library(tidyverse)
library(httr)
library(jsonlite)
library(lubridate)
library(rvest)

urls = c('https://geizhals.eu/gigabyte-geforce-rtx-3080-gaming-oc-10g-gv-n3080gaming-oc-10gd-a2366735.html',
         'https://geizhals.eu/gigabyte-geforce-rtx-3070-gaming-oc-8g-gv-n3070gaming-oc-8gd-a2406936.html?hloc=at&hloc=de&hloc=eu&hloc=pl&hloc=uk',
         'https://geizhals.eu/msi-geforce-rtx-3080-ti-gaming-x-trio-12g-a2538377.html?hloc=at&hloc=de&hloc=eu&hloc=pl&hloc=uk',
         'https://geizhals.eu/powercolor-radeon-rx-6700-xt-red-devil-axrx-6700xt-12gbd6-3dhe-oc-a2494750.html')


get_geizhals_data = function(url){
  read_html(url) %>%
    html_text() -> html_raw
  
  html_raw %>%
    str_extract('(?<=productIds: )[0-9]+')  -> id
  
  html_raw %>%
    str_extract("(?<=productName: )'.+'") %>%
    str_remove("'") -> name
  
  request_url = 'https://geizhals.eu/api/gh0/price_history'
  args = list(id = id, params = list(days = 9999, loc = 'eu'))
  body = toJSON(args, auto_unbox = TRUE)
  raw = POST(request_url, body = body, encode='json')
  tmp = fromJSON(content(raw, as='text'))
  tmp_df = data.frame(tmp$response)
  colnames(tmp_df) = c('timestamp', 'price', 'availability')
  tmp_df %>%
    mutate(date = as.Date(as.POSIXct(timestamp/1000, origin="1970-01-01"))) %>%
    select(-timestamp) -> tmp_df
  tmp_df$id_gpu = id
  tmp_df$name = name
  tmp_df$url = url
  return(tmp_df)
}



gpu_data = map_df(urls, get_geizhals_data)