library(dplyr)
library(readr)
library(tidyr)

setwd("/scratch/msserpa/microdados_ed_superior_2018/dados")

dfIES   <- read_delim("DM_IES.CSV", delim="|") %>%  as.data.frame
dfCurso <- read_delim("DM_CURSO.CSV", delim="|") %>% filter(TP_ORGANIZACAO_ACADEMICA == 1) %>% filter(CO_MUNICIPIO == 4314902) %>% filter(!is.na(QT_INSCRITO_TOTAL)) %>% filter(TP_MODALIDADE_ENSINO == 1) %>% filter(TP_NIVEL_ACADEMICO == 1) %>% as.data.frame
dfAux   <- read_delim("TB_AUX_CINE_BRASIL.CSV", delim="|") %>% as.data.frame

dfCurso <- dfCurso %>% left_join(dfAux, by = c("CO_CINE_ROTULO"))
dfCurso <- dfCurso %>% left_join(dfIES, by = c("CO_IES"))

df <- dfCurso

df <- df %>% group_by(CO_CINE_AREA_GERAL, NO_CINE_AREA_GERAL, NO_IES) %>% summarize(total=sum(QT_INSCRITO_TOTAL)) %>% as.data.frame()

df

write_csv(df, "saida.csv")