Padroniza o layout de raspadores para municípios replicados sem modif…

…icar suas informações (#1157)

resolve item 1 de #1156

data_collection/gazette/spiders/al/al_igaci.py

@@ -1,11 +1,11 @@
-import datetime as dt
+from datetime import date
 
 from gazette.spiders.base.sai import SaiGazetteSpider
 
 
 class AlIgaciSpider(SaiGazetteSpider):
     TERRITORY_ID = "2703106"
     name = "al_igaci"
-    start_date = dt.date(2015, 7, 17)
     allowed_domains = ["igaci.al.gov.br"]
     base_url = "https://www.igaci.al.gov.br"
+    start_date = date(2015, 7, 17)

data_collection/gazette/spiders/ba/ba_abare.py

@@ -1,11 +1,11 @@
-import datetime as dt
+from datetime import date
 
 from gazette.spiders.base.sai import SaiGazetteSpider
 
 
 class BaAbareSpider(SaiGazetteSpider):
     TERRITORY_ID = "2900207"
     name = "ba_abare"
-    start_date = dt.date(2007, 1, 9)
     allowed_domains = ["sai.io.org.br"]
     base_url = "https://sai.io.org.br/ba/abare"
+    start_date = date(2007, 1, 9)

data_collection/gazette/spiders/ba/ba_adustina.py

@@ -1,11 +1,11 @@
-import datetime as dt
+from datetime import date
 
 from gazette.spiders.base.sai import SaiGazetteSpider
 
 
 class BaAdustinaSpider(SaiGazetteSpider):
     TERRITORY_ID = "2900355"
     name = "ba_adustina"
-    start_date = dt.date(2017, 1, 3)
     allowed_domains = ["adustina.ba.gov.br"]
     base_url = "https://www.adustina.ba.gov.br"
+    start_date = date(2017, 1, 3)

data_collection/gazette/spiders/ba/ba_almadina.py

@@ -1,11 +1,11 @@
-import datetime as dt
+from datetime import date
 
 from gazette.spiders.base.sai import SaiGazetteSpider
 
 
 class BaAlmadinaSpider(SaiGazetteSpider):
     TERRITORY_ID = "2900900"
     name = "ba_almadina"
-    start_date = dt.date(2005, 1, 3)
     allowed_domains = ["almadina.ba.gov.br"]
     base_url = "https://www.almadina.ba.gov.br"
+    start_date = date(2005, 1, 3)

data_collection/gazette/spiders/ba/ba_anage.py

@@ -1,11 +1,11 @@
-import datetime as dt
+from datetime import date
 
 from gazette.spiders.base.sai import SaiGazetteSpider
 
 
 class BaAnageSpider(SaiGazetteSpider):
     TERRITORY_ID = "2901205"
     name = "ba_anage"
-    start_date = dt.date(2007, 1, 12)
     allowed_domains = ["anage.ba.gov.br"]
     base_url = "https://www.anage.ba.gov.br"
+    start_date = date(2007, 1, 12)

data_collection/gazette/spiders/ba/ba_andorinha.py

@@ -1,11 +1,11 @@
-import datetime as dt
+from datetime import date
 
 from gazette.spiders.base.sai import SaiGazetteSpider
 
 
 class BaAndorinhaSpider(SaiGazetteSpider):
     TERRITORY_ID = "2901353"
     name = "ba_andorinha"
-    start_date = dt.date(2013, 1, 2)
     allowed_domains = ["andorinha.ba.gov.br"]
     base_url = "https://www.andorinha.ba.gov.br"
+    start_date = date(2013, 1, 2)

data_collection/gazette/spiders/ba/ba_correntina.py

@@ -1,11 +1,11 @@
-import datetime as dt
+from datetime import date
 
 from gazette.spiders.base.sai import SaiGazetteSpider
 
 
 class BaCorrentinaSpider(SaiGazetteSpider):
     TERRITORY_ID = "2909307"
     name = "ba_correntina"
-    start_date = dt.date(2007, 11, 30)
     allowed_domains = ["sai.io.org.br"]
     base_url = "https://sai.io.org.br/ba/correntina"
+    start_date = date(2007, 11, 30)

data_collection/gazette/spiders/ba/ba_jaborandi.py

@@ -1,11 +1,11 @@
-import datetime as dt
+from datetime import date
 
 from gazette.spiders.base.sai import SaiGazetteSpider
 
 
 class BaJaborandiSpider(SaiGazetteSpider):
     TERRITORY_ID = "2917359"
     name = "ba_jaborandi"
-    start_date = dt.date(2022, 3, 4)
     allowed_domains = ["sai.io.org.br"]
     base_url = "https://sai.io.org.br/ba/jaborandi"
+    start_date = date(2022, 3, 4)

