In RobWHickman/CamStroke:

---

title: "Untitled" output: html_document

---

library(rvest)
library(tidyverse)
library(lubridate)

rowclock <- "http://www.rowclock.com/Results/Index" %>%
  read_html()

wh2h <- rowclock %>%
  html_nodes("td:nth-child(1)") %>%
  html_text() %>%
  grep("Winter Head to Head|Winter H2H|Winter Head 2 Head", .)

#download the files
rowclock %>%
  html_nodes(".table a:nth-child(1)") %>%
  .[wh2h] %>%
  html_attr("href") %>%
  paste0("http://www.rowclock.com", .) %>%
  lapply(., function(link) 
    read_html(link) %>%
      html_nodes(".btn-lg") %>%
      html_attr("href") %>%
      paste0("http://www.rowclock.com", .) %>%
      download.file(., 
                    destfile = file.path("../data", paste0("h2h", gsub(".*=", "", .), ".csv")),
                    mode = "wb")
  )

wh2h_results <- dir("./data") %>%
  .[grep("h2h[0-9]", .)] %>%
  file.path("./data", .) %>%
  lapply(., function(file) 
    read.csv(file, stringsAsFactors = FALSE) %>%
      select(., -c(Division, Handicap, Penalty, Position, Number, VetClass, BoatNumber)) %>%
      rename_at(vars(contains('End')), funs(sub('End', 'Finish', .)))
  ) %>%
  bind_rows(.id = "year") %>%
  mutate(year = as.numeric(year) + 2015) %>%
  filter(grepl("Lent", Category)) %>%
  filter(BoatType == "8+") %>%
  mutate(college = case_when(
    grepl("Caius", Club) ~ "Caius",
    grepl("Christ's", Club) ~ "Christs",
    grepl("Churchill", Club) ~ "Churchill",
    grepl("Clare Hall", Club) ~ "ClareHall",
    grepl("Clare", Club) ~ "Clare",
    grepl("Corpus", Club) ~ "Corpus",
    grepl("Darwin", Club) ~ "Darwin",
    grepl("Downing", Club) ~ "Downing",
    grepl("Emmanuel", Club) ~ "Emma",
    grepl("First", Club) ~ "FaT",
    grepl("Fitzwilliam", Club) ~ "Fitz",
    grepl("Girton", Club) ~ "Girton",
    grepl("HCBC|Homerton", Club) ~ "Homerton",
    grepl("Hall\\/Lucy|Cavendish\\/Hughes", Club) ~ "HughesHall/LucyCav",
    grepl("Jesus", Club) ~ "Jesus",
    grepl("KCBC|King", Club) ~ "Kings",
    grepl("Margaret|LMBC", Club) ~ "Maggie",
    grepl("Magdalene", Club) ~ "Magdalene",
    grepl("Murray", Club) ~ "Medwards",
    grepl("Newnham", Club) ~ "Newnham",
    grepl("Pembroke", Club) ~ "Pembroke",
    grepl("Queen", Club) ~ "Queens",
    grepl("Robinson", Club) ~ "Robinson",
    grepl("SCCBC|Catharine", Club) ~ "Catz",
    grepl("Selwyn", Club) ~ "Selwyn",
    grepl("Sidney", Club) ~ "Sidney",
    grepl("Edmund|Eddy", Club) ~ "Eddies",
    grepl("Trinity Hall", Club) ~ "TitHall",
    grepl("Wolfson", Club) ~ "Wolfson"
  )) %>%
  mutate(leg1 = as.numeric(hms(Leg.1.Finish)),
         leg2 = as.numeric(hms(Leg.2.Finish))) %>%
  filter(leg1 > 0 & leg2 > 0) %>%
  select(year, college, Gender, leg1, leg2, Name) %>%
  mutate(crew = gsub("(.*)([0-9].*)", "\\2", Name)) %>%
  mutate(crew = as.numeric(gsub(" \\Cam.*|st .*", "", crew))) %>%
  mutate(crew = ifelse(is.na(crew) & 
                         grepl("Steady|Christ's Men A|Darwin Women|Lucy/Hughes|Clare Hall Women's Team", Name), 1, crew)) %>%
  filter(!is.na(crew)) %>%
  mutate(gender = case_when(
    Gender == "Male" ~ "M",
    Gender == "Female" ~ "F"
  )) %>%
  mutate(race = "Winter Head 2 Head") %>%
  select(year, college, crew, gender, leg1, leg2, race) %>%
  reshape2::melt(id.vars = c("year", "college", "crew", "gender", "race"),
                 value.name = "seconds", variable.name = "leg")

#saveRDS(wh2h_results, ".data/wh2h_results.rds")

RobWHickman/CamStroke documentation built on Nov. 22, 2019, 12:04 a.m.