Grabbing GB5 results

library(assertthat)
library(tidyverse)
library(xml2)
library(rlang)
library(parallel)


extract.ram <- function(link) {
  doc <- read_html(url(link))

  el <- xml_find_first(doc, ".//tr[contains(., 'Size')]")

  xml_text(xml_children(el)[[2]])
}


extract.row <- function(row) {
  col <- xml_find_first(row, ".//div[contains(@class, 'col-12 col-lg-4')]")


  assert_that(xml_text(xml_children(col)[[1]]) == "\nSystem\n")
  system <- str_squish(xml_text(xml_children(col)[[2]]))
  url <- xml_attr(xml_children(col)[[2]], "href")
  #RAM <- extract.ram(paste0("https://browser.geekbench.com", url))
  description <- str_squish(xml_text(xml_children(col)[[3]]))

  cols <- xml_find_all(row, ".//div[contains(@class, 'col-6 col-md-3 col-lg-2')]")

  bind_cols(
    data.frame(System = system, Description = description, stringsAsFactors=FALSE),
    map(1:length(cols), ~ {
      title <- str_squish(xml_text(xml_find_first(cols[[.]], ".//span[contains(@class, 'list-col-subtitle')]")))
      text <- str_squish(xml_text(xml_find_first(cols[[.]], ".//span[contains(@class, 'list-col-text')]")))

      X <- data.frame(text, stringsAsFactors = FALSE)
      names(X) <- title

      X
    }) %>% bind_cols()
  )
}

extract.page <- function(page, query) {
  doc <- read_html(url(glue::glue("https://browser.geekbench.com/v5/cpu/search?page={page}&q={query}")))

  rows <- xml_find_all(doc, ".//div[contains(@class, 'list-col-inner')]")

  map(1:length(rows), ~ extract.row(rows[[.]])) %>% bind_rows()
}

grab.data <- function(query) {
  geekbench.url <- glue::glue("https://browser.geekbench.com/v5/cpu/search?&q={query}")

  # get number of pages
  doc <- read_html(url(geekbench.url))
  page.count <-
    map_chr(xml_find_all(doc, ".//a[contains(@class, 'page-link')]"), xml_text) %>%
    { suppressWarnings(as.numeric(.)) } %>%
    na.omit() %>%
    max(., 1)

  #page.count <- min(page.count, 2)
  #
  # print(page.count)

  #page.count <- min(page.count, 16)
  # grab all pages
  mclapply(1:page.count, function(page) {
    cat(glue::glue("Getting page {page} out of {page.count}..."), "\n")
    while(TRUE) {
      result <- try(extract.page(page, query))
      if(is.data.frame(result)) return(result)
      cat(glue::glue("Retrying page {page} out of {page.count}..."), "\n")
    }

  }) %>%
  bind_rows() %>%
  as_tibble()
}


options(mc.cores = 5)
data <- bind_rows(
  grab.data("4800U") %>% mutate(CPU = "Ryzen 4800U"),
  grab.data("MacBookAir10,1") %>% mutate(CPU = "Apple M1 (MacBook Air)"),
  grab.data("1165G7") %>% mutate(CPU = "Intel i7-1165G7")
) %>%
as_tibble %>%
mutate(
  `Single-Core Score` = as.integer(`Single-Core Score`),
  `Multi-Core Score` = as.integer(`Multi-Core Score`)
)


saveRDS(data, "geekbench-data.rds")
write_csv(data, "gb5-m1-4800U-1165G7.csv")


p_single <- ggplot(data) + geom_density(aes(`Single-Core Score`, color = CPU))
p_multi <- ggplot(data) + geom_density(aes(`Multi-Core Score`, color = CPU))

ggsave("gb5_m1_single.jpg", p_single)
ggsave("gb5_m1_multi.jpg", p_multi)