Day 9 of 30DayMapChallenge : « space » (previously).

Scrape and geolocate data from Wikipedia. We will map the active space launch sites.

library(tidyverse)
library(lubridate)
library(janitor)
library(sf)
library(glue)
library(rvest)
library(rnaturalearth)

# read all tables from the page
# keep onle those containing a country column
# clean and merge
launch <- read_html("https://en.wikipedia.org/wiki/List_of_rocket_launch_sites") %>% 
  html_table() %>% 
  keep(~ names(.x)[[1]] == "Country") %>% 
  map(~ .x %>% mutate(across(everything(), as.character))) %>% 
  reduce(bind_rows) %>% 
  clean_names() 

# keep only active sites : roughly "YYYY-"
# and extract coordinates
active <- launch %>% 
  filter(str_detect(operational_date, "\\d{4}s?–(\\[.*\\])?$")) %>% 
  mutate(x = str_extract(coordinates, "(?<=;\\s)\\-?\\d*\\.\\d*"),
         y = str_extract(coordinates, "\\-?\\d*\\.\\d*(?=;)")) %>% 
  select(country,location, coordinates, x, y) %>% 
  drop_na(x, y) %>% 
  st_as_sf(coords = c("x", "y"), crs = "EPSG:4326")

# map
ne_countries(scale = "small", type = "countries", returnclass = "sf") %>% 
  ggplot() +
  geom_sf(color = "darkblue", fill = "darkblue") + 
  geom_sf(data = active, color = "yellow", size = 1) +
  coord_sf(crs = "+proj=robin") +
  labs(title = "Active space launch sites",
       subtitle = year(Sys.Date()),
       caption = glue("data : https://en.wikipedia.org/wiki/List_of_rocket_launch_sites
                      r.iresmi.net - {Sys.Date()}")) +
  theme(plot.background = element_rect(fill = "grey10", color = NA),
        panel.background = element_blank(),
        panel.grid = element_line(color = "grey20"),
        plot.title = element_text(color = "grey90", family = "Waree"),
        plot.subtitle = element_text(color = "grey90"),
        plot.caption = element_text(color = "grey90", size = 6))

ggsave("space.png", width = 20, height = 12.36, units = "cm", scale = 1.1)