = browser_key.replace(value, replacement)

            browser_key = browser_key.lower()

            browsers_dict[browser_key] = get_browser_versions(browser)

            # it is actually so bad way for randomizing, simple list with
            # browser_key's is event better
            # I've failed so much a lot of years ago.
            # Ideas for refactoring
            # {'chrome': <percantage|int>, 'firefox': '<percatage|int>'}
            for _ in range(int(float(percent) * 10)):
                randomize_dict[str(len(randomize_dict))] = browser_key
    except Exception as exc:
        logger.warning(
            'Error occurred during formatting data. '
            'Trying to use fallback server %s',
            settings.CACHE_SERVER,
            exc_info=exc,
        )
        try:
            ret = json.loads(get(settings.CACHE_SERVER).decode('utf-8'))
        except (UnicodeDecodeError, TypeError, ValueError):
            raise FakeUserAgentError('Can not load data from cached server')
    else:
        ret = {
            'browsers': browsers_dict,
            'randomize': randomize_dict,
        }

    if not isinstance(ret, dict):
        raise FakeUserAgentError('Data is not dictionary ', ret)

    for param in ['browsers', 'randomize']:
        if param not in ret:
            raise FakeUserAgentError('Missing data param: ', param)

        if not isinstance(ret[param], dict):
            raise FakeUserAgentError('Data param is not dictionary', ret[param])  # noqa

        if not ret[param]:
            raise FakeUserAgentError('Data param is empty', ret[param])

    return ret
コード例 #7
0
    def load(self):
        try:
            with self.load.lock:
                if self.cache:
                    self.data = load_cached(
                        self.path,
                        use_cache_server=self.use_cache_server,
                        verify_ssl=self.verify_ssl,
                    )
                else:
                    self.data = load(
                        use_cache_server=self.use_cache_server,
                        verify_ssl=self.verify_ssl,
                    )

                # TODO: change source file format
                # version 0.1.4+ migration tool
                self.data_randomize = list(self.data['randomize'].values())
                self.data_browsers = self.data['browsers']

                for family in self.data_browsers:
                    for browser in self.data_browsers[family]:
                        if re.search('(Mobile|Android)', browser):
                            if family not in self.data_browsers_mobile:
                                self.data_browsers_mobile.update({family: []})
                            self.data_browsers_mobile[family].append(browser)
                        else:
                            if family not in self.data_browsers_desktop:
                                self.data_browsers_desktop.update({family: []})
                            self.data_browsers_desktop[family].append(browser)
        except FakeUserAgentError:
            if self.fallback is None:
                raise
            else:
                logger.warning(
                    'Error occurred during fetching data, '
                    'but was suppressed with fallback.', )