def parse_property_page(self, response):
l = BristolPropertyCentrePropertyLoader(item=PropertyItem(),
response=response)
l.add_css('area', "#propertyContent h1::text")
l.add_css('street_name', "#propertyContent h1::text")
#l.add_css('postcode', '.detailHeader > h2::text')
l.add_css('price_per_month', ".propertyPrice h2::text")
l.add_value('agent', 'Bristol Property Centre')
l.add_css('number_bedrooms', ".roomIcons .roombeds::text")
l.add_css('number_bathrooms', ".roomIcons .roombaths::text")
l.add_css('description', "#propertyDescription p ::text")
l.add_css('description', "#keyFeatures ul li ::text")
l.add_css('amenities', "#propertyDescription p ::text")
l.add_css('amenities', "#keyFeatures ul li ::text")
l.add_css('heating_type', "#propertyDescription p ::text")
l.add_css('heating_type', "#keyFeatures ul li ::text")
#l.add_css('epc_rating', "section.singProp.content p::text")
l.add_css('let_agreed', ".propertyPrice h2::text")
l.add_value('url', response.url)
l.add_value(
'image_url',
response.urljoin(
response.css("meta[property='og:image']::attr('content')").
extract()[0]))
return l.load_item()
def parse_property_page(self, response):
l = UbuLettingsPropertyLoader(item=PropertyItem(), response=response)
l.add_css('area', '#propAddress::text')
l.add_css('street_name', '#propAddress::text')
#l.add_css('postcode', '.detailHeader > h2::text')
l.add_css('price_per_month', '#propPrice::text')
l.add_value('agent', 'UBU Lettings')
l.add_value('number_bedrooms', 5)
# TODO: bathrooms
l.add_xpath('description',
"//div[@id='propDetShortDescCont1']//text()")
l.add_xpath('description',
"//ul[@id='propDetStarItemsCont1']//li//text()")
l.add_xpath('amenities', "//div[@id='propDetShortDescCont1']//text()")
l.add_xpath('amenities',
"//ul[@id='propDetStarItemsCont1']//li//text()")
l.add_xpath('heating_type',
"//div[@id='propDetShortDescCont1']//text()")
l.add_xpath('heating_type',
"//ul[@id='propDetStarItemsCont1']//li//text()")
l.add_xpath('epc_rating',
"//ul[@id='propDetStarItemsCont1']//li//text()")
l.add_value('url', response.url)
l.add_value(
'image_url',
response.urljoin(
response.css("#mainImage::attr('src')").extract()[0]))
return l.load_item()
def parse_property_page(self, response):
l = LetWithEasePropertyLoader(item=PropertyItem(), response=response)
l.add_css('area', '.address::text')
l.add_css('street_name', '.address::text')
l.add_css('postcode', '.address::text')
l.add_css('price_per_month', '.price::text')
l.add_value('agent', 'Let With Ease')
l.add_css('number_bedrooms', ".beds::text")
# TODO: bathrooms
l.add_css('description', "div.features ul li::text")
l.add_css('description', "div.description ::text")
l.add_css('amenities', "div.features ul li::text")
l.add_css('amenities', "div.description ::text")
l.add_css('heating_type', "div.features ul li::text")
l.add_css('heating_type', "div.description ::text")
#l.add_xpath('epc_rating', "//ul[@id='propDetStarItemsCont1']//li//text()")
l.add_css('let_agreed', "div.status::text")
l.add_value('url', response.url)
l.add_value(
'image_url',
response.urljoin(
response.css("#largePhoto::attr('src')").extract()[0]))
return l.load_item()
def parse_property_page(self, response):
l = ChappellAndMatthewsPropertyLoader(item=PropertyItem(),
response=response)
l.add_xpath('area',
"//div[@id='DetailsPriceAndContactContainer']//h1/text()")
l.add_xpath('street_name', "//h2[@class='DescriptionAddress']/text()")
l.add_xpath('postcode', "//h2[@class='DescriptionAddress']/text()")
l.add_xpath('price_per_month',
