add: 20 novas cidades (#1014)

#999

data_collection/gazette/spiders/sp/sp_avanhandava.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpAvanhandavaSpider(BaseInstarSpider):
+    TERRITORY_ID = "3504404"
+    name = "sp_avanhandava"
+    allowed_domains = ["avanhandava.sp.gov.br"]
+    base_url = "https://www.avanhandava.sp.gov.br/portal/diario-oficial"
+    start_date = date(2022, 1, 7)

data_collection/gazette/spiders/sp/sp_barbosa.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpBarbosaSpider(BaseInstarSpider):
+    TERRITORY_ID = "3505104"
+    name = "sp_barbosa"
+    allowed_domains = ["barbosa.sp.gov.br"]
+    base_url = "https://www.barbosa.sp.gov.br/portal/diario-oficial"
+    start_date = date(2017, 1, 5)

data_collection/gazette/spiders/sp/sp_charqueada.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpCharqueadaSpider(BaseInstarSpider):
+    TERRITORY_ID = "3511706"
+    name = "sp_charqueada"
+    allowed_domains = ["charqueada.sp.gov.br"]
+    base_url = "https://www.charqueada.sp.gov.br/portal/diario-oficial"
+    start_date = date(2009, 1, 9)

data_collection/gazette/spiders/sp/sp_eldorado.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpEldoradoSpider(BaseInstarSpider):
+    TERRITORY_ID = "3514809"
+    name = "sp_eldorado"
+    allowed_domains = ["eldorado.sp.gov.br"]
+    base_url = "https://www.eldorado.sp.gov.br/portal/diario-oficial"
+    start_date = date(2018, 12, 4)

data_collection/gazette/spiders/sp/sp_general_salgado.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpGeneralSalgadoSpider(BaseInstarSpider):
+    TERRITORY_ID = "3516903"
+    name = "sp_general_salgado"
+    allowed_domains = ["generalsalgado.sp.gov.br"]
+    base_url = "https://www.generalsalgado.sp.gov.br/portal/diario-oficial"
+    start_date = date(2022, 1, 13)

data_collection/gazette/spiders/sp/sp_iepe.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpIepeSpider(BaseInstarSpider):
+    TERRITORY_ID = "3519907"
+    name = "sp_iepe"
+    allowed_domains = ["iepe.sp.gov.br"]
+    base_url = "https://www.iepe.sp.gov.br/portal/diario-oficial"
+    start_date = date(2019, 10, 16)

data_collection/gazette/spiders/sp/sp_irapuru.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpIrapuruSpider(BaseInstarSpider):
+    TERRITORY_ID = "3521606"
+    name = "sp_irapuru"
+    allowed_domains = ["irapuru.sp.gov.br"]
+    base_url = "https://www.irapuru.sp.gov.br/portal/diario-oficial"
+    start_date = date(2023, 3, 10)

data_collection/gazette/spiders/sp/sp_itapui.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpItapuiSpider(BaseInstarSpider):
+    TERRITORY_ID = "3522901"
+    name = "sp_itapui"
+    allowed_domains = ["itapui.sp.gov.br"]
+    base_url = "https://www.itapui.sp.gov.br/portal/diario-oficial"
+    start_date = date(2017, 12, 15)

data_collection/gazette/spiders/sp/sp_itariri.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpItaririSpider(BaseInstarSpider):
+    TERRITORY_ID = "3523305"
+    name = "sp_itariri"
+    allowed_domains = ["itariri.sp.gov.br"]
+    base_url = "https://www.itariri.sp.gov.br/portal/diario-oficial"
+    start_date = date(2023, 2, 24)

data_collection/gazette/spiders/sp/sp_itobi.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpItobiSpider(BaseInstarSpider):
+    TERRITORY_ID = "3523800"
+    name = "sp_itobi"
+    allowed_domains = ["itobi.sp.gov.br"]
+    base_url = "https://www.itobi.sp.gov.br/portal/diario-oficial"
+    start_date = date(2011, 4, 1)

data_collection/gazette/spiders/sp/sp_joanopolis.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpJoanopolisSpider(BaseInstarSpider):
+    TERRITORY_ID = "3525508"
+    name = "sp_joanopolis"
+    allowed_domains = ["joanopolis.sp.gov.br"]
+    base_url = "https://www.joanopolis.sp.gov.br/portal/diario-oficial"
+    start_date = date(2013, 1, 30)

data_collection/gazette/spiders/sp/sp_junqueiropolis.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpJunqueiropolisSpider(BaseInstarSpider):
+    TERRITORY_ID = "3526001"
+    name = "sp_junqueiropolis"
+    allowed_domains = ["junqueiropolis.sp.gov.br"]
+    base_url = "https://www.junqueiropolis.sp.gov.br/portal/diario-oficial"
+    start_date = date(2020, 10, 24)

data_collection/gazette/spiders/sp/sp_macaubal.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpMacaubalSpider(BaseInstarSpider):
+    TERRITORY_ID = "3528106"
+    name = "sp_macaubal"
+    allowed_domains = ["macaubal.sp.gov.br"]
+    base_url = "https://macaubal.sp.gov.br/portal/diario-oficial"
+    start_date = date(2023, 6, 21)

data_collection/gazette/spiders/sp/sp_mirante_do_paranapanema.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpMiranteDoParanapanemaSpider(BaseInstarSpider):
+    TERRITORY_ID = "3530201"
+    name = "sp_mirante_do_paranapanema"
+    allowed_domains = ["mirantedoparanapanema.sp.gov.br"]
+    base_url = "https://www.mirantedoparanapanema.sp.gov.br/portal/diario-oficial"
+    start_date = date(2019, 5, 7)

