get MERS-2025 data using rvest

get_MERS_table.R

library(ggplot2)
library(dplyr)
library(tidyr)
library(rvest)
library(readr)

url = "https://ja.wikipedia.org/wiki/2015年韓国におけるMERSの流行"
taball <- read_html(url) |>
  html_table()

dat <- rename(
  taball[[2]],
  Date = 日付,
  Cases = `感染者数[35]`,
  Deaths = `死亡者数[35]`
) %>%
  dplyr::filter(!grepl("総", Date)) %>%
  mutate(Date = gsub("2015年", "", Date)) %>%
  mutate(Date = gsub("日", "", Date)) %>%
  mutate(Date = gsub("年|月", "-", Date)) %>%
  mutate(Date = as.Date(paste("2015", Date, sep = "-"))) %>%
  mutate(
    Cases = gsub("人", "", Cases),
    Deaths = gsub("人", "", Deaths)
  ) %>%
  mutate(
    Cases = gsub("\\[.+\\]", "", Cases),
    Deaths = gsub("\\[.+\\]", "", Deaths)
  ) %>%
  mutate(
    Cases = as.integer(Cases),
    Deaths = if_else(Deaths == "", 0L, as.integer(Deaths))
  )

write_csv(dat, file = "MERS2015.csv")

datl = pivot_longer(dat, Cases:Deaths, names_to = "variable")
ggplot(datl, aes(x = Date, y = value, colour = variable, linetype = variable)) +
  geom_line() +
  theme_bw()