data_collection/gazette/spiders/ba/ba_jeremoabo.py

@@ -1,11 +1,11 @@
-import datetime as dt
+from datetime import date
 
 from gazette.spiders.base.sai import SaiGazetteSpider
 
 
 class BaJeremoaboSpider(SaiGazetteSpider):
     TERRITORY_ID = "2918100"
     name = "ba_jeremoabo"
-    start_date = dt.date(2016, 4, 28)
     allowed_domains = ["jeremoabo.ba.gov.br"]
     base_url = "https://www.jeremoabo.ba.gov.br"
+    start_date = date(2016, 4, 28)

data_collection/gazette/spiders/ba/ba_lauro_de_freitas.py

@@ -1,11 +1,11 @@
-import datetime as dt
+from datetime import date
 
 from gazette.spiders.base.sai import SaiGazetteSpider
 
 
 class BaLauroDeFreitasSpider(SaiGazetteSpider):
     TERRITORY_ID = "2919207"
     name = "ba_lauro_de_freitas"
-    start_date = dt.date(2013, 7, 31)
     allowed_domains = ["sai.io.org.br"]
     base_url = "https://sai.io.org.br/ba/laurodefreitas"
+    start_date = date(2013, 7, 31)

data_collection/gazette/spiders/ba/ba_luis_eduardo_magalhaes.py

@@ -1,11 +1,11 @@
-import datetime as dt
+from datetime import date
 
 from gazette.spiders.base.sai import SaiGazetteSpider
 
 
 class BaLuisEduardoMagalhaesSpider(SaiGazetteSpider):
     TERRITORY_ID = "2919553"
     name = "ba_luis_eduardo_magalhaes"
-    start_date = dt.date(2017, 1, 4)
     allowed_domains = ["sai.io.org.br"]
     base_url = "https://sai.io.org.br/ba/luiseduardomagalhaes"
+    start_date = date(2017, 1, 4)

data_collection/gazette/spiders/ba/ba_riachao_das_neves.py

@@ -1,11 +1,11 @@
-import datetime as dt
+from datetime import date
 
 from gazette.spiders.base.sai import SaiGazetteSpider
 
 
 class BaRiachaoDasNevesSpider(SaiGazetteSpider):
     TERRITORY_ID = "2926202"
     name = "ba_riachao_das_neves"
-    start_date = dt.date(2010, 2, 4)
     allowed_domains = ["sai.io.org.br"]
     base_url = "https://sai.io.org.br/ba/riachaodasneves"
+    start_date = date(2010, 2, 4)

data_collection/gazette/spiders/base/atende_layoutdois.py → data_collection/gazette/spiders/base/atende_v2.py

@@ -7,7 +7,7 @@
 from gazette.spiders.base import BaseGazetteSpider
 
 
-class BaseAtendeL2Spider(BaseGazetteSpider):
+class BaseAtendeV2Spider(BaseGazetteSpider):
     """
     Base spider for Gazettes that are available from cities listed on
     https://{city_subdomain}.atende.net

data_collection/gazette/spiders/ce/ce_caucaia.py

@@ -1,12 +1,11 @@
-import datetime
+from datetime import date
 
 from gazette.spiders.base.adiarios_v1 import BaseAdiariosV1Spider
 
 
 class CeCaucaiaSpider(BaseAdiariosV1Spider):
+    TERRITORY_ID = "2303709"
     name = "ce_caucaia"
     allowed_domains = ["caucaia.ce.gov.br"]
-    start_date = datetime.date(2002, 6, 3)
-
-    TERRITORY_ID = "2303709"
     BASE_URL = "https://www.caucaia.ce.gov.br"
+    start_date = date(2002, 6, 3)

data_collection/gazette/spiders/ce/ce_horizonte.py

@@ -6,5 +6,5 @@
 class CeHorizonteSpider(DospGazetteSpider):
     TERRITORY_ID = "2305233"
     name = "ce_horizonte"
-    code = "687"
+    code = 687
     start_date = date(2023, 7, 3)

data_collection/gazette/spiders/ce/ce_juazeiro_do_norte.py

@@ -1,12 +1,11 @@
-import datetime
+from datetime import date
 
 from gazette.spiders.base.adiarios_v1 import BaseAdiariosV1Spider
 
 
 class CeJuazeiroDoNorteSpider(BaseAdiariosV1Spider):
+    TERRITORY_ID = "2307304"
     name = "ce_juazeiro_do_norte"
     allowed_domains = ["juazeirodonorte.ce.gov.br"]
-    start_date = datetime.date(2021, 1, 12)
-
-    TERRITORY_ID = "2307304"
     BASE_URL = "https://www.juazeirodonorte.ce.gov.br"
+    start_date = date(2021, 1, 12)

