scraper igreja universal

universal.R

library(httr)
library(rvest)
library(dplyr)
library(stringr)
library(tidyr)


universal_pags <- function() {
  universal_pag <- function(pag) {
    link <- sprintf('http://www.universal.org/enderecos/busca/brasil/todas/%04d/', pag)
    r <- GET(link)
    item <- function(node) {
      end <- node %>% html_node('h4') %>% html_text() %>% str_trim()
      link <- node %>% html_node('.btn-url') %>% html_attr('href')
      nome <- node %>% html_node('.link-titulo') %>% html_text()
      if(length(html_nodes(node, '.tel')) > 0) {
        tel <- node %>% html_node('.tel') %>% html_text()
      } else {
        tel <- ''
      }
      data_frame(nome = nome, endereco = end, tel = tel, link = link)
    }
    f <- failwith(data_frame(nome = NA_character_), item)
    r %>% 
      read_html() %>% 
      html_nodes('li.thumb-box.home') %>% 
      lapply(f) %>% 
      bind_rows()
  }
  pags <- 'http://www.universal.org/enderecos/busca/brasil/' %>% 
    GET() %>% 
    read_html() %>% 
    html_nodes('.pagination li') %>% 
    html_text() %>% {
      suppressWarnings(extract_numeric(.))
    } %>% 
    max(na.rm = TRUE)
  ff <- failwith(data_frame(nome = NA_character_), universal_pag)
  data_frame(pag = 1:pags) %>% 
    group_by(pag) %>% 
    do(ff(.$pag)) %>% 
    ungroup()
}

d_universal_pags <- universal_pags()

openxlsx::write.xlsx(d_universal_pags, 'd_universal_pags_brasil.xlsx')