diff --git a/kingfisher_scrapy/spiders/spain_zaragoza.py b/kingfisher_scrapy/spiders/spain_zaragoza.py index 7c1ab824..695a3ac6 100644 --- a/kingfisher_scrapy/spiders/spain_zaragoza.py +++ b/kingfisher_scrapy/spiders/spain_zaragoza.py @@ -10,32 +10,16 @@ class SpainZaragoza(SimpleSpider): """ Domain Ayuntamiento de Zaragoza - Spider arguments - from_date - Download only data from this date onward (YYYY-MM-DDTHH:mm:ss format). - If ``until_date`` is provided, defaults to '2000-01-01T00:00:00'. - until_date - Download only data until this date (YYYY-MM-DDTHH:mm:ss format). - If ``from_date`` is provided, defaults to today. Swagger API documentation https://www.zaragoza.es/docs-api_sede/ """ name = 'spain_zaragoza' - data_type = 'release_list' - date_format = 'datetime' - default_from_date = '2000-01-01T00:00:00' + data_type = 'release_package' url = 'https://www.zaragoza.es/sede/servicio/contratacion-publica/ocds/contracting-process/' def start_requests(self): # row parameter setting to 100000 to get all releases - url = self.url + '?rf=html&rows=100000' - - # check date parameters and set "yyyy-MM-dd'T'HH:mm:ss'Z'" format - if self.from_date and self.until_date: - # `before` and `after` query string parameters behave opposite in API - after = self.until_date.strftime("%Y-%m-%dT%H:%M:%SZ") - before = self.from_date.strftime("%Y-%m-%dT%H:%M:%SZ") - url = f'{url}&before={before}&after={after}' + url = f'{self.url}?rf=html&rows=100000' yield scrapy.Request(url, meta={'file_name': 'list.json'}, callback=self.parse_list) @@ -43,5 +27,6 @@ def start_requests(self): def parse_list(self, response): ids = json.loads(response.text) for contracting_process_id in ids: - url = self.url + contracting_process_id['id'] + ocid = contracting_process_id['ocid'] + url = f'{self.url}{ocid}' yield self.build_request(url, formatter=components(-1))