data_collection/gazette/spiders/sp/sp_nhandeara.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpNhandearaSpider(BaseInstarSpider):
+    TERRITORY_ID = "3532603"
+    name = "sp_nhandeara"
+    allowed_domains = ["nhandeara.sp.gov.br"]
+    base_url = "https://www.nhandeara.sp.gov.br/portal/diario-oficial"
+    start_date = date(2023, 8, 17)

data_collection/gazette/spiders/sp/sp_palmital.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpPalmitalSpider(BaseInstarSpider):
+    TERRITORY_ID = "3535309"
+    name = "sp_palmital"
+    allowed_domains = ["palmital.sp.gov.br"]
+    base_url = "https://www.palmital.sp.gov.br/portal/diario-oficial"
+    start_date = date(2005, 6, 11)

data_collection/gazette/spiders/sp/sp_pindorama.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpPindoramaSpider(BaseInstarSpider):
+    TERRITORY_ID = "3538105"
+    name = "sp_pindorama"
+    allowed_domains = ["pindorama.sp.gov.br"]
+    base_url = "https://www.pindorama.sp.gov.br/portal/diario-oficial"
+    start_date = date(2022, 4, 29)

data_collection/gazette/spiders/sp/sp_porangaba.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpPorangabaSpider(BaseInstarSpider):
+    TERRITORY_ID = "3540507"
+    name = "sp_porangaba"
+    allowed_domains = ["porangaba.sp.gov.br"]
+    base_url = "https://www.porangaba.sp.gov.br/portal/diario-oficial"
+    start_date = date(2020, 10, 6)

data_collection/gazette/spiders/sp/sp_tapirai.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpTapiraiSpider(BaseInstarSpider):
+    TERRITORY_ID = "3553500"
+    name = "sp_tapirai"
+    allowed_domains = ["tapirai.sp.gov.br"]
+    base_url = "https://www.tapirai.sp.gov.br/portal/diario-oficial"
+    start_date = date(2019, 9, 3)

data_collection/gazette/spiders/sp/sp_terra_roxa.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpTerraRoxaSpider(BaseInstarSpider):
+    TERRITORY_ID = "3554409"
+    name = "sp_terra_roxa"
+    allowed_domains = ["terraroxa.sp.gov.br"]
+    base_url = "https://www.terraroxa.sp.gov.br/portal/diario-oficial"
+    start_date = date(2022, 6, 29)

scripts/enabled_spiders.py

@@ -134,59 +134,77 @@
     "sp_aracariguama",
     "sp_aracatuba",
     "sp_arapei",
+    "sp_avanhandava",
     "sp_avare",
     "sp_barao_de_antonina",
+    "sp_barbosa",
     "sp_birigui",
     "sp_botucatu",
     "sp_braganca_paulista",
     "sp_brejo_alegre",
     "sp_campinas",
     "sp_campo_limpo_paulista",
     "sp_catanduva",
+    "sp_charqueada",
     "sp_coronel_macedo",
     "sp_dirce_reis",
     "sp_dracena",
+    "sp_eldorado",
     "sp_floreal",
+    "sp_general_salgado",
     "sp_glicerio",
     "sp_guaracai",
     "sp_guarulhos",
     "sp_ibitinga",
+    "sp_iepe",
+    "sp_irapuru",
     "sp_itapeva",
     "sp_itapevi",
     "sp_itapirapua_paulista",
     "sp_itapolis",
+    "sp_itapui",
+    "sp_itariri",
+    "sp_itobi",
     "sp_itu",
-    "sp_jandira",
     "sp_jaboticabal",
+    "sp_jandira",
     "sp_jau_2023",
+    "sp_joanopolis",
     "sp_joao_ramalho",
     "sp_jundiai",
+    "sp_junqueiropolis",
     "sp_lagoinha",
     "sp_lavinia",
     "sp_luiziania",
+    "sp_macaubal",
     "sp_marilia",
     "sp_mira_estrela",
-    "sp_monte_alto_2017",
+    "sp_mirante_do_paranapanema",
     "sp_mogi_guacu",
+    "sp_monte_alto_2017",
     "sp_monte_mor",
+    "sp_nhandeara",
     "sp_nova_luzitania",
     "sp_nova_castilho",
     "sp_osasco",
     "sp_ourinhos",
+    "sp_palmital",
     "sp_parisi",
     "sp_patrocinio_paulista",
     "sp_paulinia",
     "sp_penapolis",
     "sp_piedade",
-    "sp_poloni",
+    "sp_pindorama",
     "sp_planalto",
+    "sp_poloni",
     "sp_pontes_gestal",
+    "sp_porangaba",
     "sp_pratania",
     "sp_presidente_epitacio",
     "sp_rio_claro",
+    "sp_salto",
     "sp_santa_ernestina",
     "sp_santa_maria_da_serra",
-    "sp_salto",
     "sp_santo_andre",
     "sp_santos",
     "sp_sao_bernardo_do_campo",
@@ -200,6 +218,8 @@
     "sp_taquaral",
     "sp_turiuba",
     "sp_uniao_paulista",
+    "sp_tapirai",
+    "sp_terra_roxa",
     "sp_valinhos",
     "sp_valparaiso",
     "sp_vera_cruz",