"//label[@class='generic-price']/text()")
l.add_value('agent', 'Chappell And Matthews')
l.add_xpath('number_bedrooms', "//li[@class='Bedrooms']//h2/text()")
# TODO: bathrooms, epc rating
l.add_xpath('description',
"//div[@id='DetailsContentContainer']//text()")
l.add_xpath('amenities',
"//div[@id='DetailsContentContainer']//text()")
l.add_xpath('heating_type',
"//div[@id='DetailsContentContainer']//text()")
l.add_xpath(
'let_agreed',
"//div[@id='GlobalPropertyDetails']//div[@class='LetAgreed']/text()"
)
l.add_value('url', response.url)
l.add_value(
'image_url',
response.urljoin(
response.xpath("//img[@id='imgMainPhoto']/@src").extract()[0]))
yield l.load_item()
def parse_property_page(self, response):
l = AbodePropertyLoader(item=PropertyItem(),
response=response,
number_bedrooms=5)
l.add_css('area', '.detailHeader > h2::text')
l.add_css('street_name', '.detailHeader > h2::text')
l.add_css('postcode', '.detailHeader > h2::text')
l.add_css('price_per_month', '.detailHeader > h2 > strong::text')
l.add_value('agent', 'Abode')
l.add_value('number_bedrooms', 5)
# TODO: bathrooms
l.add_xpath('description',
"//div[@id='description']/div[@class='inner']//text()")
l.add_xpath('amenities',
"//div[@id='description']/div[@class='inner']//text()")
l.add_xpath('amenities', "//div[@class='features']//li//text()")
l.add_xpath('heating_type',
"//div[@id='description']/div[@class='inner']//text()")
l.add_xpath('heating_type', "//div[@class='features']//li//text()")
l.add_xpath('epc_rating', "//div[@class='features']//li//text()")
l.add_value('url', response.url)
l.add_value(
'image_url',
response.urljoin(
response.css("#mainImg > img::attr('src')").extract()[0]))
return l.load_item()
def parse_property_page(self, response):
number_bedrooms = to_number_bedrooms(
response.css("li.beds span::text").extract()[0])
l = PurpleFrogPropertyLoader(item=PropertyItem(),
response=response,
number_bedrooms=number_bedrooms)
l.add_css('area', "h1.property-title::text")
l.add_css('street_name', "h1.property-title::text")
l.add_css('postcode', "h1.property-title::text")
l.add_css('price_per_month', "div.prices span.price::text")
l.add_value('agent', 'Purple Frog')
l.add_css('number_bedrooms', "li.beds span::text")
#l.add_css('number_bathrooms', "ul.icons li.baths span::text")
l.add_css('description',
"div#description div.left-content-panel p::text")
l.add_css('amenities',
"div#description div.left-content-panel p::text")
l.add_css('amenities', "div.features i::text")
l.add_css('heating_type',
"div#description div.left-content-panel p::text")
l.add_css('heating_type', "div.features i::text")
l.add_css('epc_rating',
".indicators .indicator.ind-current span::text")
#l.add_css('let_agreed', "span.soldbar::text")
l.add_value('url', response.url)
l.add_value(
'image_url',
response.urljoin(
response.css(".main-img img::attr('src')").extract()[0]))
return l.load_item()
def parse_property_page(self, response):
l = TheLettingGamePropertyLoader(item=PropertyItem(),
response=response)
l.add_xpath(
'area',
"//div[@id='propertyAddress']//span[@itemprop='name']/text()")
l.add_xpath(
'street_name',
"//div[@id='propertyAddress']//span[@itemprop='name']/text()")
l.add_xpath(
'postcode',
"//div[@id='propertyAddress']//span[@itemprop='name']/text()")
l.add_xpath(
'price_per_month',
"//div[@id='propertyAddress']//span[@itemprop='price']/text()")
l.add_value('agent', 'The Letting Game')
l.add_css('number_bedrooms', "li.bedrooms::text")
# TODO: bathrooms, epc
l.add_xpath('description', "//p[@itemprop='description']//text()")
