diff --git a/data_collection/gazette/spiders/base/adiarios_v2.py b/data_collection/gazette/spiders/base/adiarios_v2.py index 0a44bdc60..85c3a0319 100644 --- a/data_collection/gazette/spiders/base/adiarios_v2.py +++ b/data_collection/gazette/spiders/base/adiarios_v2.py @@ -8,6 +8,11 @@ class BaseAdiariosV2Spider(BaseGazetteSpider): + """ + This base class deals with 'Layout 2' gazette pages, usually requested + from https://{city_website}/jornal.php + """ + def start_requests(self): start_date = self.start_date.strftime("%d/%m/%Y") end_date = self.end_date.strftime("%d/%m/%Y") diff --git a/data_collection/gazette/spiders/rj/rj_armacao_dos_buzios.py b/data_collection/gazette/spiders/rj/rj_armacao_dos_buzios.py new file mode 100644 index 000000000..4aaeca319 --- /dev/null +++ b/data_collection/gazette/spiders/rj/rj_armacao_dos_buzios.py @@ -0,0 +1,11 @@ +from datetime import date + +from gazette.spiders.base.adiarios_v2 import BaseAdiariosV2Spider + + +class RjArmacaoDosBuziosSpider(BaseAdiariosV2Spider): + TERRITORY_ID = "3300233" + name = "rj_armacao_dos_buzios" + allowed_domains = ["buzios.aexecutivo.com.br"] + BASE_URL = "https://buzios.aexecutivo.com.br" + start_date = date(2015, 9, 3) diff --git a/data_collection/gazette/spiders/rj/rj_iguaba_grande.py b/data_collection/gazette/spiders/rj/rj_iguaba_grande.py new file mode 100644 index 000000000..6e0d365d5 --- /dev/null +++ b/data_collection/gazette/spiders/rj/rj_iguaba_grande.py @@ -0,0 +1,11 @@ +from datetime import date + +from gazette.spiders.base.adiarios_v2 import BaseAdiariosV2Spider + + +class RjIguabaGrandeSpider(BaseAdiariosV2Spider): + TERRITORY_ID = "3301876" + name = "rj_iguaba_grande" + allowed_domains = ["iguaba.rj.gov.br"] + BASE_URL = "https://portal.iguaba.rj.gov.br" + start_date = date(2013, 1, 1) diff --git a/data_collection/gazette/spiders/rj/rj_quissama.py b/data_collection/gazette/spiders/rj/rj_quissama.py new file mode 100644 index 000000000..f1b0311c1 --- /dev/null +++ b/data_collection/gazette/spiders/rj/rj_quissama.py @@ -0,0 +1,11 @@ +from datetime import date + +from gazette.spiders.base.adiarios_v2 import BaseAdiariosV2Spider + + +class RjQuissamaSpider(BaseAdiariosV2Spider): + TERRITORY_ID = "3304151" + name = "rj_quissama" + allowed_domains = ["quissama.rj.gov.br"] + BASE_URL = "https://portal.quissama.rj.gov.br" + start_date = date(2017, 1, 31) diff --git a/data_collection/gazette/spiders/rj/rj_sao_jose_do_vale_do_rio_preto.py b/data_collection/gazette/spiders/rj/rj_sao_jose_do_vale_do_rio_preto.py new file mode 100644 index 000000000..8c6143d77 --- /dev/null +++ b/data_collection/gazette/spiders/rj/rj_sao_jose_do_vale_do_rio_preto.py @@ -0,0 +1,10 @@ +from datetime import date + +from gazette.spiders.base.dosp import DospGazetteSpider + + +class RjSaoJoseDoValeDoRioPretoSpider(DospGazetteSpider): + TERRITORY_ID = "3305158" + name = "rj_sao_jose_do_vale_do_rio_preto" + code = 3640 + start_date = date(2023, 5, 24)