harvesting cvs store locations

cvs_rvest.R

library(xml2)
library(rvest)
library(glue)
library(dplyr)
library(purrr)
library(purrrogress)
library(ggmap)
library(ggplot2)

srm <- function(x,pat = '\\t+|\\n+',replace = '') gsub(pat,replace,x)

get_cities <- function(x){
  x%>%
    xml2::read_html()%>%
    rvest::xml_nodes(xpath='//div[contains(@class, "states")]')%>%
    rvest::html_text()%>%
    srm()%>%
    srm('\\)',',')%>%
    srm('\\s\\([1-9]','')%>%
    strsplit(', ')%>%
    purrr::flatten_chr()%>%
    srm('^\\s+|\\s+$')%>%
    srm('\\s','-')
}

get_stores <- function(uri){

  uri%>%
    xml2::read_html()%>%
    rvest::html_nodes(xpath='//p[contains(@class, "store-address")]')%>%
    rvest::html_text()%>%
    srm()
}

register_google(key = Sys.getenv('GOOGLE_KEY'))

uri_root <- 'https://www.cvs.com/store-locator/cvs-pharmacy-locations'

cvs <- tibble::tibble(state = gsub('\\s','-',state.name))

cvs_cities <- cvs%>%
  dplyr::mutate(
    state_uri = glue::glue('{uri_root}/{state}'),
    city  = purrr::map(state_uri,purrrogress::with_progress(get_cities,total = 50))
  )%>%
  tidyr::unnest()

this <- 'West-Virginia'

cvs_stores <- cvs_cities%>%
  dplyr::filter(state==this)%>%
  dplyr::mutate(
    store = purrr::map(glue::glue('{uri_root}/{state}/{city}'),
                       purrrogress::with_progress(get_stores,
                                                  total = cvs_cities%>%
                                                    dplyr::filter(state==this)%>%
                                                    nrow()))
  )%>%
  tidyr::unnest()%>%
  ggmap::mutate_geocode(store)

m <- ggmap::get_map(
  location = this%>%srm('-',' '),
  zoom = 7, 
  scale = "auto",
  maptype = "toner-lite",
  source = "stamen")

m%>%
  ggmap::ggmap(
    extent = "device",
    legend = "topright"
  ) +
  ggplot2::geom_point(
    data = cvs_stores, 
    aes(x = lon, y = lat),
    colour='red'
    ) +
  labs(title = glue::glue('CVS store location {this}'))