l.add_xpath('amenities', "//p[@itemprop='description']//text()")
l.add_xpath('amenities', "//ul[@class='result-bullets']//li//text()")
l.add_xpath('heating_type', "//p[@itemprop='description']//text()")
l.add_xpath('heating_type',
"//ul[@class='result-bullets']//li//text()")
#l.add_xpath('epc_rating', "//ul[@id='propDetStarItemsCont1']//li//text()")
l.add_value('url', response.url)
l.add_value(
'image_url',
response.urljoin(
response.css(".carousel-inner img::attr('src')").extract()[0]))
return l.load_item()
def parse_property_page(self, response):
l = PropertyConceptPropertyLoader(item=PropertyItem(),
response=response)
l.add_css('area', "div.introBox h1::text")
l.add_css('street_name', "div.introBox h1::text")
#l.add_css('postcode', '.detailHeader > h2::text')
l.add_css('price_per_month', "div.introBox p.pPrice::text")
l.add_value('agent', 'Property Concept')
l.add_css('number_bedrooms', "ul.icons li.beds span::text")
l.add_css('number_bathrooms', "ul.icons li.baths span::text")
l.add_css('description', "section.singProp.content p::text")
l.add_css('description', "div.left ul li::text")
l.add_css('amenities', "section.singProp.content p::text")
l.add_css('amenities', "div.left ul li::text")
l.add_css('heating_type', "section.singProp.content p::text")
l.add_css('heating_type', "div.left ul li::text")
l.add_css('epc_rating', "section.singProp.content p::text")
l.add_css('let_agreed', "span.soldbar::text")
l.add_value('url', response.url)
l.add_value(
'image_url',
response.urljoin(
response.css("#slider img::attr('src')").extract()[0]))
return l.load_item()
def parse_property_page(self, response):
l = OceanPropertyLoader(item=PropertyItem(), response=response)
l.add_xpath(
'area',
"//span[@id='ctl00_cntrlCenterRegion_cntrlProperties_ctl00_ctl00_lblFormattedAddress']/text()"
)
l.add_xpath(
'street_name',
"//span[@id='ctl00_cntrlCenterRegion_cntrlProperties_ctl00_ctl00_lblFormattedAddress']/text()"
)
# l.add_css('postcode', '.detailHeader > h2::text')
l.add_xpath(
'price_per_month',
"//span[@id='ctl00_cntrlCenterRegion_cntrlProperties_ctl00_ctl00_lblFormattedPrice']/text()"
)
l.add_value('agent', 'Ocean')
l.add_xpath(
'number_bedrooms',
"//span[@id='ctl00_cntrlCenterRegion_cntrlProperties_ctl00_ctl00_lblBedrooms']/text()"
)
# TODO: bathrooms
l.add_xpath(
'description',
"//div[@id='ctl00_cntrlCenterRegion_cntrlProperties_ctl00_ctl00_cntrlFullDescription']//text()"
)
l.add_xpath(
'amenities',
"//div[@id='ctl00_cntrlCenterRegion_cntrlProperties_ctl00_ctl00_cntrlFullDescription']//text()"
)
l.add_xpath(
'heating_type',
"//div[@id='ctl00_cntrlCenterRegion_cntrlProperties_ctl00_ctl00_cntrlFullDescription']//text()"
)
l.add_xpath(
'epc_rating',
"//div[@id='ctl00_cntrlCenterRegion_cntrlProperties_ctl00_ctl00_cntrlFullDescription']//text()"
)
l.add_value('url', response.url)
l.add_value(
'image_url',
response.urljoin(
response.xpath(
"//img[@id='ctl00_cntrlCenterRegion_cntrlProperties_ctl00_ctl00_cntrlGallery_cntrlPhotos_ctl00_imgPhoto']/@src"
).extract()[0]))
return l.load_item()
def parse_property_page(self, response):
l = AbsolutePropertyLoader(item=PropertyItem(), response=response)
l.add_css('area', '.property-area::text')
l.add_css('street_name', 'h3.property-address::text')
l.add_css('postcode', 'h3.property-address::text')
l.add_css('price_per_month', '.property-price::text')
l.add_value('agent', 'Absolute Property')
l.add_css('number_bedrooms', '.property-bedrooms::text')
l.add_css('let_agreed', '.property-status::text')
