Habilita 13 novos raspadores Instar em produção (#959)

resolve #948

data_collection/gazette/spiders/sp/sp_alto_alegre.py

@@ -1,8 +1,11 @@
+from datetime import date
+
 from gazette.spiders.base.instar import BaseInstarSpider
 
 
 class SpAltoAlegreSpider(BaseInstarSpider):
     TERRITORY_ID = "3501103"
     name = "sp_alto_alegre"
     allowed_domains = ["altoalegre.sp.gov.br"]
-    start_urls = ["http://www.altoalegre.sp.gov.br/portal/diario-oficial"]
+    start_date = date(2018, 7, 3)
+    base_url = "http://www.altoalegre.sp.gov.br/portal/diario-oficial"

data_collection/gazette/spiders/sp/sp_aracariguama.py

@@ -1,8 +1,11 @@
+from datetime import date
+
 from gazette.spiders.base.instar import BaseInstarSpider
 
 
 class SpAracariguamaSpider(BaseInstarSpider):
     TERRITORY_ID = "3502754"
     name = "sp_aracariguama"
     allowed_domains = ["aracariguama.sp.gov.br"]
-    start_urls = ["https://www.aracariguama.sp.gov.br/portal/diario-oficial"]
+    start_date = date(2019, 9, 6)
+    base_url = "https://www.aracariguama.sp.gov.br/portal/diario-oficial"

data_collection/gazette/spiders/sp/sp_coronel_macedo.py

@@ -1,8 +1,11 @@
+from datetime import date
+
 from gazette.spiders.base.instar import BaseInstarSpider
 
 
 class SpCoronelMacedoSpider(BaseInstarSpider):
     TERRITORY_ID = "3512605"
     name = "sp_coronel_macedo"
     allowed_domains = ["coronelmacedo.sp.gov.br"]
-    start_urls = ["https://www.coronelmacedo.sp.gov.br/portal/diario-oficial"]
+    start_date = date(2017, 5, 29)
+    base_url = "https://www.coronelmacedo.sp.gov.br/portal/diario-oficial"

data_collection/gazette/spiders/sp/sp_glicerio.py

@@ -1,8 +1,11 @@
+from datetime import date
+
 from gazette.spiders.base.instar import BaseInstarSpider
 
 
 class SpGlicerioSpider(BaseInstarSpider):
     TERRITORY_ID = "3517109"
     name = "sp_glicerio"
     allowed_domains = ["glicerio.sp.gov.br"]
-    start_urls = ["https://www.glicerio.sp.gov.br/portal/diario-oficial/"]
+    start_date = date(2019, 1, 8)
+    base_url = "https://www.glicerio.sp.gov.br/portal/diario-oficial/"

data_collection/gazette/spiders/sp/sp_itapirapua_paulista.py

@@ -1,8 +1,11 @@
+from datetime import date
+
 from gazette.spiders.base.instar import BaseInstarSpider
 
 
 class SpItapirapuaPaulistaSpider(BaseInstarSpider):
     TERRITORY_ID = "3522653"
     name = "sp_itapirapua_paulista"
     allowed_domains = ["itapirapuapaulista.sp.gov.br"]
-    start_urls = ["https://www.itapirapuapaulista.sp.gov.br/portal/diario-oficial"]
+    start_date = date(2019, 5, 24)
+    base_url = "https://www.itapirapuapaulista.sp.gov.br/portal/diario-oficial"

data_collection/gazette/spiders/sp/sp_lavinia.py

@@ -1,8 +1,11 @@
+from datetime import date
+
 from gazette.spiders.base.instar import BaseInstarSpider
 
 
 class SpLaviniaSpider(BaseInstarSpider):
     TERRITORY_ID = "3526506"
     name = "sp_lavinia"
     allowed_domains = ["lavinia.sp.gov.br"]
-    start_urls = ["https://www.lavinia.sp.gov.br/portal/diario-oficial"]
+    start_date = date(2018, 8, 2)
+    base_url = "https://www.lavinia.sp.gov.br/portal/diario-oficial"

data_collection/gazette/spiders/sp/sp_monte_alto.py

@@ -1,14 +1,6 @@
-from gazette.spiders.base.instar import BaseInstarSpider
 from gazette.spiders.base.sigpub import SigpubGazetteSpider
 
 
-class SpMonteAltoSpider(BaseInstarSpider):
-    TERRITORY_ID = "3531308"
-    name = "sp_monte_alto"
-    allowed_domains = ["montealto.instaridc.com.br"]
-    start_urls = ["http://montealto.instaridc.com.br/portal/diario-oficial"]
-
-
 class SpMonteAltoSigpubSpider(SigpubGazetteSpider):
     name = "sp_monte_alto_sigpub"
     TERRITORY_ID = "3531308"

data_collection/gazette/spiders/sp/sp_monte_alto_2017.py

@@ -0,0 +1,11 @@
+from datetime import date
+
+from gazette.spiders.base.instar import BaseInstarSpider
+
+
+class SpMonteAltoSpider(BaseInstarSpider):
+    TERRITORY_ID = "3531308"
+    name = "sp_monte_alto_2017"
+    allowed_domains = ["montealto.instaridc.com.br"]
+    start_date = date(2017, 9, 11)
+    base_url = "http://montealto.instaridc.com.br/portal/diario-oficial"

