scraping_zap.R

library(rvest)
library(stringr)

url_zap <- 'https://www.zapimoveis.com.br/aluguel/apartamentos/sp+sao-paulo/?__zt=ad:a'

xml2::read_html("https://www.zapimoveis.com.br/")


html_obj <- httr::GET(url_zap) %>% 
  read_html()

# ult pagina
num_paginas <- html_obj %>% 
  html_nodes(".num-of") %>% 
  html_text()

num_paginas <- str_extract_all(num_paginas, "[0-9]+")[[1]] %>% 
  paste0(collapse = "") %>% 
  as.numeric()

### dados a extrair
# minificha de cada imovel
fichas <- html_obj %>% 
  html_nodes(".minificha")



# url


# bairro
fichas[1] %>% 
  html_nodes(".pull-right strong") %>% 
  html_text() %>% 
  str_squish()

# logradouro
endereco <- fichas[1] %>% 
  html_nodes("h2 span") %>% 
  html_text()

rua <- endereco[1]
cidade <- endereco[2]
uf <- endereco[3]

# preco aluguel
fichas[1] %>% 
  html_nodes(".preco strong") %>% 
  html_text() %>% 
  str_extract_all("\\d+") %>% 
  unlist() %>% 
  str_c(collapse = "") %>% 
  as.numeric()

# iptu + condominio | valor total
fichas[1] %>% 
  html_nodes(".preco span") %>% 
  html_text()
  
# quartos
fichas[1] %>% 
  html_nodes(".icone-quartos") %>% 
  html_text() %>% 
  str_extract_all("\\d+") %>% 
  unlist() %>% 
  as.numeric()

# vagas
fichas[1] %>% 
  html_nodes(".icone-vagas") %>% 
  html_text() %>% 
  str_extract_all("\\d+") %>% 
  unlist() %>% 
  as.numeric()


# m2
fichas[1] %>% 
  html_nodes(".icone-area") %>% 
  html_text() %>% 
  str_remove("m2") %>% 
  as.numeric()