library(stringi)
library(patchwork)
library(hrbrthemes)
library(tidyverse)
library(jsonlite)

fontname <- "PT Sans"
Sys.setlocale("LC_ALL", "fi_FI.utf8")

cols(
  Alue = col_character(),
  Aika = col_date(format = ""),
  val = col_double()
) -> cov_cols

xdf_raw <- read_csv2("https://sampo.thl.fi/pivot/prod/fi/epirapo/covid19case/fact_epirapo_covid19case.csv?row=dateweek2020010120201231-443702L&column=hcd-444832#", col_types = cov_cols)
xdf <- xdf_raw %>% 
  # filter(!grepl("Kaikki", Alue)) %>% 
  rename(date = Aika, 
         shp = Alue, 
         day_cases = val) %>% 
  group_by(shp) %>% 
  arrange(shp,date) %>% 
  filter(!is.na(day_cases)) %>% 
  mutate(total_cases = cumsum(day_cases))

xdf %>% 
  filter(total_cases != 0) %>%
  arrange(shp,date) %>% 
  group_by(shp, date) %>% 
  arrange(date) %>% 
  mutate(
    idx = 1:n()
  ) %>% 
  ungroup() %>% 
  arrange(shp, idx) -> gdf

gdf %>% 
  group_by(shp) %>% 
  filter(date == max(date)) %>% # last observation
  ungroup() %>% 
  arrange(desc(total_cases)) %>% # largest to smallest
  pull(shp) %>% # make a combo graph for each shp
  map(~{
    
    cur <- filter(gdf, shp == .x)
    rest <- filter(gdf, shp != .x)
    
    # log10 cumulative line
    
    ggplot(cur) +
      geom_path(
        data = rest, 
        aes(date, total_cases, group = shp),
        color = ft_cols$gray, size = 0.35, alpha = 1/2
      ) +
      geom_path(data = cur, aes(date, total_cases), size = 1) +
      scale_x_date(position = "top") +
      scale_y_log10(
        breaks = c(1, 10, 100, 1000, 10000),
        label = c("1", "10", "100", "1K", "10K")
      ) +
      labs(x = NULL, y = NULL, subtitle = .x) +
      # theme_ipsum_es(grid="XY", subtitle_family = font_es_bold, subtitle_face = "bold") +
      theme_ipsum_es(grid="XY", subtitle_family = fontname, base_family = fontname, subtitle_face = "bold") +
      theme(panel.spacing = unit(0, "lines")) +
      theme(panel.spacing.y = unit(0, "lines")) +
      theme(axis.text.x = element_text(size = 12)) +
      theme(plot.margin = margin(10, 10, 6, 10)) -> gg2
    
    # individual day counts bars 
    
    ggplot(cur) +
      geom_segment(aes(date, 0, xend = date, yend = day_cases)) +
      scale_x_date(limits = range(gdf$date)) +
      labs(x = NULL, y = NULL) + 
      # theme_ipsum_es(grid="Y", axis_text_size = 6) +
      theme_ipsum_es(grid="Y", axis_text_size = 6, base_family = fontname) +
      theme(axis.text.x = element_blank()) +
      theme(panel.spacing = unit(0, "lines")) +
      theme(panel.spacing.y = unit(0, "lines")) +
      theme(plot.margin = margin(0, 10, 18, 10)) -> gg3
    
    # combine them
    
    gg2 + gg3 +
      plot_layout(
        ncol = 1,
        heights = c(5, 1)
      ) 
    
  }) -> gg_x

tf1 <- "./shp.svg"
file.create(tf1)

# svglite::svglite(tf1, width = 14, height = 25)
wrap_plots(gg_x, ncol = 3) +
  plot_annotation(
    title = "SARS-Cov-2 -tapausten  aikasarjat sairaanhoitopiireittäin",
    subtitle = sprintf("Jokaisen paneelin ylempi kuva esittää vahvistettujen koronatapausten kertymän sairaanhoitopiireittäin logaritmisella asteikolla\nAlempi kuva näyttää näyttää päivittäiset tapausmäärät (huomaa vaihteleva y-akseli).\nSairaanhoitopiirit järjestetty tapausmäärien mukaan suurimmasta pienimpään\nPäivitetty: %s (Päivittyy joka aamu 9.30)\nDatan uusin päivä: %s",Sys.time(),max(gdf$date)),
    caption = "Data: THL <https://sampo.thl.fi/pivot/prod/api/epirapo/covid19case.json>;\nKoodin lähde: <https://rud.is/>",
    theme = theme_minimal(base_family = fontname, base_size = 15) + theme(plot.title = element_text(face = "bold", size = 20))
  ) -> pp
ggsave(filename = tf1, plot = pp, width = 14, height = 32)  

frs::svg_googlefonts(tf1, "PT Sans", new_svgfile = "./shp-with-fonts.svg")