data_collection/gazette/spiders/sp/sp_parisi.py

@@ -1,8 +1,11 @@
+from datetime import date
+
 from gazette.spiders.base.instar import BaseInstarSpider
 
 
 class SpParisiSpider(BaseInstarSpider):
     TERRITORY_ID = "3536257"
     name = "sp_parisi"
     allowed_domains = ["parisi.sp.gov.br"]
-    start_urls = ["https://www.parisi.sp.gov.br/portal/diario-oficial"]
+    start_date = date(2015, 2, 27)
+    base_url = "https://www.parisi.sp.gov.br/portal/diario-oficial"

data_collection/gazette/spiders/sp/sp_patrocinio_paulista.py

@@ -1,8 +1,11 @@
+from datetime import date
+
 from gazette.spiders.base.instar import BaseInstarSpider
 
 
 class SpPatrocinioPaulistaSpider(BaseInstarSpider):
     TERRITORY_ID = "3536307"
     name = "sp_patrocinio_paulista"
     allowed_domains = ["patrociniopaulista.sp.gov.br"]
-    start_urls = ["https://www.patrociniopaulista.sp.gov.br/portal/diario-oficial"]
+    start_date = date(2017, 8, 18)
+    base_url = "https://www.patrociniopaulista.sp.gov.br/portal/diario-oficial"

data_collection/gazette/spiders/sp/sp_pratania.py

@@ -1,8 +1,11 @@
+from datetime import date
+
 from gazette.spiders.base.instar import BaseInstarSpider
 
 
 class SpPrataniaSpider(BaseInstarSpider):
     TERRITORY_ID = "3541059"
     name = "sp_pratania"
     allowed_domains = ["pratania.sp.gov.br"]
-    start_urls = ["https://www.pratania.sp.gov.br/portal/diario-oficial"]
+    start_date = date(2019, 5, 13)
+    base_url = "https://www.pratania.sp.gov.br/portal/diario-oficial"

data_collection/gazette/spiders/sp/sp_santa_ernestina.py

@@ -1,8 +1,11 @@
+from datetime import date
+
 from gazette.spiders.base.instar import BaseInstarSpider
 
 
 class SpSantaErnestinaPaulistaSpider(BaseInstarSpider):
     TERRITORY_ID = "3546504"
     name = "sp_santa_ernestina"
     allowed_domains = ["santaernestina.sp.gov.br"]
-    start_urls = ["https://www.santaernestina.sp.gov.br/portal/diario-oficial"]
+    start_date = date(2019, 8, 19)
+    base_url = "https://www.santaernestina.sp.gov.br/portal/diario-oficial"

data_collection/gazette/spiders/sp/sp_sao_manuel.py

@@ -1,8 +1,11 @@
+from datetime import date
+
 from gazette.spiders.base.instar import BaseInstarSpider
 
 
 class SpSaoManuelSpider(BaseInstarSpider):
     TERRITORY_ID = "3550100"
     name = "sp_sao_manuel"
     allowed_domains = ["saomanuel.sp.gov.br"]
-    start_urls = ["https://www.saomanuel.sp.gov.br/portal/diario-oficial"]
+    start_date = date(2016, 6, 7)
+    base_url = "https://www.saomanuel.sp.gov.br/portal/diario-oficial"

data_collection/gazette/spiders/sp/sp_sarutaia.py

@@ -1,8 +1,11 @@
+from datetime import date
+
 from gazette.spiders.base.instar import BaseInstarSpider
 
 
 class SpSarutaiaSpider(BaseInstarSpider):
     TERRITORY_ID = "3551207"
     name = "sp_sarutaia"
     allowed_domains = ["sarutaia.sp.gov.br"]
-    start_urls = ["https://www.sarutaia.sp.gov.br/portal/diario-oficial"]
+    start_date = date(2020, 3, 27)
+    base_url = "https://www.sarutaia.sp.gov.br/portal/diario-oficial"

scripts/enabled_spiders.py

@@ -94,28 +94,41 @@
     "sc_joinville",
     "se_nossa_senhora_do_socorro",
     "sp_adolfo",
+    "sp_alto_alegre",
+    "sp_aracariguama",
     "sp_barao_de_antonina",
     "sp_birigui",
     "sp_braganca_paulista",
     "sp_campinas",
     "sp_catanduva",
+    "sp_coronel_macedo",
+    "sp_glicerio",
     "sp_guaracai",
     "sp_guarulhos",
     "sp_ibitinga",
     "sp_itapevi",
+    "sp_itapirapua_paulista",
     "sp_jaboticabal",
     "sp_jandira",
     "sp_jundiai",
+    "sp_lavinia",
     "sp_marilia",
+    "sp_monte_alto_2017",
     "sp_osasco",
+    "sp_parisi",
+    "sp_patrocinio_paulista",
     "sp_paulinia",
     "sp_penapolis",
     "sp_piedade",
+    "sp_pratania",
     "sp_rio_claro",
+    "sp_santa_ernestina",
     "sp_santo_andre",
     "sp_santos",
     "sp_sao_bernardo_do_campo",
+    "sp_sao_manuel",
     "sp_sao_roque",
+    "sp_sarutaia",
     "sp_sumare",
     "sp_valinhos",
     "sp_vera_cruz",