l.add_css('description', '.property-description')
l.add_css('amenities', '.property-description')
l.add_css('heating_type', '.property-description')
# TODO: bathrooms, epc rating
l.add_value('url', response.url)
l.add_value(
'image_url',
response.urljoin(
response.css("img.attachment-full::attr('src')").extract()[0]))
return l.load_item()
def parse_property_list(self, response):
for property in response.xpath("//table[@width='96%']"):
l = TerryOlpinPropertyLoader(item=PropertyItem(),
selector=property)
l.add_xpath('area', ".//td[@width='31%']/text()")
#l.add_css('street_name', '.detailHeader > h2::text')
#l.add_css('postcode', '.detailHeader > h2::text')
l.add_xpath('price_per_month', ".//td[@width='20%']/text()")
l.add_value('agent', 'Terry Olpin')
l.add_xpath('number_bedrooms', ".//td[@width='17%']/text()")
# TODO: bathrooms
l.add_xpath('description', ".//p//text()")
l.add_xpath('amenities', ".//p//text()")
l.add_xpath('heating_type', ".//p//text()")
l.add_xpath('epc_rating', ".//p//text()")
l.add_value('url', response.url)
l.add_value(
'image_url',
response.urljoin(property.xpath(".//img/@src").extract()[0]))
yield l.load_item()
def parse_property_page(self, response):
l = CliftonPropertyServicesPropertyLoader(item=PropertyItem(), response=response)
l.add_xpath('area', "//h3[@class='fulladdress']/text()")
l.add_xpath('street_name', "//h3[@class='fulladdress']/text()")
# l.add_css('postcode', '.detailHeader > h2::text')
l.add_xpath('price_per_month', "//h2[@class='dtaddress']/em/text()")
l.add_value('agent', 'Clifton Property Services')
l.add_xpath('number_bedrooms', "//span[@class='rs-bedrooms']/text()")
# TODO: bathrooms
l.add_xpath('description', "//div[@id='s-dtintrodesc']//text()")
l.add_xpath('description', "//ul[@id='s-dtbullets']//text()")
l.add_xpath('amenities', "//div[@id='s-dtintrodesc']//text()")
l.add_xpath('amenities', "//ul[@id='s-dtbullets']//text()")
l.add_xpath('heating_type', "//div[@id='s-dtintrodesc']//text()")
l.add_xpath('heating_type', "//ul[@id='s-dtbullets']//text()")
#l.add_xpath('epc_rating',"//div[@id='ctl00_cntrlCenterRegion_cntrlProperties_ctl00_ctl00_cntrlFullDescription']//text()")
l.add_xpath('let_agreed', "h2[@class='dtaddress']/em/span[@class='PropStatus']/text()")
l.add_value('url', response.url)
l.add_value('image_url', response.urljoin(response.xpath("//img[@id='mainimage']/@src").extract()[0]))
return l.load_item()
def parse(self, response):
for property in response.xpath("//div[@id='content']//div[@class='lightboxGallery']"):
l = CityPropertyLetsPropertyLoader(item=PropertyItem(), selector=property,
number_bedrooms=CityPropertyLetsNumberBedroomsProcessor()(property.xpath(".//ul/li[3]/text()").extract()[0]))
l.add_xpath('area', ".//h1/text()")
l.add_xpath('street_name', ".//h1/text()")
# l.add_css('postcode', '.detailHeader > h2::text')
l.add_xpath('price_per_month', ".//li/b/text()")
l.add_value('agent', 'City Property Lets')
l.add_xpath('number_bedrooms', ".//ul/li[3]/text()")
# TODO: bathrooms
l.add_xpath('description', ".//p//text()")
l.add_xpath('amenities', ".//p//text()")
l.add_xpath('amenities', ".//ul//li//text()")
l.add_xpath('heating_type', ".//p//text()")
l.add_xpath('heating_type', ".//ul//li//text()")
# l.add_xpath('epc_rating', ".//p//text()")
l.add_value('url', response.url)
l.add_value('image_url', response.urljoin(property.xpath(".//img/@src").extract()[0]))
yield l.load_item()
def parse_property_page(self, response):
l = JacksonPropertyPropertyLoader(item=PropertyItem(), response=response,
