nettoyage

crawler/perils/pipelines.py

@@ -5,9 +5,3 @@
 
 
 # useful for handling different item types with a single interface
-from itemadapter import ItemAdapter
-
-
-class PerilsPipeline:
-    def process_item(self, item, spider):
-        return item

crawler/perils/spiders/scraperils.py

@@ -9,13 +9,6 @@ class ScrapePerils(scrapy.Spider):
     def parse(self, response):
         for adresses in response.xpath('//div[@class="card"]//li|//div[@class="card"]//li/p'):
                 item = PerilsItem()
-                # l = ItemLoader(item = PerilsItem(), selector=adresses)
-
-                # l.add_xpath('adrs', './text()')
-                # l.add_xpath('dernierA', './a[not(contains(translate(.,"MODIF","modif"),"modif"))][last()]/text()')
-                # if adresses.xpath('./a[not(contains(translate(.,"MODIF","modif"),"modif"))][last()]/text()').get() is None:
-                #     l.add_xpath('dernierA', './a[last()]/text()')
-
                 item['adrs'] = adresses.xpath('./text()').get(),
                 item['dernierA'] = adresses.xpath('./a[not(contains(translate(.,"MODIF","modif"),"modif"))][last()]/text()').get(),
                 if adresses.xpath('./a[not(contains(translate(.,"MODIF","modif"),"modif"))][last()]/text()').get() is None:
@@ -24,7 +17,3 @@ class ScrapePerils(scrapy.Spider):
                 item['raw'] = adresses.xpath('.').get()
 
                 yield item
-
-
-
-#response.xpath('//div[@class="card"]//li/text()[1]|//div[@class="card"]//li/p/text()[1]').getall()

crawler/perils/spiders/scraperils.py.bak

@@ -1,13 +0,0 @@
-import scrapy
-
-class ScrapePerils(scrapy.Spider):
-    name = "perils"
-    start_urls = ["https://www.marseille.fr/logement-urbanisme/am%C3%A9lioration-de-lhabitat/arretes-de-peril"]
-
-    def parse(self, response):
-        for adresses in response.xpath('//div[@class="card"]//li'):
-            yield {
-                'adresse': adresses.xpath('./text()').get(),
-                'dernier arrêté hors modificatif' : adresses.xpath('./a[not(contains(translate(.,"MODIF","modif"),"modif"))][last()]/text()').get(),
-                'dernier arrêté' : adresses.xpath('./a[last()]/text()').get(),
-                }