gila-buttes-parcel-details-scrape.R

library(tidyverse)
library(rvest)

# read in parcel list
url = 'https://gist.githubusercontent.com/erikgregorywebb/741b472a8ae1bb97d617eda3d6aaea3d/raw/aef7b5083670659788a009185e0e055d0ac1b350/gila-buttes-parcel-search.csv'
parcels_df = read_csv(url)
parcels = parcels_df %>% pull(`PARCEL NUMBER`) %>% str_replace_all(., '-', '')

# scrape details
datalist = list()
for (i in 1:length(parcels)) {
  Sys.sleep(1)
  # read page
  url = paste('https://app1.pinal.gov/Search/Parcel-Details.aspx?parcel_ID=', parcels[i], sep = '')
  print(url)
  print(paste(i, ' of ', length(parcels), sep = ''))
  page = read_html(url)
  
  # grab fields
  parcel_number = page %>% html_node('#lblParcelNum') %>% html_text2()
  s_t_r = page %>% html_node('#lblSTR') %>% html_text2()
  property_description = page %>% html_node('#lblPropDesc') %>% html_text2()
  subdivision = page %>% html_node('#lblSubdivision') %>% html_text2()
  unit = page %>% html_node('#lblUnit') %>% html_text2()
  block = page %>% html_node('#lblBlock') %>% html_text2()
  lot = page %>% html_node('#lblLot') %>% html_text2()
  phase = page %>% html_node('#lblPhase') %>% html_text2()
  cabinet = page %>% html_node('#lblCabinet') %>% html_text2()
  slide = page %>% html_node('#lblSlide') %>% html_text2()
  owner_1 = page %>% html_node('#lblOwner1') %>% html_text2()
  owner_2 = page %>% html_node('#lblOwner2') %>% html_text2()
  in_care_of = page %>% html_node('#lblCareof') %>% html_text2()
  mailing_address = page %>% html_node('#lblMailingAddress') %>% html_text2()
  property_address = page %>% html_node('#lblPropAddress') %>% html_text2()
  recording_date = page %>% html_node('#lblRecDate') %>% html_text2()
  sale_amount = page %>% html_node('#lblSaleAmt') %>% html_text2()
  exemption_widow = page %>% html_node('#lblWidow') %>% html_text2()
  exemption_widower = page %>% html_node('#lblWidower') %>% html_text2()
  exemption_disabled = page %>% html_node('#lblDisabled') %>% html_text2()
  exemption_senior_freeze = page %>% html_node('#lblSrFreeze') %>% html_text2()
  
  temp = tibble(
    parcel_number = parcel_number,
    s_t_r = s_t_r,
    property_description = property_description,
    subdivision = subdivision,
    unit = unit,
    block = block,
    lot = lot,
    phase = phase,
    cabinet = cabinet,
    slide = slide,
    owner_1 = owner_1,
    owner_2 = owner_2,
    in_care_of = in_care_of,
    mailing_address = mailing_address,
    property_address = property_address,
    recording_date = recording_date,
    sale_amount = sale_amount,
    exemption_widow = exemption_widow,
    exemption_widower = exemption_widower,
    exemption_disabled = exemption_disabled,
    exemption_senior_freeze = exemption_senior_freeze
  )
  datalist[[i]] = temp
}
raw = do.call(rbind, datalist)

# clean, export
setwd("~/Downloads")
write_csv(raw, 'gila-buttes-parcel-details.csv')