gbbc-results.R

library(tidyverse)
library(rvest)

url <- "https://en.wikipedia.org/wiki/The_Great_British_Bake_Off_(series_3)"
page <- read_html(url)

table <- page %>% 
  html_nodes("table.wikitable") %>% 
  .[[2]]

rows <- table %>% html_nodes("tr")

# Figure it out for one row
baker_elimination <- function(row) {
  cells <- row %>% html_nodes("td")
  
  baker <- cells %>% .[[1]] %>% html_text()
  
  results <- cells %>% .[-1]
  colspan <- results %>% html_attr("colspan") %>% as.numeric() %>% coalesce(1)
  colour <- results %>% html_attr("style") %>% str_match("background:(.*?);") %>% .[ ,2]
  
  tibble(
    baker = baker,
    colour = rep(colour, colspan),
    week = seq_len(sum(colspan))
  )
}

results <- rows[-c(1:2)] %>% map_dfr(baker_elimination)

status <- c(
  "lightblue" = "ok", 
  "lemonchiffon" = "star", 
  "plum" = "bottom", 
  "cornflowerblue" = "top", 
  "yellow" = "winner", 
  "limegreen" = "runner-up", 
  "orangered" = "out", 
  "silver" = NA
)

results %>% 
  mutate(status = unname(status[tolower(colour)]), colour = NULL) %>% 
  filter(!is.na(status))