number_bedrooms=to_number_bedrooms(
response.xpath("//div[@class='searchHeaderRight']/text()").extract()[
0]))
# l.add_xpath('area', "//div[@class='searchHeaderLeft']/text()")
l.add_xpath('street_name', "//div[@class='searchHeaderLeft']/text()")
# l.add_css('postcode', '.detailHeader > h2::text')
l.add_xpath('price_per_month', "//div[@class='searchHeaderRight']/text()")
l.add_value('agent', 'Jackson Property')
l.add_xpath('number_bedrooms', "//div[@class='searchHeaderRight']/text()")
# TODO: bathrooms
l.add_xpath('description', "//div[@class='propertyDetailInfo']//p//text()")
l.add_xpath('amenities', "//div[@class='propertyDetailInfo']//p//text()")
l.add_xpath('heating_type', "//div[@class='propertyDetailInfo']//p//text()")
l.add_xpath('epc_rating', "//div[@class='propertyDetailInfo']//p//text()")
l.add_value('url', response.url)
maybe_image = response.xpath("//div[@class='preview']//img/@src").extract()
l.add_value('image_url', response.urljoin(maybe_image[0]) if maybe_image else "")
return l.load_item()
def parse_property_page(self, response):
l = AccommodationUnlimitedPropertyLoader(item=PropertyItem(), response=response)
l.add_xpath('area', "//div[@class='row property-detail']//p//text()[5]")
l.add_xpath('street_name', "//div[@class='row property-detail']//p//text()[5]")
#l.add_css('postcode', '.detailHeader > h2::text')
l.add_xpath('price_per_month', "//div[@class='row property-detail']//p//text()[1]")
l.add_value('agent', 'Accommodation Unlimited')
l.add_xpath('number_bedrooms', "//div[@class='row property-detail']//p//text()[2]")
#l.add_css('number_bathrooms', ".roomIcons .roombaths::text")
l.add_xpath('description', "//div[@class='row property-detail']//p//text()")
l.add_xpath('amenities', "//div[@class='row property-detail']//p//text()")
l.add_xpath('heating_type', "//div[@class='row property-detail']//p//text()")
#l.add_css('epc_rating', "section.singProp.content p::text")
#l.add_css('let_agreed', ".propertyPrice h2::text")
l.add_value('url', response.url)
maybe_image = response.css("div.large-6.columns.text-center img::attr('src')").extract()
if maybe_image:
l.add_value('image_url', response.urljoin(maybe_image[0]))
return l.load_item()
def parse_property_page(self, response):
l = BerkeleyEstatesPropertyLoader(
item=PropertyItem(),
response=response,
number_bedrooms=to_number_bedrooms(
response.xpath(
"//table[@class='bannertable']//tr[1]//td[2]//b/text()").
extract()[0]))
l.add_css('area',
'.bannertable > tr:first-child > td:first-child b::text')
l.add_css('street_name',
'.bannertable > tr:first-child > td:first-child b::text')
#l.add_css('postcode', '.detailHeader > h2::text')
l.add_xpath('price_per_month', u"//b[contains(text(), '£')]//text()")
l.add_value('agent', 'Berkeley Estates')
l.add_xpath('number_bedrooms',
"//table[@class='bannertable']//tr[1]//td[2]//b/text()")
l.add_xpath('number_bathrooms',
"//table[@class='bannertable']//tr[1]//td[2]//b/text()")
# TODO: bathrooms, epc
l.add_xpath('description',
"//td[.//*[contains(text(),'Full Description')]]//text()")
l.add_xpath('description', "//ul//text()")
l.add_xpath('amenities',
"//td[.//*[contains(text(),'Full Description')]]//text()")
l.add_xpath('amenities', "//ul//text()")
l.add_xpath('heating_type',
"//td[.//*[contains(text(),'Full Description')]]//text()")
l.add_xpath('heating_type', "//ul//text()")
l.add_value('url', response.url)
l.add_value(
'image_url',
response.urljoin(
response.css(".image img::attr('src')").extract()[0]))
return l.load_item()
def parse_property_page(self, response):
