In Vintea01/openNDB: NDBオープンデータをtidyにする

knitr::opts_chunk$set(echo = TRUE)
knitr::opts_knit$set(root.dir = "~/Projects/openNDB/")
library(tidyverse)
library(readxl)

tidy_syohouyaku_prefecture <- function(xlsx) {
  raw <-
    readxl::read_xlsx(xlsx,
                      na = c("", "-"))
  col_name_1 <-
    c(
      "typecode",
      "typename",
      "drugcode",
      "drugname",
      "tanni",
      "yakkakijun",
      "price",
      "generic",
      "total"
    )
  col_name_2 <- raw[3, 10:56]
  col_type_list <-
    c("numeric",
      "text",
      "numeric",
      rep("text", 3),
      rep("numeric", 50))
  data <-
    suppressWarnings(readxl::read_xlsx(
      xlsx,
      skip = 4,
      col_names = FALSE,
      col_type = col_type_list,
      na = c("", "-")
    )) %>%
    tidyr::fill(1, 2) %>%
    purrr::set_names(c(col_name_1, col_name_2)) %>% 
    tidyr::pivot_longer(10:56, names_to = "prefecture", values_to = "count") %>%
    dplyr::mutate(generic = as.factor(generic),
                  prefecture = as.factor(prefecture))
  return(data)
}

tidy_syohouyaku_prefecture("NDBdata/original/syohouyaku/h29_chusya_prefecture.xlsx")

col_name_1 <-
    c(
      "typecode",
      "typename",
      "drugcode",
      "drugname",
      "tanni",
      "yakkakijun",
      "price",
      "generic",
      "total"
    )
col_name_2 <- read_xlsx("NDBdata/original/syohouyaku/h29_chusya_prefecture.xlsx")[3,10:56]
c(col_name_1,col_name_2)