Merge branch 'main' into staging/br_ms_sia

models/br_camara_dados_abertos/br_camara_dados_abertos__orgao_deputado.sql

@@ -1,12 +1,25 @@
 {{ config(alias="orgao_deputado", schema="br_camara_dados_abertos") }}
-select distinct
-    regexp_extract(uriorgao, r'/orgaos/(\d+)') as id_orgao,
-    safe_cast(nomeorgao as string) nome,
-    safe_cast(siglaorgao as string) sigla,
-    safe_cast(nomedeputado as string) nome_deputado,
-    safe_cast(cargo as string) cargo,
-    safe_cast(siglauf as string) sigla_uf,
-    safe_cast(datainicio as date) data_inicio,
-    safe_cast(datafim as date) data_final,
-    safe_cast(siglapartido as string) sigla_partido,
-from `basedosdados-staging.br_camara_dados_abertos_staging.orgao_deputado` as t
+with
+    orgao_deputado as (
+        select distinct
+            regexp_extract(uriorgao, r'/orgaos/(\d+)') as id_orgao,
+            safe_cast(nomeorgao as string) nome,
+            safe_cast(siglaorgao as string) sigla,
+            safe_cast(nomedeputado as string) nome_deputado,
+            safe_cast(cargo as string) cargo,
+            safe_cast(siglauf as string) sigla_uf,
+            safe_cast(datainicio as date) data_inicio,
+            safe_cast(datafim as date) data_final,
+            safe_cast(siglapartido as string) sigla_partido,
+        from `basedosdados-staging.br_camara_dados_abertos_staging.orgao_deputado`
+    )
+select *
+from orgao_deputado
+where
+    not (
+        nome_deputado = 'Hélio Leite'
+        and cargo = 'Titular'
+        and sigla_uf is null
+        and data_inicio = '2022-05-03'
+        and data_final = '2023-02-01'
+    )

models/br_camara_dados_abertos/br_camara_dados_abertos__proposicao_microdados.sql

@@ -6,43 +6,87 @@
         partition_by={
             "field": "ano",
             "data_type": "INT64",
-            "range": {"start": 1935, "end": 2023, "interval": 1},
+            "range": {"start": 1935, "end": 2024, "interval": 1},
         },
     )
 }}
 
