def parse_property_page(self, response):
        l = TruliaItemLoader(item=TruliaItem(), response=response)
        self.load_common_fields(item_loader=l, response=response)

        item = l.load_item()
        self.post_process(item=item)
        return item
Пример #2
0
    def parse_property_page(self, response):
        l = TruliaItemLoader(item=TruliaItem(), response=response)
        self.load_common_fields(item_loader=l, response=response)

        listing_information = l.nested_xpath('//span[text() = "LISTING INFORMATION"]')
        listing_information.add_xpath('listing_information', './parent::div/following-sibling::ul[1]/li/text()')
        listing_information.add_xpath('listing_information_date_updated', './following-sibling::span/text()', re=r'^Updated: (.*)')

        public_records = l.nested_xpath('//span[text() = "PUBLIC RECORDS"]')
        public_records.add_xpath('public_records', './parent::div/following-sibling::ul[1]/li/text()')
        public_records.add_xpath('public_records_date_updated', './following-sibling::span/text()', re=r'^Updated: (.*)')

        item = l.load_item()
        self.post_process(item=item)
        return item
Пример #3
0
    def parse_property_page(self, response):
        item_loader = TruliaItemLoader(item=TruliaItem(), response=response)
        trulia.TruliaSpider.load_common_fields(item_loader=item_loader, response=response)

        details = item_loader.nested_css('.homeDetailsHeading')
        taxes = details.nested_xpath('.//*[text() = "Property Taxes and Assessment"]/parent::div')
        taxes.add_xpath('property_tax_assessment_year', './following-sibling::div/div[contains(text(), "Year")]/following-sibling::div/text()')
        taxes.add_xpath('property_tax', './following-sibling::div/div[contains(text(), "Tax")]/following-sibling::div/text()')
        taxes.add_xpath('property_tax_assessment_land', './following-sibling::div/div/div[contains(text(), "Land")]/following-sibling::div/text()')
        taxes.add_xpath('property_tax_assessment_improvements', './following-sibling::div/div/div[contains(text(), "Improvements")]/following-sibling::div/text()')
        taxes.add_xpath('property_tax_assessment_total', './following-sibling::div/div/div[contains(text(), "Total")]/following-sibling::div/text()')
        taxes.add_xpath('property_tax_market_value', './following-sibling::div/div[contains(text(), "Market Value")]/following-sibling::div/text()')

        item = item_loader.load_item()
        trulia.TruliaSpider.post_process(item=item)
        return item