Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- # -*- coding: utf-8 -*-
- import scrapy
- from scrapy.spider import Spider
- from scrapy.selector import Selector
- from zvjs.items import Website
- class TutorialSpiderSpider(scrapy.Spider):
- name = "zvjs-spider"
- allowed_domains = ["zvjs.sk"]
- start_urls = [
- "http://www.zvjs.sk/index.php?fa_obj&type=fa&id=98273",
- "http://www.zvjs.sk/index.php?fa_obj&type=fa&id=98270",
- "http://www.zvjs.sk/index.php?fa_obj&type=fa&id=10845",
- ]
- def parse(self, response):
- sel = Selector(response)
- sites = sel.xpath('//strong')
- items = []
- for site in sites:
- item = Website()
- item['evcislo'] = site.xpath('//strong/text()').extract()[0].encode("utf-8")
- item['evcisloval'] = site.xpath("//td/text()").extract()[1].encode("utf-8")
- item['prinalezi'] = site.xpath("//strong/text()").extract()[1].encode("utf-8")
- item['prinalezival'] = site.xpath("//td/text()").extract()[2].encode("utf-8")
- item['popisfa'] = sel.xpath("//strong/text()").extract()[2].encode("utf-8")
- item['popisfaval'] = sel.xpath("//td/text()").extract()[3].encode("utf-8")
- item['hodnotafa'] = sel.xpath("//strong/text()").extract()[3].encode("windows-1250")
- item['hodnotafaval'] = sel.xpath("//td/text()").extract()[4].encode("windows-1250")
- item['datumdoruceniaval'] = sel.xpath("//td/text()").extract()[5].encode("windows-1250")
- item['datumzaplateniaval']= sel.xpath("//td/text()").extract()[6].encode("windows-1250")
- items.append(item)
- return items
- ####################################
- items.py
- ####################################
- # -*- coding: utf-8 -*-
- import scrapy
- from scrapy.item import Item, Field
- class Website(Item):
- evcislo = Field()
- evcisloval = Field()
- prinalezi = Field()
- prinalezival = Field()
- popisfa = Field()
- popisfaval = Field()
- hodnotafa = Field()
- hodnotafaval = Field()
- datumdoruceniaval = Field()
- datumzaplateniaval = Field()
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement