def parse_item(self, response):
hxs = HtmlXPathSelector(response)
data = {"url": response.url, "source": self.source}
recipe = RecipeItem.from_dict(parse_recipe(hxs, data))
loader = RecipeItemLoader(item=recipe)
loader.add_value("image", select_class(hxs, "post_image").select("@src").extract())
loader.add_value("description", hxs.select('//meta[@name="description"]/@content').extract())
loader.add_value("name", select_class(hxs, "entry-title").select("text()").extract())
return [loader.load_item()]
def parse_item(self, response):
hxs = HtmlXPathSelector(response)
data = {'url': response.url, 'source': self.source}
recipe = RecipeItem.from_dict(parse_recipe(hxs, data))
loader = RecipeItemLoader(item=recipe)
loader.add_value('image', select_class(hxs, 'post_image').select('@src').extract())
loader.add_value('description', hxs.select('//meta[@name="description"]/@content').extract())
loader.add_value('name', select_class(hxs, 'entry-title').select('text()').extract())
return loader.load_item()
def parse_item(self, response):
hxs = HtmlXPathSelector(response)
data = {'url': response.url, 'source': self.source}
recipe = RecipeItem.from_dict(parse_recipe(hxs, data))
loader = RecipeItemLoader(item=recipe)
loader.add_value(
'image',
select_class(hxs, 'post_image').select('@src').extract())
loader.add_value(
'description',
hxs.select('//meta[@name="description"]/@content').extract())
loader.add_value(
'name',
select_class(hxs, 'entry-title').select('text()').extract())
return [loader.load_item()]
def parse_recipe(scope, data={}):
root = select_class(scope, 'hrecipe')
data['name'] = select_class(root, 'fn').select('.//text()').extract()
data['yield'] = select_class(root, 'yield').select('.//text()').extract()
data['published'] = select_class(root, 'published').select('.//text()').extract()
data['description'] = select_class(root, 'summary').select('.//text()').extract()
data['duration'] = select_class(root, 'duration').select('.//text()').extract()
data['prepTime'] = select_class(root, 'preptime').select('.//text()').extract()
data['cookTime'] = select_class(root, 'cooktime').select('.//text()').extract()
data['ingredients'] = []
for ingredient in select_class(root, 'ingredient'):
data['ingredients'].append(''.join(ingredient.select('.//text()').extract()))
return data