l = SureMovePropertyLoader(item=PropertyItem(), response=response)
l.add_css('area', "p.address::text")
l.add_css('street_name', "p.address::text")
#l.add_css('postcode', '.detailHeader > h2::text')
l.add_css('price_per_month', "span.property-price::text")
l.add_value('agent', 'Sure Move')
l.add_css('number_bedrooms', "ul.guide li.beds::text")
l.add_css('number_bathrooms', "ul.guide li.bath::text")
l.add_css('description', ".property-description p ::text")
l.add_css('amenities', ".property-description p ::text")
l.add_css('heating_type', ".property-description p ::text")
#l.add_css('epc_rating', "section.singProp.content p::text")
l.add_css('let_agreed', "div.banner-agreed p::text")
l.add_value('url', response.url)
l.add_value(
'image_url',
response.urljoin(
response.css("img.main-prop-image::attr('src')").extract()[0]))
return l.load_item()
def parse_property_page(self, response):
number_bedrooms = int(
response.xpath(
"//span[@id='ctl00_cntrlCenterRegion_cntrlProperties_item_0_ctl00_lblBedrooms']/text()"
).extract()[0])
l = CliftonsPropertyLoader(item=PropertyItem(),
response=response,
number_bedrooms=number_bedrooms)
l.add_xpath(
'area',
"//span[@id='ctl00_cntrlCenterRegion_cntrlProperties_item_0_ctl00_lblFormattedAddress']/text()"
)
l.add_xpath(
'street_name',
"//span[@id='ctl00_cntrlCenterRegion_cntrlProperties_item_0_ctl00_lblFormattedAddress']/text()"
)
# l.add_css('postcode', '.detailHeader > h2::text')
l.add_xpath(
'price_per_month',
"//span[@id='ctl00_cntrlCenterRegion_cntrlProperties_item_0_ctl00_lblFormattedPrice']/text()"
)
l.add_value('agent', 'Cliftons')
l.add_xpath(
'number_bedrooms',
"//span[@id='ctl00_cntrlCenterRegion_cntrlProperties_item_0_ctl00_lblBedrooms']/text()"
)
l.add_xpath(
'number_bedrooms',
"//span[@id='ctl00_cntrlCenterRegion_cntrlProperties_item_0_ctl00_lblBathrooms']/text()"
)
# TODO: bathrooms
l.add_xpath(
'description',
"//table[@id='ctl00_cntrlCenterRegion_cntrlProperties_item_0_ctl00_cntrlBulletedList']//li//text()"
)
l.add_xpath(
'description',
"//div[@id='ctl00_cntrlCenterRegion_cntrlProperties_item_0_ctl00_lblBriefDescription']//text()"
)
l.add_xpath(
'description',
"//div[@id='ctl00_cntrlCenterRegion_cntrlProperties_ctl00_ctl00_cntrlFullDescription']//text()"
)
l.add_xpath(
'amenities',
"//table[@id='ctl00_cntrlCenterRegion_cntrlProperties_item_0_ctl00_cntrlBulletedList']//li//text()"
)
l.add_xpath(
'amenities',
"//div[@id='ctl00_cntrlCenterRegion_cntrlProperties_item_0_ctl00_lblBriefDescription']//text()"
)
l.add_xpath(
'amenities',
"//div[@id='ctl00_cntrlCenterRegion_cntrlProperties_ctl00_ctl00_cntrlFullDescription']//text()"
)
l.add_xpath(
'heating_type',
"//table[@id='ctl00_cntrlCenterRegion_cntrlProperties_item_0_ctl00_cntrlBulletedList']//li//text()"
)
l.add_xpath(
'heating_type',
"//div[@id='ctl00_cntrlCenterRegion_cntrlProperties_item_0_ctl00_lblBriefDescription']//text()"
)
l.add_xpath(
'heating_type',
"//div[@id='ctl00_cntrlCenterRegion_cntrlProperties_ctl00_ctl00_cntrlFullDescription']//text()"
)
l.add_xpath(
'epc_rating',
"//table[@id='ctl00_cntrlCenterRegion_cntrlProperties_item_0_ctl00_cntrlBulletedList']//li//text()"
)
l.add_xpath(
'let_agreed',
"//span[@id='ctl00_cntrlCenterRegion_cntrlProperties_item_0_ctl00_cntrlStatus_lblStatus']/text()"
)
l.add_value('url', response.url)
l.add_value(
'image_url',
response.urljoin(
response.xpath(
"//img[@id='ctl00_cntrlCenterRegion_cntrlProperties_item_0_ctl00_cntrlGallery_cntrlMainPhoto_ctl00_cntrlAsset']/@src"
).extract()[0]))
return l.load_item()