PricePerSqm Pipeline

ImmoScrap/pipelines.py

@@ -10,11 +10,8 @@
 
 class PricePerSqmPipeline:
     def process_item(self, item, spider):
-        '''
-        pricefloat = float(price.replace(',', '.').replace(' ', '').replace('€', ''))
-        areafloat = float(area.replace(',', '.').replace(' ', '').replace('m²', ''))
-        pricesqmint = round(pricefloat / areafloat)
-        pricepersqm = float(item['price']) / float(item['area'])
-        item['price/sqm'] = pricepersqm
-        '''
+        area = item['area']
+        if area is not None:
+            pricepersqm = item['price'] / float(item['area'])
+            item['price/sqm'] = round(pricepersqm)
         return item

ImmoScrap/spiders/PAP.py

@@ -37,10 +37,10 @@ class PAPSpider(scrapy.Spider):
             yield {
                 'url': self.baseurl + url,
                 'type': 'Unknown',
-                'price': ad.css(self.price_css_sel).get().replace('.', '').replace('\xa0€', ''),
+                'price': int(ad.css(self.price_css_sel).get().replace('.', '').replace('\xa0€', '')),
                 'rooms': rooms,
                 'bedrooms': bedrooms,
-                'area': infos.xpath('./li[contains(small, "m")]/text()').get()
+                'area': infos.xpath('./li[contains(small, "m")]/text()').get().replace(' ', '')
             }
 
         next_page = response.xpath(self.next_page_xpath).get()

ImmoScrap/spiders/Seloger.py

@@ -57,10 +57,10 @@ class SelogerSpider(scrapy.Spider):
             yield {
                 'url': ad.css(self.url_css_sel).get().split('?')[0],
                 'type': ad.css(self.type_css_sel).get().replace('/Villa', ''),
-                'price': ad.css(self.price_css_sel).get().replace('\xa0', '').replace(' €', ''),
+                'price': int(ad.css(self.price_css_sel).get().replace('\xa0', '').replace(' €', '')),
                 'rooms': rooms,
                 'bedrooms': bedrooms,
-                'area': area
+                'area': area.replace(',', '.')
             }
         active_page = response.url
         active_page_nb = int(active_page.split('LISTING-LISTpg=')[1])

ImmoScrap/spiders/leboncoin.py

@@ -87,7 +87,7 @@ class LeboncoinSpider(scrapy.Spider):
             yield {
                 'url': ad['url'],
                 'type': type,
-                'price': ad['price'][0],
+                'price': int(ad['price'][0]),
                 'rooms': rooms,
                 'bedrooms': 'Unknown',
                 'area': area