#1113 Cabo Frio passa a usar base instar a partir de set/2024

data_collection/gazette/spiders/base/instar.py

@@ -8,6 +8,8 @@
 
 
 class BaseInstarSpider(BaseGazetteSpider):
+ power = "executive_legislative"
+
  def start_requests(self):
  page = 1
  start_date = self.start_date.strftime("%d-%m-%Y")
@@ -58,7 +60,7 @@ def parse(self, response, page, start_date, end_date):
  date=gazette_date,
  edition_number=edition_number,
  is_extra_edition=False,
- power="executive_legislative",
+ power=self.power,
  )
 
  yield scrapy.Request(

data_collection/gazette/spiders/base/ptio.py

@@ -7,9 +7,9 @@
 
 class BasePtioSpider(BaseGazetteSpider):
  def start_requests(self):
- yield scrapy.Request(self.BASE_URL)
+ yield scrapy.Request(url=self.base_url, callback=self.ptio_parse)
 
- def parse(self, response):
+ def ptio_parse(self, response):
  for gazette_div in response.xpath("//div[@class='edicoes']"):
  raw_gazete_date = gazette_div.xpath(
  ".//div[@class='data-caderno hidden-phone']/text()"
@@ -41,4 +41,6 @@ def parse(self, response):
  "//ul[@class='paginacao']//a[@class='proximo']/@href"
  )
  if next_page:
- yield scrapy.Request(response.urljoin(next_page.get()))
+ yield scrapy.Request(
+ response.urljoin(next_page.get()), callback=self.ptio_parse
+ )

data_collection/gazette/spiders/rj/rj_areal.py

@@ -7,5 +7,5 @@ class RjArealSpider(BasePtioSpider):
  name = "rj_areal"
  TERRITORY_ID = "3300225"
  allowed_domains = ["portaldatransparencia.com.br"]
- BASE_URL = "http://rj.portaldatransparencia.com.br/prefeitura/areal/"
+ base_url = "http://rj.portaldatransparencia.com.br/prefeitura/areal/"
  start_date = date(2006, 8, 1)

data_collection/gazette/spiders/rj/rj_cabo_frio.py

@@ -1,11 +1,36 @@
-from datetime import date
+from copy import copy
+from datetime import date, timedelta
 
+from gazette.spiders.base.instar import BaseInstarSpider
 from gazette.spiders.base.ptio import BasePtioSpider
 
 
-class RjCaboFrioSpider(BasePtioSpider):
+class RjCaboFrioSpider(BaseInstarSpider, BasePtioSpider):
  name = "rj_cabo_frio"
  TERRITORY_ID = "3300704"
- allowed_domains = ["portaldatransparencia.com.br"]
- BASE_URL = "http://rj.portaldatransparencia.com.br/prefeitura/cabofrio/"
+ allowed_domains = [
+ "portaldatransparencia.com.br",
+ "cabofrio.instartecnologia.com.br",
+ ]
+ base_url = "https://www.cabofrio.instartecnologia.com.br/portal/diario-oficial"
  start_date = date(2020, 7, 29)
+ power = "executive"
+
+ def start_requests(self):
+ ptio_url = "http://rj.portaldatransparencia.com.br/prefeitura/cabofrio/"
+ ptio_end_date = date(2024, 8, 31)
+ if self.start_date > ptio_end_date:
+ yield from BaseInstarSpider.start_requests(self)
+ else:
+ if self.end_date <= ptio_end_date:
+ self.base_url = ptio_url
+ yield from BasePtioSpider.start_requests(self)
+ else:
+ ptio = copy(self)
+ ptio.end_date = ptio_end_date
+ ptio.base_url = ptio_url
+ yield from BasePtioSpider.start_requests(ptio)
+
+ instar = copy(self)
+ instar.start_date = ptio_end_date + timedelta(days=1)
+ yield from BaseInstarSpider.start_requests(instar)

data_collection/gazette/spiders/rj/rj_comendador_levy_gasparian.py

@@ -7,7 +7,7 @@ class RjComendadorLevyGasparianSpider(BasePtioSpider):
  name = "rj_comendador_levy_gasparian"
  TERRITORY_ID = "3300951"
  allowed_domains = ["portaldatransparencia.com.br"]
- BASE_URL = (
+ base_url = (
  "http://rj.portaldatransparencia.com.br/prefeitura/comendadorlevygasparian/"
  )
  start_date = date(2013, 11, 26)

data_collection/gazette/spiders/rj/rj_sapucaia.py

@@ -7,5 +7,5 @@ class RjSapucaiaSpider(BasePtioSpider):
  name = "rj_sapucaia"
  TERRITORY_ID = "3305406"
  allowed_domains = ["portaldatransparencia.com.br"]
- BASE_URL = "http://rj.portaldatransparencia.com.br/prefeitura/sapucaia/"
+ base_url = "http://rj.portaldatransparencia.com.br/prefeitura/sapucaia/"
  start_date = date(2019, 1, 16)