First try at web-scraping with rvest to plot age-tenure of the Regents of Sweden

swe-regents.R

#webscraping livrustkammaren to create a age-tenure graph of Swedish regents
#inspired by https://gist.github.com/acoppock

library(rvest)
library(tidyverse)
library(stringr)
library(lubridate)
library(ggrepel)
library(stringr)

# lets scrape data from regentlängden
name <- read_html("http://livrustkammaren.se/sv/utforska/regentlangden") %>%
  html_nodes("h3") %>% 
  html_text() %>% 
  str_sub(start = 3) %>% 
  data.frame(regent = .)

df <- read_html("http://livrustkammaren.se/sv/utforska/regentlangden") %>%
  html_nodes("#node-349 p") %>% 
  html_text() %>% 
  gsub("[^0-9]", "", .) %>% 
  data.frame(years = .) %>% 
  filter(years != "" & years != "23") %>% 
  mutate(born = as.numeric(str_sub(years, end = 4)), 
         term_start = as.numeric(str_sub(years, start = 9,  end = 12)),
         term_end = as.numeric(str_sub(years, start = 13,  end = 16)),
         term_start = ifelse(is.na(term_start), 1973, term_start), #current king who by not having died ruins the pattern...
         term_end = ifelse(is.na(term_end), year(today()), term_end), 
         regent_num = 1:length(term_start)) %>% 
  select(-years) %>%
  bind_cols(name) %>% 
  gather(key, year, term_start, term_end) %>% 
  mutate(age = year - born, 
         torch = if_else(key == "term_start", regent_num, as.integer(regent_num + 1))) 

df_lab <- df %>% filter(key == "term_start")

ggplot(df, aes(x = year, y = age, group = regent)) +
  geom_point(aes(color = age), size = 2) +
  geom_line(aes(color = age), size = 1) +
  geom_text_repel(data = df_lab, aes(label = regent), size = 3, nudge_y = -1, nudge_x = 10) +
  geom_line(aes(group = torch), linetype = "dotted", alpha = 0.5) +
  scale_color_viridis_c(begin = .5) +
  theme_bw() +
  ylab("Age")+
  xlab("Year") + 
  theme(legend.position = "none") +
  labs(title = "How old was the Regent of Sweden?",
       subtitle = "Code at: https://gist.github.com/tannenberg", 
       caption = "Source: livrustkammaren.se")