data_collection/gazette/spiders/ma/ma_buriticupu.py

@@ -1,12 +1,11 @@
-import datetime
+from datetime import date
 
 from gazette.spiders.base.adiarios_v1 import BaseAdiariosV1Spider
 
 
 class MaBuriticupuSpider(BaseAdiariosV1Spider):
+    TERRITORY_ID = "2102325"
     name = "ma_buriticupu"
     allowed_domains = ["buriticupu.ma.gov.br"]
-    start_date = datetime.date(2018, 5, 7)
-
-    TERRITORY_ID = "2102325"
     BASE_URL = "https://www.buriticupu.ma.gov.br"
+    start_date = date(2018, 5, 7)

data_collection/gazette/spiders/mg/mg_campo_belo.py

@@ -8,4 +8,4 @@ class MgCampoBeloSpider(BaseInstarSpider):
     name = "mg_campo_belo"
     allowed_domains = ["campobelo.mg.gov.br"]
     base_url = "https://www.campobelo.mg.gov.br/portal/diario-oficial"
-    start_date = date(2016, 1, 5)  # edition_number 001
+    start_date = date(2016, 1, 5)

data_collection/gazette/spiders/mg/mg_carmo_da_cachoeira.py

@@ -3,7 +3,7 @@
 from gazette.spiders.base.instar import BaseInstarSpider
 
 
-class MgCarmodaCachoeiraSpider(BaseInstarSpider):
+class MgCarmoDaCachoeiraSpider(BaseInstarSpider):
     TERRITORY_ID = "3113909"
     name = "mg_carmo_da_cachoeira"
     allowed_domains = ["carmodacachoeira.mg.gov.br"]

data_collection/gazette/spiders/mg/mg_crucilandia.py

@@ -7,5 +7,5 @@ class MgCrucilandiaSpider(BaseInstarSpider):
     TERRITORY_ID = "3120607"
     name = "mg_crucilandia"
     allowed_domains = ["prefeituradecrucilandia.mg.gov.br"]
-    base_url = "http://www.prefeituradecrucilandia.mg.gov.br/portal/diario-oficial"
+    base_url = "https://www.prefeituradecrucilandia.mg.gov.br/portal/diario-oficial"
     start_date = date(2015, 3, 31)

data_collection/gazette/spiders/mg/mg_itajuba.py

@@ -6,5 +6,5 @@
 class MgItajubaSpider(DospGazetteSpider):
     TERRITORY_ID = "3132404"
     name = "mg_itajuba"
-    code = "1908"
+    code = 1908
     start_date = date(2023, 5, 15)

data_collection/gazette/spiders/mg/mg_itauna.py

@@ -8,4 +8,4 @@ class MgItaunaSpider(BaseInstarSpider):
     name = "mg_itauna"
     allowed_domains = ["itauna.mg.gov.br"]
     base_url = "https://www.itauna.mg.gov.br/portal/diario-oficial"
-    start_date = date(2013, 6, 20)  # edition_number 907
+    start_date = date(2013, 6, 20)

data_collection/gazette/spiders/mg/mg_taiobeiras.py

@@ -3,9 +3,9 @@
 from gazette.spiders.base.instar import BaseInstarSpider
 
 
-class MgTaiobeiras(BaseInstarSpider):
+class MgTaiobeirasSpider(BaseInstarSpider):
     TERRITORY_ID = "3168002"
     name = "mg_taiobeiras"
     allowed_domains = ["taiobeiras.mg.gov.br"]
     base_url = "https://www.taiobeiras.mg.gov.br/portal/diario-oficial"
-    start_date = date(2022, 12, 31)  # Primeira edicao
+    start_date = date(2022, 12, 31)

data_collection/gazette/spiders/mg/mg_uberaba_2021.py

@@ -1,4 +1,4 @@
-import datetime as dt
+from datetime import date
 
 from gazette.spiders.base.dosp import DospGazetteSpider
 
@@ -7,4 +7,4 @@ class MgUberabaSpider(DospGazetteSpider):
     TERRITORY_ID = "3170107"
     name = "mg_uberaba_2021"
     code = 2364
-    start_date = dt.date(2021, 9, 1)
+    start_date = date(2021, 9, 1)

data_collection/gazette/spiders/mg/mg_varzea_da_palma.py

@@ -8,4 +8,4 @@ class MgVarzeaDaPalmaSpider(BaseInstarSpider):
     name = "mg_varzea_da_palma"
     allowed_domains = ["varzeadapalma.mg.gov.br"]
     base_url = "https://www.varzeadapalma.mg.gov.br/portal/diario-oficial"
-    start_date = date(2022, 9, 30)  # Primeiro diario registrado
+    start_date = date(2022, 9, 30)

