Cria Base Ptio e adiicona 3 municípios do RJ

data_collection/gazette/spiders/base/ptio.py

@@ -0,0 +1,44 @@
+import dateparser
+import scrapy
+
+from gazette.items import Gazette
+from gazette.spiders.base import BaseGazetteSpider
+
+
+class BasePtioSpider(BaseGazetteSpider):
+    def start_requests(self):
+        yield scrapy.Request(self.BASE_URL)
+
+    def 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()"
+            ).get()
+            gazette_date = dateparser.parse(raw_gazete_date).date()
+
+            if gazette_date > self.end_date:
+                continue
+            if gazette_date < self.start_date:
+                return
+
+            gazette_edition = gazette_div.xpath(
+                ".//span[@class='edicao']/strong/text()"
+            ).get()
+            gazette_edition_number = gazette_edition.split()[1].replace(".", "")
+
+            sub_dir = gazette_div.xpath(".//button[1]/@href").get()
+            gazette_url = response.urljoin(sub_dir[sub_dir.index("?") :])
+
+            yield Gazette(
+                date=gazette_date,
+                edition_number=gazette_edition_number,
+                is_extra_edition=False,
+                file_urls=[gazette_url],
+                power="executive",
+            )
+
+        next_page = response.xpath(
+            "//ul[@class='paginacao']//a[@class='proximo']/@href"
+        )
+        if next_page:
+            yield scrapy.Request(response.urljoin(next_page.get()))

data_collection/gazette/spiders/rj/rj_areal.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.ptio import BasePtioSpider
+
+
+class RjArealSpider(BasePtioSpider):
+    name = "rj_areal"
+    TERRITORY_ID = "3300225"
+    allowed_domains = ["portaldatransparencia.com.br"]
+    BASE_URL = "http://rj.portaldatransparencia.com.br/prefeitura/areal/"
+    start_date = date(2006, 8, 1)

data_collection/gazette/spiders/rj/rj_comendador_levy_gasparian.py

@@ -0,0 +1,13 @@
+from datetime import date
+
+from gazette.spiders.base.ptio import BasePtioSpider
+
+
+class RjComendadorLevyGasparianSpider(BasePtioSpider):
+    name = "rj_comendador_levy_gasparian"
+    TERRITORY_ID = "3300951"
+    allowed_domains = ["portaldatransparencia.com.br"]
+    BASE_URL = (
+        "http://rj.portaldatransparencia.com.br/prefeitura/comendadorlevygasparian/"
+    )
+    start_date = date(2013, 11, 26)

data_collection/gazette/spiders/rj/rj_sapucaia.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.ptio import BasePtioSpider
+
+
+class RjSapucaiaSpider(BasePtioSpider):
+    name = "rj_sapucaia"
+    TERRITORY_ID = "3305406"
+    allowed_domains = ["portaldatransparencia.com.br"]
+    BASE_URL = "http://rj.portaldatransparencia.com.br/prefeitura/sapucaia/"
+    start_date = date(2019, 1, 16)