scrape20

library(tidyverse)
library(rvest)

hp_link <- "https://www.mountainproject.com/"

hp_links <- read_html(hp_link) |>
  html_nodes("a") |>
  html_attr("href")

route_links <- hp_links |>
  keep(str_detect(hp_links, "/route/")) |>
  unique() |>
  head(-1)

base <- route_links |>
  str_split("/route/") |>
  pluck(1) |>
  pluck(1) |>
  paste0("/route/stats")

stat_pages <- route_links |>
  str_split("/route") |>
  map_chr(pluck(2)) |>
  (\(x) paste0(base, x))()


get_users <- function(link) {
  route <- link |>
    read_html() |>
    html_nodes("table") |>
    pluck(4) |>
    html_nodes("a") |>
    html_attr("href") |>
    str_extract("user/\\d+") |>
    str_remove("user/")
  route
}

users <- map(stat_pages, get_users)

route_names <- route_links |>
  str_extract("\\d+/.+") |>
  str_remove("\\d+/")

users |>
  map(as.data.frame) |>
  setNames(route_names) |>
  bind_rows(.id = "route") |>
  setNames(c("route", "user")) |>
  mutate(user = as.numeric(user)) |>
  pivot_wider(names_from = user, values_from = route, values_fn = length) |>
  pivot_longer(everything(), names_to = "user", values_to = "count") |>
  View()