Ejemplo n.º 1
0
    def setUp(self):
        database_factory = DatabaseFactory(
            settings.SQL_HOST, settings.SQL_PORT, settings.SQL_USER,
            settings.SQL_PASSWD, settings.SQL_DB, settings.SQL_COLLECTION_NAME)

        self._cache_db = database_factory.get_database(CollectionTypes.cache)
        self._history_db = database_factory.get_database(
            CollectionTypes.history)
        self._item_db = database_factory.get_database(CollectionTypes.item)

        self._page_url = "https://sa.opensooq.com/ar/search/30002057/استراحة-سديم-للايجار-اليومي-والشهري-والسنوي-حي-الأمانة-شمال-الرياض"
        self._page_from = WebsiteTypes.opensooq.value

        self._ads_id = 70
        self._ads_item = Ad.get_default(
            section_item=SectionItem.get_default(),
            ads_title="اسكاليد موديل 2016 بسعر جي",
            city_id='44',
            ads_contact="123454321",
            ads_body="test mysql db",
            image_link=
            "https://img1cdn.haraj.com.sa/userfiles30/2015-07-18/55aa1ba3366cd.jpeg,https://img1cdn.haraj.com.sa/userfiles30/2015-07-18/55aa1bab49a3b.jpeg,https://img1cdn.haraj.com.sa/userfiles30/2015-07-18/55aa1bb0b6ca6.jpeg",
            His_announcement_id=60,
            type_ads_or=1,
            _close_ads=0)

        self._member_id = 60
        self._memberName = "djzhang"

        self.opensooq_phone_id = 24
Ejemplo n.º 2
0
    def setUp(self):
        database_factory = DatabaseFactory(settings.SQL_HOST, settings.SQL_PORT,
                                           settings.SQL_USER, settings.SQL_PASSWD,
                                           settings.SQL_DB, settings.SQL_COLLECTION_NAME)

        self._cache_db = database_factory.get_database(CollectionTypes.cache)
        self._history_db = database_factory.get_database(CollectionTypes.history)
        self._item_db = database_factory.get_database(CollectionTypes.item)

        self._page_url = "https://sa.opensooq.com/ar/search/30002057/استراحة-سديم-للايجار-اليومي-والشهري-والسنوي-حي-الأمانة-شمال-الرياض"
        self._page_from = WebsiteTypes.opensooq.value

        self._ads_id = 70
        self._ads_item = Ad.get_default(
            section_item=SectionItem.get_default(),
            ads_title="اسكاليد موديل 2016 بسعر جي",
            city_id='44',
            ads_contact="123454321",
            ads_body="test mysql db",
            image_link="https://img1cdn.haraj.com.sa/userfiles30/2015-07-18/55aa1ba3366cd.jpeg,https://img1cdn.haraj.com.sa/userfiles30/2015-07-18/55aa1bab49a3b.jpeg,https://img1cdn.haraj.com.sa/userfiles30/2015-07-18/55aa1bb0b6ca6.jpeg",
            His_announcement_id=60,
            type_ads_or=1, _close_ads=0
        )

        self._member_id = 60
        self._memberName = "djzhang"

        self.opensooq_phone_id = 24
Ejemplo n.º 3
0
    def setUp(self):
        database_factory = DatabaseFactory(settings.SQL_HOST, settings.SQL_PORT,
                                           settings.SQL_USER, settings.SQL_PASSWD,
                                           settings.SQL_DB, settings.SQL_COLLECTION_NAME)

        self._item_db = database_factory.get_database(CollectionTypes.item)
        self.section_data = SectionData()
        self.expect = self.section_data.expect

        self.section_mgr = HarajsSection(self.section_data.sections, self._item_db)
    def __init__(self, name=None, **kwargs):
        from cwharaj.database_factory import DatabaseFactory, CollectionTypes
        database_factory = DatabaseFactory(kwargs['host'], kwargs['port'],
                                           kwargs['user'], kwargs['passwd'],
                                           kwargs['db'], kwargs['collection_name'])

        self._cache_db = database_factory.get_database(CollectionTypes.cache)
        self._history_db = database_factory.get_database(CollectionTypes.history)
        self._item_db = database_factory.get_database(CollectionTypes.item)

        from cwharaj.parser.opensooq_parser import OpensooqParse
        self.opensooq_parse = OpensooqParse()

        super(OpensooqDebugCommentDateSpider, self).__init__(name, **kwargs)
Ejemplo n.º 5
0
    def __init__(self, name=None, **kwargs):
        from cwharaj.database_factory import DatabaseFactory, CollectionTypes
        database_factory = DatabaseFactory(kwargs['host'], kwargs['port'],
                                           kwargs['user'], kwargs['passwd'],
                                           kwargs['db'],
                                           kwargs['collection_name'])

        self._cache_db = database_factory.get_database(CollectionTypes.cache)
        self._history_db = database_factory.get_database(
            CollectionTypes.history)
        self._item_db = database_factory.get_database(CollectionTypes.item)

        from cwharaj.parser.opensooq_parser import OpensooqParse
        self.opensooq_parse = OpensooqParse()

        super(OpensooqDebugWatchSpider, self).__init__(name, **kwargs)
    def __init__(self, name=None, **kwargs):
        from cwharaj.database_factory import DatabaseFactory, CollectionTypes
        database_factory = DatabaseFactory(kwargs['host'], kwargs['port'],
                                           kwargs['user'], kwargs['passwd'],
                                           kwargs['db'], kwargs['collection_name'])

        self._cache_db = database_factory.get_database(CollectionTypes.cache)
        self._history_db = database_factory.get_database(CollectionTypes.history)
        self._item_db = database_factory.get_database(CollectionTypes.item)

        from cwharaj.spiders.dispatch.spider_watch_dispatch import SpiderWatchDispatch
        self.watch_dispatch = SpiderWatchDispatch()

