Brazilian cuture ministry data - Dados do ministério da cultura - SalicNet

salicnet.R

salicnet<-function(cnpj){
  url<-"http://sistemas.cultura.gov.br/salicnet/conDadosCadastraisProponente/conDadosCadastraisProponente.php"
  
  cnpj<-stringr::str_remove_all(cnpj,"\\D+")
  
  nmgp_parms=paste0("cgccpf?#?",cnpj,"?@?NM_btn_insert?#?S?@?NM_btn_update?#?S?@?NM_btn_delete?#?S?@?NM_btn_navega?#?S?@?")
  
  
  h <- curl::new_handle()
  curl::handle_setheaders(h, .list=(charset="iso-8859-1"))
  
  lista<-vector("list",length(cnpj)) ## coloca o n\u00famero total aqui.
  
  for(i in seq_along(lista)){
    
    Sys.sleep(2)
    tryCatch({
    curl::handle_setform(h,
                         nmgp_parms=nmgp_parms[i],
                         nmgp_url_saida= "/salicnet/ctrDadosCadastraisProponente/ctrDadosCadastraisProponente.php",
                         script_case_init= "1")
    
    req <- curl::curl_fetch_memory(url, handle = h)
    
    conteudo<-req$content %>% 
      rawToChar() %>% 
      iconv(from = "ISO-8859-1", to = "UTF-8") %>% 
      xml2::read_html()
    
    
    
    identificacao<-rvest::html_nodes(conteudo,xpath='//table[@id="hidden_bloco_0_1"]/tr[position() > 1]') %>% 
      rvest::html_text() %>% 
      stringr::str_squish() %>% 
      stringr::str_split("\r\n")
    
    cnpj<-identificacao[[1]] %>% stringr::str_extract("(?<=CPF\\s).*?(?=\\s)")
    nome<-identificacao[[1]] %>% stringr::str_extract("(?<=Nome\\s).*")
    responsavel<-identificacao[[2]] %>% stringr::str_extract("(?<=Respons\u00e1vel.).*")
    
    
    logradouro<-rvest::html_nodes(conteudo,xpath='//table[@id="hidden_bloco_1_1"]/tr[position() > 1]') %>% 
      rvest::html_text() %>% 
      stringr::str_squish() %>% 
      stringr::str_split("\r\n")
    
    
    endereco<-logradouro[[1]] %>%stringr::str_extract("(?<=Logradouro ).*")
    uf<-logradouro[[2]] %>% stringr::str_extract("(?<=Proponente ).*?(?= CEP)")
    cep<-logradouro[[2]] %>% stringr::str_extract("(?<=CEP ).*")
    email<-logradouro[[3]] %>% stringr::str_extract("(?<=Email ).*")
    
    telefones<-rvest::html_nodes(conteudo,xpath='//table[@id="hidden_bloco_2_1"]/tr[position() > 1]') %>% 
      rvest::html_text() %>% 
      stringr::str_squish() %>% 
      stringr::str_split("\r\n")
    
    
    residencial<-telefones[[1]] %>% stringr::str_extract("(?<=Residencial ).*?(?=Comercial)")
    comercial<-telefones[[1]] %>% stringr::str_extract("(?<=Comercial ).*?(?= Celular)")
    celular<-telefones[[1]] %>% stringr::str_extract("(?<=Celular ).*?(?= Fax)")
    fax<-telefones[[1]] %>% stringr::str_extract("(?<=Fax ).*")
    
    df<-tibble::tibble(cnpj,nome,responsavel,endereco,uf,cep,email,residencial,comercial,celular,fax)
    lista[[i]]<-df
    
    }, error=function(e){
      e
    }, finally={
      next
    })
  }
  
  dplyr::bind_rows(lista)
  
}