data_collection/gazette/spiders/ms/ms_inocencia.py

@@ -1,4 +1,4 @@
-import datetime
+from datetime import date
 
 from gazette.spiders.base.instar import BaseInstarSpider
 
@@ -8,4 +8,4 @@ class MsInocenciaSpider(BaseInstarSpider):
     name = "ms_inocencia"
     allowed_domains = ["inocencia.ms.gov.br"]
     base_url = "https://www.inocencia.ms.gov.br/portal/diario-oficial"
-    start_date = datetime.date(2013, 10, 17)  # edition_number 01
+    start_date = date(2013, 10, 17)

data_collection/gazette/spiders/ms/ms_maracaju.py

@@ -1,4 +1,4 @@
-import datetime
+from datetime import date
 
 from gazette.spiders.base.instar import BaseInstarSpider
 
@@ -8,4 +8,4 @@ class MsMaracajuSpider(BaseInstarSpider):
     name = "ms_maracaju"
     allowed_domains = ["maracaju.ms.gov.br"]
     base_url = "https://www.maracaju.ms.gov.br/portal/diario-oficial"
-    start_date = datetime.date(2013, 3, 14)  # edition_number 01
+    start_date = date(2013, 3, 14)

data_collection/gazette/spiders/pe/pe_cabrobo.py

@@ -6,5 +6,5 @@
 class PeCabroboSpider(DospGazetteSpider):
     TERRITORY_ID = "2603009"
     name = "pe_cabrobo"
-    code = "3190"
+    code = 3190
     start_date = date(2019, 4, 8)

data_collection/gazette/spiders/pr/pr_cafelandia.py

@@ -6,5 +6,5 @@
 class PrCafelandiaSpider(DospGazetteSpider):
     TERRITORY_ID = "4103453"
     name = "pr_cafelandia"
-    code = "4748"
+    code = 4748
     start_date = date(2017, 6, 12)

data_collection/gazette/spiders/pr/pr_sao_mateus_do_sul.py

@@ -1,11 +1,11 @@
-import datetime
+from datetime import date
 
 from gazette.spiders.base.instar import BaseInstarSpider
 
 
-class PrSaoMateusdoSUlSpider(BaseInstarSpider):
+class PrSaoMateusDoSulSpider(BaseInstarSpider):
     TERRITORY_ID = "4125605"
     name = "pr_sao_mateus_do_sul"
     allowed_domains = ["saomateusdosul.pr.gov.br"]
     base_url = "https://www.saomateusdosul.pr.gov.br/portal/diario-oficial"
-    start_date = datetime.date(2010, 12, 10)  # edition_number 01
+    start_date = date(2010, 12, 10)

data_collection/gazette/spiders/rj/rj_casimiro_de_abreu.py

@@ -4,8 +4,8 @@
 
 
 class RjCasimiroDeAbreuSpider(BaseAdiariosV2Spider):
-    name = "rj_casimiro_de_abreu"
     TERRITORY_ID = "3301306"
-    start_date = date(2017, 1, 4)
+    name = "rj_casimiro_de_abreu"
     allowed_domains = ["casimirodeabreu.rj.gov.br"]
-    BASE_URL = "https://transparencia.casimirodeabreu.rj.gov.br"
+    BASE_URL = "https://www.transparencia.casimirodeabreu.rj.gov.br"
+    start_date = date(2017, 1, 4)

data_collection/gazette/spiders/rj/rj_cordeiro.py

@@ -4,8 +4,8 @@
 
 
 class RjCordeiroSpider(BaseAdiariosV2Spider):
-    name = "rj_cordeiro"
     TERRITORY_ID = "3301504"
-    start_date = date(2017, 10, 5)
+    name = "rj_cordeiro"
     allowed_domains = ["cordeiro.rj.gov.br"]
-    BASE_URL = "https://transparencia.cordeiro.rj.gov.br"
+    BASE_URL = "https://www.transparencia.cordeiro.rj.gov.br"
+    start_date = date(2017, 10, 5)

data_collection/gazette/spiders/rs/rs_bento_goncalves.py

@@ -1,11 +1,10 @@
 from datetime import date
 
-from gazette.spiders.base.atende_layoutdois import BaseAtendeL2Spider
+from gazette.spiders.base.atende_v2 import BaseAtendeV2Spider
 
 
-class RsBentoGoncalvesSpider(BaseAtendeL2Spider):
+class RsBentoGoncalvesSpider(BaseAtendeV2Spider):
     TERRITORY_ID = "4302105"
     name = "rs_bento_goncalves"
-    start_date = date(2019, 4, 1)  # Edição 1124
     city_subdomain = "bentogoncalves"
-    # power
+    start_date = date(2019, 4, 1)