        # Dynamic the domains and start url.
        self.allowed_domains = self.watch_dispatch.get_allowed_domains()
        self.start_urls = self.watch_dispatch.get_pagination_websites()

        super(HarajsSpiderWatch, self).__init__(name, **kwargs)
Ejemplo n.º 7
0
    def __init__(self, name=None, **kwargs):
        from cwharaj.database_factory import DatabaseFactory, CollectionTypes
        database_factory = DatabaseFactory(kwargs['host'], kwargs['port'],
                                           kwargs['user'], kwargs['passwd'],
                                           kwargs['db'],
                                           kwargs['collection_name'])

        self._cache_db = database_factory.get_database(CollectionTypes.cache)
        self._history_db = database_factory.get_database(
            CollectionTypes.history)
        self._item_db = database_factory.get_database(CollectionTypes.item)

        from cwharaj.spiders.dispatch.spider_watch_dispatch import SpiderWatchDispatch
        self.watch_dispatch = SpiderWatchDispatch()

        # Dynamic the domains and start url.
        self.allowed_domains = self.watch_dispatch.get_allowed_domains()
        self.start_urls = self.watch_dispatch.get_pagination_websites()

        super(HarajsSpiderWatch, self).__init__(name, **kwargs)
    def __init__(self, name=None, **kwargs):
        self.allowed_domains = [websites_allowed_domains.get(self.url_from)]

        if is_pagination:
            self.start_urls = [WebsiteTypes.get_pagination_url(self.url_from)]
        else:
            self.start_urls = self.details_urls

        from cwharaj.database_factory import DatabaseFactory, CollectionTypes
        database_factory = DatabaseFactory(kwargs['host'], kwargs['port'],
                                           kwargs['user'], kwargs['passwd'],
                                           kwargs['db'], kwargs['collection_name'])

        self._cache_db = database_factory.get_database(CollectionTypes.cache)
        self._history_db = database_factory.get_database(CollectionTypes.history)
        self._item_db = database_factory.get_database(CollectionTypes.item)

        from cwharaj.parser.opensooq_parser import OpensooqParse
        self._parser = OpensooqParse()

        super(OpensooqDebugSpider, self).__init__(name, **kwargs)
Ejemplo n.º 9
0
    def __init__(self, name=None, **kwargs):
        from cwharaj.database_factory import DatabaseFactory, CollectionTypes
        database_factory = DatabaseFactory(kwargs['host'], kwargs['port'],
                                           kwargs['user'], kwargs['passwd'],
                                           kwargs['db'], kwargs['collection_name'])

        self._cache_db = database_factory.get_database(CollectionTypes.cache)
        self._history_db = database_factory.get_database(CollectionTypes.history)
        self._item_db = database_factory.get_database(CollectionTypes.item)

        from cwharaj.spiders.dispatch.spider_dispatch import SpiderDispatch
        self.spider_dispatch = SpiderDispatch()

        # Dynamic the domains
        self.allowed_domains = self.spider_dispatch.get_allowed_domains()

        # Get the start urls from the cache database
        row = self.get_row_from_cache("", "")
        if row:
            self.start_urls = [row['url']]

        super(HarajsSpider, self).__init__(name, **kwargs)
Ejemplo n.º 10
0
    def __init__(self, name=None, **kwargs):
        self.allowed_domains = [websites_allowed_domains.get(self.url_from)]

        if is_pagination:
            self.start_urls = [WebsiteTypes.get_pagination_url(self.url_from)]
        else:
            self.start_urls = self.details_urls

        from cwharaj.database_factory import DatabaseFactory, CollectionTypes
        database_factory = DatabaseFactory(kwargs['host'], kwargs['port'],
                                           kwargs['user'], kwargs['passwd'],
                                           kwargs['db'],
                                           kwargs['collection_name'])

        self._cache_db = database_factory.get_database(CollectionTypes.cache)
        self._history_db = database_factory.get_database(
            CollectionTypes.history)
        self._item_db = database_factory.get_database(CollectionTypes.item)

        from cwharaj.parser.mstaml_parser import MstamlParse
        self._parser = MstamlParse()

        super(MstamlDebugWatchSpider, self).__init__(name, **kwargs)
Ejemplo n.º 11
0
    def __init__(self, name=None, **kwargs):
        from cwharaj.database_factory import DatabaseFactory, CollectionTypes
        database_factory = DatabaseFactory(kwargs['host'], kwargs['port'],
                                           kwargs['user'], kwargs['passwd'],
                                           kwargs['db'],
                                           kwargs['collection_name'])

        self._cache_db = database_factory.get_database(CollectionTypes.cache)
        self._history_db = database_factory.get_database(
            CollectionTypes.history)
        self._item_db = database_factory.get_database(CollectionTypes.item)

        from cwharaj.spiders.dispatch.spider_dispatch import SpiderDispatch
        self.spider_dispatch = SpiderDispatch()

        # Dynamic the domains
        self.allowed_domains = self.spider_dispatch.get_allowed_domains()

        # Get the start urls from the cache database
        row = self.get_row_from_cache("", "")
        if row:
            self.start_urls = [row['url']]

        super(HarajsSpider, self).__init__(name, **kwargs)