-select
-    safe_cast(ano as int64) ano,
-    safe_cast(
-        split(
-            format_timestamp('%Y-%m-%dT%H:%M:%E*S', timestamp(dataapresentacao)), 'T'
-        )[offset(0)] as date
-    ) data,
-    safe_cast(
-        split(
-            format_timestamp('%Y-%m-%dT%H:%M:%E*S', timestamp(dataapresentacao)), 'T'
-        )[offset(1)] as time
-    ) horario,
-    safe_cast(id as string) id_proposicao,
-    safe_cast(uri as string) url,
-    safe_cast(numero as string) numero,
-    safe_cast(siglatipo as string) sigla,
-    safe_cast(descricaotipo as string) tipo,
-    safe_cast(ementa as string) ementa,
-    safe_cast(ementadetalhada as string) ementa_detalhada,
-    safe_cast(keywords as string) palavra_chave,
-    safe_cast(uriorgaonumerador as string) url_orgao_numerador,
-    safe_cast(uripropprincipal as string) url_principal,
-    safe_cast(uripropposterior as string) url_posterior,
-    safe_cast(urlinteiroteor as string) url_teor_proposicao,
-    safe_cast(ultimostatus_datahora as string) data_hora_ultimo_status,
-    safe_cast(ultimostatus_urirelator as string) url_relator_ultimo_status,
-    safe_cast(ultimostatus_siglaorgao as string) sigla_orgao_ultimo_status,
-    safe_cast(ultimostatus_regime as string) regime_ultimo_status,
-    safe_cast(ultimostatus_descricaotramitacao as string) tramitacao_ultimo_status,
-    safe_cast(ultimostatus_descricaosituacao as string) situacao_ultimo_status,
-    safe_cast(ultimostatus_despacho as string) despacho_ultimo_status,
-    safe_cast(ultimostatus_apreciacao as string) apreciacao_ultimo_status,
-    safe_cast(ultimostatus_sequencia as string) sequencia_ultimo_status,
-    safe_cast(ultimostatus_url as string) url_ultimo_status,
-from `basedosdados-staging.br_camara_dados_abertos_staging.proposicao_microdados` as t
+with
+    table as (
+        select
+            safe_cast(ano as int64) ano,
+            safe_cast(
+                split(
+                    format_timestamp(
+                        '%Y-%m-%dT%H:%M:%E*S', timestamp(dataapresentacao)
+                    ),
+                    'T'
+                )[offset (0)] as date
+            ) data,
+            safe_cast(
+                split(
+                    format_timestamp(
+                        '%Y-%m-%dT%H:%M:%E*S', timestamp(dataapresentacao)
+                    ),
+                    'T'
+                )[offset (1)] as time
+            ) horario,
+            safe_cast(id as string) id_proposicao,
+            safe_cast(uri as string) url,
+            safe_cast(numero as string) numero,
+            safe_cast(siglatipo as string) sigla,
+            safe_cast(descricaotipo as string) tipo,
+            safe_cast(ementa as string) ementa,
+            safe_cast(ementadetalhada as string) ementa_detalhada,
+            safe_cast(keywords as string) palavra_chave,
+            safe_cast(uriorgaonumerador as string) url_orgao_numerador,
+            safe_cast(uripropprincipal as string) url_principal,
+            safe_cast(uripropposterior as string) url_posterior,
+            safe_cast(urlinteiroteor as string) url_teor_proposicao,
+            safe_cast(ultimostatus_datahora as string) data_hora_ultimo_status,
+            safe_cast(ultimostatus_urirelator as string) url_relator_ultimo_status,
+            safe_cast(ultimostatus_siglaorgao as string) sigla_orgao_ultimo_status,
+            safe_cast(ultimostatus_regime as string) regime_ultimo_status,
+            safe_cast(
+                ultimostatus_descricaotramitacao as string
+            ) tramitacao_ultimo_status,
+            safe_cast(ultimostatus_descricaosituacao as string) situacao_ultimo_status,
+            safe_cast(ultimostatus_despacho as string) despacho_ultimo_status,
+            safe_cast(ultimostatus_apreciacao as string) apreciacao_ultimo_status,
+            safe_cast(ultimostatus_sequencia as string) sequencia_ultimo_status,
+            safe_cast(ultimostatus_url as string) url_ultimo_status,
+        from
+            `basedosdados-staging.br_camara_dados_abertos_staging.proposicao_microdados`
+            as t
+    ),
+    query_total as (
+        select
+            ano,
+            case when data >= current_date() then null else data end as data,
+            horario,
+            id_proposicao,
+            url,
+            numero,
+            sigla,
+            tipo,
+            ementa,
+            ementa_detalhada,
+            palavra_chave,
+            url_orgao_numerador,
+            url_principal,
+            url_posterior,
+            url_teor_proposicao,
+            data_hora_ultimo_status,
+            url_relator_ultimo_status,
+            sigla_orgao_ultimo_status,
+            regime_ultimo_status,
+            tramitacao_ultimo_status,
+            situacao_ultimo_status,
+            despacho_ultimo_status,
+            apreciacao_ultimo_status,
+            sequencia_ultimo_status,
+            url_ultimo_status,
+        from table
+    )
+select *
+from query_total

models/br_ms_cnes/br_ms_cnes__estabelecimento.sql

@@ -96,7 +96,9 @@ select
     safe_cast(clasaval as string) tipo_classificacao_acreditacao_hospitalar,
     safe_cast(substr(cast(dt_acred as string), 1, 4) as int64) as ano_acreditacao,
     safe_cast(substr(cast(dt_acred as string), 5, 2) as int64) as mes_acreditacao,
-    safe_cast({{ clean_cols("AV_PNASS") }} as int64) tipo_avaliacao_pnass,
+    safe_cast(
+        cast({{ clean_cols("AV_PNASS") }} as string) as int64
+    ) tipo_avaliacao_pnass,
     safe_cast(substr(cast(dt_pnass as string), 1, 4) as int64) as ano_avaliacao_pnass,
     safe_cast(substr(cast(dt_pnass as string), 5, 2) as int64) as mes_avaliacao_pnass,
     safe_cast(nivate_a as int64) indicador_atencao_ambulatorial,