harrier_league_actual_time.R

library(tidyverse)
library(htmltab)

raw_results = htmltab::htmltab(doc = "http://www.harrierleague.com/results/2017-18/druridge/SenF.htm")

clean_results = raw_results %>%
  as_data_frame() %>%
  rename(race_time = `Race Time`, actual_time = `Actual Time`) %>%
  filter(Cat != "guest") %>%
  mutate(division = as.numeric(stringr::str_sub(Club, start = 2, end = 2)), 
         club = stringr::str_sub(Club, start = 5)) %>%
  select(Name, actual_time, division, club) %>%
  mutate(actual_seconds = as.numeric(lubridate::ms(actual_time)))

# Concatenation helper function
concat <- function(strings) {
  Reduce(f = function(string_1, string_2) paste(string_1, string_2, sep = ", "), x = strings)
}

get_results_one_division = function(raw_results, counters = 4) {
  incomplete_teams = raw_results %>%
    count(club) %>%
    filter(n < counters)
  
  raw_results %>%
    anti_join(incomplete_teams, by = "club") %>%
    group_by(club) %>%
    top_n(n = counters, wt = desc(pos)) %>%
    group_by(club) %>%
    summarise(total_points = sum(pos), total_counters = n(), counters = concat(paste0(Name, " (", pos, ")"))) %>%
    mutate(position = min_rank(total_points))
}

actual_time_results = clean_results %>%
  group_by(division) %>%
  mutate(pos = min_rank(actual_seconds)) %>%
  do(get_results_one_division(.)) %>%
  arrange(division, position)
  
actual_time_results %>% 
  filter(division == 1)